EATING BEHAVIOUR, PERSONALITY TRAITS AND BODY MASS

Promotoren: dr. P.B. Defares, hoogleraar in de sociale psychologie.

dr. J.G.A.J. Hautvast, hoogleraar in de leer van de voedingen
de voedselbereiding.
 l 7Z
t)M*U>\, *

Tatjana vanStrien

EATING BEHAVIOUR,
PERSONALITY TRAITS AND
BODY MASS

PROEFSCHRIFT

ter verkrijging van de graad van

doctor in de landbouwwetenschappen,
op gezag van de rector magnificus,
dr. C.C. Oosterlee,
in het openbaar te verdedigen
op woensdag 5maart 1986
des namiddags te vier uur in de aula
van de Landbouwhogeschool te Wageningen

BZ SWETS.&ZEITLINGER L I
^
Omslag ontwerp: H. Veltman
Gedrukt bij Offsetdrukkerij Kanters B.V., Alblasserdam
Copyright ©1986 T. van Strien en Swets &Zeitlinger, B.V.
Alle rechten voorbehouden. Niets uit deze uitgave mag worden verveelvoudigd, opgeslagen in een
geautomatiseerd gegevensbestand, of openbaar gemaakt, in enige vorm of op enige wijze, hetzij
elektronisch, mechanisch, door fotokopieën, opnamen, of op enige andere manier, zonder vooraf-
gaande schriftelijke toestemming van de uitgever.

All rights reserved. No part of thispublication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording,
or otherwise, without the prior written permission of the publisher.

ISBN 902650702X
 jJMOÏÊOl, HfrV

STELLINGEN

1.Psychometrischeschalenvooreetgedragingenzijnonmisbaarvooronderzoek
naardebehandelingvanovergewicht.
Ditproefschrift.

2.Hetheeftgeenzineen"emotioneleeter"eenvermageringsdieetvoorte
schrijven.
Ditproefschrift.

3.Sommigevrouwenmakenzichdikdoordatzezichaandelijnlatenleggen.
Ditproefschrift.

4.Ineenvrouwelijkeonderzoekspopulatieiserinlongitudinaalonderzoekeen
groterekansvoorhetoptredenvanresponsesetsdanineen
cross-sectioneelonderzoek.
Ditproefschrift.

5.Hetadviesinzakeadipositasuitgebrachtdooreencommissievande
Gezondheidsraadwaarinslechtséénvandenegenaanbevelingenvoor
onderzoeknaarpsychologischefactorenrefereert,wekttenonrechtede
indrukdatderolvanpsychologischefactorenbijhetontstaanenhandhaven
vanovergewichtmarginaalis.
AdviesinzakeAdipositasaandeGezondheidsraad,1984.

6.Probleemgerichtonderwijsbevordertdeintegratievannieuwekennisin
bestaandekennisstructuren.
H.G.Schmidt.Activatievanvoorkennis,intrinsiekemotivatieende
verwerkingvantekst,1982"T

7.Hetvergrotenvanmondigheidvanvrouwenverkleinthunmedischeconsumptie.
Gezondheidszorgenvrouwen.Eaancipatieraad,1984.

8."Feminismisnotconcernedwithagroupofpeopleitwantstobenifit,but
withatypeofinjusticeitwantstoeliminate".
J.RadcliffeRichards.TheScepticalFeminist,1982.

9.Hetopzettenvanonderzoekdatvoorhetmerendeeldrijftoptijdelijke
wetenschappelijkpersoneelloopthetgevaargelijktestaanaanhetbouwen
vaneenbrugzonderpijlers.

10.Degebrekkigebegeleidingvanpassagierstijdenstreinstoringenwektde
indrukdatdeNederlandseSpoorwegeneenSocialeDienstzijnnochhebben.

11.Indubioproarte.
D.Bonte.MuseumModernerKunstPalaisLichtenstein.Museumdes20
Jahrhunderts,Wien,Museum,1982,3,19.

Tatjana van Strien:"Eating Behaviour, Personality Traits and Body Mass",

Wageningen, 5 maart 1986.
AANMIJNOUDERS
 biz.
Contents

Dankwoord
Chapter 1 Introduction

PartI. Thethreetheoriesofeatingbehaviour
Reviewof literature
Chapter 2 Psychological theoriesofovereatingandweightgain 5
Chapter 3 Psychosomatic theory 6
Chapter 4 Externalitytheory 14
Chapter 5 Theoryof restrainedeating 21
Chapter 6 Thethreepsychological theoriesofovereating and
weightmanagement:therapy 33
Chapter 7 Summaryandoutlineoftheresearch 36

PartII.Psychoaetricstudies
Chapter 8 TheDutchBatingBehaviourQuestionnaire (DEBQ)for
assessmentof restrained,emotionalandexternal
eatingbehaviour 43
Chapter 9 ThepredictivevalidityoftheDutchRestrained
EatingScale 62
Chapter 10 Eatingbehaviour,personalitytraitsand
bodymassinwomen 69
Chapter 11 Eatingbehaviourandsex-roleorientationinwomen 82
Chapter 12 Lifeevents,emotionaleatingandchange
inbodymass index 98
Chapter 13 Onlongitudinalversuscross-sectional studiesof
obesity:possibleartefacts 107
Chapter 14 Generaldiscussion 119

Summary 137

Samenvatting 142

References 149

Curriculumvitae 167
DANKWOORD
Het inditproefschriftbeschrevenonderzoeknaarpsychologischeaspectenvan
overetenenovergewichtwerdopdevakgroepHumaneVoeding,innauwesamenwer-
kingmetdevakgroepPsychologie,vandeLandbouwhogeschool teWageningenuit-
gevoerd.Hetonderzoekmaaktdeeluitvaneenmultidisciplinair onderzoek naar
overgewichtenwerd financieelgesteunddoorhetPraeventiefonds.
Vandevelemensendieaanditonderzoek eenbijdragehebbengeleverd,wil
ikgraagenkelebijnamenoemen.
MijnpromotoresProf.Dr.P.B.Defares enProf.Dr.J.G.A.J.Hautvastwil ik
graagdankenvoorhetinmijgesteldevertrouwen.Prof.Defaresdank ikvoor
zijnbereidheid steedstijdvrijtemakenvooroverlegenzijnconstructief
kritischcommentaar.Prof.Hautvastbenikzeererkentelijkvoordemogelijk-
hedendiehijheeftgebodenditonderzoek optezettenenuittevoeren.
ZeerveeldankbenikverschuldigdaanJanFrijters.Zijnpersoonlijkebe-
trokkenheidbijhetonderzoeksproces-dathijopdevoetvolgdezondermijin
mijnvrijheidtebeperken-, zijnkritischekanttekeningenennauwgezettele-
zingvandemanuscriptenendediscussiesdiewijhaddenwareneengrotesteun.
VeeldankgaatuitnaarGerardBergersvoordestatistischeanalysesen
MattiRookusvoordevruchtbarediscussiesenhetkritischdoorlezenvande
manuscripten.Aandesamenwerkingmethenhebikeenzeerprettigeherinnering.
GraagnoemikWillyKnuiman-HijlenHelmivanderWiel-Wetzelsvoorhunad-
ministratieveenorganisatorische ondersteuningenClaryBaerends-Bos,Marian
Fortuin,EllenFrijters-vanderHagen,RinaKleinjan,EvelynOosterhaven-Rap-
pard,WilPinksterenHiaRuys-Quint,voorhuninzetbijhetveldwerk.Ookwil
ikdeprettige samenwerkingmetBenSchölte,GebcaVelema,WijavanStaveren,
JanBurema,JaapSeidellenPaulDeurenberghiernietonvermeldlaten.
LeovanderKamp,GretaNoordenbos,MarjaBrandjesenJanKnuimanwilik
graaghartelijkdankenvoorhunwaardevolleopmerkingenbijenkelemanuscrip-
ten.GrotewaarderinghebikvoordebijdragevanHelenWestenHilaryvan
Strien-Marland bijdecorrectievanhetengels,envoorde inzetvanAnnemarie
ZijlmansenWillyKnuiman-Hijlbijhetuittypenvandemanuscripten.
Tenslottedank ikdebestuurdersvanEdeenenkelehuisartsen inWageningen
voorhetterbeschikking stellenvanhetgevraagde adressenbestand. Zeerveel
dankbenikverschuldigd aandevele inwonersvanEdeenWageningendieaandit
onderzoekhunmedewerkingverleendenenbereidwarendedikwijlsomvangrijke
vragenlijsten intevullen.Zonderhunmedewerkingwasditonderzoek niet
mogelijkgeweest.
 -1-

CHAPTER1

This study formsapartofamulti-disciplinary researchproject on overweight

whichwas initiated in1979 attheDepartmentofHumanNutrition,TheAgricul-
turalUniversity,Wageningen, theNetherlands.This study focuseson psycholo-
gical aspectsof overeating andweight gain.Others involved inthe project
have covered suchaspects asepidemiology, healthandnutrition (e.g.,Baecke,
1982;Rookus,Burema,Deurenberg andHautvast, inpreparation; Seidell,Bakx,
deBoer,Deurenberg andHautvast, 1985;Van Staveren,Deurenberg, Burema,de
Groot andHautvast, submitted).
Three psychological theories,dealingwithdevelopment andmaintenance of
humanobesity, are central tothe research.Associatedwith these theories are
three typesof eatingbehaviour.The first ofthese ispsychosomatic theory,
which focuseson thephenomenon ofemotional eating, that is,eating in respon-
setoarousal states suchasanger, fear oranxiety. Secondly, externality
theory concentrates on thephenomenon ofexternal eating, that is,eating in
response to food-related stimuli regardless of the internal stateofhunger or
satiety.Thirdly, the theoryof restrained eating focuseson the psychological
side-effects ofdieting, that is,thepossiblebreakdown of restrictive con-
trol,which results inthedisinhibition of suppressed eatingbehaviour and
excessive food intake.Themainproblemassociatedwith the three theories is
thattheydiffer intheirexplanationsofexactlywhyindividuals overeat,
whichmakes itdifficult todetermine howovereating oroverweight canbest be
treated.
Theprincipal aimof this study, therefore,was toexamine thevalidity of
eachof these theories.Thiswascarried out intwoways (PartsOne andTwo).
Thepublished studiescentral tothethreetheoriesareexamined inPartOne.
Thenature of theprocesses described ineachof these theories,anda review
of themost central research findings,are covered inthispart.Inaddition, a
brief description of thedifferent principles ofweightmanagement advocated by
the three theories isgiven.PartOne concludeswithanoutline of the
empirical researchwhichhasbeencarriedout inorder totest the three
theories.InPartTwoa seriesofpsychometric studieswhich looked at the
relationships between the three typesof eatingbehaviour (emotional eating,
external eating and restrained eating)andbodyweight, personality characte-
risticsand lifeevents,were conducted.

Theauthor joined theproject inSeptember, 1982.

PARTI. THETHREETHEORIESOFEATINGBEHAVIOUR:

REVIEWOF LITERATURE
 -5-

CHAPTER 2

PSYCHOLOGICALTHEORIESOFOVEREATINGANDWEIGHTGAIN

Theintroductionhasalready referred inbrieftothethreepsychological

theoriesofovereatingandweightgain.Inthispart thetheorieswillbe
furtherexamined.As thethreetheoriesarecentraltothe researchconducted
inthisthesis,thenatureofthedifferentprocessesoutlined inthemwillbe
described indetail.Eachofthesetheorieshasbeen subjected toempirical
testingbutonlypartialvalidationcould sofarbeenobtained.Thoughthe
threepointsofviewwillbepresented astheories,theymightalsobedepicted
asasetofhypotheses.The theorieswillbepresented inchronologicalorder
todemonstratehowtheexternalitytheoryoriginated fromthepsychosomatic
theory,andhowthetheoryof restrainedeatingoriginated initsturnfromthe
externalitytheory.Onedangerofthisprocedure isthatitmaygivethe
impression thatthedevelopmentofanewtheory refutestheveracityofthe
earlier theories.The truth,however,isthatallthree theoriesstillhave
great influence,not only intheareasoftherapyandweightmanagement,but
alsoupon research.Recently,anumber ofattemptshavebeenmade tocombine
theviewpointsexpressed inthethree theoriessoastoformulateone single
theory (e.g.,Slochower, 1983a;HermanandPolivy, 1984).
Fromahistoricalpointofview, itisnot surprisingthatthe first
attempts toexplainoverweight inpsychological termsweremade by
psychoanalysts.Overweightwaslongconsideredasanexclusivemedical problem.
But theconclusionthat theoverwhelmingmajorityofcasesofobesitywerenot
causedbyanyorganicdisorderofthemetabolism, butwere simply the resultof
overeating,moved theareaof interest frombiochemical regulatorymechanisms
tothepsychological factorswhichcauseanincrease infood intake.However,
tieswiththe somaticdeterminantsofobesitywerenot completelybroken.
Obesitywas firstdealtwithbyusingpsychoanalysis,anoffshootofmedical
practice,bypsychoanalysts,whooftenhadamedical training.Hostofthose
active inthisfieldworkedaspractical therapists,andatheoryofovereating
originated fromtheirexperiencewithclients.Asa result,avarietyofview-
points,all fallingwithinthecategoryofpsychoanalytic thought,were
developed;thesewillbe summarisedunderthegeneralheadingof "psychosomatic
theory".
Asa resultofan increased interest incarefullyconducted empirical
research,anattemptwasmade totestthepsychosomatic theoryempirically,
 -6-

withan initiallynegativeoutcome.Inanendeavour tobetterexplainthe

observed results,theexternalitytheorywasformulated,andthistheory,in
itsturn,was "replaced"bythetheoryofrestrainedeating.

CHAPTER 3

PSYCHOSOMATICTHEORY

A normal responsetoemotionalarousal statessuchasanger,fearoranxiety is

lossofappetite.Emotionalarousal inhibitsgastricmotility (Carlson, 1916),
and leadstothesuppressionofgastricmovementandtheliberationof sugar
fromtheliver intothebloodstream (Cannon,1915).Asthesephysiological
statesaresimilar tothechiefperipheralphysiological correlatesof
anxiety,emotionalarousalandstressgenerallyleadtodecreasedeatingand
subsequentweightloss.However,forsomeindividualsemotionalarousaland
stressleadtoanexcessive intakeoffood.Thisso-designated"emotional
eating"is,according topsychosomatic theory,foundinindividualswhoeatin
responsetoemotional states,especiallyanxietyanddepression,ratherthanin
responsetointernalcuesofhungerandsatiety.
Classicalpsychoanalystshaveexplainedemotionaleatingintermsof
fixationattheoralstage.As foodintakeistheprincipal sourceofpleasure
during thefirstyearoflife,themouthfunctionsasthepredominanteroge-
neouszone.Duringthisperiod,themother constitutesthecentral loveobject
ofthechild,assheisthesourceofmany instinctualgratifications,and the
child learnstoassociate foodconsumptionwithmaternalcare.Inanormal
developmentalprocessthefocusonoralneedsbecomesgraduallyintegratedwith
other sensoryprocesses.However,ifsometraumaticeventoccursduring this
process,thechildmaybecome fixatedatoneoftheearlystagesof
development.Ifitbecomesfixatedattheoralstage,itispossible thatthe
childwillnoteasily relinquish itsneed fororalgratification,and in
adulthoodwill continuetorecognisefoodasasymbolofmaternal care.When
undergoingemotional stress,suchindividualsarelikelytoturntofood inan
attempt torecapture thesecurityandcomfortexperienced ininfancy
(Slochower,1983 a ,pp. 12-13).
Thepsychoanalytic explanationofthegenerationofemotionaleatinghas
beencriticisedonthegroundsthatitdoesnotidentifytheprecisemechanism
which transformsoral fixations intothespecific symptomofovereating
 -7-

(Slochower,1983 ).Somanyunconsciousmeaningshavebeenassociated with

overeating, foodandobesitybyvariouspsychoanalytic authors,thatKaplanand
Kaplan conclude that it seems that"...anyemotional conflictmay eventually
result inovereating" (KaplanandKaplan, 1957,pp. 196-197). For this reason,
Kaplan and Kaplanmaintain thatovereating seeminglyneednot alwayshave
resulted fromoral fixation,anddescribe emotionaleatersas emotionally
disturbed personswhohave learned touseovereating asameansof copingwith
theirpsychological problems (I.e.,p.197).InasimilarwayBruch (1961,
1964)moved fromaclassical psychoanalytic interpretation ofeating disorders
toapsychodynamic interpretation based onadistinction betweenwhat she
termed "developmental"and "reactive"typesof obesity.
Developmental obesity isassociatedwithsevere emotional and personality
disturbances,and isconceptualised asbeginning ininfancy, andasbeing
caused bya fundamental feeling of rejection on thepartof themother towards
her child.According toBruch (1961,pp. 470-471), themotherwould often feel
that the childwas "toomuch" forher,andcouldpayvery little attention to
it,with theexception of stuffing itwith food. Ifthe response of themother
iscontinuously inappropriate, be itneglectful,oversolicitous, inhibiting or
indiscriminativelypermissive, theoutcome for the childwill bea perplexing
confusion ofhisbiological cues,and ofhisperceptions and conceptualisa-
tions.When thechild isolder, itwillbeunable to recognizewhether it is
hungry or satiated, or suffering from someother discomfort, andmayovereat in
response tovirtually anyarousal state.Asanadult itsuffers fromadeficit
inhunger awareness.
Reactive obesity isseenasoccurring primarily inadultsin response to
traumatic environmental circumstances (Bruch, 1964,pp.270).Ulistypeof
obesity hasnotbeenassociatedwith fundamental life relationships.Rather it
isseenasoccurring following situations that areakin tothose usually
accompanying grief anddepression, suchasseveremental shock, thedeathofa
beloved person, separation from thehome,orother events involving fearof
desertion and loneliness.
Although psychoanalytic theorists agree thatunconscious conflictsare
central totheemotional distress that leadstoovereating, theydisagree about
themeaning ofovereating for the individual.Nevertheless,all of them regard
overeating asa response to internalemotional arousal.
 -8-

RESEARCH

Studies of thepsychosomatic theoryhaveexamined firstly,whether obese

individuals experience negative emotionsmore oftenandwitha greater
intensity thandonormal-weight individuals,and, secondly,whether obese
individuals increase food intake in response toemotional stress,and whether
they showanxiety reduction asa resultofemotional eating.
Emotional reactivity. Studies ofemotional reactivityhave consistently
found thatobese individualsexperience negative emotionsmoreoften andwitha
greater intensity thandonormal-weight individuals.Leonand Chamberlain
(1973)noted that formerlyand currentlyobesewomen reported ahigher
frequencyof traumatic eventsand feelingsoftensionduring threedays of
recording thandidnormal-weight subjects.Similarly,LoweandFisher (1983)
found a significant association betweennegative emotional arousal and body
weight,when subjects self-monitored theirmood intheir home environment
immediately before eachintakeof food for twelve consecutive days.Finally,
experimentsbyAbramsonandWunderlich (1972), Pliner,Meyer and Blankstein
(1974), andRodin, ElmanandSchachter (1974)indicated that obese subjects
weremore distressed byavariety ofnoxious stimuli than normal-weight
subjects.
However, intwostudieswhereauditory feedback onheart ratewas utilised
as thearousalmanipulation, nodifferences inthe responsiveness ofobese and
normal-weight subjects,as recorded inself-reported manipulation checks,were
found (Slochower, 1976;Slochower andKaplan, 1980).Similarly,no obese/normal
differenceswere found inthe intensity of self-reported mood in response to
the stressof takingexamsor the relief of finishing them (Slochower, Kaplan
andMann, 1981). Inall three studies,however, obese subjects showed less
variability intheir responses tomanipulation check scalesthannormal-weight
subjects.This indicates,according toSlochower (Slochower and Kaplan, 1980),
thatoverweight subjects respondmore consistently to experimental
manipulations thandonormalweight subjects.Slochowermaintains that this is
consistentwiththenotionof the increased external responsivenessof the
obese.
Emotional eating andanxiety reduction,interviewandquestionnaire studies
ofobese individuals in (mostly)psychiatric settings support thenotion that
themajority ofobese individuals overeatwhen anxious,depressed or lonely
(Freed, 1947;Hamburger, 1951;Hecht,1955;Stunkard, 1959a;Atkinson and Rin-
guette, 1967;Holland, Masling andCopley, 1970;LeonandChamberlain, 1973;
 -9-

Kalucyand Crisp, 1974;Castelnuovo-Tedescoand Schiebel,1975;Plutchik, 1976,

andRand andStunkard, 1978). Incontrast, testsofarousalmanipulations on
eatingbehaviour showed either slight increasesorno increases in food intake
byobese subjectstestedunderhighversuslowanxietyconditions.Schachter,
Goldman andGordon (1968)conducted the first testofthepsychosomatic hypo-
thesis, and for thepurpose ofillustration itwillbedescribed insomedetail.
Schachter andhiscolleagues (e.g.,Schachter, 1964;Schachter and Singer,
1962)carried out a seriesofexperimentsdealingwith the interactionof the
cognitive and physiological determinants ofemotional states.Their experiments
involvedmanipulating bodily stateby injectionsofadrenaline orplacebo.
Simultaneously, theymanipulated cognitiveand situational variableswhichwere
presumed toaffect a subject'sinterpretation ofhisbodily state.The results
showed, thatphysiologically aroused subjectsattached specific emotional la-
bels totheir arousal states,whichare consistentwith concurrently experien-
ced situational cuesor cognitive factors.This suggests that there isnot a
direct relationshipbetweena setofphysiological symptoms and a psychological
state. Schachter concluded fromtheseobservations that, "Obviously, attaching
a particular label toanyparticular internalorvisceral syndrome isa lear-
ned, cognitivelyand sociallydetermined act", (Schachter etal., 1968,p.91).
Bruch (1961)madea similar observation concerning obesityduring the same
period, suggesting that obese individualshavenot learned to discriminate
betweenphysiological symptoms ofhunger orsatietyandaccompanying emotional
internal arousal states,suchasanger, fearoranxiety.When Stunkard and Koch
(1964)demonstrated that someoverweight individualsweremuch less likely than
normal-weight individuals to relate their stomachcontractions to the conscious
experience ofhunger, Schachter became interested inthe idea (that ina
similarwaytothe interpretationofemotional states)subjectsmaydiffer in
theextent towhich theyinterpret physiological states,suchasgastric
motility, as "hunger".
To test thishypothesis.Schachter andhisco-workers conducted an
experiment inwhich theyattempted tomanipulate gastricmotility and other
correlates of fooddeprivation intwoways (Schachter et al., 1968). Firstly,
theydirectlymanipulated fooddeprivation, sothat some subjectshad empty
stomachs and others full stomachsbefore entering theexperimental eating
situation.Secondly, through themanipulation of fear, some subjects entered
theeating situation frightened andotherscalm. Fooddeprivation was
manipulated by feeding the subjects roastbeef sandwiches (the "preload
condition"),orbygiving themnothing toeat (the "nopreload condition").
 -10-

Fearwasmanipulated by threatening the subjectswith averypainful or avery

mild shock (highor lowfear conditions).Intheeating situation, following
themanipulation, the subjects tastedand rated crackers for 15minutes,under
theguise ofa "taste test",and Schachter andhiscc—workers simply counted
thenumber of crackers that a subject ate.Bothnormal-weight and overweight
(male)subjectsparticipated intheexperiment.Thenormal-weight subjects
consumedmore crackerswhen theyhadhadnopreload andwere experiencing low
anxiety, thanwhen their stomachswere fullandunder highanxiety. Incon-
trast, theobese subjectsate the sameamount of crackers independently of the
preload or fear conditions.Itwas found, inaddition, that theobese subjects
didnot report any reduction in fearoftheelectric shock following eating.
The findings,therefore,didnotlend support tothepsychosomatic theory,
asobese subjectsdidnot increase consumption underhighanxiety conditions,
andeatingdid not function toalleviate their fearofapainful shock.
Remarkably, aswith thethreat of shock,apreload didnotaffect the amount
eatenbyoverweight subjects.Boththeseconditions,however,did affect the
amount eatenbynormal-weight subjects.From thisobservation, Schachter
concluded that the internal state seems irrelevant tothe food intakeof the
obese,andhe suggested that their eatingbehaviour maybe triggered instead by
external food-related cues.Thisso-called "externality hypothesis"was
examined ina seriesofexperiments,whichwill bedescribed inthe following
chapter.
Schachter's réfutaiof the resultsobtained through clinical experience
provided the impetus forawhole seriesof studies relating tothevalidity of
thepsychosomatic theory,but the initial resultswere generally negative.
Abramson andWunderlich (1972)attempted to refine thetypeof stimulusused to
induce arousal,byusing the threat of shock for themanipulation of objective
fear,and faulty feedback of scoreson the Interpersonal Stability
Questionnaire tomanipulate anxiety.However, they foundno increase in
consumption byobese subjectsunder either of thehigharousal conditions.
(Data relating tonormal-weight subjectswasnotdiscussed, as normal-weight
subjectswere found tobeunresponsive totheanxiety manipulation.)
Both Schachter etal. (1968)andAbramson andWunderlich (1972)had
employed "neutral"crackersasthecriterion food.McKenna (1972)manipulated,
inaddition toarousal conditions, thequality of test foodby offering
subjects eitherbland cookiesor "extremely appetizing and tasty" chocolate
chipcookies.Itwas found that obese subjectsate slightlymore of the tasty
cookiesunder thehigh rather thanunder the lowarousal conditions (p<10),
 -11-

while they atenearly identical amountsofbland cookiesunder both arousal

conditions. Incontrast, normal-weight subjectsate significantly fewer of both
thegood andbland tasting cookieswhen aroused thanwhen calm.As overeating
didnot result inarousal reduction intheobese,onlymarginal support was
obtained for thepsychosomatic theory.
Utilising a fooddispenserwhichdisguised thevisual feedback of the food
intake of subjects,Meyer and Pudel (1972,1977)obtained similar results.They
found that a significant increase in food intakeonlyoccurred inolder obese
womenwhen theyhadbeenexposed tostressors suchasnoise, flickering lights
and insoluble anagrams.Jung (1976)also supplied limited support for the
psychosomatic theory.While the stressof removing toys from four-to six-year
old children reduced their food intake (Jung, 1973), suspense films seemed to
enhance "oral responsiveness"and increase foodconsumption. This response was
most significant inolder, overweight girls (Jung, 1976).
Finally, inanexperiment undertaken byAbramsonand Stinson (1977)obese
subjects atemore crackerswhilstundertaking aboring rather thanan interes-
ting task.This tendency,however,wasalso found innormal-weight subjects,
the results indicating nodifference between theweight groups.
Reznickand Balch (1977), however, foundevidence contradictory tothe
psychosomatic theory. Intheirexperiment ahigher number of overweight
subjects ate test food (candies)under low rather thanhighanxiety conditions.
Contrary toclinical observations, therefore,experimental findings
indicate thateither no increase oronlyaminimal increase infood intake
occurs inobese individualsunder higharousal conditions.Eatingdoesnot
appear to reduceanxiety.Host experiments showed, on theotherhand, that
normal-weight subjects ate lessunderhigharousal conditions.
Slochower (1976)has suggested that thediscrepancybetween experimental
and clinical observations mayhave resulted fromdifferences intheoretical and
experimental definitions ofanxiety.Thesediscrepancies maybe reflected in
theobese person'sdistinctive responses todifferent typesofanxiety states.
Thepsychosomatic explanation ofobesity focuseson the roleofdiffuse (often
unconscious) internal conflicts inproducing theanxiety state that leads to
overeating. Thisanxiety isprobably experienced as "free-floating", rather
thanashaving aclear source.Incontrast, experimental testsof the
psychosomatic hypothesis have induced anxiety through themanipulation of the
external environment (e.g., through the threat of shock), which enabled
subjects toclearly label the sourceof their affective state.Following this
lineof thought, Slochower hypothesised thatemotional eatingwould occur only
 -12-

whenthe source of theobese person'semotional arousalwasunclear.

inanexperimental testof thishypothesis Slochower (1976)demonstrated
thatoverweight subjects increased their food intake onlywhen they could not
label the source of theiremotional arousal.Theamount eatenby normal-weight
subjectswasnot affected by theavailability ofa label toexplain their
arousal state.Slochower and Kaplan (1980)found that theabsence ofa label
and the inability tocontrol the source ofemotional arousal served indepen-
dently toaugment eating inobese,butnot innormal-weight, subjects.Similar
resultswere obtained ina studywhere theeffectsofanaturally occurring
stressor (final examinations)were studied ina repeated measures design
(during and threeweeksafter examinations) (Slochower, Kaplan andMann, 1981).
Similarly, Loweand Fisher (1983)found evidence insupport of the
emotional eating theory ina study conducted inanatural setting. Inthis
study subjects self-monitored theirmoodprior toeach instance of food intake.
Itappeared that obese subjectsweremore likely toengage ineating following
negative emotions thannormal-weight individuals.These findingsapplied only
tosnacks,andnot tomeals,andnoobese/normal differenceswere registered
with regard toconsumption following positive emotions.Inaddition, it
appeared that "negative"emotional eating, associatedwith bothsnacks and
meals,was related tothe subjects'percentage overweight.
Therefore, ithasbeen seen thatobese individualsovereat in response only
tocertain typesof stressors.Theabovementioned studiesof Slochower showed
thatoverweight individuals overate inresponse todiffuseemotional states,
butnotwhen the source ofemotional arousalwas clear.
Slochower observed inlater studies (Slochower and Kaplan, 1980;Slochower,
KaplanandMann, 1981;Slochower, 1983 ;Slochower and Kaplan, 1983)that a
crucial aspect of the "diffuse"anxietyexperience isthat itleaves the
individual feeling relatively powerless toalter hisorher internal state.
These studies indicated thatoverweight subjects overatewhilst experiencing
highuncontrollable anxiety, butnotwhen theanxietywas controllable.Twoof
these studies (Slochower, KaplanandMann, 1983;Slochower, 1983 )also
demonstrated thatavariety ofuncontrollable emotional statesmay result in
overeating on thepartofobese individuals,apositive relationship being
observed between low self-esteem andovereating inthisgroup. Finally,
findings obtained instudiesofuncontrollable emotional arousal (Slochower,
1976;Slochower and Kaplan, 1980;Slochower and Kaplan, 1983)suggested that
eating in response touncontrollable stresshasananxiety-reducing function
for theoverweight. The resultsof these studies revealed apositive
 -13-

association between emotional eatingand arousal reduction inthe overweight

subjects,butnot inthe subjectsofnormalbodyweight.
According toSlochower, theobservation ofapositive relationship between
overeating and lowself-esteem suggests, firstly, thatavariety of
uncontrollable emotional statesmay result inovereating inobese subjects,
and, secondly, that thedimension ofthecontrollability ofemotions isof
relevance tovirtually anyemotional state.As confirmed byKaplan and Kaplan
(1957),what thismeans forpsychosomatic theory isthatovereating and obesity
donot seemtobe related toany singledevelopmental fixation.The concept of
thecontrollability ofemotionsdoesnot seemtobe statespecific ina
developmental sense,as itcanbearoused atanytimeduring thedevelopment of
the individual.

SUMMARY

While agreeing that overeating results from internal emotional arousal,

psychoanalytic theoristsdisagree about themeaningofovereating for the
individual.Studiesofemotional reactivityhave consistently found that obese
individuals experience negative emotionsmoreoftenandwith greater intensity
thandonormal-weight individuals.Interviewandquestionnaire studiesof obese
individuals, inmostly clinical settings,have found support for the
psychosomatic theory inthat themajority oftheobesewere seentoovereat
when feeling emotionally aroused.However,experimental tests,examining the
effects ofarousalmanipulations on theeatingbehaviour ofobese and non-obese
subjects,have initially foundnooronlyweak indications thatobese persons
overeat in response toemotional arousal.Moreover,eatingwasnot shown to
reduce anxiety.
Inanattempt toexplain thediscrepancy between "clinical"and
"experimental"observations, Slochower introduced theconceptsof "diffuse"as
opposed to "clearly labelled" emotions,and controllability asopposed to
"uncontrollability" ofemotional states.Findings revealed that overweight
subjects overate only in response toemotional stateswhichareexperienced as
diffuse and out ofcontrol.Inaddition, itwasshownthateating in response
touncontrollable stresshasananxiety-reducing function for the overweight.
Thus, itappears that obese individuals overeat in response to certain
typesof stressors,and that thisovereating occursbecause "itworks",that
is,because ithasananxiety-reducing function.
 -14-

CHAPTER 4

EXTERNALITY THEORY

Ashasalready been stated, Schachter formulated theexternality theorywhen he

didnot find support for thepsychosomatic theory inhisexperiment with
Goldman andGordon (Schachter,Goldman andGordon, 1968). Itwasobserved that
neither apreload nor a threat of shock affected theamounts eatenby
overweight subjects.Therefore, their internal state seemed tohaveno impact
upon their eatingbehaviour.Reasoning fromthis,Schachter suggested that the
eating behaviour of theobesemaybe triggered byexternal food-relevant cues,
suchas the smell,sightand availability of food.
Ina similarway tothepsychosomatic theory, therefore, the externality
theory alsoposits that theeating behaviour of overweight individuals is
comparatively unresponsive tointernal physiological signals, suchas gastric
motility. But incontrast to theemphasis placed on internal,emotional factors
inthepsychosomatic theory, theexternality theory focuseson the external
food environment asadeterminant ofeatingbehaviour.Overweight individuals
are considered asbeinghyperresponsive toexternal food-related cues.The
sensitivity ofobese persons toexternal cueshigh insaliencewas later
extended toinclude externality asageneral personality trait, rather than as
a trait specific toeating behaviour (Schachter andRodin, 1974).

RESEARCH

Sensitivity toexternal food-related stimuli. Ina seriesof experiments,

Schachter and his colleagues (Schachter andGross,1968;Goldman, Jaffa and
Schachter, 1968;Nisbett, 1968 a , )explored the experimental and theoretical
implications of theexternality theory.Two studies revealed that the eating
behaviour of the obesewas influencedmore by time than that of normal-weight
subjects.Ina studyundertaken bySchachter andGross (1968) "dinnertime"was
manipulated by theuse ofdoctored clocks,and the results indicated that the
obese atemorewhen they thought that theywere eating after their regular hour
than theydidwhen they thought theywere eating before their dinner hour.
Similarly, a studyofGoldman etal. (1968)demonstrated that fat fliershad
lessdifficulty inadjusting totime-zone changeson longdistance East-West
travel thandid their normal-weight counterparts.
 -15-

Two further studies revealed that theeatingbehaviour of overweight

subjects ismore influenced bythe sensoryattributes than thatof
normal-weight subjects.ResearchbyNisbett (1968a)showed that overweight
subjects ate lessquinine-adulterated icecream and slightly more of an
excellent andgood tastingvariety ofvanilla icecream than individuals of
normalweight.A similar observationwasmadebyGoldman etal. (1968), who
demonstrated that fat studentsappeared tobe less tolerant ofdormitory food
than their normal-weight counterparts.
Finally, itappeared that theeatingbehaviour of overweight subjects was
more influenced by thevisualityof food than that ofnormal-weight subjects.
Nisbett (1968 )found that overweight subjects ateanaverage ofalmost one
sandwich lesswhen one sandwichwasvisible,andmorewere available ina
nearby refrigerator, thanwhenthree sandwicheswerevisuallyprominent and
within easy reachon the table.Meanwhile, thevisual prominence of presented
fooddid not affect theeatingbehaviour ofthenormal-weight subjects.
Similarly, Goldman, Jaffa and Schachter (1968)observed that fat, religious
Jewsweremore likely tofastonYomKippur,when food cueswereminimal, than
their normal-weight counterparts.
The evidence obtained inthese studieswas summarised bySchachter (1967)
inthe formof twogeneralisations. Firstly, overweight individuals seemed to
behighly responsive to "external cues",that is,tocuesinherent in foodor
theenvironment. Secondly, overweight individuals seemed toa largedegree to
beunresponsive to "internal"physiological cuesofhunger and satiety.
Implicit inthisdistinctionbetweenexternal and internaldeterminants of
eating behaviour, isthenotion that overresponsiveness toexternal variables
andunresponsiveness to internal variables on thepart of obese individuals
should lead toovereating and subsequent overweight (Nisbett, 1968 a ).
Sensitivity toexternal non-food related stimuli.Thenext seriesof
experimentswasundertaken inorder to investigate whether the heightened
responsiveness of theobese alsoholds goodwith respect to non-food-related
stimuli,and totest thehypothesis that overweight subjects are stimulus bound
and therefore generally sensitive toanysortof event.
Intwoexperiments (Rodin,1973)overweight subjectswere found tobemore
disrupted thannormalweight-subjects by interesting, emotionally-toned
material inmeasures ofproof-reading accuracy and reaction time latency. In
contrast, theywere found toperformbetter than thenormal-weight groupwhen
therewere nodistracting events.The same study revealed that overweight
subjects reflected moreabout the tasks inhand thannormal subjectswhen there
 -16-

werenodistractions,which suggeststhat thedifferencebetweennormal-weight

andobesesubjectsintaskperformancemayhavebeenduetodifferencesin
attention.
A number ofstudiesalsodemonstrated thatoverweight subjectsweremore
emotionally responsivethannormal-weight individualsto highly salient,
affective stimuli,and lessemotionally responsivetoneutral stimuli (Pliner,
1974; Rodin,ElmanandSchachter,1974;Rodin,Slochowerand Fleming, 1977).
Thisaccordswith the resultofpreviouslymentioned studiesconductedwithin
the framework ofthepsychosomatic theory,inwhichitwasconsistently found
that obese individualsweremoredistractedbyavarietyofnoxiousstimuli
thannormal-weight subjects (e.g.,AbramsonandWunderlich, 1972;Pliner,Heyer
andBlankstein, 1974). Inother studies,obese subjectsappeared tohavea
quickermeanlatencyin respondingtocomplexexternal stimuli,lower
tachistoscopic recognition thresholdsandbetter immediate recallof food-and
non-food-relevant cuesthannormal subjects (Rodin,HermanandSchachter,1974;
Rodin,Slochower and Fleming, 1977). thisexternal responsivenesswas shownto
extend tosuchareasastimeperception (Rodin,1975)and information
processing (Rodin,HermanandSchachter,1974;RodinandSlochower, 1974).
Inconclusion, itappearsthat theobesediffer fromnormal subjectsinthe
way information isattended toandprocessed,and thatovereating isjustone
aspectofexternalityasapersonalitytrait (RodinandSinger, 1976).

THEEXTERNALITY THEORYCHALLENGED

The stimulusbindingconcept.ResultsobtainedbyNisbett andTemoshok (1976)

strongly challenged thecontention thatobesepersonsdiffer fromnormal-weight
individuals intheir styleof response toexternal stimuli,inthat they
exhibit ageneral tendencytowards "stimulusbinding",an "external"cognitive
style. Intheir studyavarietyoftasksmeasuringthestyleof response to
external stimuliwereadministered toapopulation,toassesswhether the
variousmeasuresofcognitive styleare interrelated. Itwastheir contention
that itwouldonlybesensibletospeakofasingleorbroad "external"
cognitive styledimension, ifthecorrelationsbetween themeasureswere found
tobeuniformlyhigh.The results indicated thatthecorrelationsamongst the
measuresofcognitive styleweregenerallylowor zero.Even responsestothe
taskswhichhadbeen foundbySchachterandRodin (1974)todifferentiateobese
andnormal-weight individualswerenotcorrelated inthepopulationexamined by
NisbettandTemoshock.Inasimilarway, IsbitskyandWhite (1981)also failed
 -17-

to findapatternof intercorrelationsamongst thevariousmeasuresof

external-cue sensitivityandlocusofcontrol,whichwouldhaveoffered support
tothestimulus-binding conceptoftheexternality theory.
Brand andClotz (1982)similarly failedtofindasingle instanceofa
positive relationshipbetweenRotter'smeasureoflocusofcontroland
Schachter'smeasuresofexternality, intheir reviewof research investigating
the relationshipbetweenthesetwovariables.Theauthorscontended that this
absenceofa relationshipwasnotsurprising,asthetwoconceptsare
theoretically incompatible.Schachter'sstimulus-binding conceptcanbe
considered asaStimulusModel.Becauseof itsemphasisonthe individual's
responsetoexternal stimuli,anditslackofattentiontocognitiveprocesses,
suchasthedecisionnot toeat in responsetotempting external stimuli or
internal cuesofhunger,itmustbeclassified asabehaviouristic learning
theory (note 1)*. Incontrast,Rotter'sconceptoflocusofcontrol canbe
consideredasacontingencymodel,andmustbeclassified asacognitive theory
(note2 ) ,for reasonof itsemphasisonthe individual's cognitive expectation
that futureeventscanbecontrolledbytheindividualhimself,orare totally
determinedbyoutwardcircumstances.BrandandClotzconcludethat the
usefulnessofaglobalconcept likeexternality ishighlyquestionnable.
External controlof feeding.Variousstudiesfailedto replicate the
findingsofSchachterandhisco-workers (e.g.,Milich,AndersonandHills,
1976;MilichandFisher,1979;Meyers,StunkardandColl, 1980;Isbitskyand
White, 1981), findingnoevidence tosuggest thatoverweight subjectsdiffered
fromnormalsubjectsintheir responsivenesstoexternal foodcues.
Rodin,togetherwithSchachteroneofthemainadvocatesofthe externality
theory,attributed thisdiscrepancy intheresultstothearbitrarinessof
sampling. InastudycarriedoutwithSlochower (1976)sheobserved that in
everyweight categorytherewere individualswhowere responsive toexternal
stimuli,and individualswhowerenot.Her studywithWooley etal (1976)
demonstrated thatthesameappeared tobetruewith respectto internal
responsiveness.Thus,according toRodin (1981 a ), thediscoveryof
overweight/normal-weight differences inanyaparticular studyisdependent on
".... howmany individualsofeachtypewindup insamplesdividedaccording to
weight" (I.e.p.383).However,sheobserved inaddition,that,acrossall
weightgroups,thedegreeofoverweightwasnot strongly related tothedegree
ofexternalor internal responsivenessdemonstrated inthese studies (Rodin,

* seepage 39fornotes
 -18-

Slochowerand Fleming, 1977), and that "Infact,atextremedegreesofobesity,

someindividuals showedverylittle responsiveness toexternal cues"(Rodin,
1981 a ,p.363).TheseobservationsledRodintoseriouslyquestion thevalidity
of theexternalitytheory,and stating thattheuseof the internalversusthe
externaldistinction toexplaindifferences intheeatingbehaviour ofobese
andnormal-weight individualsseems fartoosimplean interpretation (Rodin,
1978).
Internal controlof feeding.Early researchonthecausal factorsof
overeatinghasbeengreatly influencedbytheimpression thattheeating
behaviour ofnormal-weight individuals isdeterminedbyinternalphysiological
signalslikegastricmotility,while these signalsdonot functionasahunger
cue for theobese.Cannon (1915)andCarlson (1916)pointed togastric hunger
contractionsasdeterminantsoftheexperienceofhungerand the regulationof
food intake.Following thislineof thought,Stunkardandhis co-workers
investigated thedegreeofcorrespondencebetweenthestateof the stomachand
self-reportsofhunger inobeseandnormal-weight individuals,usingagastric
balloon tomeasuregastriccontractions (Stunkard, 1959 ;Stunkard andKoch,
1964).Ashasalreadybeenmentioned,within the frameworkof thepsychosomatic
theory, these studiesdemonstrated that someoverweight individualsweremuch
lesslikelythannormal-weight individualsto relatetheir stomachcontractions
to theconsciousexperienceofhunger.
However, thereare indicationsthatalackof internal responsivenessalone
doesnot leadautomatically toovereatingorweight gain.Firstly,obese/normal
differences inthelabellingofgastriccontractionsdonot seencrucial for
thepredictionofobese/normaldifferences ineatingbehaviour and food intake.
Stunkard and Fox (1971)utilisedamore sophisticatedmeasureofgastric
motility inalater study,but failedtofinda relationshipbetweengastric
motilityand reportsofhunger insubjectsofnormalweight. Itevenappeared
that gastricmotilityplayedonlya small role inthehunger experienceofboth
obeseandnon-obese subjects.This isconsistentwith thecontentionof
Janowitz (1967)that " atpresentnopublisheddataexistsaboutany
species that indicatewhetherornot food intakeorother hunger-motivated
behavioral parametersareactuallycorrelatedwithgastrichunger contractions"
(I.e.,p.220).Asaconsequence, theearlyobservationsofStunkard andhis
cc—workersconcerningobese/normaldifferences inthelabellingofgastric
contractionsdonot tellusanythingaboutobese/normaldifferenceswith
respect totheir responsiveness tointernalcuesofhungeror satiety.
 -19-

Secondly, thereace indications thatnormal-weight individuals alsohave

difficulties inadjusting their eatingaccording to internal signals (Wooley et
al., 1976).When reporting hunger,bothnormalweight andoverweight subjects
were influencedmore bytheir initialbeliefsconcerning energyvalue thanby
theactual energyvalueof the food,and showedalmost noability to identify
mealsashighor lowinenergy content (Wooley,WooleyandDunham, 1972).

CONFOUNDERSOFTHE INTERNAL/EXTERNAL DISTINCTION

Dieting.Pudel andhisco-workers (Pudel,MetzdorffandOetting, 1975;Meyer

and Pudel,1977;Pudel,1978)studied the satietycurvesofnormal-weight and
obese individuals,byusing a fooddispenserwhichdisguised thevisual
feedback of the intakeof food.Theyobserved that the rateof consumption ofa
20-minutetestmealdifferentiated obese fromnon-obesepersons (Meyerand
Pudel, 1977).Theeating rateofnon-obese individuals slowedduring themeal,
while thatof theobesedidnot, suggesting an impairment of feelings of
satietyamongst theobese.Thedistinction between obese and non-obese
individuals,however,wasblurred bythediscovery ofagroupof non-obese
personswhose rateofeatingdidnot lessenduring themeal.Further
questioning ofall the subjects revealed that, incontrast to the remaining
subjects,this so-called "latent obese"group-biologically programmed tobe
obese,but ofnormal bodyweight -hadoncebeenoverweight, buthadmanaged to
maintain anormal bodyweight atthe timeoftheexperiment bymeansofa
deliberate restriction of food intake.Thisobservation ledPudel to conclude
thatbodyweight resulting fromthedegree of internal responsiveness isnot
necessarily thesameastheactualbodyweightofan individual,asbodyweight
canbe controlled tosomeextent bydieting.
Inanendeavour toidentify those individualswhohaveobese eating
patterns,butareofnormal bodyweight due toaconscious restriction of food
intake,Pudel,MetzdorffandOetting (1975)constructed theFragenbogen für
LatenteAdipositas (FLA). Thisquestionnaire proved successful in identifying
individuals ofnormalweightwho failed toslow their eating during ameal,
that is,the "latent obese" (Pudel, 1978).
Thediscovery of "latent obese"subjectswithin a sample of normal-weight
subjects appeared tooffer apossible explanation of thedifficulties of
replicating theexperiments ofSchachter andhiscc—workers.Thepossibility of
identifying the "latent obese"bymeansofaquestionnaire wasconsidered asa
major step forward inovercoming problems insampling subjects (Pudel, 1978).
 -20-

Physiologicalandfurtherpsychological factors.RodinandSlochower (1976)

observed, inastudyoftheweightchangesofnormal-weightgirlsinaneight
week summer campwhere foodwasabundantand freelyavailable,thatgirlswho
werehyperresponsive toallkindsofexternalstimuli,asdetermined ina
pre-test,were thosewhogainedmostweight inthis "landofmilkandhoney".
However,70%of thisgroupofexternally responsivenormal-weightgirlswho
gainedweightduringthesummer camp, reachedtheirhighestweightbefore the
finalweekof thecamp,and thereafterbegan tolooseweight.According to
Rodin (1981 a ), thissuggested thatphysiological andpsychologicalvariables,
suchasbody image,areultimatelymore important thanexternal responsiveness
indetermining an individual's finallevelofbodyweight.Rodindidnot refute
thecontention thatexternal responsiveness istheprimaryand, therefore,the
possible causal factor ofovereatingandweight gain,givenaplentiful food
environment.However, shedidproposeamoredifferentiatedmodel inwhichthe
regulationof foodalsodependsuponthe interactionofphysiological,sensory,
cognitive-motivational and socio-culturalvariables (Rodin, 1981 a ).

SUMMARY

According toSchachter'sexternalitytheory,overweight individualsovereat,

because theyareunresponsive tointernalcuesofhungeror satiety,and
hyperresponsive toexternal food-relatedcues.Thisexternal responsivenesswas
later extendedtonon-food-related stimuli,obesepersonsbeing consideredto
exhibitan "external"cognitivestyle.
ExperimentscarriedoutbySchachterandRodinhavediscovered consistent
obese/normaldifferencesinresponsivenesstofood-andnon-food-related
stimuli.However,thecontentionthatthereexistsan "external"cognitive
stylehasbeenchallengedbyvariousinvestigators.AlsoSchachterandRodin's
experimentsdealingwithresponsivenesstofood-related stimulihavebeen
difficulttoreplicate.Furtherstudiesrevealedthatthe internal/external
distinctionoftheexternalitytheorymayconstitutetoosimpleanexplanation
ofdifferencesintheeatingbehaviourofobeseandnormal-weight individuals -
itappeared that ineveryweightcategorythereareindividualswhoare
responsivetoexternal stimuliorunresponsivetointernalstimuli.
Pudel'sdiscoveryofnormal-weightindividualswithobeseeatingpatterns
demonstratedthat theinternal/externaldistinctionmaybeconfoundedby
variablessuchasdieting.Alongthesamelines,Rodinproposedthatphysiolo-
gicalvariablesmayhavean influenceonthefinal levelofan individual's
 -21-

bodyweight.As aconsequence, theexternality theorywas replaced byanew

model, inwhich the regulationof fooddepended not onlyon external
responsiveness,but alsoon theinteractionofphysiological, sensory, cogni-
tive-motivationaland socio-culturalvariables.

CHAPTER 5

THEORYOFRESTRAINED EATING

Set point theory. In1972Nisbettput forward analternative explanation of

obese/normaldifferences ineatingbehaviour.He suggested that a state of
relative physiological deprivation, rather thanexternality as personality
trait,accounted for theheightened external sensitivity oftheobese,as
demonstrated by Schachter andRodin (1974). Fromthisnewperspective, Nisbett
advanced the "setpoint theory"ofobesity,whichposits that some individuals
arephysiologically "programmed" tobeobesebecause ofanexcessnumber of fat
cells,a resultofgenetic factorsorofoverfeeding duringpre-adult
development. Inorder tonourish thesecells such individuals are
physiologically predisposed bytheir hypothalamic "feeding center" to overeat.
However,at the sametime,social and cultural constraintsdeter the individual
from fully satisfying hisbiological requirements.The result couldbea person
who, inabsolute terms,isoverweight,butwho isphysiologically farbelow the
setpoint atwhich the fat cellsareadequately nourished.This compromise of
biological and socialdemandswouldproduce an individualwho isboth
overweight and chronically hungry.
Nisbett'slineof reasoningwasgreatly influenced bytheobservations of
Cabanac andhisco-workers.While researching obese/normal differences in
sensitivity to internal stimuli,CabanacandDuclaux (1970)had investigated
theeffectsofglucose loadsontheaffective responsiveness to sweet
solutions.Thepleasantnessorunpleasantness ofthe sweet sensation is
controlled inpartby internal signalsofhunger and satiety.Therefore the
affective response ofobese individuals togustative stimulation isof
relevance tostudiesoftheir responsiveness tointernal stimuli.The results
indicated thatnormal-weight individuals foundotherwise pleasant tasting sweet
solutions tobeaversive,when theysampled thema fewminutesafter a gastric
loadofglucose.Thisso-calledalliesthesiawas found tobeabsent inthe
obese.For themaglucose load onlyslightlydecreased theattractiveness of
 -22-

the sweet solutions.Ina similarway totheobese group, subjectswhohad

lost 10%bodyweight alsofailed toshowalliesthesia following aglucose load.
Thisphenomenonwasalsoobserved inastudy inwhichCabanac and twoofhis
cc—workers (Cabanac,Duclaux and Spector, 1971)investigated their own (sic!)
response toglucose loadsbefore,duringand afterweight loss.Prior toloss,
and afterweightwas regained, alliesthesia wasobserved.During theweight
lossand themaintenance of lowbodyweight, alliesthesia was found tobe
absent.
Cabanac et al. (1971)interpreted these resultsasevidence of the
existence of a "ponderostat",that isa regulatory systemwhichhomoeostati-
callycontrolsbodyweight.Theponderostat detectswhen theorganism isbelow
itssetpoint,with the results that eventswhichwould signal satiety tothe
organism at setpoint level are ignored, andalliesthesia isabsent.Thus,
according toCabanac and hisco-workers, thepresence orabsence of alliesthe-
sia following a glucose loadmay indicatewhether the subject isat or below
setpoint.As aconsequence, theabsence ofalliesthesia intheobese subjects
studied byCabanac andDuclaux (1970)was seenasan indication that these
persons are, for reasonsofhealthoraestheticism,below set point.
According toNisbett, thecontention thatbodyweight is homoeostatically
regulated implies thatapersonwho isbelowsetpoint should show "obese
eatingbehaviour", regardless ofwhether thisperson isobeseorof normal
weight.Moreover, thebehaviour ofanobesepersonwho isat setpoint should
resemble thatofnormal-weight individuals.According tothis lineof thought,
deviation from the setpoint isamore fundamentaldeterminant, and abetter
predictor ofbehaviour thanbodyweight. Itmayalsoexplainwhy heightened
external responsiveness anda lack of internal responsivenesswas observed
across allweight categories.
Cognitive restraint of food intake:dieting.Withthecontention that an
individual maybeable tosuppress hisorher bodyweight belowsetpoint bya
conscious restriction of food intake,anewelementwas introduced into the
complex rangeofdeterminantswhich governeatingbehaviour, theessential men-
talelement of self-control.
Herman, Polivyand their co-workersexpanded the ideaof self-control, and
made cognitive restraint, that is,the cognitive resolvenot toeat, the focal
point of their theoryof restrained eating.These investigators contended, in
linewithNisbett, thatobese/normal differences ineatingbehaviour may re-
flect a conditionwhere themajorityofobese individualsarechronicdieters,
somethingwhich isprobably not true fornormal-weight individuals.However,
 -23-

they latermodifiedNisbett'soriginal explanation astowhydieting influences

behaviour (e.g.,Herman,Polivy, Pliner,Threlkeld andMunie, 1978).Nisbett
had suggested thatmost obese individuals,becauseof their extensive dieting,
and their resultingmaintenance ofbodyweight below "setpoint", are hungry.
Alternatively, Hibscher andHerman (1977)suggested thatdietingmayactasa
stressor,and the stressofdieting couldpossibly result inthe heightened
externality andemotionality of theobese.Thismeans thatevenwhen Jieting
failstoproduce substantialweight loss,andbodyweightdoesnot fall below
setpoint, side-effectsmay stilloccur.

RESEARCH

Inorder tobeable tomeasure an individual's level of cognitive restraint,

Herman, Polivyand their co-workersdeveloped theRestraint Scale (Hermanand
Hack, 1975;Herman and Polivy, 1975;Herman,Polivy, Pliner,Threlkeld and
Munie, 1978).At oneendof the continuum are restrained eaterswho restrict
their food intake inorder tocontrol bodyweight.At theother endare totally
unrestrained eaterswho seldomgivea thought totheamount of food theyeat.
This scale isadministered topopulations ofpredominantly normal-weight indi-
viduals,anda classification of subjectsasdietersornon-dieters isachieved
bymeansof amedian split of scoreson theRestraint Scale.
Effects ofdietingoneating.The firstexperiment (HermanandHack, 1975)
wasdesigned totest thehypothesis that ifthechronic restraint ofdieters is
overcome, theywould showa "latent" formofexternality, and eatmorewhen
exposed toprominent food cuesthan individualswhodonot chronically restrain
their eating.Theexperiment waspresented tothe (female)subjectsunder the
guise of a tastetest.Subjectswere required toconsume 0,1or 2milkshakes
(preload),and then to "sample" some icecream. The results revealed that
non-dietersate less following apreload. Incontrast,dietersatemuch more
icecream after having consumed apreload excessive inenergy.
Herman andMack suggested that themilkshakes serve tobreakdown the
cognitive restraint thatdietersnormallyexhibit inthe faceof tempting
stimuli.Theexcessive consumption of icecreamby thedieters following a
preloadmustbe attributed tothebreakdownof their normal restraint, a result
of exceeding their caloric threshold; ".... without a realistic hope of staying
within the caloric confines imposed forherself, thedieterwas leftwithout a
sufficient reason fordieting" (Hermanand Polivy, 1980 a ,p.245).Later
studies (Polivy, 1976;Spencer and Fremouw,1979;RudermanandWilson,1979;
 -24-

tion"does indeed haveacognitive effect.Whenperceived calories inthe

preloadweredistinguished fromtheactual calories, itwas found that dieters
overate onlywhen theythought that theyhad consumed ahighcaloric preload.
The subsequent study investigated theeffect ofanxiety (threat of shock)
upon theconsumption oficecream innormal-weight dietersand non-dieters
(Hermanand Polivy, 1975).The results suggested twothings.Firstly, the
resultsobtainedwhere the restraint/non-restraint distinctionwasapplied were
found toparallel thoseobtained using theobese/normal distinction in studies
suchasthoseundertakenbySchachter etal. (1968)andMcKenna (1972).
Normal-weight dieters, likeobese subjects,ate slightlymorewhen anxious.In
contrast, normal-weight non-dieters, likenormal-weight dieters,asassessed by
emotional self-reportmeasures,were found likeobese subjects,to over-react
totheanxietymanipulation. Itseemed, therefore,thatbehaviourwhichhad
formerlybeen thought onlytocharacterise theobese,also characterised
normal-weight dieters.Secondly, thedata suggested thatanxiety, inthe same
wayasapreload, serves todisrupt cognitively-mediated self-control proces-
ses, suchasdietary restraint.Thedisinhibition ofcognitive restraint
results inan increased icecream consumption by the restrained eaters,but not
by theunrestrained eaters,thelatter groupnotbeing inhibited in their
eatingbehaviour tobeginwith (note3 ) .
The same interpretative schemewasapplied todata collected ina studyof
the relationshipbetweenweight change and clinicaldepression (Polivyand
Herman, 1976 a ).A totalof12clinicallydepressed patientswasdesignated as
restrained orunrestrained eaters,and itwas found that the former group
tended togainweight astheirdepression increased,while the latter group
tended toloseweight.Thus,depression, likeanxiety, appeared todisrupt
self-control processeswith respect tofood intake in restrained eaters.
Thenotion ofthedisinhibition ofself-controlprocesseswas further
bc
exploredusing theclassicdisinhibitor, alcohol (PolivyandHerman, 1976 ').
Itappeared that alcohol served todisrupt cognitive restraint indieters only
when they fullyunderstood that theywere consuming alcohol,that is,when both
the cognitive andpharmacological pre-conditions foradisinhibitory experience
were inforce.Under these circumstances, the ingestion ofalcohol increased
the food intakeof restrained, butnot ofunrestrained, eaters.
Further research revealed thatunder the influence ofsocial factors
restrained eatersare capableof "sensible eating"andunrestrained eatersof
"counterregulatory eating".A studyofHerman,Polivyand Silver (1979)showed
that thepresence ofa researchworker served to induce "sensible eating" in
 -25-

that thepresence ofa researchworker served to induce "sensible eating" in

restrained eaters,inthat theyshoweda compensatory response,eating less
after a largepreload andmoreafter a smallpreload. Italso appeared,
however, that this sensible eating ".... iseasily superseded bymore habitual
(andless sensible)eatingpatterns" (Hermanand Polivy, 1980 a , p.221).When
left alone tocarryout a further, thistimeunobserved, eating task, the same
subjectswhohad justdemonstrated sensibleeatingbehaviour, abruptly reverted
totheir counterregulatory eatingpatterns.A studyofPolivy,Herman, Younger
andErskine (1979), found thatthepresenceofadietingmodelwhobrokeher
diet, served to inducea counterregulatory response inunrestrained eaters.
StudiesbyKirschenbaumandTomarkenalso indicated that restrained eaters
are capable of sensible eatingandunrestrained eatersof counterregulatory
eating. Inone study (Kirschenbaum andTomarken, 1982) restrained eaters
demonstrated sensible eating behaviourwhen stimuliwere provided that improved
their self-monitoring abilities (presenting food ina smallbowl,and informing
the subjects about itsenergy content),unrestrainedeaters demonstrated
counterregulatory eating patternswhen these stimuliwerenotprovided.Another
study (Tomarken and Kirschenbaum, 1984)showed that both restrained and
unrestrained eatersatemore test foodwhentheyhad anticipated adinner of
high energy content, thanwhen expecting adinner of lowenergycontent or no
dinner at all.Fromtheobservation thatunrestrained eatersarealso capable
of counterregulation, itwas concluded that restrained eating isa continuous
rather thanadichotomousvariable (HermanandPolivy, 1980 a ): unrestrained
eatersdiffer only from restrained eaters inthat theyhaveahigher threshold
of counterregulation.
One general conclusion that canbedrawn from the foregoing research is
that,as they tend toovereatwhen their cognitive restraint isbrokendown,
restrained eaters riskgainingweight.Theabilitytomaintain restraint was
found todepend on such factorsasemotional state,theperception of having
broken thediet,and thebehaviour ofotherpersonspresent, inshort, factors
".... oftheworld around them" (PolivyandHerman, 1983,p.153).Itwasalso
indicated thatbehaviourwhich had been thought formerly tocharacterise only
theobese,also characterised normal-weight dieters.The study of the effects
ofanxiety on theconsumption of icecream showed thatnormal-weight dieters,
like theobese,appeared insensitive to internal signalsof satiety, and ate
slightlymorewhen anxious.Inaddition,normal-weight dieters, like obese
subjects,appeared emotionally reactive, inthat theyover-reacted tothe
anxiety manipulation.
 -26-

Sideeffects ofdieting. Further studies revealed that the

restraint/hon-restraintdistinction alsoparalleled the obese/normal
distinction innon-eating situations.Replicating a studyofPliner,Meyer and
Blankstein (1974), itwasobserved thatdieters,like theobese, responded more
strongly toemotional stimuli thannon-dieters.Similarly,by replicating
Rodin's studyof theeffectsofdistraction and emotional arousal onproof-
readingperformance (1973), itwas found that theperformance ofdieters, like
thatof theobese,was impaired bydistracting stimuli (Herman,Polivy, Pliner,
Threlkeld andMunie, 1978).
Hermanandhisco-workers interpreted thedataobtained inthe studyof
proof-reading performanceasbeing indicative ofanelevated arousal level in
dieters.Thearousal/performance curve,whichportrays the relationship between
performance andarousal asan invertedU-function,withoptimal performance
occurring during intermediate levelsofarousal,was takenasa starting point.
Dieterswere ina stateofarousal optimal forperformance before the
imposition of adistractor (arouser), showing superior performance on
proof-reading tasksunder conditions ofnodistraction, and impaired
performance under conditions ofdistraction.As theirperformance improved
under conditionsofdistraction, non-dieterswere atanarousal level less than
optimal forperformance before the imposition of adistractor (arouser).
According toHermanandhis co-workers, theelevated arousal level amongst
dieters results fromthedieting,which actsasa stressor.Some support for
thisassumption wasoffferedbyHibscher andHerman (1977),who found that
dieters haveanelevated level of free fattyacids,which ispresumed tobea
physiological indication that theorganism isexperiencing stress.
Theobservation thatdieters,like theobese,demonstrated elevated
emotionality anddistractability, was takenasan indication that ".... an
external orientation appears tobea traitofdieters ingeneral,and isnot
confined tothe obese, thevastmajorityofwhom, of course,are chronic if
unsuccessful dieters" (Polivy, HermanandWarsh, 1978,p.498)and that "....
dieting, rather thanobesity isthe critical determinant of at least some
"obese/normal"differences" (Hermanetal., 1978,p.536).Inaddition, the
hyperemotionality anddistractibility ofdieters (andobese persons)were
regarded asconsequences of the stressof dieting.
 -27-

THETHEORYOFRESTRAINED EATING CHALLENGED

Itsapplicability tooverweight persons.According toRudermanandWilson

(1979), thereare twomajor problems associatedwith the theoryof restrained
eating. Firstly, thenotion that restraint isamore fundamental dimension than
weighthasneverbeenadequately tested.Secondly, ithasnever been
demonstrated that counterregulation isaneating style typical to theobese.
A studymadebyHibscher andHerman (1977),which included a sample of
obese individuals,did lead totheobservation that consumption following a
preloaddepended onthedegreeofdieting,andnotonobesity,with dieters
exhibiting counterregulation, irrespective ofweight classification. In
addition, elevated levelsof free fattyacids (asexplained above,ameasure of
stress), which arenormally found inobese individuals,appeared tobe
associatedwithdieting, rather thanobesityper se.Itshouldbe stressed,
however, thatHibscher and Hermandidnotexamine the interaction between
weight and restraint,which iscrucial totheassumption that restraint isa
better predictor ofconsumption thanbodyweight.Thus,according toRuderman
andWilson, thenotion that restraint predicts thebehaviour ofobese and
normal-weight individualsequallywell remainstobe demonstrated.
Inorder toexamine themainpremise of restrained theory, that restraint
isamore fundamental determinant ofeatingpatterns thanweight, Ruderman and
Wilson (1979)modified theparadigmused inearlier studiesof restrained
eating theoryby includingweight asa factor.Theyexamined the findingsof
their own study togetherwitha reanalysis ofdata taken fromtwoother studies
of restrained eatingbehaviour (Hibscher andHerman, 1977;Spencerand Fremouw,
1979).Thedata indicated that counterregulation isnot apatternof behaviour
typical toobese individuals.Neither restrained norunrestrained obese
individualsexhibited counterregulation, althoughunrestrained obese subjects
appeared tobe thebetter regulators,eating lessfollowingapreload than in
theabsenceofapreload. The restrained obese groupate similar amountswith
orwithout apreload.This findingwas replicated ina second study (Ruderman
andChristensen, 1983).
Thus, itappears that restraint scorescanpredictdifferent behaviour
patterns inobese andnormal-weight groupsofindividuals:counterregulatory
eatingbehaviour seems typical of restrained normal-weight individuals,but not
of restrained obese individuals.This raisesthequestionastowhether these
results reflect inconsistencies inthetheoryof restrained eating,or
difficulties inthemeasurement ofcognitive restraint, that is,inthe
 -28-

Restraint Scale.As theRestraint Scale iscentral tothe theoryof restrained

eating, thenext sectionswill examinetheRestraint Scale,asdeveloped by
Herman andPolivy, inmoredetail.
TheRestraint Scaleand itsmulti-dimensionality.Threedifferent measures
of cognitive restraint havebeenused instudiesof restrained eating behaviour
(HermanandMack, 1975;Herman and Polivy, 1975;Polivy,HermanandWarsh,
1978).The firstmeasurewasdesigned toassess theextent towhich chronic
dieters,whether normal-weight orobese,tried tosuppress their bodyweight
below "setpoint", andwasmadeupof 5items.The secondmeasurewas designed
".... toassess theextent towhich individuals exhibit behavioural and attitu-
dinal concernabout dieting andkeeping theirweight down" (Hermanand Polivy,
1975, p.668).No referencewasmade tothe "setpoint", and itconsisted of 11
items, the 5original items,plus6additional itemsconcernedwithweight
fluctuation.The rationale for including itemsonweight fluctuationwas that
individualswhoare concerned about limiting their food intake,are vulnerable
toboutsofovereatingwhen their cognitive restraint isbroken,and, there-
fore,demonstrate ahistoryof considerableweight fluctuation.The third and
finalmeasurediffered fromthe secondmeasureonly inthedeletionof one
item, and ingiving all items,which inearlierversionshad beensetup to
allowa free response,aclosed response format.
Therefore,both the 11-and the10-itemversions of themeasure of
restraint, the so-called Restraint Scale, includedquestionsdealingwith fluc-
tuations inbodyweight and conscious concernwithdieting.Rodin (1981 )was
the first tocriticise themulti-factorialityof the scale.Shepointed out
that the inclusion of itemsonweight fluctuation inthe scalemay result in
individualshaving ahigh scoreontheRestraint Scale,simply on thebasis of
largeweight fluctuations inthepast,without currentlywatching their weight
or consciously restraining their eating.Thismaybe especially true for the
obese,as their bodyweight tends to fluctuatemore than thatof normal-weight
individuals,due tochanges inactivity level or spontaneousdiuresis (Bray,
1976).Thus ".... a 5%weight fluctuationina 300lbsubjectwould naturally
lead toahigher "restraint" score thanacomparable fluctuation ina 110 lb
college student" (Drewnowski,RiskeyandDesor, 1982,p.274).Drewnowski et
al. (1982)conducted a study todetermine, firstly,whether thetotal scoreson
theRestraint Scalewere anaccurate reflection of concernwithdieting, and
secondly,whether thedegree of concernwithdieting changedwith self-reported
overweight.A principal component analysis (oblique rotation) revealed two
underlying factors,which reflectedweight history and fluctuations inbody
 -29-

weight, and actual concernwithdieting. Inaddition, itwas found that

overweight subjectshadhigherweight history scores thannormal-weight
subjects,while having lower scoresontheDietaryConcern factor.The authors
concluded that ". at least some self-reported overweight subjectswho score
highon theRestraint Scaledo sonotbecause theyareexcessively concerned
with dieting, butbecause theyareoverweight tobeginwith" (I.e.p.278).
Herman and Polivy (1982)responded tothiscriticism bypointing cothe
fact that theWeight History factor ofDrewnowskietal. (1982)included the
items "howoftenareyoudieting?"and "doyoueat sensibly infrontof others
and splurge alone?".According toHerman and Polivy, these two items pinpoint
the sort ofundereating/overeating that characterises theoverweight, which,
according tothem, accounts forweight fluctuation.
Lowe (1984)replicated the studyofDrewnowski etal. (1982)this time
usingapopulationwhichmore closely resembled theoneused inprevious
research on restraint.Weight suppressionwasusedasa second validational
criterion of restraint scores.Factor analysis revealed a factor structure
composed of three factors.The first two factors corresponded with those
obtained byDrewnowski et al.,but theweight fluctuation factor didnot
include the items "howoftenareyoudieting?"and "doyoueat sensibly in
front of othersand splurge alone".Instead, the firstof these two items
loaded onto theDietary Concern factor,andthe second item loaded, together
with the item "doyougive toomuch timeand thought tofood"ona separate
third factor.This finding contrastswith that obtained byDrewnowski et al.
(1982),andweakens Hermanand Polivy'sargument (1982)that the "dieter
instigated behaviour"of "undereating/overeating" isinextricably linked with
weight fluctuation.Ofmore significance, however,wasLowe'sobservation that
Dietary Concernwasmore strongly related tooverweight thanweight fluc-
tuation,which contradicted theobservationsmadebyDrewnowski et al.More-
over,both factors appeared tobe correlatedwithweight suppression.
Blanchard and Frost (1983)alsoconducted factor analyses (Varimax rota-
tion)of the 10-itemRestraint Scale.Intheir twostudies,they identified a
similar structure composed of twodistinct factors.The first factorwasmade
up of 6 items, representing concernwithdieting,while the second factorwas
composed of the remaining items,all ofwhichdealtwithweight fluctuation. In
a similarway toLowe,Blanchard and Frost'sanalysis of theweight fluctuation
factor didnot include the items "howoftenareyoudieting?"and "doyou eat
sensibly in frontof othersand splurge alone?".Incontrast toLowe (though
similarly toDrewnowski et al.), BlanchardandFrost found thatweight fluctua-
 -30-

tionwasmore closely related tooverweight thanwas concernwithdieting.They

also found thatConcern forDietingwaspositively correlatedwith Public
self-consciousness and SocialAnxiety,whileWeight Fluctuationwasnot.
According toBlanchard and Frost, thisindicated that " the two factors are
sufficiently distinct topreclude combination ina single scale" (I.e.259).
Ruderman (1983)andJohnson, LakeandMahan (1983)also found the10-item
Restraint Scale tobemulti-factorial.Ruderman obtained a two-dimensional
(Varimax)solution for thenormal-weight subjects,and afour-dimensional
solution for theobese.Thetwo-dimensional solutionwas similar tothe one
obtained byBlanchard and Frost (1983).The first factor inthe four-dimensio-
nal solution reflectsweight fluctuation, the second factor,bingeing, the
third tendency todiet,and the fourthoverconcernwithdieting.Johnson et al.
(1983)found similar structures composed of three factors (Varimax rotation)
within theRestraint Scale ingroupsofobesedieters,obese non-dieters and
normal-weight subjects.The first factorwasheterogeneous incontent,while
the second and the third factors referred respectively topreoccupation with
eating andweight fluctuation.Bothgroupsof investigators found, inaddition,
that thepatternof item-restcorrelation coefficientsdiffered according to
weight category, and that the relationshipbetween restraint scoresand
measuresofdefensiveness or socialdesirabilitydiffered amongst thevarious
groupsof subjects.This indicated that the reliability and construct validity
of the scalediffers forobeseandnormal-weightpopulations.
Thus it seems that theRestraint Scale ismulti-dimensional,and measures
different constructs inobese andnormal-weight populations. Inaddition, there
are indications thatnumerically equivalent scores represent less restraint in
obese subjects than innormal-weight individuals, though findingson this
subject are contradictory.
TheRestraint Scaleasameasureof cognitive restraint. The Restraint
Scalemaynot onlybean invalidmeasureof cognitive restraint inobese
individuals: it ispossible that theRestraint Scale isalso an invalid measure
of cognitive restriction of food intake inpopulations of normal-weight
subjects.Thispoint requires someexpansion.
Wardle (1980)foundno relationshipbetween the11-itemversionof the
Restraint Scale and the intakeofenergy inapopulation of normal-weight
subjects.BothWardle (1980), andHawkins IIand Clement (1980)found, in
addition, a strong relationshipbetween theRestraint Scaleandmeasures of
bingeeating.Theseobservations suggest that,while theRestraint Scaledoes
measure thedisinhibition of restraint, itdoesnotmeasure thedegree of
 -31-

restciction of food intake.Itcanbeargued that thisobservation is inline

withthecontention ofHerman,Polivyandtheir co-workers, that individuals
whoare concernedwith limiting their food intake,arevulnerable tobouts of
overeating,when their cognitive restraint isbroken. Following this lineof
thought, itwould be logical tosuggest that theRestraint Scaledoesnot
predict the restriction of food intake,but rather the inability tomaintain
cognitive control over food intake,that is, counterregulation.
More recently,however, ithasbeenshown that eatingbehaviour isnot
necessarily the consequence ofdietingper se (Stunkard andNessick, 1985).
These findingswere obtainedusing theStunkard andnessickquestionnaire, the
Three Factor EatingQuestionnaire (TFEQ)which assesses restrained eating. The
TFEQ separatelymeasures cognitive restraint of food intake (FactorI),
disinhibition of cognitive restraint (Factor II)andhunger (Factor III).The
resultsof a laboratory test on food intake (Shrager,Wadden,Niller, Stunkard
and Stellar, 1983), showed that Factor II,rather than Factor I,was closely
associated withovereating. The importance ofthese findings isdemonstrated by
the fact that the correlationbetween theHermanand PolivyRestraint Scale and
the restraint subscale of scoresontheTFEQ (Factor I)was statistically
insignificant (r-.168), while thecorrelation between scoreson this scale and
thedisinhibition subscale (Factor II)of theTFEQwashighly significant
(r-.840) (StunkardandMessick, 1985 ;Weissenburger, Rush,Gilesand Stunkard,
submitted).From this itcanbe suggested that thecounterregulatory eating
behaviour ofHerman and Polivy's "restrained eaters"is related to
disinhibition of restraint rather than to restraint per se,and, also,that the
Restraint Scale isnotameasureof restriction of food intake,but of
disinhibition of cognitive restraint.

SUMMARY ANDCONCLUSION

The theory of restrained eating attributes overeating todieting. This

paradoxical notion isbased on the contention thateach individual hashis own
rangeofbodyweightwhich ishomoeostatically regulated.Attempts to lower
bodyweight bya conscious restriction of food intakemay result inpersistent
hunger.When self-control processes areundermined bydisinhibitorssuchas
alcohol,anxiety,depression or the consumption ofenergydense food, excessive
food intake or counterregulation mayoccur.The stressofdietingmay also
result inanelevated level ofemotionality ordistractability. It isassumed
thatwhile themajorityofobese individualsare chronicdieters, this isnot
 -32-

the case fornormal-weight individuals,and that thisfact explains

obese/normaldifferences ineating behaviour.
Herman, Polivyand their co-workers found support for the theoryof
restrained eating, by showing that individualswho restrain their food intake -
inthe samewayasobese individuals presumablydo-behave like theobese.
However, these results couldnotbe replicated instudiesof restrained obese
individuals.Counterregulation doesnot seem tobea typeofeating behaviour
typical totheobese.Thismeans that themainpremise ofthe theoryof
restrained eating, that theeatingpatterns ofobese individuals are caused by
a high level of restraint, hasnotbeen corroborated.
This result couldwell reflectdifficulties inthemeasurement of cognitive
restraint, that is,intheRestraint Scale.TheRestraint Scalewas found to
contain at least twofactors, fluctuations inbodyweight andactual concern
with dieting.Thereare indications thatwhile overweight subjects achieve
higher scoreswith respect toweight history, theyscore lower onconcernwith
dieting than individuals ofnormal bodyweight.Thismeans that numerically
equivalent scores represent less restraint inobese individuals than in
normal-weight individuals.However, findingson thisare contradictory. The
Restraint Scalewasalso found tomeasuredifferentconstructs inobese and
normal-weight populations.Finally, it ispossible that theRestraint Scale
doesnotmeasure cognitive restriction of food intake,but rather the
disinhibition of cognitive restraint.Thismeans that counterregulatory eating
behaviour isnot caused bydieting per se,butbya tendency towards the
disinhibition ofdieting.
Thus,problems linkedwith theuseoftheRestraint Scalemayexplain the
fact that restraint scoresdidnotpredict thebehaviour ofobese individuals.
However, thepossibility that thesenegative findings reflect problems inthe
theory of restrained eating cannotbe ruled out,anda fair testof the theory
calls for anew instrument tomeasure restrained eating.

Postscript:As apostscript to thissection theboundarymodel proposed by

Herman and Polivy for the regulation ofeatingwill bedescribed. In 1984
Herman and Polivy (1984)proposed anewmodel,which attempted tobetter
explain the couterregulatory eatingbehaviour ofdieters- the "boundarymodel"
for the regulation ofeating.According tothismodel, theconsumption of food
is regulated by twoboundaries, corresponding tohunger and satiety.The
assumption that consumption is regulated byboundaries (rather thana point)
gives risetoa "zoneof indifference", that is,a zonewhere food intake is
 -33-

notconstrainedbyphysiological factorsofhungeror satiety,butwhere

non-physiologicalagents,suchascognitive,socialandotherpsychological
factors,are inforce.Dietersareassumed tohavea lowerhungerboundaryand
ahigher satietyboundarythannon-dieters,forthereasonthatonnormal
occasionstheyeatlessthannon-dieters,butonotheroccasions considerably
more. Inaddition,dietersareassumed tohaveadiet-boundary,which is
entirelycognitiveincharacter,andwhichfallswell shortofthe satiety
boundary.Innon-dietersthesatietyboundaryconstitutestheupperboundary
forfoodintake.Indietersthediet-boundaryprovidestheupperboundaryfor
theregulationoffood intake,solongasithasnotbeenbreached.Once ithas
beensurpassed,thatis,whenthedieter thinksthatishasbeentransgressed,
itnolongerhasa regulatory function,andthesatietyboundarybecomesthe
boundaryofreference.
Thus,usingthismodel,thecounterregulatoryeatingbehaviourofdieters
followinganenergydensepreload isnotunregulated. Instead,itis regulated
onadifferentbasis,bythesatietyboundary ratherthanthedietboundary.
Severehungerandsatiety remainasthedeterminantsofeatingbehaviour.
HermanandPolivynotethatinthis respectdietersdiffer frompatientswith
eatingdisorders,suchasanorecticsandbulimics,asthese individuals
regularlytransgressthehungerorsatietyboundary.

CHAPTER6

THETHREEPSYCHOLOGICALTHEORIESOFOVEREATINGANDWEIGHTMANAGEMENT:
THERAPY

All threepsychological theoriesattributeobesitytoovereating.However,they

differ intheirassumptionsastopreciselywhy individualsovereat,and this,
ofcourse,hasimplicationsfortherapy.According topsychosomatic theory,
whichassumesthatovereating resultsfromaconfusionofphysiological
symptomsofhungerandsatiety,andtheaccompanimentofemotional stressby
internalarousal states,overweightcanneverbetreatedper se.Instead,
therapyfocusesonpsychicconflicts,ratherthanoneatingbehaviouror
weight,andwhileweightlossmayoccurduringthecourseoftreatment,itis
notthefocusofthepsychoanalyticprocess (Rand,1982).Theusefulnessof
psychotherapyandpsychoanalysisinthetreatmentofobesityhasforsomeyears
beenseriouslyquestionned.However,theresultsofa recent large-scale
 -34-

assessraentof obesepatientsundergoing psychoanalysis indicatednot only an

improvement inpsychological well-being, butalso considerableweight loss,and
a satisfactory maintenance of this loss (Rand andStunkard,1978).While it is
not clearwhatmechanism inpsychotherapy causesweight loss,itseems,
according toStunkard (1980), likely that ifpsychotherapy helps individuals to
lead less stressful andmore contended lives,that theyhave lessoccasion to
overeat.Patientsmayeven looseweight, and thisweight losscouldbe perma-
nent.Yet,psychoanalysis isanexpensive formof treatment toachieve weight
reduction,andpsychotherapy maybewarranted only inthecaseof individuals
whohave repeatedly failed tolooseweightwith simpler and lesscostly methods
(Stunkard, 1980).
Externality theory, attributes overeating toahypersensitivity to external
food cues,theoutcome ofan "external cognitive style",and treating external
responsiveness constitutes themainmethod ofdealingwith overweight. Patients
areprovidedwith techniques forcontrolling their foodenvironment and their
food-related thoughtsby stimulus control procedures (suchaseating inonly
one place, separating eating fromother activities,pre-planning, and
minimising food cues). Itdoesnot seem that theuseof stimulus control
procedures as single treatment modality iseffective inproducing significant
weight loss (Loro,Fisher and Levenkron, 1979).However, it isusually combined
with otherbehaviour modification procedures (suchasself-monitoring of food
intake, calories,weight andphysical activity, slowing the rateof eating, and
increasing physical activity) (see forexample Stuart andDavis,1972). The
data indicates that thesemulti-faceted behavioural treatment programmes are
effective inproducing significantweight reduction ona short-termbasis,and
a satisfactory maintenance ofweight lossduring the firstyear following
treatment. There are indications,however, thatbeyond oneyear the maintenance
ofweight loss is far less satisfactory (Stunkard andPennick, 1979).
Slochower (1983a)suggested that thepsychosomatic theoryand the
externality theorymaybe related, inthata stateofhigh, uncontrollable
anxietymayenhance theoverweight person's reactivity toexternal cues.The
results of the (previouslymentioned) studyofMcKenna (1972)showed an
interaction between levelsofanxietyand the salience of food- obese
individuals ate somewhatmore foodwhen ittasted good andwhen theywere
anxious.Two studiesundertaken bySlochowerand Kaplan (1983)and Slochower
(1983 )produced similar results,obese individualsbeing found tobe
responsive tolevelsof food salienceonlywhen theywere anxious. Similarly,
obese individuals were found toovereatwhenanxiousonlywhenhighly salient
 -35-

foodwasavailable.Slochower concluded that ".... bothhigh uncontrollable

anxiety andprominent foodmust bepresent for theobese person toovereat"
(Slochower, 1983 a ,p.79).
These resultsindicate thatbothpsychotherapy andbehaviour modification
treatment programmes maybewarranted inorder toproduceweight loss.However,
Slochower hastens topoint at the importance of treating emotional problems,
forthe reasonthat it isdifficult toalter the factthatprominent food cues
areusuallypresent inour food-abundantWesternworld.Moreover, stimulus
control proceduresmay lose importance forapersonwhenhe isanxiousor
emotionallyupset, and thismayexplain the less satisfactory long-term results
of behaviour modification programmes.
The theoryof restrained eating seesovereating asoneof the side-effects
ofdieting.Teaching individuals toaccept their own "naturalweight" and to
return to "naturaleating", bymeansof the "naturalweightundiet" procedure,
constitutes themajor approach inthe treatment ofovereating (Polivyand
Herman, 1983). PolivyandHerman contend that those individualswhohave always
beenoverweight, the 30- 40%so-called "juvenile onset obese",dobetter to
accept their bodyweight, instead of trying tolooseweight by "chronic
semi-starvation". Incontrast, individualswhohavebecome overweight later in
life throughovereating or following a sedentary lifestyle, the60-70%
so-called "adultonset obese",mayexpect tolooseweight bymeansof the
"naturalweightundiet"procedure.
The "naturalweightundiet"procedure aimstoteachindividuals tobecome
more responsive topressures ofhunger and satiety.Their boundariesof hunger
and satiety,which hadbeenpushed apart because of continuous attempts todefy
thesephysiological controls,maybythismethod, drawmore closely together
again, recreating the situation that existed before theperson started todiet.
Individualsusing thisapproachare allowed toeatasmuch ofwhatever food
theywant inorder tobecome satieted.Toprevent counterregulation, there
shouldbeno forbidden food,and small portions ofenergydense food should be
incorporated inweeklydiets.Individuals areadvised tothink intermsofa
week rather thanaday.Asyet,however,no research data isavailable on the
resultsof this approach.Therefore, itseffectiveness remains tobe assessed.
 -36-

CHAPTER7

SUMMARY ANDOUTLINEOFTHE RESEARCH

There are threepsychological theories ofovereating andweight gain.The psy-

chosomatic theory focuseson internalemotional factors,and attributes
overeating toa confusiondue tothe incapability todistinguishbetween the
internal arousal statesaccompanying hunger and satiety, and the physiological
symptoms ofhunger and satiety.Theexternality theory focusesonexternal food
cues,and attributes overeating toahyperresponsiveness tofood-related cues
in theenvironment, combined with anunresponsiveness to internal cuesof
hunger and satiety.The theoryof restrained eating attributesovereating to
the conscious restriction of food intake.Theessential factor distinguishing
thepsychosomatic theory from theexternality theory istheir differing
emphasis on the impact of internal instigation (psychosomatic theory)and
external instigation (externality theory)oneating.What essentially
distinguishes the theoryof restrained eating frombothpsychosomatic and
externality theory isthe contention thatdietingmay cause (theoryof
restrained eating) rather than result from (psychosomatic and externality
theory)overeating andweight gain.
The theories,therefore,differ intheir assumption astowhy individuals
overeat, and this,naturally has implications for therapyandweight
management. Thepsychosomatic therapy focusesonpsychic conflicts and the
solution ofemotional problems.Therapybased on theexternality theory
concentrates on stimulus controlprocedures,while therapybased on the theory
of restrained eating focuseson "naturalweightundiet"procedures.
All three theorieshavebeenlaboratory tested, but thehypotheses derived
fromeach theoryhave notalwaysbeen confirmed by the tests.The psychosomatic
hypothesiswas supported bymost clinical studies,inthe sense that the
majority ofobese individualswere found toovereatwhen feeling emotionally
upset, but these findingshavebeendifficult to replicate in controlled
laboratory studies.Only recently, following the introduction of the concepts
of clearly labelled versusdiffuse emotions,was italsopossible to
demonstrate incontrolled studies thatoverweight individuals overeat in
response tocertain typesof stressors,and that thisovereating hasan
anxiety-reducing function.
Similarly, the internal/external distinction of theexternality theorywas
found tobe toosimple anexplanation ofobese/normal differences ineating
 -37-

behaviour,as itwas found that ineveryweight category there are individuals

whoarehyperresponsive toexternal stimuli orunresponsive to internal
stimuli.Thediscovery ofnormal-weight individualswith "obese"eating
patterns,and theobservation that an individual's final level ofbody weight
is influenced byphysiological variables,made itclear that the relationship
between external responsiveness and food intakeandbodyweight ismediated by
cognitive andphysiological variables.
The theoryof restrained eatingwas supported by studiesof restrained
non-obese individuals,butnotbystudiesof restrained obese individuals.
There are indications that thisinconsistency infindingsmayhave resulted
fromdifficulties inthemeasurement ofcognitive restraint, that is,inthe
Restraint Scale.However, thenotion that thesenegative findings reflect
problems inthe theoryof restrained eating cannotbe ruled out.
All three theories, therefore,havebeenpartiallyvalidated, but innoway
permit definite conclusions.Thismakes itdifficult todetermine on the basis
of research of literaturewhich theory isthemostvalid, and further to
specify thebest procedure for thetreatment ofovereating and overweight.

OUTLINEOF THERESEARCH

Theprincipal aimoftheempirical researchwas toexaminewhich theorymay

prove tobe themostvalid, psychosomatic theory,externality theory or the
theoryof restrained eating. Itwas the intention todetermine thisbymeansof
psychometric studiesof the relationships between scalesofemotional eating,
external eating and restrained eatingon theonehand, andvariables such as
measures ofbody fatness (note 4)andpersonality scaleson theother hand. It
was reasoned thaton thebasisofthe ideasexpressed inpsychosomatic theory,
externality theoryand the theoryof restrained eating, different findingswere
tobeexpectedwith regard tothe relationships between thethree typesof
eatingbehaviour and/or personality traits.Thus,fromthedegree towhich
observed findingsare similar toanticipated findingswithin one single
theoretical viewpoint, itwould bepossible todrawtentative conclusions
regarding thevalidity ofeach specific theoretical point ofview. Inthisway,
insight would begained into thepotential causesofovereating and into
strategies for thetreatment of overeating.
A further aimofthe researchwas todetermine,whichvariables are
associatedwitheachparticular typeofeatingbehaviour, andalso,which
variablesareassociatedwith successful weight loss.Not onlywould this
 -38-

provide further insight intothepotential causesofovereating and successful

weightmanagement,but thiswould alsogivean indicationastowhich
categoriesof individuals shouldbe given specialattention inthe treatment
andprevention of overweight.
Asno satisfactory scaleswere availablewhen theprojectwas initiated,
the first taskwas toconstruct unidimensional scalesapplicable tothe three
typesofeatingbehaviour.The construction of theDutch Eating Behaviour
Questionnaire (DEBQ),with scalesofemotional eating, external eating and
restrained eating, isdiscussed inChapter 8.Thepredictive validityof the
restrained eating scaleof theDEBQ,asameasure of restriction of food
intake, isstudied ina sample ofwomen inChapter 9.
Chapter 10constitutes themain coreof the research.Thevalidity of the
three psychological theoriesof overeating, asapplied toa sample ofwomen,
were examined. This chapter alsoattempted toassesswhichpersonality traits
are related to successfulweight loss.Thiswasdetermined by inspecting the
relationships between scalesof the threetypesofeatingbehaviour, ameasure
of body fatness,andvariouspersonality scales,and, inaddition, by exploring
thedifference between "obese"and "latent obese"women,with respect toeating
behaviour and personality,unfortunately, the final scalesof eating behaviour
as constructed inChapter 8werenotyetavailablewhen this studywasunder-
taken, soonlypreliminary scalescouldbeused tomeasure thetypesof eating
behaviour (note5 ) .
Chapter 11describes the relationships between the threeeating behaviour
scales,and scales for sex-roleorientation andpsychological adjustment as
based ona studyofwomen. InChapter 12,themodifying effect of emotional
eating onthe relationship between stressand change inbodyweight, based ona
studyofmenandwomen,was described.
Chapter 13 isofanother nature.Itiscomposed ofa methodological
treatisewhichexplores thepossibility that longitudinal studiesaremore
susceptible tothe generation of response sets than cross-sectional ones.This
treatisewas inspired by theobservation that resultsobtained forwomen onone
assessment date ofa longitudinal studyof themunicipality ofEdewere found
todiffer significantly from thoseofa cross-sectional study. It isof
relevance to thepresent research tonote that subjects for the Final study in
Chapter 8,and for the studies reported inChapter 11and 12,arederived from
this longitudinal study.While it isnot clear totheauthorwhat effects these
réponsesetsmayhave hadupon the observed results, itcannot be ruled out
 -39-

that in these studies inparticular the resultsobtained forwomenmayhave

beenconfounded by responsesets.
Host chapters ofPartTwoof this thesis consist oforiginal papers.Some
chapters show slight overlap,but thiswas inevitable, inviewof the fact that
theprojectdealtwitha continuous seriesof seperate studies.

NOTES

1. Itseems remarkable that Schachter -advocate of the cognitive-physiological

theoryofemotional arousal -didnotgivemore emphasis to the roleof
cognition inhistheoryofovereating.Schachter'semphasis on external
stimuli mustbeexplained fromthe fact thathe foundnosupport for the
psychosomatic hypothesis,asthisalsomeant thatnoevidencewas obtained
for the "... labellingviewofmatters" (Schachter etal., 1968,p.97)with
regard totheaetiology of overweight.
2.Rotter's concept of locusof control canalsobe classifiedwithin social
learning theories, for reason that individual differences in expectancies
for internal versusexternal locusofcontrol of reinforcement are
considered tobedue todifferences in socialisation.
3.Itshouldbenoted thatHerman and Polivy (1980 )maintain that this
interpretation accounts for theanxiety-induced increase ineating found
also inobese subjects.However, theydonotbelieve, incontrast tothe
psychosomatic theorists,that this increase ineatingactsasa sourceof
emotional comfort or relief. "Overeating under stress isbetter regarded as
"released" rather than reinforced, aperspective thatdoesnot require that
anxiety-induced eatingbe relaxing" (Herman.andPolivy, 1980 ,p.219).
4.That is,thebodymass index (weight/height (kg/m).
5.These scaleshadbeenconstructed on thebasisof the factor structure of
thepreliminary itempool asused in "Preliminary StudyOne" (seeChapter8 ) .
 -41-

PARTII. PSYCHOMETRIC STUDIES

 -43-

CHAPTER8

THEDUTCH EATINGBEHAVIOURQUESTIONNAIRE (DEBQ)FORASSESSMENTOFRESTRAINED,

EMOTIONALANDEXTERNAL EATING BEHAVIOUR

TatjanavanStrienwithJanE.R.Frijters,GerardP.A. BergersandPeterB.
Defares*.

According tobothpsychosomatic theoryand externalitytheorythedevelopment

andmaintenanceofhumanobesity isattributed toovereating. Psychosomatic
theory (e.g.,KaplanandKaplan,1957)focuseson thephenomenonofemotional
eating.While anormal responsetoarousalstates,suchas,anger, fearor
anxiety, islossofappetite (Carlson,1916;Cannon, 1915), some individuals
respondbyexcessiveeating.Bruch (1961,1964), themainadvocateof this
theory,attributesexcessiveeatingtoconfusionbetween internalarousal
statesandhunger,probablybecauseofearlylearningexperiences.Externality
theoryconcentrateson thephenomenonof "externaleating"which iseating in
response tofood-related stimuli regardlessofthe internal stateofhunger or
satiety (e.g.,Schachter,GoldmanandGordon, 1968).SchachterandRodin (1974)
viewexternally regulatedeatingbehaviourasapersonalitydisposition.
Inbothpsychosomaticandexternalitytheoryan individual'smisperception
ofhisinternal stateprior toeating isconsidered tobeacausal factor in
thedevelopmentofobesity (Robbinsand Fray, 1980). But,ahighdegreeofex-
ternalityora strongtendencytoemotionaleatingdoesnotnecessarily lead to
overweight.Individualsmay react tobeingoverweightbyconsciously restricting
food intake irrespectiveofwhether theyareemotionalorexternal eaters
(Rodin, 1975;RodinandSlochower,1976;Rodin, 1978).Dieting, canbeused to
correctbodyweight, thereforeahighdegreeofemotionalandexternal eating,
canbe found inallweight categories,andthusnotonly inthe"obese".
However,according toHerman,Polivyandcoworkers (e.g.,Hermanand
Polivy, 1980 a )eachindividualhashisown rangeofbodyweight,whichis
homeostatically regulated (PolivyandHerman, 1983).

* Thischapter isinpressinthe InternationalJournalofEatingDisordersand

isprinted herewithkindpermissionofJohnWiley&Sons,inc.,copyright.
 -44-

Some individualswithaweight rangeatahighlevel (ahigh "naturalweight"

range according toPolivyandHerman,1983)areunder strong socialpressure to
loseweight and thereforemaydiet,but intensedieting inthese individuals
may result inpersistent hunger.When self-control processeswhich monitor
dieting behaviour areundermined by, forexample, consumption ofalcohol or
high caloric food, or anxietyordepression, "counterregulation" may occur.
This isthebreakdown of restrictive control sothat suppressed eating
behaviour isdisinhibited,and excessive food intakeoccurs.A further
consequence of thecontinuous struggle against hunger sensationsmayalso be
lossof contactwith internal feelingsofhunger and satiety.Thus in this
theoryof restrained eating,both "external"and "emotional eating" are
considered tobe consequences of intensedieting (HermanandMack, 1975;Herman
and Polivy, 1975;PolivyandHerman, 1976 a , 1976 b , c ).
All three theories,externality theory, psychosomatic theoryand restraint
eating theory,havebeen tested inlaboratory studies,but the hypotheses
derived fromeachhavenotalwaysbeenconfirmed. Themainproblem in testing
thepsychosomatic theoryand theexternality theory isthe classification of
individuals inweight groupsonthebasisofmeasured bodyweight.As body
weight canbe controlled tosomeextent bydieting, measured bodyweight does
not necessarily represent thebodyweight ofan individual onthebasisof his
emotional or external tendency. Inanendeavour toovercome this problem,
Pudel,MetzdorffandOetting (1975)constructed the Fragenbogen für Latente
Adipositas (FLA)toidentify those individuals having obese eatingpatterns but
being ofnormal bodyweightdue toconscious restriction of food intake.This
group referred toasthe "latent obese",were individualsofnormalweight who
had similar satiety curvesasobese individuals for theconsumption of liquid
formula diets inthe laboratory (Meyerand Pudel,1977;Pudel, 1978). Although
thisquestionnaire represents amajor improvement, inspection of it reveals
that it isaheterogeneous setof itemscontaining questions on, for example,
weight history,weight consciousness,dieting, hunger and satiety, and external
eating.Principal componentanalysis (Varimax)of the FLA ina sample of 108
subjectshas revealed noless than ten factorswhenusing the Eigenvalue one
criterion for factor extraction (note 1*).Even though this multifactorial
questionnaire hasproved tobe suitable for identification of the "latent
obese", itprecludes examination of the relative importance ofeach factor
associated with overweight.

*seepage 61 for notes

 -45-

The theoryof restrained eatinghasalsobeenconfrontedwith conflicting

results.Findings for restrained non-obese subjectswere consistent, inthat
after apreload, non-restrained eaters consumed lessand restrained eaters
consumedmore.But these resultscouldnotbe replicated instudieson
restrained obese subjects (e.g.,RudermanandWilson, 1979;Rudermanand
Christensen, 1983). This inconsistency infindingsmay result from the
Restraint Scale notbeing avalid operationalization of the concept of
restrained eating.As the FLA, theRestraint ScaleusedbyHerman,Polivyand
co-workershasalsobeen found tobemultifactorial. Inaddition tothe factor
reflecting concern aboutdieting, another factor reflecting fluctuations in
bodyweight hasbeendistinguished (e.g.,Drewnowski,RiskeyandDesor,1982;
Ruderman, 1983;Blanchard and Frost, 1983;Johnson, LakeandHahan, 1983). This
multifactoriality may lead toerroneous classification of theobese especially.
Evenwhennot restricting food intake,obese individualsmayobtain higher
scoreson theRestraint Scale,because theyhavegreater fluctuation inbody
weight.Consequently, numerically equivalent scores can indicate less genuine
restraint inoverweight than innormalweight individuals and thismay explain
why resultsof restrained non-obese subjects couldnotbe replicated in
populations of restrained obese subjects.
It seems that the recentlypublished (1981)Three Factor Eating
Questionnaire developed byStunkard andHessick (Stunkard,1981)tomeasure
restrained eating represents amajor improvement ofbothHerman's Restraint
Scale and Pudel'sFIA.However, thisquestionnaire wasnotyetavailable when
thepresent studystarted (May, 1980).Thusat the timeofthepresent studyno
homogeneous scale on restrained eatingwasavailable totest these theories.
NOtonlya scaleof restrained eatingbutalso scales foremotional eating
andexternal eatingare required to improveunderstanding ofobese eating
patterns.Therefore themainpurpose of thisstudywas todevelopa
questionnaire containing three scales,thequestionnaire developed is referred
toas theDutchEating Behaviour Questionnaire (DEBQ).

PRELIMINARY STUDIES (note 2)

Aim and Method
Themain aimof thepreliminary studieswastoselect items for scaleson
restrained, emotional and external eating.Questions on the three typesof
eatingpatternswere administered topopulationsofnormalweight and
overweight subjects,datawere factoranalysed, andthe resulting factor
structurewasused asthebasis for revisionof thequestionnaire. This process
 -46-

wasthen repeatedand itemsachievingstabilitywere retained forthefinal

questionnaire.
StudyOne
The initial itempoolconsisted of100items,derived fromthreeexisting
questionnaires translated intoDutchbyFrijtersandRoosen (note 3 ) :the
EatingPatternsQuestionnaire (EPQ)(Wollersheim, 1970): TheFragenbogen für
LatenteAdipositas (FLA) (Pudel,MetzdorffandOetting, 1975); andtheEating
Behavior Inventory (EBI) (O'Neil,Currey,Hirsch,Malcolm, Sexauer,Riddleand
Taylor, 1979). The EPQwasselectedbecause,at firstsight, itcontains items
relatedtoemotionaleating;theFLA,because itcontains items related to
restrainedeatingandexternaleating;and theEBIbecause itcontainsa series
of statementsonbehaviour related totheassessment ofexternal stimulus
control. ItemsintheEPQandtheEBIhavea five-point response format (1-5),
andthose intheFLAatwopoint responseformat (YesandNo).
The itempoolwasadministered toatotalof 120subjects,40menand 80
women,all inhabitantsofWageningen.Theywere selectedbytheir general
practitioners sothatthesample containedbothnormalweightandoverweight
2
subjects.Themeanbodymassindex (BMI;weight/height )ofthemenwas 26.2
(s-5.4)andtheirmeanage 30.8years (s-5.2); andthemeanBMI ofthewomen
was25.2 (s-4.8), andtheirmeanage,31.1years (s-8.4).
After eliminating thoseitemsendorsed inonedirectionbymore than85%
of the sample,the remaining72itemswere factoranalysedusingVarimax
rotation.AminimumEigenvalueof1.0wasused forfactorextraction.A large
proportionofthevariance (44.7%)wasaccounted for inthefirstthree factors
thatemerged:the remaining20factorscontainedonlyafewitemseachandwere
uninterpretable.The first factorcontained itemsonemotionaleating;the
second, itemson restrainedeating;and thethird, itemsonexternal eatingand
perceivedhunger.

StudyTwo
The resultsobtained inStudyOnewereused to revise itemsandtodevelopnew
items.All itemsdealingwithperceived hungerandsatietyweredeletedbecause
theywereconsidered torepresent theinternalcomponent ofbothemotionaland
external eating.Sincethemainaimwastodevisedistinctive scalesfor
emotionalandexternaleating,onlythoseitems referringtotheexternal as
opposed totheinternalcomponent ofeatingbehaviourwere included inthe
external eatingscale.
 -47-

The itempoolused inStudyTwoconsistedof21 itemson restrained

eating, 15onemotionaleating,and15onexternal eating.All itemshad the
five-point response format;never (1);seldom (2);sometimes (3);often(4);
andveryoften (5).The 51 itemswereadministered toasecond sample
consisting oftwosubsamples.One subsamplewasdrawn randomly frommenand
womenagedbetween 34and 36yearsliving infourvillages inthemunicipality
ofEdeand isreferred toastheEdeSample.ThemeanBMIandmeanageofthe
161women inthissamplewere 22.9 (s-4.1)and 29.9years (s-4.7) respectively,
and themeanBMIandmeanageofthe103menparticipantswere 23.6 (s-2.8)and
26.8years (s-4.5)respectively.Participants forthesecond subsamplewere
obtainedwith thecooporationofdieticianswhohad invited their clientsto
complete thequestionnaireon their firstvisit.Thissubsample referred toas
DieticianSampleconsistedof75womenwithameanBMIof 32.8 (s-6.2)and a
meanageof 31.1years (s-8.6)andof19menwithameanBMIof 31.1 (s-2.9),
andameanageof 31 (s-8.3).
Anorthogonal (Varimax)factoranalysiswascarriedout for subjectsof
theEdeSample.TheEigenvalueone criterion revealedanorthogonal structure
of ten factors.Factorone contained itemson restrained eating, factors two
and three,itemsonemotional eating;and factors fourand five,itemson
external eating.Together these five factorsexplained 88%ofthe total
variance.Eachof the51 itemswas then reviewed toascertain thedegree of
correspondencewith thefive factors.Twelve itemsweredeletedwhichhad
either alowfactor loading,orasdecidedbythe investigators,an item
contentwhichdidnot correspond closelywiththeparticular typeofeating
behaviour.The reduced itempoolwasagain subjected toanorthogonal (Varimax)
factoranalysis.A factoranalysiswasalsoundertaken forsubjectsofthe
DieticianSample.Inboth samplesthepatternof factor loadings seemed tobe
very similar for itemsonemotional and restrainedeating,butquitedifferent
for itemsonexternaleating.Also,inbothsamples, itemsonemotional eating
were representedbytwo factors,onecontaining those referring toclearly
labelled emotions (e.g.,eating in responsetoangeror irritation),and the
other items referring toeating in response todiffuseemotions (e.g.,eating
when feeling lonely, idleorbored (items1,2and 7)).Further inboth
samples, itemson restrained eating loadedhighonone factor,although item
loadingswere ingeneralhigher intheEdeSample than intheDietician Sample.
Itemsonexternal eatinghadhighloadingsonone factor intheDietician
Sample,butwere spreadover three factorsintheEdeSample.Evidently, eating
 -48-

inresponse toexternal foodcuesisconceivedasamorehomogeneoustypeof

eatingbehaviourbyobese subjectsthanbynormalweightsubjects.

FINAL STUDY
Aim
The FinalStudyaimed todevelopafinalitempoolandalsotoassessthe
dimensional stabilityofthisitempool insubsamples ofobeseandnon-obese
subjects,andmenandwomen,and thento replicate the factor structures
obtained inthepreliminary studies.Consistencyofthe factor structures in
thevarious studies,andalsostabilityofdimensionsinvarious subsamples
wouldwarrant theconstructionof scalesoneatingbehaviour,andalso
assessment of the reliability,means,standarddeviationsand intercorrelations
of thesescales.

SubjectsandProcedure
At the timeofthe6thof sevenassessmentsofalongitudinal studybeing
carried out inthemunicipality ofEde,the revisedquestionnaire fromthe
preliminaryStudyTwowasadministered to653womenand 517men.Detailsof the
studypopulation andprocedurehavealreadybeendescribed elsewhere (Baecke,
Burema,Frijters,Hautvast andVanderWiel-Wetzels,1983;VanStrien, in
press).

Itempool
On thebasisof the resultsofpreliminaryStudyTwoitemswere revised andnew
itemsdeveloped.Eight itemswereaddedtotheitemsonemotional eating,
because thelatter referred onlytohavingadesire toeatwhenexperiencinga
particular emotion,and itwasnotclearwhether theydistinguished between
desire for foodand theactofeating.
The itempoolofthefinal studyconsisted of48items.All itemshad the
same response formatnever (1),seldom (2),sometimes (3),often (4),andvery
often (5).Anot relevant response categorywasadded toall itemswhichwere
cast inaconditional format forexample:Doyouhaveadesire toeatwhenyou
feelbored or restless;whenyouhaveeaten toomuch,doyoueat lessthan
usual the followingday;andwhenyouhaveputonweight,doyoueat lessthan
usual.Thiswasdonebecause somesubjectsneverexperienceaparticular
emotion,never eat toomuch,orneverbecomeheavier.
 -49-

RESULTS
Developmentofthe final itempool
Except forthose itemsgivenanot relevant response,each itemwas subjected
toanorthogonal (Varimax)factoranalysisforallsubjects,thusintotal616
subjects.TheEigenvalueonecriterion revealedanorthogonal structureof
sevenfactors.As inStudyTwo,itemsonemotionaleatingwere representedby
twofactors,onecontaining items referringtoclearlylabelledemotions,and
theotherto items referring todiffuseemotions.Inaddition,those items
referringtotheactofeatingwere found tohavehigh loadingsonthesame
factorsasthoseitemsreferringtothedesiretoeat.Thus,itmaybe
tentatively concluded, that thetwodomainsofitems refer tothe sameconcept.
All itemson restrainedeating loadedonone factorandthreeofthese items
hadhigh loadingsonaseparate factor.Itemsonexternaleating loadedon two
factors.
Further inspectionofthefactor loadingsand theitemscontent,led to
sevenitemsbeingdeleted fromthe itempoolon restrainedeatingandexternal
eating.Alleight itemsonactualemotionaleatingweredeleted,because they
were foundtoloadonthesamefactorsastheoriginal itemsonemotional
eatingandthereforewereconsideredtorefer tothesametypeofeating
behaviour astheoriginal items, inaddition,when scalescoreswere obtained
the "actualemotionaleating scale"correlatedhighlywiththethreeother
scalesonemotionaleating inallpopulations (r-.80).

Factor structuresofsubsamplesandtotal sampleof thefinal itempool

All 33items inthe final itempoolwere subjected tovariousorthogonal
(Varimax)factoranalysesforsubsamplesfromthelongitudinalstudyas
described inthefinalstudy.TheEigenvalueonecriterion revealed five
factors forboth theobese (men:BMI >27;women:BMI >26;inanalysis:n-91),
andnon-obese subjects (men:BMI <27;women:BMI <26;intheanalysis:
n-566). Inboth subsamples,all itemson restrainedeatinghadhighloadingson
one factor.Aspreviously, itemsonemotionaleatinghadhighloadingsontwo
factors,one containing thefour itemsondiffuseemotions,and theother the
remaining itemsonclearlylabelledemotions.Itemsonexternaleatinghadhigh
loadingsontwofactors.
Three-andfour-factor solutionswerealsocarriedout for the subsamples
ofobeseandnon-obese subjects.The four-factor solutionappeared tobethe
best interprétablesolution inbothsubsamplesandwasverysimilar tothe five
factor "Eigenvalueone"solutionforitemson restrainedeatingand emotional
 -50-

eating,butmost itemsonexternaleating loaded onone insteadof twofactors.

At facevalue,the four-factor solution seemed tobe similar forbothobeseand
non-obese subjects.
Tosubstantiate this,thegoodnessof fitof resultsof the factor
analysis forbothsubsampleswasassessedbyanorthogonalcongruence rotation
towards thebest least squares fit (ROTA01;BorgersandRoskam, 1967;Roskam
andBorgers,1969). Inthefirstsolutionthecorrelationmatrixofobese
subjectswas rotated tothecriterion factorpatternofthenormalweight
subjects.This resultwaschecked inthesecond solution, inwhichthe
correlationmatrixof thenormalweight subjectswas rotated tothe criterion
factor patternof theobesesubjects.
Table 1showsthat thepercent lostvariance,the total stressand the
stressvaluesof theindividual factorsislowinboth solutions,which isan
indicationofaclosesimilarityof factor structuresofobeseandnormal
weight subjects.
Inorthogonal (varimax)four-factor solutionsobtained forthe subsamples
ofmen (inanalysis:n-237)andwomen (inanalysis:n-348), all itemson
restrained eatingandexternal eatingalsohadhighloadingsonone factor,and
itemsonemotionaleatinghad loadingsontwofactors.At facevalue,the four-
factor solutionwasverysimilar formenandwomen.Again, thegoodnessof fit
for the resultsof the factoranalysiswasassessedbyaleast squares
orthogonal rotation. Inthe first solution, thecorrelationmatrixof thewomen
was rotatedtothecriterion factorpatternofthemen.Inthesecond solution,
the correlationmatrixofthemenwas rotated tothecriterion factor pattern
ofthewomen.As canalsobe inferred fromTable1,thepercent lostvariance,
the total stressvalueand thestressvaluesoftheindividual factorswere in
both solutionsvery low,which isan indication foraclose similarity of
factor structuresofmenandwomen.
As thestructure forthesubsamples issimilar,the resultsof factor
analysisoftheVarimax four-factor solutionare reported for thesample ofall
subjectscombined (inanalysis,n=657).Table2showsthatall itemshave
loadingsabove .45ontheappropriate factor.

Constructionof thefinal scales

BothStudyTwoand the FinalStudyhave shown thatemotional eating comprised
twodimensions,onedealingwitheatingin response todiffuseemotionsand the
otherwitheating in response toclearlylabelledemotions.
 -51-

CO >\J2 00
O) tH 4J c
O O)
•H > O
U «H 4J
U 4-1
e#«
o. u
B co co
o o u
•H M O
4J M
CO fi
H u gj
M h
h «
U "O
o a >.
ü (AH
01
«I -U >
JS J5 -H
U 00 4-1
•H O
c ai oi
ai » a
x: a
> i-H ai
« u
co e
u E s
O O 01

> 4 J II
• H M »
•O O.
C "H
•H l-i tfl
O S co
1-4
CM 4-1 u • o
O O o
a e
m CM
ai ai
3 C J5
riOUi
co «H
!> H »M

Vb u o
1-1 u u u co <r
4-1 -HO) CM ^H
co ai M co o o
• Ö 4J CO
e a fr-
co o i-t
IJ e •
0) o >

ö
•-I
CO 01 01
s
S*. 4J e r-1 O
4J ^ S 5 m co
C 0) B
01 > 01
o o
o •H j : B •o
U 4J » CO
01 o
CX 0) "O 01
CX 0} co
« co B 0)
CO 01 T-t
CO 1-1 co
01 4J
«o
M C - C UJ
4J 01 O
co a
o
O 01
H M c
14
CO O 01
W O £ 4->
o e 4J 4J
4J CO CO
o a co B
CM CO CO 01 B O B
O H H ii O •H
CO CO & •H 4-1 •H o
co 3 o
4J o
1 4-1 3 4-1
01 * U T3 O B 3 rH
O s
01 *H
3
3 -H S CO O 1-4
i-H > CO B O CO O
CO - H
CM
co a
o CO
> t) ^ B •o 4-> •o » 4-1
i-C
B 6
•H O •H
O B
CO s CO
1-1
co 0 •o
B kl
01 01 4-1 01 0) u
•H u CO •H
i-H
.O
co 3 4-1
01 i-H • H
CO
01
b « G
b

CO Ä O H JO 01
H O co U O
a
 -52-

Table2VarimaxRotatedFactorMatrix (fourfactorsolution)onthe33itemsof
theDutchEatingBehaviourQuestionnaire forthesampleofall subjects
combined (inanalysisn-657)**.

R 1.Ifyouhaveputonweight,doyoueatlessthanyouusuallydo?*
2.Doyoutrytoeatlessatmealtimesthanyouwouldliketoeat?
3.Howoftendoyou refusefoodordrinkofferedbecauseyouare
concernedaboutyourweight?
4.Doyouwatchexactlywhatyoueat?
5.Doyoudeliberatelyeatfoodsthatare slimming?
6.Whenyouhaveeatentoomuch,doyoueatlessthanusualthefollowing
days?*
7.Doyoudeliberatelyeatlessinordernottobecomeheavier?
8.Howoftendoyoutrynottoeatbetweenmealsbecauseyouarewatching
yourweight?
9.Howoftenintheeveningdoyoutrynottoeatbecauseyouarewatching
yourweight?
10.Doyoutakeintoaccountyourweightwithwhatyoueat?
E 11.Doyouhave thedesire toeatwhenyouareirritated?*
12.Doyouhaveadesiretoeatwhenyouhavenothingtodo?*
13.Doyouhaveadesiretoeatwhenyouaredepressedordiscouraged?*
14.Doyouhaveadesiretoeatwhenyouarefeelinglonely?*
15.Doyouhaveadesiretoeatwhensomebodyletsyoudown?*
16.Doyouhaveadesiretoeatwhenyouarecross?*
17.Doyouhaveadesiretoeatwhenyouareapproaching something
unpleasant tohappen?*
18.Doyougetthedesiretoeatwhenyouareanxious,worriedortense?*
19.Doyouhaveadesiretoeatwhenthingsaregoingagainstyouorwhen
thingshavegonewrong?*
20.Doyouhaveadesiretoeatwhenyouarefrightened?*
21.Doyouhaveadesiretoeatwhenyouaredisappointed?*
22.Doyouhaveadesire toeatwhenyouareemotionallyupset?*
23.Doyouhaveadesiretoeatwhenyouareboredorrestless?*
Ext.24. Iffood tastesgoodtoyou,doyoueatmorethanusual?
25. Iffood smellsandlooksgood,doyoueatmorethanusual?
26. Ifyou seeorsmell somethingdelicious,doyouhaveadesiretoeat
it?
27. Ifyouhave somethingdelicioustoeat,doyoueatitstraightaway?
28. Ifyouwalkpast thebakerdoyouhavethedesiretobuysomething
delicious?
29. Ifyouwalkpastasnackbaroracafe,doyouhavethedesiretobuy
somethingdelicious?
30.Ifyouseeotherseating,doyoualsohavethedesiretoeat?
31.Canyou resisteatingdeliciousfoods?***
32.Doyoueatmore thanusual,whenyouseeotherseating?
33.Whenpreparingamealareyouinclinedtoeat something?

R -Restrained Eating
E -Emotional Eating
Ext-ExternalEating
* - itemswithanon relevant responsecategory inadditiontothecategories
never (1),seldom (2),sometimes (3),often (4),andveryoften(5).
** -Dutchversionof thescalesmaybeobtained fromthefirstauthor.
*** -Response tothisitemhastobe reversed.
 -53-

Factor1 Factor2 Factor3 Factor4

Eigenvalue 1Ô.Ô47 4.447 2.078 0.Û03
PCTofVar 59.7 24.5 11.4 4.4
CumPCT 59.7 84.1 95.6 100.0

R1 0.795 0.042 0.118 0.075

2 0.710 0.210 0.124 0.086
3 0.809 0.118 0.001 0.058

4 0.741 0.142 0.024 0.006

5 0.752 0.198 0.002 0.068
6 0.711 0.105 0.170 0.044

7 0.895 0.113 0.025 0.031

8 0.823 0.172 0.094 0.083

9 0.823 0.215 0.099 0.091

10 0.883 0.145 0.010 0.055

E1 0.233 0.628 0.172 0.275
2 0.140 0.332 0.385 0.559
3 0.252 0.711 0.185 0.220
4 0.151 0.467 0.255 0.588
5 0.217 0.682 0.118 0.340
6 0.161 0.736 0.166 0.174
7 0.090 0.775 0.218 0.103

8 0.121 0.791 0.149 0.050

9 0.146 0.821 0.197 0.003

10 0.132 0.736 0.147 0.026

11 0.210 0.800 0.142 0.142
12 0.095 0.677 0.239 0.127
13 0.142 0.563 0.241 0.482
Ext. 1 0.039 0.077 0.575 0.087
2 0.087 0.144 0.599 0.021
3 0.118 0.073 0.583 0.088

4 0.054 0.131 0.451 0.034

5 0.114 0.185 0.465 0.143

6 0.059 0.004 0.516 0.057

7 0.126 0.151 0.551 0.163

8 0.015 0.164 0.516 0.210
9 0.177 0.255 0.468 0.069
10 0.040 0.168 0.496 0.000
 -54-

Therefore separate scaleswere constructed foreachdimensioncontaining four

andnine itemsrespectivelyandalsoanadditional scalecombining thetwo
dimensionscontainingall13items.
The final scalesonRestrained eatingandExternaleatingcontained ten
itemseach.Scoresoneachofthe fivescales,were obtainedbydividing the
sumof items scoresby thetotalnumber of itemson thatscale.

Statisticsand subscale correlations

Means, standarddeviations (sd),standarderror (se)andmaximumandminimum
value foreach scalearepresented separately forthesamplesofall subjects
combined, the subsamples ofobeseandnon-obese subjects;menandwomen;obese
menandwomen;andnon-obesemenandwomen (Table3 ) .
Subscale correlations forthevarious subsamplesofsubjectsoftheFinal
Studyarepresented inTable 4.All correlationcoefficientswere significant,
onlythePearson correlationcoefficientassessingthe relationshipbetween
Restrained eatingand Externaleatingwasmuchlower.

Reliability
Table 5containsthe item-total correlation coefficients foreachofthe items
ofthe fivescales,inthesample ofall subjectscombined, andthe subsamples
ofobeseandnon-obesesubjects,andofmenandwomen.Cronbach'salpha
coefficients foreachof thesepopulations,andalsoforobesemenandwomen,
andnon-obesemenandwomen reflectedadequate internalconsistency inall
subsamples (Table 3). The item-total correlationcoefficients,werehigh inall
fivescalesand inall subsamples.Inaddition,theitemuniformitywasvery
high forallgroupsofsubjects.
 -55-

r*. r** oo oo ^ • ï N O N r^-i-^. oo es

u i -4- • * - * - * m sr • * * * ** j ~ t - t f * t f -tf • *
I I I I I. I I I I. I I I 1 I.

en es es en es
o o o o o

I
n tu
e s •-! <-i e s e s es es H es es eu a
.o
O o
O
k so o <r -sr es N N N O O h r^ • * I-* en es er» vo r-. «-* oo
eji es o> vo o
CT\ CJN CJv v D i - 1
• * O fO • * iO
vO vO vO • * sO s MOO N ( s - J
<fr en - * es - *
»o en \D cjN r ^
m m m en i n

r s N ao oo • ï m > o o ifi r^ e s s t m oo
m <r -a- <r <r es <t m - * >st en en < • s t **<t<t <* -tf
l l i i l I l 1l 1
9 I
I
1
CM - * <ï - * vO

en en en es
O O O o O
CO

f^ r* f » o en en o es i-i *o N O M n O O O
r^ m m m 0 0 %£> i n m N I N O O O0 vD
^ CO

m o o> M3 en C r-* i-s - * e j \ cr»

e n Cf« r*. •-H r^. c ct> e y * e n t—i v D ^
0>
CM i - l i - I CM CS es es .-i es es i-t i-H i-t e s e s es I-H r-i e s e s es es es es es

er» i-i rH en 03
so i n en en *o OJ <t i - i r-^ en oo cy> ^ en en
(-1 O i-t * o «-I X* r-i en co 0"\ * t ey» - * ej» e s I-H
i—i <-H i—i r— i—i o H H
>-t •—1 H
i—O i H«-<
1 CJ\ a c <r - * -d- en i n

co
FI m en a œ »
0) S FI ai a a ai a a
4J eu CU 4J a) a) 4J 0) at
4J •H y u
1 •H •H | TH 1 1 | *H «H
en 1 en i i
CT« <t <-( o\ - *
>*• v >
ex
•H g>g>g> s?
•H
J? S? S? S?, sas*.
u i-t
W
CA 4.J U 4J•rt
0) et) U W oö q) et) 4J W cd co ra
T3
«
M W W et)
M •O
W W Pd cd
W
WWW .
a) l l l H r t H •o Ui
0) i-l r-l

5Ssaa "g
i-i l-l l-l
C cd cd cd I-H 0) i—l *—l i—I
C
S §o g c Sa
o s
1-1 C td tö tg i
ce o o CÖ o
•H o et) O O O c ctNO O O
iH 1-< U ft 1-4 «H M l-l VH T-l «H
j j 4J 4J
0) 4J u 4J 4J 4J ai u u iJ u v Ü U 4J U 01
en O O O +J co o
eu
o o 4J tO O O O 4J to o o o u to o o o
a
ssa s sâ a CE H [3 u l3
(S w w w
 -56-

o O
CTI
e S
w —' u
 -57-

e eg
ai ^o

IT
e
•H
4J
M
W

c n co so H N \0 m m r-. s o CM
<u c f-s. so N \ Û *Û m co i ^ r*. co
S

0)
to
Q) 0 0
.o n
O 00
1 M
C C oo en sT 00 co CS r-f SO
O r-* r-* r-* t>. r^ SO CO 00 CO 0 0
e

o> o
OS -3-
0* - - I
£i H
O ß
en es r^. so m oo es CO r-* s f
r%. r^ N N N sO CO r^ r^ 0 0

.es oo • * -a- oo r*. es I-H so en r - er* f^> oo r^ er* e s f-* I-H e s

sO 00

B
C" Ol
ai J3
co
•O
ai
s
ai
»
14
a) 00 4J O
t.
a
a 60
143
01
h
u
3
B
•H O O
3 O SN
» O g «
3
O
> i
U
CO
4J
04
•O
ai 4-1
4-1
0)
Ui

O e - •H e^« p- S • H CO eu
) eu •" ai c» O •O O 14 B
•* 01 >
ai
* C0
3
4J
co C"
•O
u
!><S
i H CO
O
14 • H » •
T) î>. eu
J->
ai
o
»«
> rH g co o B 00 00 <u
3 a ai •O O
o •o B CO
e*-
4J u
u ai 4J •B
•2 ° co
u
ai
.£>
3
O
4J
a
•o
00 a>
i H 3
O
•H
SI CO 00
ai
B
C cd
C o
c- je ai >s 4-1 B CO 00 U 3 ß ai o
B 3 j3
: o ^ C
M 4-1 .o 4J
a 4J
•H •H
k> j e
CO
ai
B
•H
>. co
CO
e - cO
O
o •H
•H o
-u
£
a
a
o
T3
co a> CO b 4J u tH 4J co h X 00 00
' S -a r-c
o *»
•H O a ai
a>
ai O a
U a
e T H «
co o
E
>-i
O
» = iï
c»
<u 4-1 U
s ai
•o 14-1
i-H
o co
CO Ol
14
14
144 •o a> J3
e s: U ai Ol
a
o o S 4J 4-1 CO
> ua>a a) ^ 14
• o ai ai 14 ai at
ai

i S° 01 01
B
ai
O
B
•H CO
3 J=
14
o
co J 3 co co
14
«
CO
OO
14
co
u
ed
J-i

Vi
m 3
a »
U
O
3 3 3 S
ai
3 3
3
O •H
B
3 3 3
3
*
u >-.
U
4-1
J=
00
>-.
B
O O O o O O t»x: o O O
O
>*co >, >
ai t»% 4-1 S",
j o
3
•H ai >,
S B S B C
> B
B
>>>* >\
C C C
4-1 O
ai J3 C Ol
s
» J0)= a> J3 £ai JC3 ai 0) <u 0)
ai tu J= Ol ai
W S». »
14 ci 4J
j e j =
S »*
J= f X f
» ? 4J» 4J* s
ai
• H (.. •*
x o
O
•a
3
O
o co
•H ai 4-1 4-1 4-1 4-1
4-1
CO 4-1 4J>
»» •u J-t 4-1
I O J=
4J >N CO co CO CO co CO co Ol CO CO CO cd
1
•"." « ' co U o ai ai ai ai a i a i Ol «
0)
ai a> ai

« »J « O
4-1
O
00
B
4-1 (U 4-i
o o o o o o
O
4J O C- o o o o
i g » -S (0 o a e<« •H § > ai 4-1 4-1 4J 4-1 4-1 4J 4J 00 4J 4J 4-1 4J
01 4J B e - O S u e>-
4-1 ai 4J O a» •H a i a i a> ai aai B 14
Ol
ai
s
o ai a> 0) 0)
S cd ^ =5"
P iJ H
>,
u £ .e U
!>
4J 00 a i 00 CD
•u co v i H
• H ai • H • H
14 14 14 u
• H • H • H •H
ai • H
a CO
14 14 14
•H •H
u J-J J-J

3
•H •H 1 T ) co CO co CO co co a Ol co w
CO » ai
(A
co
0) 0 ) a>
O
ai
s>
X S u
S> O m
i-i
ai ai
a i •o • o
ai a i a i a i
TJ •V T J •a
co •O
.C
Ol Ol
•o B •a T3 T D T)
>, »
I h es* u 4J B v_^ . B Ol o
0) 0) >*. 0) •H 4J <o CO co co co O JB CO 00 co (0 CtJ eu
ta
b b
: , n > to Xi
> - H (0 * 0 «H
O
• a
3B 3
O •H O a i
00
Ci a> ai ai
ai
« ai ai ai
4-1 4J
Ol ai ai ai a> ai
I H £ iH >. M •H 4J 4-1
C T3
a) a) |CU 00 0) B ^B
0) 00 ai 00 4-1
a •U >>>>>>>
co CO CO CO O co co a O l CO co CO
.B J= JS j = XI JS JS co 00 X J = -C
> > > x;>
a) C >Ö >
x:
cfl

4J C O -H
I 3 > , J
4-1 B 4-1 B
3 i 14-1 •4-1 • H 3
S 3 3 3 3 3 3 3 co 3 3 CO 3
co
3 3 3
^
> O O O O J5 O -C O r+ O O O O O O O ai O O 00 O o 0 O
*» S*» C i H >» U U CO > . r H B >•>

ai 4J >*
» S » SS
4-1 B >>>>!><>>>, >> ex •H >".!», >,^ O
zzzüss o o
00
O
•H
4J
O
S S 5 5 a SS 3
JB
4J SS ss Q O

O
CO a\ ô l-l <N eo -sr m vO r^ 00 o \ o CN n
es es e s es
 -58-

o \ r«- • * r^ CM IT» CN oo co

e— CO
a cu cd oo x>
•o 3 » C
01 II u O cd -H o
00 4-1 o 4J JZ 4-t
CO 00 4J 4J
u c u 3 CU J 3 CU 01
3
O
•H
JS
o O
s>* e-
u oo a
• H «H O
1-.
•H
^*
4-t
cd
j j
Ü 4J •o co c - CO Cd CO (0 0)
al
CO 01 01 4J co tH 0 ) 1-1 CU
•H o
^ N •O a
o
f )
14
CO
B Cu
e—
O
e-* e— 01
i H
cd • O
3
u
CQ
t s
3
*Ö
4J P"
e—
00
co
c h
co u
o
•H
co
e-*
•o
3
•O ^g
r-4
O
01
4J
CO
«••3
i H CO cd
4J
J3 CU
X CU
00
G
G
-H
o
•H e«
O 00
e
3 OO
CO
e*.
•o
II
4J
». O
4J
B
O
T> 01
14
«
3
00C
CU « H
>
O
4J
4-1 1-1
•H
•H JS
4J 4J
4J •O T ) •H II B CO
t-H 3 3 cd Cd 4-1 OJ co CO 0 1
O 0) II JS co 0 0 B •H 00 O 14 j s X CS CU 0> u a
g
V
4J
co
CO
CO C - '
o
CO
3 s
O
II
4J
O B
O. 0 4 ^ %
G
•H
00
G >>•o O C
cd
•u
3
OJ l - l
-H
>
co
JS
•o o
co co
4J II co o •H o JS a 00 J5 •H CO j s CU O 3 CO CU 14
•o •H 14 co M X 00 00 <o 0 C 4J i H 4J a> 4-1u >•. O OJ 3 J3 Ol 4J
Cl) h a o ex c •H CO
S O O a> II 14 o ^ - O O 4J JS CO
t-H h 01 14 o. 01 h •H •H G a> <-t eu a O • ^ 4J 01

X
0) • H •o O CO « co
01
14 u-t • O
II
4-1
O II
CH 8 b
O 4-t
•d o o)
T ) J3 O
01 o
o
co ai i > 0 1 I I
.-H u u u u
u
CO co
II
h u
II
U
S
oi
ai •o m
>
co h o 14
>, s cd
0) (0
«
«
4J T3 >
cd
J=
01
o
4-1

CD co co CO 00 co co J Û CO
JS 4-1 3 4-> OJ C" CO " O
>ï
i H a 3 3 3
3
O
B
T4 3 3
« 3
0)
co 3 3 S 3
ai cd 3 o cd >
eu O • H CU CO m
CU o co
co •ö 3
OJ
G
O
U o O P o >, js O O 3 O O o O O , £ cd tH o O -H

« :* >
CU >>»>> B
4->
>* >*
<W
H-l >>>,
CO >ï 3
O
>>
O
•H
i H 4J
O
3
o co 0
U-I
> M - I
CJ
i H e a c
G II B G B B T H B C G G > * T 3 CU O cd 3 S B
o 01 u u 01 JS 01 D II T3 ai I I 0 1
II T3 CO >% o CO II T4
j s js u
•s a
JS
a JS JS JS JS JS ^ 3 JS

a * •s
.G JS O 3
3 3 3 *o •o o 0 a 3
c »
•• 4-> a •• a
0 a o
00 O O
C - r t - O O •H o
o 4-1 4J 4J 4J 4J 4J 4J O 4J 4J 4J 4J o •H O h *o 0
•H
CO
cd co co co
CO
01 CO CO co cfl •H CO co CO CO 3 Ö O - C «H U cd •H i H - >.
01 Ol 01 01 II II 01 to ai D II II O 4-1 i H CU
•3 r-t CO 01
U o h CO CU CU M A: 3 14
CU O
4J
O
4-1
o
4-1
o
4J
4J O
u
C"
00
O
4J
0
4J
O OJ 0
•M
O
4-1
O
4-1
O
4-1 O
>.^ a *o cd
O £1
o
cd
Ö
0 0 01

•H
-Ö co co
3
> ni a> I I
e -
c u
01 G
o QJ
>
1
B
4J o
O
CO 00
e eu
c
co e-«
4J 00
c C co
rH

e
01 u
Ol
u u 14 II • H
I)
14 14
II
U
01
u U CU
II
u
II
U
II
14
II
14 T3
r-\
t H T4 J3 n
cd
CU 1 - ( co 01
4J • H • H • H •H a co •H •H *o 3 js e
•H CO CO ( 0 CO a II
•H
CO a CO CO W
•U •H •H •H •H
•H CO CO CO CO
O
O c
H £
CU 4-t
U
ffl O co
4J
cd 4J
QJ
1 0) Ol II II co •o II u II 01
T3
1 II OJ I I II 00 cd B CU 4J 4-t •H u a> CO
C* •o • o
•O •o j s •o B TS •o •<r • G • O • O •o co a « co O CU 01
II O CO CO o n) cd • H x. 4J 14 00
•o ^^ co co CO CO o JS CO 00 CO CO cd •>»-' CO CO co co CU t-t U 10 O. ftrH co o
4-1 c
3! 00 4-1 4 J
0)
00 4-t i - i o CU o • H a -H
3 e OJ II 01 0J II II II II e ai ai 01 01 00 CO CU eu M A i *o co 14
C
>>>> B S
•H 4-1 4J •H cd CU > i H e« tH 01 CU 4-1 CO
•H
4J
4J > > > co> e
CO CO CO
JS JS JS
co > > > >
>
j s£
II
OOJS
CO CO 0 cd •U CO CO CO CO
J3 JS JS JS 4J
4J CQ cu cd cd
co 4= 3
co ctt 0 0
3 c
01
CO
u CO
dl
O.
OJ
c •3 CO
CO-fi S
JS JS
•Ö *Ö O > •H 3 14
o
o t-H
3
o
3 3
O o
3
O
3 3
o O 00 o
3 3 3
CO
II
3 3 3 3
O O
S oo
CO
O O r4 O O O
O 3 3
o o
3
o
•H
O
3
o
J=
4-1 o
3 o
o
3 O.

!»!>, 0 co. !*> >* B •« ><> X m

>. a >>3>*>* G
CS
m e >,>
> i H
> B >.>* :*•H
CO i-i
cd
M-l
>» >^ >> •H
tH
CU
II
o O O 0 C M-t « W W W CU «4-t
5
01
tH
JS
•H
4-1
O
a a s 3 SS 88 p u 8888
Q 4J
eu
CS
o
U

4J
M M M M M
•o M .3 as
co
M

a
s -* m
v£> r~ oo <?* o t-t B m sr m vO l-N. 00 O O H CNJ co < • m vo
H u CN CN CN CN CM CM PO po co m PO PO PO PO ä CO CO CO - î t ST •<r si- ^r <r -a-
 -59-

SUMMARYANDDISCUSSION
IntheDutchEatingBehaviourQuestionnaire (DEBQ)constructed, emotional
eatingwas showntocomprise twodimensions,onedealingwitheating in
response todiffuseemotionsandtheotherwitheating in response toclearly
labelledemotions.
Thethree scalesinthequestionnaire (Restrainedeating,Emotionaleating
andExternal eating)haveahigh internalconsistencyandalsoahigh factorial
validity.The similarityofthepatternofcorrected item-total correlation
coefficientsandofthe factorpattern inall subsamplesalsoindicatesahigh
dimensional stabilityoftheeatingbehaviourscales.
In recent studies reportedbyStunkardandMessick (Stunkard,1981,
Stunkard andMessick, 1985), alsothreeeatingbehaviourdimensionswere
obtained, i.e. (1)"cognitive restraintofeating", (2)"disinhibition"and (3)
"hunger".OurRestrainedEatingScale ishighly similar totheStunkardand
Messick CognitiveRestraint factor.Thisisnot surprising,becausebothhad
items fromPudel'sFragenbogen fürLatenteAdipositas (Pudelet al., 1975)in
the initial itempool.Theprecise relationbetweentheStunkardandMessick
DisinhibitionandHunger scalesandourEmotionaleatingandExternaleating
scales remainstobeassessed,however.
Thetwo-dimensionalityof itemsonemotionaleating isinaccordancewith
theobservationofSlochower (1983a)thatvarioustypesofemotional states
producedifferent typesofeating response:onlydiffuseemotional states
trigger theovereating response,andclearlylabelledemotionsdonotaffect
eatingbehaviour.However,thereare indicationsthatobese individuals
perceivethoseemotionswhichareclearlylabelled inmostnormalweight
individualsasdiffuse.Inapilot study,Slochower (1983a)found that
overweight subjectshadmoredifficultythannormalweight subjectsin
describingor labellingemotionalexperiences.Thepresentdatamaybe
considered tosubstantiatethisfindingonthebasisofthefollowinglineof
thought.Althoughthe factor structuresoftheitemsonemotional eatingwere
very similar inall subsamplesofobeseandnormalweight subjects,inthat
these itemsweredivided intotwodimensionsinboththesamples,posthoc
inspectionofthe itemloadingsshowedthatthedifferencebetween thesetwo
dimensionswas lessdistinct intheobese subsamples.Intheobese,threeof
the four itemsdealingwith thediffuseemotionsseemedtohavehigher loadings
onthedimensiondealingwithclearlylabelledemotions,indicating that these
emotionsarealsoexperiencedasdiffusebytheobese.
 -60-

To investigate this infuture studies,three scales foremotional eating

havebeen constructed: one isatwo-dimensional scale recommended for general
assessment; and twoarehomogeneous scalesdealing separatelywith "clearly
labelled" and "diffuse"emotions tobeused formore specific hypotheses.
Thepositive relationship between theEmotional eating scalesand the
External eating scale isinlinewith thepsychosomatic theory, that emotiona-
lityand food cuescanoperate together toelicit eating behaviour:a state of
highuncontrollabe anxietymayenhance the reactions of theoverweight toex-
ternal cues (Slochower, 1983 ).However, the same observation canalsobe ex-
plained intermsofexternality theory, inwhich ahighdegree of emotionality
isconsidered tobeamanifestation of the general traitofexternality, postu-
lated asbeing characteristic ofobese individuals (Schachter andRodin, 1974).
The significant relationship between theEmotional eating scalesand the
Restrained eating scalemaybedue tothe sideeffects of intense dieting,
which generate stress,andmay result inemotional instability and
hyperemotipnality (Hibscher andHerman, 1977;Herman, Polivy, Pliner, Threlkeld
andMunie, 1978;Polivy,Herman andWarsh, 1978).
However, theobserved relationshipbetween emotional eatingand restrained
eating inthis longitudinal study shouldbe interpreted with caution, because
intwoearlier cross-sectional studiesno relationshipwas foundbetween these
twotypesofeatingbehaviour (VanStrien, Frijters,Roosen,Knuiman-Hijl and
Defares, inpress;Van Strien, 1984 a ).Although asyet there isno clear
explanation for thisdiscrepancy itmaybeexplained by response setswhich may
havebeenmoreprevalent inthe longitudinal studydue toprolonged contact
with the investigators (VanStrien, 1984).
Endorsement of itemsontheemotional eating andexternal eating scales is
probably less sociallydesirable forobese than fornormal-weight individuals.
But, incontrast, the reverse isprobably thecase forendorsement of itemson
the restrained eating scale.Thiscanbederived from the fact that scores on
theHermanand PolivyRestraint Scalewere foundbyRuderman (1983)tobe
negatively related toscoresontheEysenck Lie scale inthenormalweight, but
intheobese,no relationshipwas foundbetween these two scales.This suggests
thatobese individuals consider endorsement of restraint scoresnot tobe
negative,which isconsistent with theobservation inthisstudy, that the
obesehad significantly higher scoreson restrained eating thannormal weight
subjects.Incontrast, however,Johnson, LakeandHahan (1983)found
significant negative relationships between restrained eating and theMMPI Lie
scale,and theCrowne andMarlowe SocialDesirability scale ina sample of
 -61-

obesedieters,and no such relationship insamplesofnormalweight and obese

non-dieters.
Inamixed sample ofobeseandnormalweight subjects (VanStrien,
Frijters,Roosen,Knuiman-Hijland Defares, inpress), a significant negative
relationshipwas foundbetween theDutchversionof theCrowne and Marlowe
Social Desirability scaleand a shortversionof thepresent Emotional eating
scale,butno relationshipswere foundbetween socialdesirabilityai<dearlier
versions of thepresent scales for restrained eatingand external eating.No
datawere available on the relationshipsbetween socialdesirability and the
present eatingbehaviour scales insubsamples ofobeseandnormal weight
subjects.
The extent towhich socialdesirability tendencies affect scoreson the
eatingbehaviour scales,and the consequence forconstructvalidity isnot
known and thedegree of external validity oftheeating behaviour scales has
yet tobe investigated.Nevertheless the threeeatingbehaviour scales,
constructed inthis study, permit exploration of thevalidity of themain
theories on thedevelopment andmaintenance ofhuman obesity.

NOTES

1.Unpublished data set.

2.A more detailed description of thePreliminary Studies isgiven inVan
Strien,T., Frijters,J.E.R.,Bergers,G.P.A.andDefares,P.B. TheDutch
Eating Behaviour Questionnaire (DEBQ).Assessment ofRestrained Eating,
Emotional Eating and External Eating. Internalpublication, rapport nr.
0-1185, Department ofHumanNutrition,Agricultural University, Wageningen,
1984.
3.Frijters,J.E.R. andRoosen, R.G.F.M.
Translation of threeeatingbehavior questionnaires.
Internal publication, Department ofHumanNutrition, Agricultural
University,Wageningen, 1980.
 -62-

CHAPTER9

THEPREDICTIVEVALIDITYOFTHEDUTCHRESTRAINEDEATINGSCALE

TatjanavanStrienwithJanE.R. Frijters,WijaA.vanStaveren,PeterB.
DefaresandPaulDeurenberg*.

HermanandPolivyconstructed theRestraintScale (e.g.,Herman,Polivy,

Pliner,ThrelkeldandMunie,1978),which inourviewhastwomainproblems.
Firstly,thescaledoesnotseemtobeunidimensional,andsecondly,it
probablydoesnotmeasurecognitive controloffoodintake.
Drewnowski,RiskeyandDesor (1982)andBlanchardandFrost (1983)have
shownthatthisscalemeasurestwounderlying factorswhichreflect fluctua-
tionsinbodyweightandconcernaboutdieting,respectively.Lowe (1984)found
a third factorcontaining itemsonsplurgingandthoughtaboutfood.In
separateanalysesofdata fromnormalweightandoverweight subjects,Ruderman
(1983)obtained thesametwo-dimensionalsolution inthenormalweightasin
thestudiesmentionedpreviously,butafour-dimensionalsolutionintheobese.
Thefirst factor inthisfour-dimensionalsolutionreflectsweight fluctuation,
thesecondbingeing,thethirdatendencytodiet,andthefourthoverconcern
withdieting.Finally,Johnson,LakeandHahan (1983)foundthree factors
withintheRestraintScaleingroupsofobesedieters,obesenon-dieters,and
normalweight subjects.Thefirstfactorwasheterogeneous incontent,and the
secondandthird factorsreferred topreoccupationswitheatingandweight
fluctuations,respectively.
The factthattheRestraintScalecanbereducedtoatleastthetwofac-
tors,fluctuations inbodyweightandactualconcernwithdieting,mayhave the
consequence thatindividualsscorehighontheRestraintScale,simplyonthe
basisoflargeweight fluctuations inthepast,withoutcurrentlywatching
theirweightorconsciously restraining theireating.
Other indicationshavebeenfoundthattheRestraintScalemaynotmeasure
cognitive restrictionoffoodintake.Wardle (1980)foundno relationshipbe-
tweentheRestraintScale,andintakeofenergy.

*Thischapter isinpressintheInternational JournalofEatingDisorders,

and isprintedherewithkindpermissionofJohnWiley&Sons,Inc.,copyright.
 -63-

inaddition,bothWardle (1980)andHawkinsIIandClement (1980)founda close

relationshipbetween theRestraint Scale,andmeasuresofbingeeating.These
observations suggest thattheRestraintScaledoesnotpredictthedegreeof
restrictionof food intake,butdisinhibitionof restraint tosomedegree.It
maybeargued, thatthisobservation isinlinewithHermanandPolivy (1980),
that theessential featureofRestrained Eatingtheory isnot the relationship
betweentheRestraint Scaleand food intake,but the inabilitytomaintain
cognitive controlover food intake (i.e.,susceptibility todisinhibition).
According tothisinterpretation, intensedieting resultsincounterregulatory
eatingbehaviour orexcessive food intake,when selfcontrol isundermined by
disinhibitors,suchasalcohol,anxietyordepression (HermanandHack,1975;
HermanandPolivy, 1975;PolivyandHerman,1976 a ' ,c
) . Howevermore recently,
counterregulatoryeatingbehaviour hasbeen showntobenotnecessarilythe
consequenceofdietingper se (StunkardandHessick, 1985).This findingwas
obtainedwithaquestionnaire toassess restrained eating, theThree Factor
EatingQuestionnaire (TFEQ),whichmeasuresseparatelycognitive restraintof
food intake (Factor I),disinhibitionofcognitive restraint (Factor II)and
hunger (Factor III).Resultsofalaboratorystudyonfood intake,obtained by
Shrager,Wadden,Miller,Stunkard andStellar (1983)showed thatnot Factor I,
butFactor IIwascloselyassociatedwithovereating.Thesignificanceofthese
findingsbecomes clear fromthe fact thatthecorrelationbetween theHerman
andPolivyRestraintScale andthe restraintsubscaleofscoreson theTFEQ
(Factor I)wasnot statistically significant (r«.168),while the correlation
between scoresonthisscaleand thedisinhibition subscale (Factor II)ofthe
TFEQwashighly significant (r-.840) (StunkardandMessick,1985;
Weissenburger,Rush,GilesandStunkard, submitted).Thus,itcanbe
tentatively concluded thatthecounterregulatoryeatingbehaviour of theHerman
andPolivy "restrained eaters"isrelatedtodisinhibitionof restraintandnot
to restraintper se,andalso,thattheRestraint Scale isnotameasureof
restrictionof food intake,butofdisinhibitionofcognitive restraint.

PURPOSEOFTHEPRESENTSTUDY

TheStunkardandHessick scalewiththreedimensionsof restrainedeating isa

major improvementbutwasnotavailablewhenthepresent studywas initiated.
Hence,TheDutchEatingBehaviourQuestionnaire (DEBQ)withscalesonEmotional
Eating, External EatingandRestrained Eatingwasdeveloped independently (Van
 -64-

Strien,Frijters,BergersandDefares,inpress). The focusofthepresent

studyistheRestrainedEating scaleoftheDEBQ.Thisscalehasbeen shownto
havehigh internalconsistencyand factorvalidity (VanStrienetal., in
press). However,thedecisive factordetermining itsusefulness isitsdegree
ofpredictivevalidity.As thescale isverysimilar tothecognitive restraint
subscale (Factor I)of theStunkardandMessickTFEQ (VanStrienetal., in
press), ithasahighpredictivevalidityifitpredictsadequately restriction
of food intake.However ingeneral,determinationofthe relationshipbetweena
scoreonaparticular restrainedeatingscaleand food intake (Wardle,1980)
doesnot seemtobe thebestway toestablishthepredictivevalidityofthe
scale. Inourview,a restrainedeating scaleisonlyvalid ifascore reflects
thedegree towhichan individualeatslessthanheor sheactuallywould like
toeat,becauseeatinglessthandesired istheverynatureof restrained
eatingbehaviour.Absolute intakeofenergydoesnotmeasure thedegree to
whichan individualeatslessthandesired.However, itisverycomplicated, if
not impossible, toconductaneatingbehaviour studytoinvestigate simulta-
neouslyactual foodconsumptionand restrictionoffoodintake.This results
fromthepracticaldifficultyofdetermining thequantityofenergy,or food
not eatenby the individualbutwhich the individualwouldhaveeatenwithout
cognitive inhibition.Onederivative,and thereforean indirectway toassess
thedifferencebetweenactual intakeofenergyanddesired intakeofenergyis
as follows.Theenergy requiredbyanindividualdependslargelyonhisbody
weight,bodycompositionandphysicalactivity. Inmostpeople,physical
activitydoesnot influence toalargeextent theenergy required,andalsothe
energy required foractivitiesdependsonbodyweight.Fromthispostulate,it
followsthatunder thestatedconditionsthenecessitytoeatacertain
quantityof food ismainlycontrolledbythe individual'sbodyweight.Heavy
individualsneedmoreenergythanleanindividualsinorder to remainbody
weight constantand tobeable toperformthesamephysicalactivity.Therefore
whenan individual isinenergybalance,theenergy requirement isan indirect
estimateof theenergycontent ofthefooddesired.Thus,ifthisassumption is
correct,thedifferencebetweenenergy requirementestimated frombodyweight
andactivitypattern,andenergy intakeasassessed ina food consumption study
canbe takenasan indirectmeasureofdegreeof restraint.IftheDutch
RestrainedEatingScalehasahighpredictivevalidity, there shouldbeahigh
correlationbetweenscoresonthescaleandestimationofdeviation inenergy
required. Ifthisiscorrect,thenahighnegativecorrelationmayalsobe
expected between scoresonthescaleand intakeof fatand sugars (mono-and
 -65-

disaccharides),because theseenergysourcesareperceived asbeing fattening

bymanydieters.

METHOD

Overview
Estimatesofvaluesofactual intakeofenergy, fatand sugarwere obtained
fromthoseparticipating inanongoing studyonseasonalvariations inenergy
intake inwomen (VanStaveren,Deurenberg,Burema,DeGroot andHautvast,
submitted).Fromthisstudy,three24-hour foodintake recallstakenat
three-month intervalswereundertaken toassessvariousmeasuresofenergy.The
DEBQwasadministered tothesubjectsduringthesecondassessmentdate.
Thesedatawereexamined todeterminewhether therewere relationshipsbe-
tween restrainedeating scoresandthemagnitudeofthedeviation fromenergy
requirement,andalsobetween restrained scoresandthe intakeof fatand sugar
(mono-anddisaccharides).

Subjectsandprocedure
Three 24-hour food recallswereobtained for 110womenparticipating ina
longitudinal studyon seasonalvariation inenergy intakecarriedout in
Renkum,theNetherlands.Allwomenwere intheagegroup31to34yearsasat
1 January1983.Noonewasonadietprescribedbyadoctor,orwaspregnant.
2 2
At June 1983,themeanbodymassindex (BMI;weight/height )was22.3kg/in
(s-2.7).The 24-hour food recallshadbeenobtainedondifferentdaysinthe
monthsofApril,June,andSeptember 1983bytraineddieticiansonunannounced
homevisits.Toensure thatthe recordswereasaccurateaspossible,they
weighed theportionsof foodmost frequentlyconsumed (otherportionswere
estimated intermsofhouseholdmeasures),andthe reportedfood intakeofthe
previousdaywas recorded.The followingday,after risingandurinating, the
women themselvesmeasured theirbodyweightoncalibrated scalestothenearest
0.5 kg (note 1)*. Onthefirstvisit,bodyheightwasmeasuredbythedietician
tothenearest0.1 cm.Onthehomevisit inJune,theRestrainedEatingScale
wasadministered bythedieticians.

* seepage68fornotes
 -66-

Measures
Restrained EatingScale.TheRestrainedEatingScaleconsistsoften items
havingaLikert response format:never,seldom, sometimes,often,andvery
often.Twoitems (seeTable 2,chapter 8)haveanadditional category, "not
relevant", toindicate thata respondenthasnevereaten toomuch,or never
gainedweight inadult life.A scoreforthisscale isobtainedbydividing the
sumofitemscoresbythetotalnumber of items,thusahighscoreindicatesa
highdegreeof restrainedeating.Those (n»6)whogaveanot relevant response
wereexcluded fromanalysis.Table 1showstheitemsand their item-rest
correlation coefficientsof theDutchRestrainedEatingScale.
Conversion intonutrients.Intakeofdailyenergy, fatand sugar (mono-and
disaccharides)was calculatedbymultiplying theestimated amountsof foods
consumedbytheappropriatevaluesobtained fromacomputerized food
composition table (Hautvast, 1975).
Meanone-dayintake.Foreachsubject,themeanone-dayintakeof energyand
nutrientswasobtained fromthemean intakeofenergyandnutrient intakeof
thethree 24-hour recallperiods.
Deviation fromthe required energyintake.A totalof2200Kcal isthe
averagedaily requirement forwomen intheagegroup20-35yearsandof 60kg
bodyweighthavingmoderatelystrenuoushabitualactivitylevels (Committee on
Caloric Requirements,1950) (note2 ) .Correction forbodyweightwasmadeby
eitherdecreasing or increasing the required 2200Kcalby150Kcal forevery5
kgbeloworabove 60kg, respectively (DutchExpertCommittee onEnergyand
Nutrient Requirements, 1983).Deviationofan individual'smeanone-day caloric
intake from the required caloric intakewasobtained asfollows:

Caloric intakendays /60kg-bodyweight ]

2200Kcal- ƒ x150KcalJ
5 /

RESULTS

Themean scoreon theRestrained EatingScalewas 2.52 (s-,89).Thisvalue is

similar tothat found inanearlier study (VanStrienetal.,inpress). The
meandeviationofenergyintake fromenergy requirementwas-278Kcal (s-641).
Fromthenegative signofthisvalue itcanbeconcluded that themean intake
ofenergywas lessthanthecalculatedmeanenergy requirement.Themean intake
of fatand sugarwas85g (s-28)and 114g (s-39), respectively.
 -67-

The correlation coefficient showing the relationshipbetween restrained

eatinganddeviationofenergy intakefromenergy requirementwas-.37 (p<.01).
After correction forattenuation (note3),thiscorrelation increased to-.45.
Thismeansthatasubjectate lessthanrequiredthehigher shescoredonthe
RestrainedEatingScale.Thecorrelationcoefficientbetween restrainedeating
and intakeof fatwas-.28 (p<.01),and thiscorrelation increased to-.38
after correctionforattenuation.The correlationbetween restrainedeatingand
intakeof sugarwas-.38 (p<.01),andthiscorrelation increased to-.46after
correction forattenuation.Thus,all correlationsbetween restraint scoresand
themeasuresof food intakewere significant,and inthedirectionpredicted.

CONCLUSION

About 20%ofthevariance inthescoresontheDutchRestrained EatingScalewas

explainedbymostmeasuresof food intake.These resultsshouldbe considered
fromtheperspective that thepresentmeasuresmaybe contaminatedbyanumber
ofunwanted sourcesoferrors.Firstly, themeasuresof food intakeareonlyes-
timatesof theactual food intake (Block, 1982). Secondly,thederived estimate
ofthestabilityofthemeanover-three-day food intakemaynothavebeenthe
bestpossible.Thirdly, theestimateofthe required intakeofenergy isonlyan
approximationoftheactualenergy requirement,which foran individual couldbe
biasedbymany factors.Finally,thedeviation fromenergy requirement isonlya
derivativeof thediscrepancybetweenan individual'sactual food intakeandhis
orherdesired foodintake.
Another sourceoferrormaybe that food intakesassessed inthemonthsof
April andSeptember maynothavebeenaccurateestimatesofthedegreeof res-
trained eatingassessed inthemonthofJune.An individual'sdegreeof restric-
tionof food intakemayhave changed intheperiodbetweenAprilandJuneorbe-
tweenJuneandSeptember. Itwouldhavebeenpreferable tohaveobtainedall
food recallswithinatimespanclosetotheasssessmentof restrainedeating.
Taking these factorsintoconsideration, itcanbe concluded that the results
ofthepresent studysuggestthattheDutchRestrainedEatingScalehasmoderate
togoodpredictivevaliditybut thattheeffectsofpossible sourcesoferror
shouldbe investigated further.
 -68-

NOTES

1.Weightwas reportedbythewomen themselvesbecause itwasnotpossible for

thedieticians tomeasure thewomen'sbodyweightwithout clothes,before
breakfast,andwithemptybladder.Although relyingonself-reportingmay
have introduced adegreeofunreliability toweightmeasurements,a recent
studybyStunkard andAlbaum (1981)suggests that thiserrormayhave been
minimal.
2.Theseconditionshold forthegroupunder study (VanStaverenet al.,
submitted).
3.Correction forattenuationwasobtainedbyapplicationof theequation:
r
xy

V rel
x
. r,
3s

inwhich,
r -thecorrelation coefficientbetween theRestrained EatingScaleand
™ ameasure of food intake
rel. -the reliabilitycoefficient oftheRestrained EatingScale (.94)
r, -the reliabilitycoefficientofameasure forfood intake (deviation
from required intake- .73;intakeof fat- .58;intakeof sugar -
.72)
The reliability coefficient ofameasure forfoodintake (r,)wasobtained by
application of theSpearman-Brown equationasfollows:
3(r„)
1 + 2r
s

inwhich,
r - thestabilitycoefficient forone-day intake
r, - thestability coefficient forthemeanover-three-daysfood intake.
The stabilitycoefficient forone-dayintake (r )wasobtained asfollows:
2 2 2
r + r + r
l,2 l,3 2,3

inwhich,
r
s
r.1,4, -the correlation coefficientbetweenquantityconsumed onday1and
day2
r.l ,
,i
» the correlationcoefficientbetweenquantityconsumedonday1and
day3
r,. - thecorrelation coefficient betweenquantityconsumedonday2and
*•* day3
 -69-

CHAPTER10

EATINGBEHAVIOUR,PERSONALITYTRAITSANDBODYMASSINWOMEN

TatjanavanStrienwithJanE.R.Frijters,RenéG.F.H.Roosen,Wil J.H.
Knuiman-HijlandPeterB.Defares*

THEORY

Inbothpsychosomaticandexternality theoryobesity isattributed toover-

eating.Inpsychosomatic theoryeatingisconsidered tobea responseto
emotionalarousal.Intrapsychicconflicthasbeenputforwardasacentral
element inducingtheeatingpatternofemotionaleaters (Bruch,1957,1961,
1973; Slochower,1983 a ).Overeating,hasbeendiscussedasameansof:
diminishinganxiety;sedation;counteractinga feelingofbeingunloved;an
expressionofhostility;andtheavoidanceofcompetition (KaplanandKaplan,
1957). Ingeneral,suchindividualsareconsidered tobelessadjusted,to
exhibitdistinctivepersonalitytraits,andacharacteristic emotionality
(HcReynolds,1983).Externalitytheoryaccountsforthephenomenonofeating in
responsetofoodcues,whichisconsidered tobeonlyonemanifestationofa
generalized stimulussensitivity.Thispointofviewisbasedonobserveddif-
ferences innormalweightandobesesubjectsintasks,including reactiontime,
short-termmemory,tachistoscopic thresholds (Rodin,HermanandSchachter,
1974),distractability (Rodin,1973),timeestimation (Pliner,1973a), focused
thinking (Pliner,1973b), emotional reactivity (Pliner,1974;Rodin,1973;Ro-
din, ElmanandSchachter,1974)andclassicalconditioning (Yaremko,Fisherand
Price,1975).Recently,however,thisgeneralized stimulusbindingconcepthas
beencriticized,becausenopatternofintercorrelationshasbeenfoundbetween
thevariousmeasuresofexternality (IsbitskyandWhite,1981).Eventhose
measuresusedbySchachterandRodinandshownbythese investigatorsto
differentiatebetweenobeseandnormalweightsubjectswerenot intercorrelated
(NisbettandTemoshok, 1976).

*Thischapterhasbeenpublished (inslightlydifferentform,see footnotes)

inAddictiveBehaviors (Vol10,(1985)- )and isreprintedherewithkind
permissionofPergamonPress,copyright1985.
 -70-

Common tobothpsychosomatictheoryandexternality theoryisthe

misperceptionofhungerand satietyfor food intakeasacausal factor in
overeating (Robbinsand Fray, 1980).Recently, Slochower (1983a)hasproposed
thatemotionalityand foodcuesoperateconjointly toeliciteatingbehaviour:
a stateofhighuncontrollable anxietymayenhance reactionstoexternalcues.
Inexternalitytheory,aclose relationshipbetweenemotionalityand
externality isalsoassumed, ifahighdegreeofemotionality isconsidered to
beamanifestation ofthegeneral traitofexternality, asassertedby
Schachter andRodin (1974).
Theobservation, thatsomenormalweight individualshaveahighdegreeof
externality (e.g.,Rodinand Slochower,1976)hasled tothecontention that
othermechanismsunrelated toexternal responsivenessdeterminebodyweight
(Rodin, 1978).Thesemechanismsmaybeeitherphysiological orpsychological,
forexample, conscious restrictionof food intake.Even though this construct
isnot incorporated inpsychosomatic theory, thesamemayhold for emotional
eaters.External responsivenessandhighemotionaleatingmay thusbeprevalent
inallweight categories.
Pudel (e.g.,1978)devisedaquestionnaire,theFragenbogen für Latente
Adipositas (FIA)fortheidentificationofnormalweight subjectshaving
"obese"eatingpatternsthat isthe "latentobese". The "latentobese"were
shownnot to reduce theireating rateduringatestmealof20minutes
duration,which isaneatingpatternsimilar toobese subjects.Thisphenomenon
suggestsadisturbance ofthehunger-satietymechanism,which,asalready
stated, isinlinewithbothpsychosomatic theoryandexternality theory
(Pudel,MetzdorffandOetting,1975;MeyerandPudel,1977;Pudel, 1978).
Pudel stressed thetheoretical importance ofassessing personality
differencesbetween "obese"and "latentobese"regardingbodyweightcontrol,
and referred to "PrometheischerWille" (willpower)asoneof thedeterminants
of successfulweight loss.Thedegreeof internal control hasalsobeen
suggestedasapredictor ofsuccessfulweight lossinobesity therapy (e.g.,
Salzer,1982;BalchandRoss,1975;Weiss, 1977).
Incontrast tobothpsychosomaticandexternality theory,the theoryof
restrained eatingattributesovereating todieting (HermanandPolivy, 1960a;
PolivyandHerman, 1983).Thisparadox isbasedontheconceptofnatural
weight, a rangeofbodyweightwhich ishomoeostaticallypreservedbythe
individual.Attempts tolowerbodyweightbyconscious restrictionof food
intake initiatephysiological defenses,suchas,lowering themetabolic rate
andarousal ofpersistent hunger.When self-control isundermined bydisinhi-
 -71-

bitocs, suchas,alcohol,anxiety,depression, orevenby theconsumption of

high calorie foods,counterregulationmayoccurand inthisexcessive food in-
take (HermanandMack, 1975;Herman and Polivy,1975;Polivyand Herman,
bc
1976a' ).Inaddition, continuousdenial ofhungermay result inlossof
contactwith feelings ofhunger and satiety (PolivyandHerman, 1983).Thus,
intensedietingmayultimately result inobese eatingpatterns, since both
arousal andexternal stimuli disrupt thecognitive restraint normally exercised
bydieters facedwithpersistent hunger (Hermanand Polivy, 1980b).

Aim of the study

Theprincipal aimof this studywas toexaminewhich theory seemsmost valid,
psychosomatic theory, externality theoryor the theoryof restrained eating. It
was the intention todetermine thisbymeansofpsychometric studiesof the
relationships between scalesofemotional eating, external eating and
restrained eatingon theonehand, andvariables suchasbody fatnessandper-
sonality scaleson theother hand. Itwas reasoned thatonthebasisof the
ideasexpressed inpsychosomatic theory, externality theoryand the theory of
restrained eating, different findingswere tobeexpectedwith regard tothe
relationships between thethree typesofeatingbehaviour and/or personality
traits.Thus, from thedegree towhich observed findings are similar toantici-
pated findingswithinone single theoreticalviewpoint, itwouldbepossible to
drawtentative conclusions regarding thevalidity ofeach specific point of
view.
The theoryof restrained eating suggests thataheightened degree ofexter-
nality and emotionality isthe consequence ofdieting. Ifthislineof reaso-
ning iscorrect, thena significant relationshipmaybeexpected inthe present
studybetween restrained eatingandemotional eating andbetween restrained
eating andexternal eating.
Incontrast tothe theoryof restrained eating, inwhich "obese"eating
patterns are considered tobeaconsequence ofdieting, inboththe psychosoma-
ticand the externality theory,dieting isconsidered tobe one of the possible
reactions tooverweight caused byovereating.These theoretical positions imply
that "obese"eating patternspreceed restrained eating. Ifthis supposition is
correct, nodifference shouldbe found between "obese"and "latent obese"women
inthedegree of emotional and external eating.
As a consequence ofboth theories,butondifferent grounds,a significant
relationship istobeexpected between emotional eatingand external eating.
From theviewpoint ofexternality theory, this relationship seemsmost proba-
 -72-

ble,becauseahighdegreeofemotionality isconsidered tobeamanifestation

of thegeneral traitofexternality.Fromtheviewpointofpsychosomatic theo-
ry,this relationshipisalsotobeexpected, sincehigharousaland tempting
foodmustbepresent forapersontoovereat.
The followingexplorationmaywarrant tentativeconclusionsastowhich
theory istobe favoured,psychosomatic theoryorexternality theory.An indi-
cationofthevalidityofthestimulus-bound conceptofexternality theorymay
beobtained if relationshipsbetweenexternaleatingandpersonalitymeasures
reflectingaspectsofexternality, forexample,Rotter I-Escalesand social
approval,andpersonality scalesmeasuringaspectsofemotionalityare foundto
be significant.An indicationoftheimpact oftheemotionalityofemotional
eatersinpsychosomatic theorymaybeobtained ifrelationshipsbetweenemotio-
naleatingandscales indicatingaspectsofemotional instabilityand lowad-
justmentare found tobe significant.
Inaddition tothesetheoreticalconsiderations,also personalitydiffe-
rencesbetween "obese"and "latentobese"womenhavebeenexplored for indica-
tionsastowhichtraitsare related toahighdegreeof restrained eatingand
successfulweightloss.

METHOD

Subjects
The sample consisted of80women,all inhabitantsofWageningen,whowere
selected bytheir generalpractitioners.The samplewas selected insuchaman-
ner, that itconsistedofbothnormal-weightandoverweight subjects.Seventeen
had aBMIof 30.0orhigherandameanageof 31.4years (s-3.0). The remai-
ning63womenhadaBMIlessthan 30.0andameanageof 31.1years (s-9.4).
ThemeanBMI ofthewhole samplewas25.2 (s-4.8), and themeanagewas 31.1
years (s-8.4).

SessionandQuestionnaires
Duringanevening session (7.00pmto10.30pm),thesubjectswere requested to
complete threetypesofquestionnaires.As timelimitswere imposed, some
subjectsdidnot completeallquestionnaires.At theendof the session,body
weightandheightweremeasured.Thefollowingquestionnaireswereused:
- aquestionnaire onage,marital status,height,weightanddegreeofweight
lossorweightgainintheprecedingtwoyears,
- threescalestoassessemotionaleating,externaleating,and restrained
 -73-

eating,derived fromthefirstversionoftheDutchEating Behaviour

Questionnaire; (DEBQ) (seePreliminary StudyOne,chapter 8). The "Emotional
Eating"scaleconsisted ofseven five-point items ( 1 - 5 ) (Cronbach's alpha
-.89); the "External Eating/PerceivedHunger"scale,sevendichotomous items
(1,3) (KR-20-.83); and the "Restrained Eating"scale,sixdichotomous items
(1,3), (KR-20-.78) (SeeTable 1)
- asetofDutchpersonalityquestionnaires (seeTable2 ) .

Table1Itemsofthe threeeatingbehaviour scalesand their corrected

item-totalcorrelation coefficients.

Emotional Eating corrected item-

total
correlation
coefficients
1.Doyouhaveadesire toeatwhenyouhavenothing todo? .65
2.Doyouhaveadesire toeatwhenyouarebored and restless? .70
3.Doyouhaveadesire toeatwhenyouareanxious,worriedor tense? .76
4.Doyouhaveadesire toeatwhenyouaredepressed ordiscouraged? .79
5.Doyouhaveadesire toeatwhenyouarefeelinglonely? .66
6.Doyouhaveadesire toeatwhenyouareworried about something
orwhenyouareunder stress? .60
7.Doyouhaveadesire toeatwhenyouareexcited? .68

Restrained Eating

1.Doyounot eat certain foodsbecause theymakeyou fat? 751

2.Doyouweighyourself at leastonceaweek? .61
3.Soasnot tobecomeheavier,doyoudeliberately eat less? .54
4.Doyouoften readadvice onhowtoslim inthe newspapers
andmagazines? .49
5.Doyouwatchyour figurecarefully? .46
6.Doyouwatchpreciselyhowmuchyoueat? .47

External Eating/PerceivedHunger

1.Whenyou seefood,doyouhaveadesire toeat? 755

2.Whenyou seeotherseating,doyouhaveadesire toeat? .80
3.Eventhoughyoueatat regular times,doyou feelhungryat
other timesduring theday? .47
4.Doyoueatevenwhenyouarenot feelinghungry? .42
5.Whenyousee something that isreallydeliciousdoyouwant
toeat itimmediately? .60
6.Canyou letanopenpacketofdelicious food lieinthe
cupboard fora fewdayswithout eating it? .56
7.Whenyou smellfreshlybaked food,doyouhaveadesire to
eat,even ifyouhave justhadameal? .55
 -74-

Table2Personality scales administered tothestudy sample

Personality scale used

Abbreviation Dutch English translation
NPV Nederlandse Persoonlijkheids- Dutch Personality Questionnaire
Vragenlljst (Luteyn, Starren
enVanDijk,1975)
-IN Scales:lnadekwatie Scales: inadequacy
-SI sociale lnadekwatie social inadequacy
-RG rigiditeit rigidity
-VE verongelljktheid f e e l i n g wronged
-ZE zelfgenoegzaamheid self-sufficiency
-DO dominantie dominance
-ZW zelfwaardering self-esteem
GLTS Guilford LTP Temperament Guilford Zimmerman Temperament
Survey Survey (1949)edited inDutch
by: Laboratorium voor Toege-
paste Psychologie (1974)
-E Scales: sociale extraversie Scales:sociability
-G gevoelsmatigheid sentimentality,
emotionele labiliteit emotional Instability
-R reagibiliteit, temperament impatience
-A autoritair gedrag, authorltarism,
geldingsdrang assertiveness
-0 onbezorgdheid, optimisme carefree, optimistic,
oppervlakkigheid superficiality
-M masculiniteit masculinity ofemotions
and i n t e r e s t s
SA Sociale Angst (Willems,Tuendei Questionnaire t o measure s o c i a l
de HaanenDefares,1973) anxiety as a t r a i t
SD Sociale Goedkeuring'schaal Social Approval Scale (Crowne
and Marlowe, 1964) e d i t e d I n t o
Dutch by Groffen and Van Kreveld
(1972)
ZOET Punt 9vandeEPQ Item 9 of the Eating P a t t e r n s
Questionnaire (EPQ)
(Wollershelm, 1970)
Bent ugekopzoetigheid(1-5) Do you l i k e sweet foods?
AS Sociale stijgingschaal A s c a l e t o measure s o c i a l
ascendancy from the Guilford
Zimmerman Temperament Survey,
Guilford & Zimmerman, 1949)
I-E Interne-Externe Beheersingschaal I n t e r n a l - E x t e r n a l Control Scale
( R o t t e r , 1966) Edited i n Dutch
by Andriessen (1972)
-18 Scales:I-E18 S c a l e s : 18 items i n t e r n a l -
external control
-RuwSC I-E Ruwe score Raw s c o r e s of I n t e r n a l -
e x t e r n a l c o n t r o l on
remaining items
PMT* Prestatie Motivatie Test Achievement Motivation Test
(Hermans,1968)
-P_ Scales:prestatie motivatie S c a l e s : achievement m o t i v a t i o n
-F~ negatieve faalangst disfunction!ng in unstructu-
red t a s k s i t u a t i o n s
positieve faalangst f u n c t i o n i n g well In u n s t r u c -
tured t a s k s i t u a t i o n s

* R e s u l t s on the PMT have not been presented In t h e o r i g i n a l paper f o r reason

of missing d a t a (n=55).
 -75-

RESULTS

RelationshipsbetweenBMIandeatingbehaviour,andinterrelationships between
thethreeeatingbehaviour components
ItiePearson correlation coefficients showingthe relationshipbetweenBMI and
eachof thethreeeatingbehaviour componentsaregiveninTable 3,andalsofor
thecorrelationsbetween theeatingbehaviour components.Partial correlation
coefficientsarealsopresented toexclude confoundingeffectsofoneofthe
other twoeatingbehaviour components,andofBMI.

Table 3Pearson correlation coefficients showing the relationshipsbetween the

threeeatingbehaviour componentsandbodymassindex (BMI)andpartial
correlation coefficientsadjusting fortheeffectofconfounding factors.

Motional EatingandBKt 735**

adjusted fortheeffectofRestrained Eating .46**
adjusted fortheeffectofExternalEating/PerceivedHunger .34**

Restrained EatingandBMI .09

adjusted fortheeffectofEmotional Eating .11
adjusted fortheeffectofExternalEating/PerceivedHunger .10

External Eating/PerceivedHungerandBMI .38**

adjusted for theeffectofEmotional Eating .20*
adjusted for theeffectofRestrained Eating .38**

Restrained EatingandEmotionalEating .00

adjusted for theeffectofBMI -.05

External Eating/PerceivedHungerandEmotional Eating .49**

adjusted fortheeffectofBMI .39**
External Eating/PerceivedHungerandRestrainedEating .01
adjusted fortheeffectofBMI -.03

**p < .01

* p > .05

Both "Emotional Eating"and "External Eating/Perceived Hunger"were foundtobe

significantly related toBMI,but "Restrained Eating"wasnotshowntobe
related toBMI.A significant relationshipwas foundbetween "EmotionalEating"
and "ExternalEating/PerceivedHunger",but "RestrainedEating"wasnotfound to
be related toeither "Emotional Eating"or "External Eating/Perceived Hunger".

Relationshipbetweeneatingbehaviour componentsandpersonality traits

Pearsoncorrelation coefficients showingthe relationshipbetweentheeating
behaviour components scalesandpersonality scalesarepresented inTable 4.
 -76-

Partial correlation coefficientsadjusting fortheeffectofBMIarealsogiven

toallow forassessment ofthe relationshipbetweenbehavioural components and
personality traitsafterexclusionoftheBMIvalueasaconfoundingvariable.
Theanalysiswas restricted tothepartial correlation coefficients only.

Table 4Pearson'scorrelation coefficients showing the relationshipsbetweenthe

threeeatingbehaviour componentsandthepersonality scalesandpartial corre-
lationcoeffcients ()adjusting fortheeffectofthebodymassindex (BMI)

Personality EmotionalEating Restrained Eating External Eating

a
scale Perceived Hunger
NPV-IN 51** .50**) 7Ü7- ^59~ 09 .05 )
-SI 13 .05 ) .14 -.16 12 .05 )
-RG 10 -.13 ) .15 .14 08 -.10 )

-VE 13 .09 ) .01 .00 00 -.05 )

-ZE 10 -.20* ) .14 -.15 04 -.10 )
-DO 07 -.00 ) .16 .17 10 -.05 )
-ZW 32** -.29**) .18* .20* 13 -.07 )
GLTS-E 25* -.12 ) .13 .17 17 -.05 )
-G 32** .30**) .13 .12 01 -.04 )
-R 10 .20*) .06 -.05 04 .03 )
-A 05 .03 ) .22* .22* 04 -.07 )
-O 20* -.19* ) .06 .07 03 .07 )
-M 21* -.15 ) .06 -.04 05 .13 )
SA 23* .14 ) .02 -.04 13 .05 )
SD 21* -.28**) .16 -.16 13 -.18 )
ZOET 30** .28**) .25* -.27** 43** .43**)
AS 20* -.10 ) .20* .23** 14 -.04 )
I-E-18 02 -.03 ) .03 .02 01 -.06 )
-RuwSC 10 .02 ) .06 -.08 01 -.09 )
PMT-pf 31** -.28* ) .12 .14 15 -.10 )
" F 29* .32**) .15 -.15 04 -.04 )
-F + 04 .06 ) .04 -.03 25* .28* )

* P<.05 seeTable2
**P<.01 seefootnoteTable2

"Emotional Eating":Thehigherawoman scoredon "Emotional Eating", themore

she reported feelinganxiousandacting inadequately (NPV-IN), lacking in
self-sufficiency (NPV-ZE), lacking inself-esteem (NPV-ZW),being sentimental
andemotionallyunstable (GLTS-G),beingworried (GLTS-O),lacking patience
(GLTS-R), showingno signsof socialdesirability (SD)andthemore she reported
tohaveahighpreference for sweet foods (ZOET),tohavelowachievement
motivation (PMT-P)andthemoreshe reported tobeunable tofunctionwell in
unstructured task situations (PMT-F-).

"ExternalEating/PerceivedHunger":Thehigherawoman scoredon "External

Eating/PerceivedHunger",themore she reported tohaveahighpreference for
 -77-

sweet foods (ZOET),and to functionwell inunstructured task situations

(PMT-F + ).

"Restrained Eating":Themore restrained ineatingawoman indicated tobe,the

more she reported high self-esteem (NPV-ZW),ofbeing authoritarian (GLTS-A),
having a lowpreference for sweet foods (ZOET),andhavinganeed for social
ascendance(AS).

Differences inpersonalityand eating behaviour between "latent obese"and

"obese"women
Inthepresent study, a subjectwas classified asbeing "obese" if shehad aBMI
of 30.0orhigher, inthemanner proposed byPudel (e.g.,Pudel, 1978),a
subject was classified "latent obese"onthe following criteria:
- ifherBMIwas less than30.0;
- if she considered herself tohavebeenoverweight at least once inher life;
thiswas considered tobe the case, ifshegaveanegative answer tothe
question: Ihavenever been overweight;
- ifher score on the "Restrained Eating" scalewashigher than theaverage of
the sample (x-2.3).
On thebasisof these criteria, 17womenwere classified as "obese"and 10as
"latent obese".The significance ofthedifferences between themeanof each
group onvariouspersonality scaleswasassessesbya t-test, and the results
aregiven inTable 5.
"Latent obese"subjects reported feelingmore socially adequate (NPV-SI),
less social anxiety (SA)andbeingmore outgoing (GLTS-E)than "obese"subjects.
Furthermore, they reportedbeingmoredominant (NPV-DO),havingagreater need
for social ascendance (AS),andbeing able to functionbetter inunstructured
task situations (PMT-F - ).They showed ahigher preference formale occupations,
were lesssentimental (GLTS-M),andhadahigherdegreeof self-esteem (NPV-zw)
than "obese"women. Finally, "latent obese"women reported ahigher degree of
internal control, asmeasured ontheRotter internal-external control scale
(IE-ROW).Nodifferenceswereobeserved between the "latentobese"and the
"obesewomen, in respect of "Emotional Eating"and "External Eating/Perceived
Hunger" (seeTable6 ) .
 -78-

Table 5Personalitydifferencesbetween "latent obese"and "obese"women

Personality* Latent obese Obese

scales Mean SD n Mean SD n P
NPV-IN 11.8 7.5 10 14.9 9.6 17 .3é
-SI 6.9 3.7 10 13.7 5.8 17 .00
-RG 25.7 7.8 10 26.4 6.4 17 .80
-VE 14.9 7.7 10 17.8 8.5 17 .37
-ZE 10.6 3.9 10 11.9 5.8 17 .50
-DO 15.3 7.1 10 8.9 5.3 17 .03
-ZW 29.7 3.6 10 24.9 6.7 17 .02
GLTS-E 23.0 6.1 10 10.8 6.7 16 .00
-G 21.8 8.4 10 25.6 7.2 17 .24
-R 17.7 8.6 10 16.5 5.8 17 .71
-A 24.4 10.4 10 22.8 9.3 17 .70
-0 18.2 8.0 10 12.8 7.4 17 .10
-M 14.6 6.6 10 9.4 4.5 17 .05
SA 40.1 12.4 10 64.9 11.9 17 .00
SD 8.0 3.1 10 10.2 3.7 14 .12
ZOET 2.7 1.1 10 3.2 1.2 17 .30
AS 16.6 4.3 10 10.3 3.1 14 .00
I-E-18 25.0 3.9 10 27.9 4.4 16 .09
-RUWSC 22.2 6.1 10 27.4 4.9 16 .04
PMT-P 19.0 8.7 8 16.6 6.8 9 .53
-F 12.5 4.4 8 17.8 5.7 9 .05
-F+ 7.8 4.2 8 5.9 2.9 9 .32

* seeTable 2
see footnoteTable 2

Table6Differences inthethreeeatingbehaviour components between the "latent

obese"and the "obese"women

Eating Behaviour Latentobese Obese

component Mean SD Mean SD
Emotional Eating 2.8 1.1 10 3.2 .8 17 .24
External Eating/
Perceived Hunger 2.1 .6 9 2.5 .7 15 .16
Restrained Eating* 2.8 .2 9 2.1 .5 17 .00

*Thisdifference isfoundbydefinitionof "latentobesity"asa resultof the

classification

DISCUSSION

"RestrainedEating"wasnot foundtobe related toBMI.No relationshipwas

foundbetween "Restrained Eating"andeither "External Eating/Perceived Hunger"
or "Emotional Eating",which isnot inaccordancewiththe findingsofHerman
 -79-

and coworkers.They found thataheighteneddegreeofemotionality and

externalitywascharacteristic ofdieters,butnot ofnon-dieters,and
hypothesized that restrained eatingelicitsexternalityoremotionality (Herman
andPolivy, 1975;Herman, Polivy, Pliner,Threlkeld andMunie, 1978;Polivy,
Herman andWarsh, 1978).
Recently, ithasbeen shownbymeansof factor analysis that the Herman
Restraint Scale canbe reduced toat least twounderlying factors reflecting
fluctuations inbodyweight, andactual concernwithdieting (e.g., Drewnowsky,
Riskey andDesor, 1982;Ruderman,1983;Blanchard and Frost, 1983).Thus,in
retrospect, the relationship foundbyHerman and coworkersbetween restrained
eatingandexternality, andbetween restrained eating and emotionality may have
been caused by thedualeffect ofaweighthistory component anda dieting
component intheir scale,because ahigh scoreon restrained eating canbe the
result of ahigh subscore oneither of these twodimensions.
Incontrast toHerman's scale,the "Restrained Eating" scaleused inthe
present study contained only statementsaboutdieting andweight consciousness.
Theabsence ofa significant relationship between "Restrained Eating" and either
"Emotional Eating" or "External Eating/Perceived Hunger" inthis study isnot in
agreementwith thehypothesis ofHermanand Polivy, thatdieting isthe causal
agent of increased externalityor emotionality.
No significant differenceswere foundbetween "obese"and "latent obese"
women inthedegree of "Emotional Eating" and "External Eating/Perceived
Hunger".At the timeofassessment, the "obese"womenwereoverweight, and the
"latent obese"womenwere ofnormalweight buthadbeenoverweight at least once
intheir livesandwerehighly restrained eaters.Thus itmaybe tentatively
concluded that these "obese"eatingpatterns arepotential causal factors for
overweight andalsofor restrained eating.Thisfindingdoesnotviolate the
theoretical standsof either externality theoryorpsychosomatic theory.
Significant relationshipswere foundbetween BMI and "External Eating/
Perceived Hunger",BMI and "Emotional Eating"andalsobetween the twoeating
behaviour components.Thisalso isnot incontradictionwitheither externality
theoryorpsychosomatic theory.As statedpreviouslybutondifferent grounds,a
significant relationship canbeexpected betweenemotional eating and external
eating.According toexternality theory, sucha relationship shouldbe found,
because ahighdegree of emotionality isconsidered tobeamanifestation of the
general trait ofexternality. Psychosomatic theoryalso requires sucha rela-
tionship becausehigharousal and foodmustbepresent forovereating tooccur.
 -80-

Inthepresent study, significant relationshipswerenot found between

"External Eating/PerceivedHunger"andpersonality scales reflecting aspects of
externality, for example, Rotter I-Escalesand socialapproval (SA) (Table4 ) .
Furthermore, significant relationshipswerenot foundbetween "External Eating/
Perceived Hunger"and personality scalesmeasuring aspectsofemotionality, for
example, emotional instability (GLTS-G), socialanxiety (SA)andworry (GLTS-O)
(seeTable 4 ) .Theabsence ofsignificant relationshipsbetween "External Eating
/ Perceived Hunger"andpersonalitymeasures ofexternality andemotionality is
not inaccordancewith thegeneralized stimulus-binding concept ofSchachter and
Rodin (1974).However, the resultsare inaccordancewith findings reported by
Nisbett andTeraoshok (1976)and IsbitskyandWhite (1981).Significant relation-
shipswere foundbetween "Emotional Eating"andanumber ofpersonality scales
indicating aspectsofemotional instability, forexample, low self-esteem
(NPV-ZW),emotional instability (GLTS-G),anxiety (NPV-IN)andworry (GLTS-O).
Thus insummary, theabsence ofa significant relationship between
"Restrained Eating"and either "Emotional Eating"or "External Eating/Perceived
Hunger" inthepresent study isnot inaccordancewith the theoryof restrained
eating, thatdieting results inincreased externality oremotionality. The
presentdata suggest that "External Eating/PerceivedHunger"and "Emotional
Eating" arepotential causal factors foroverweight andalso for restrained
eating,which isinlinewith both thepsychosomatic theoryand the externality
theory.BMIwas found tobe significantly related toboth "External
Eating/PerceivedHunger"and "Emotional Eating". Inaddition these two eating
behaviour components werenot found tobe significantly different in "obese"and
"latent obese"women.However,onthebasisof the relationships between eachof
these twoeating behaviour components and thepattern of relationships between
these and personality scales,thepresent authorsaremore inclined to stress
the importance of "Emotional Eating"asacausal factor of overweight.
Therefore,of the threeeating behaviour components studied, "Emotional Eating"
seems tobe themostpromising for abetterunderstanding of causal factorsof
overweight, and thepresent datapoint topsychosomatic theoryasbeing most
valid.
Thepersonality traitsshownby "latentobese"and "obese"women may
indicatewhich traitsare related toahighdegree of restrained eating and
successfulweight loss.Bothgroupscanbe considered tohavebeen confronted
withweight problems.Bydefinition the "latentobese"hadanormal BMIasa
result of subjecting themselves toa severeweight reduction regime.As compared
with the "obese"women, they indicated thatthey functionedmoreeffectively in
 -81-

social situations andweremoreoutgoing inattitude;andalso they demonstrated

a higher degree of self-esteem and stronger self-control.
Thehigher degree of internal control bythe "latent obese" is inlinewith
findings inother studies (BalchandRoss,1975;Weiss,1977;Stuart andGuire,
1978; Salzer, 1982;Hartigan,Baker-Strauch andMorris, 1982). It isdifficult
todeterminewhether thedifference inotherpersonality traitsbetween these
twogroups,precedesor followsweight loss.Itmaywell be,thatawoman who
successfully reducesherweightno longeranticipatesnegative evaluation ofher
appearance byothers,and consequently, feelsmore atease insocial situations.
On theother hand, ifawoman ismore outgoing inattitude,and isstriving for
acceptance byothers,shemaywell continuewitha strictweight reduction
programme.
Karpovitz andZeis (1975)have suggested thatan improvement inmental and
emotional stateprecedesweight loss.They found that individualswho refused to
takepart inaweight reductionprogramme hadmoreemotional problems than those
whoparticipated; "... thenonparticipating subjectswereusing somuch energy
tocopewith themanyotherverytroublesomeproblems,thattheydidnotwant to
work onweight control" (Karpovitzand Zeis, 1975).
Thus, inatherapeutic context, itwould seemuseful toassess the
structure ofan individual's eatingbehaviour beforedeciding on the typeof
weight reductionprogramme tobe recommended. For example, foran individual
demonstrating ahighdegreeofemotional eating, itislikelytobemore
productive tocounteract feelingsof inadequacy, emotional instability, and low
self-esteem, beforeprescribing a strictweight reduction programme.
 -82-

CHAPTER11

EATINGBEHAVIOURANDSEX-ROLEORIENTATIONINWOMEN

TatjanavanStrienwithGerardP.A.Bergers

Thehypothesisthatthedegreetowhichparticular typesofeatingbehaviour is
relatedtosex-roleorientationwastestedinastudyonwomen.Threetypesof
eatingbehaviouraredistinguished:emotionaleating,eating inresponseto
arousal statessuchasanger,fearoranxiety (e.g.,Slochower,1983) ;
externaleating,eatinginresponsetofoodrelated stimuli regardlessof
internal feelingsofhungerorsatiety (SchachterandRodin,1974)and
restrainedeating,theconsciousattempttocontrolbodyweightbycognitive
restrictionoffoodintake (e.g.,HermanandPolivy,1980 a ).

Eatingbehaviour andsex-role orientation

Variousstudiesoneatingbehaviourhaveconsideredthepossibleeffectof
adherence tothefemalestereotypeonvarioustypesofeatingbehaviour.Leon
and Finn (1984)notedwith respecttoemotionaleating,thatthefemale
sex-role stereotype ofdependency,passivityandlackofassertiveness "can
functionto reinforcenon-effective responsestointerpersonal situations
requiringassertivebehaviours" (p.328).Ameansofcopingwith these
situationsand feelingsofangeroranxietymaybeconsumptionoffood. "Given
particular learningexperiencesinrelationtofood....somewomenwill overeat
andgainweightaspartofcopingwithvariousstressfuleventsintheirlife"
(p.330).HallandHavassy (1981)statedwith respecttoexternaleating that
thecentral roleof foodintraditionalwomen'sworkmayleadtoapermanent
seductiontoeat,especially forthosewomenwhoeatinresponsetoexternal
foodcuesratherthaninresponsetointernalphysiological states.Also,the
socialexpectationthatwomenshouldbesensitivetotheneedsofothersmay
result inrefusal tostoppurchasingtempting foodsonthegroundsthatthe
restofthefamilyenjoysthem.
 -83-

Thus,nurturance, lack ofassectivenessand lack ofadaptive coping

behaviour associated with female stereotype traitsare thought toplaya
central role inthegeneration ofemotionaleatingandexternal eating.
Opposing hypotheses havebeen formulated with respect tothe relationship
between adherence tofemale stereotype traitsand restrained eating (Rodin,
Silberstein and Striegel-Moore, 1985).Attempts tolooseweight and restrict
food intakemaybeassociated with female stereotype traits,because women
adhering tothe female sex-role status, referred toas sex-typed women,were
found tohave lower body satisfaction (Kimlicka,CrossandTarnai,1983)and to
showmore conformity ininterpersonal situations (Bern,1975;Brehony and
Geller, 1981)thannon-sex typedwomendid.Thismayextend toconformity to
socialpressure for female slenderness,because according toa study of Playboy
centre-folds andHissAmerica Pageant contestants,beauty hasbecome equated
increasinglywith slimness (Garner,Garfinkel,Schwartz andThompson, 1980).
Traditionally beautyhasbeen thecentral asset ofwomen toobtain upward
socialmobility bymarriage (HaccobyandJacklin, 1974). Therefore, particular-
lysex-typed womenmaydesire tobe slimbecausemost of themdonot pursue
high statuscareers or financial independence (Clareyand Sanford, 1982).
Non-sex typedwomenmayalso strife for slimness.Fallon and Rozin (1985)
suggest that thepursuit of thinness bywomenmaynot simply derive froman
attempt topleasemen. Intheir study they found thatwomenwanted tobe
thinner than they thoughtmenwant them tobe,andeven their estimate ofmen's
ideal female shapewas significantly slimmer thanmen'sactual preferences.
Also,womenwhohaveperfectionistic standards andhigh expectations for
personal performance were found tohavea tendency tobemoredissatisfied with
their ownbodies than lessachievement-orientedwomen of identical weight
(Rodinand Striegel-Moore, 1984). Beck,Ward-Hull andMcLear (1976)found that
womenwhovalued non-traditional rolesand greater options forwomen preferred
a smaller thinner femalebody, andassociated amore ample female body with
"wifeandmother".This finding isinlinewith theobservation of Chemin
(1981)thatbyemulating theangular male figure,women symbolically attempt to
secure social rightsthatare traditionally viewed asbeingmale (thisbeing
analogouswithGeorge Sandwearingmaledress inorder togainacceptance in
male society intheearly 1800's)and so reject earlier female formsand their
associated roles.
Insummary, itisproposed thatemotional eating and external eating are
associatedwith female stereotype traits.Opposing hypotheses have been
formulated about restrained eating.Although each seems tobe intuitively
 -84-

plausible,atpresent neither hasbeen testedempirically (note 1 ) * .Thus, in

thepresent study, the relationships between eachofthe three typesof eating
behaviour and sex-roleorientationwas investigated.

Sex-role orientation, adjustmentand eating behaviour

Bern(1974)introduced the concept ofpsychological androgyny inreaction tothe
conventional assumption thathealthy functioning impliesmasculinity formen
and femininity forwomen.Shesuggested that it ismore advantageous for an
individual tohave a relative balance of sex-typed characteristics than to
adhere rigidlyto rolecharacteristics ofhisorherown sex,asthe former
condition results inamore flexiblebehavioural repertoire,higher self-esteem
andpositive self-evaluation.Based on theobservations ofSpence, Helmreich
and Stapp (1975), that subjects scoring lowonmasculinity and femininity are
lesswell-adjusted thanthose scoringhighonbothdimensions, theone
dimensional balancemodel ofandrogynywas replaced by the two-dimensional
masculinity-femininity model (Bern,1977). Inthismodel, subjects are
characterized asandrogynous iftheyhaveahighdegree ofbothmasculinity and
femininity, but asundifferentiated ifthey score lowonbothdimensions.
Subjects scoring highoneithermasculinity or femininity arecharacterized as
masculine or feminine sex-typed, respectively.
The androgynous sex rolehasbeenshowntobeassociatedwith flexible
behavioural repertoire (Bern,1975;Bernand Lenney, 1976);high self-esteem
(Spenceet al., 1975;Bern,1975);adjustment (Bern,1975;Deutschand Gilbert,
1976; Heilbrun, 1976); andpositive self-statements (Kelly,Caudill,Hathorn
andO'Brien, 1977;Wiggins andHolzmuller, 1978). However,various studieshave
failed todifferentiate consistently androgynous andmasculine-typed females
(e.g.,Jones,Chernovetz andHansson, 1978;Bernard, 1980).According toKelly
andWorell (1977), thislatter finding suggests thatadjustment is related
mainly tomasculine-typed characteristics.
A similar observationwasalsomadebyTaylorand Hall (1982)who
re-examinedevidence ofandrogynyusing the two-wayanalysis ofvariance
(ANOVA)model.The two-dimensional masculinity-femininity androgyny is tested
by examining themaineffects ofmasculinity and femininity onthe relevant
dependent variable,and thebalancemodel istestedbyexamining themasculi-
nity-femininity interaction term.No support hasbeenobtained foreither model
ofandrogyny, inaddition,masculinityhasbeen shown tobemore highly related

* seepage 97 for notes

 -85-

tomeasuresofpsychological health than femininity has (note2 ) .

Further support forapositive relationshipbetweenendorsement of female
stereotype traits,andemotional eating andexternal eating isthe observation,
that femininity islessadvantageous forpsychological health than masculinity
is. Inanearlier study (VanStrien, Frijters,Roosen,Knuiman-Hijland
Defares, inpress)lowself-esteem andemotional instabilitywere shown tobe
associated withemotional eating, andemotional eating associated with external
eating.Thus,lowadjustment,which iscommon toboth femininity and the two
typesofeatingbehaviours,maybethe reason for the interrelationship between
thesevariables.
Thepresent study isapsychometric studyon the relationship between the
adherence ofwomen tosex-role stereotypes, their eating behaviour, and their
anxiety, and self-concept. Themainhypothesis investigated isthat emotional
eating and external eatingarepositively associatedwith female stereotype
traitsbutnotwithmasculine stereotype traits.Because itwasexpected that
thisobservation isdue tolowadjustment associatedwith female stereotype
traits, itwaspredicted that the relationships betweenbothtypesof eating
behaviour and feminine stereotype traitswould be reducedwhen scores for the
scales for anxietyand self-conceptwerepartialed out.
No specific hypotheseswere formulated for theeffectsof sex-typed
characteristics on restrained eating, inaddition, itwasleftasanopen
empirical questionwhether eatingbehaviour,anxietyandendorsement of
negative self-statements areassociatedwithandrogyny, conceived of aseither
the two-dimensionalmodel ofmasculinity and femininity, orasbalance of
masculinity and femininity.

METHOD

Subjects and procedure

Subjectswere 540 female participants inanongoing longitudinal studyon
overweight carried out intheMunicipality ofEde,theNetherlands, inthree age
groups (20-21,25-27and 30-32yearsasat 1January 1981). Details of themuni-
cipality, the studypopulation and theprocedure followed have been described
elsewhere (Baecke,Burema, Frijters,Hautvast andVanderWiel-Wetzels,1983;
Van Strien, 1985).All participants inthis study completed theDutch Eating
Behaviour Questionnaire (DEBQ;VanStrien, Frijters,BergersandDefares, in
press)with scalesonemotional eating, external eating and restrained eating; a
Dutchversionof the trait anxiety scaleof theSpielberger StateTraitAnxiety
 -86-

Inventory (STAI;Spielbecger, GorsucheindLushene, 1970;Vander Ploeg, Defares

and Spielberger, 1980,1981);theself-concept scaleof theDutch Self-Partner
Scale (Preventie-project, 1976); and theGroningerAndrogyny Scale (GRAS;De
Graaf, 1984),which isameasureof sex-role identification.
The first threemeasureswere administered ononedate of the longitudinal
study (Autumn, 1983); the last sixmonths later onanother assessment date of
the same study (Spring, 1984).

Instruments
GroningerAndrogyny Scale (GRAS).
As adescription of theGroningerAndrogyny Scale (GRAS) ispublished only in
theDutch language, this scale isdiscussed indetail.
Both theBSRIand theGRAShavebeendesigned tomeasuremasculinity and
femininity separately. However, theGRASdiffers fromtheBSRI intwoways (De
Graaf, 1984). Firstly, themasculinity (M)and femininity (F)scalesof the BSRI
assess onlypositively valued personality characteristics,while on the GRAS
these scalesalso containnegativelyvalued personality characteristics. This
hasbeendonebecause a scale containing exclusively positivelyvalued persona-
lity characteristics maynot take intoaccount apotentially important component
of sex roles, that is,the extent towhichnegativelyvalued sex-correlated
traits formpart ofan individuals stereotype ofmasculinity and femininity
(Kellyandworell, 1977;DeGraaf, 1984).
Secondly, theBSRI itemswere selected froman initial itempool of persona-
lity characteristics that "seemed ...positive invalue and eithermasculine or
feminine intone" (Bern,1974,p.156).Potential itemswere rated by university
students,whoevaluated thedesirability ofeach characteristic foreither an
American man orwoman.The final itemson theBSRI scalewere those characteris-
ticswhich bothmale and female judgesconsidered tobe significantly more
desirable for females than formales (masculinity scale items), and significant-
lymore desirable for females than formales (femininity scale items). The items
on theGRASwere selected from the initialpool ofvirtually allDutch persona-
lity characteristics.University students rated themselves and theirpartner on
these items.Those items selected bymalesand femalesasbeingused signifi-
cantlymore for females than formales,were considered tobe feminine persona-
lity characteristics, and similarly those itemsused significantly more for
males than femaleswere considered tobemasculine personality characteristics.
Thus, the initial itempool of theGRAS contained notonly stereotype characte-
ristics,but all personality characteristics. Further, the final stereotype
 -87-

traitswerenotderived from ratingsofthehypothetical "Americanmanorwoman"

but fromactual selfandpartnerratings.
TheGRAS contains29masculine,29feminineand24neutral (e.g.,social
desirability)personalityadjectives.Inasampleof 500studentsof theUniver-
sityofGroningen (304menand 196women),Cronbach'salphacoefficientsappea-
redsatisfactory forthemasculineand feminine scale (Mscale:alpha- .79;F
scale:alpha- .86),butnot fortheneutral scales (positive socialdesirabili-
tyscale:alpha- .60;negative socialdesirability scale:alpha- .67).In
addition,a relationshipwas foundbetween thesescalesinmen (r--.29;p<
.01), butnot inwomen (r-0.3;NS).Inthepresent sampleofwomen, similar
Cronbach'salphavalueswereobtained (Mscale:alpha- .83;Fscale:alpha-
.83;positive socialdesirability scale:alpha- .63;negative socialdesirabi-
lity scale:alpha- .74), but incontrast tothesampleofwomen studentsin
Groningen, the relationshipbetweenthemasculineandthe feminine scalesappea-
redtobe significant (r- .14;p < .05).Inbothsamples,orthogonal factor
analysesofall82items revealed similar structuresof "best interprétable"
solutionsof four factors,and itemsofthemasculineand feminine scalesdonot
loadon seperate factors (DeGraaf,1984;note 3 ) .Thisobservation issimilar
tofactoranalysesoftheMand theFscalesoftheBSRI (e.g.,Bernard,1980;
Gaudreau, 1977;PedhazurandTetenbaum, 1979),and indicatesthatmasculinity
and feminityarenot independentsubsetoftraits,butaremultidimensional
constructs.
A respondent indicatesona fourpoint (1-4)bipolar scalehowwelleachof
the82characteristicsdescribeshimorherself (e.g.,notatall independent/
very independent;notatall romanticAery romantic).Scoresforeach scaleare
obtainedbydividing thesumofscoresoneachscalebythetotalnumber of
itemsendorsed, sothateachscalehasapossible score rangeof1-4.Inthe
present study,neutral itemswerenot consideredbecauseoftheir low
reliability.

DutchEatingBehaviourQuestionnaire.ThisQuestionnaire (DEBQ)contains33
items;tenon restrained eating (e.g.,doyoutrytoeatlessatmealtimes than
youwould liketoeat;doyou takeintoaccountyouweightwithwhatyoueat
(note 4 ) ;tenonexternaleating (e.g., iffood smellsand looksgood,doyou
eatmore thanusual;ifyouwalkpast thebakerydoyouhave thedesire tobuy
somethingdelicious;and 13onemotionaleating (e.g.,doyouhaveadesire to
eatwhenyouare irritated;doyouhaveadesire toeatwhenyouhavenothing to
do).
 -88-

Eachofthethreescaleshasbeenshewntohave internalconsistencyand
factorialvalidity (VanStrienetal., inpress).A respondent indicatesona
five-pointLikertscale (never (1) ...veryoften (5))howoftenheorshe
resortstoeachofthe33typesofeatingbehaviour.Scoresforeachscaleare
obtainedbydividing thesumofitemscoresbythetotalnumberof items
endorsed,ahighscoreindicatingahighdegreeofeatingbehaviour inquestion.
Thuseachscalehasapossiblescore rangeof1-5.

Spielberger StateTraitAnxietyInventory.Theoriginal inventory,theDutch

versionoftheSpielbergerStateTraitAnxiety Inventory (STAI,Spielberger et
al,, 1970),whichwasdevelopedbyVanderPloegetal.(1980)contains separate
scalesforstateandtraitanxiety.Inthepresentstudy,onlythetraitanxiety
scalewasused,becauseonlyanxietyasdispositionwasconsidered tobe
relevantforthepresenthypothesis.
Thetrai*-.anxietyscalecontains20items,tenon "anxietypresent"andten
on "anxietyabsent".ValidationstudiesoftheSTAI indicatethatscoresonthe
traitanxiety scale remainstable invarioussituations,forexample inbotha
lectureandanexam (VanderPloeg,1979;VanderPloeg,Defaresand
Spielberger, 1979). Thisindicatesthatthetraitanxiety scalemeasuresa
stabledisposition.InasampleofUniversitystudentsoftheUniversityof
Leiden (205menand202women)theCronbach'salphacoefficientofthetrait
anxiety scalewas .90andinthepresentsampleofwomen inhabitantsofEdeit
was.85.
A respondent indicatesonafour-pointbipolar scalehowwelleachofthe
characteristicsdescribehimorherself (e.g.,Ifeelsecureveryseldom/mostof
thetime).A scoreforthisscaleisobtainedfromthesumofscoresonthe
"anxietypresent"andthe recededscoreonthe "anxietyabsent"items,sothata
highscore indicatesahighdegreeofanxiety.Thepossible score rangeis
20-80.

DutchSelf-Partner Scale.Thisscale (PreventieProject,1976)measures the

personalityoftherespondentasratedbytherespondenthimselfandbyhisor
herpartner.Inthepresentstudy,onlyself ratingswereused,asmeasured by
theself-concept scaleoftheDutchSelf-Partner scale.Thisscalecontainstwo
subscalesoftenitemseach,measuringpositive?self-conceptandnegative
self-concept respectively. Inthepresentsampleofwomen inhabitantsofEde,
Cronbach'salphacoefficientsofthesescaleswere .87and .88respectively.
A respondent indicatesonafour-pointbipolar scalehowwell eachofthe
 -89-

characteristicsdescribeshimorherself (e.g., Ithink Iama firmperson/nota

firmpersonatall.Ithink Iama restlessperson/not a restlesspersonat
all). Scores for thepositive and for thenegative self-concept scalesare
obtained bydividing the sumofscoresonthat scalebythetotalnumber of
itemsendorsed.The score rangeofeach scale is1-4,withahigh score
indicating ahighdegree ofnegative self-esteem.

Analyses
Product-moment correlation coefficientswerecalculatedbetween themasculinity
and femininity scales,the three eatingbehaviour scalesand scalesof anxiety
and self-concept. Inmoredetailed analysesof the relationshipsbetween eating
behaviour, anxietyand self-concept andandrogyny,ahierarchical multiple
regressionwasused foreachof thedependent variables,aswasemployed by
Feather (1984). Inthisanalysis,thescoresforeach subject on themasculinity
and the femininity scaleswere entered asindependentvariables inthe
regression equation,and theireffectswereassessed.Subsequently, these
variables togetherwithan interaction termwereentered and theeffectof the
interaction termassessed (note5 ) .
This analysispermits thetwo-dimensionalmasculinity-femininity (or
additive)model ofandrogyny tobe tested byexamining themaineffectsof
masculinity and femininityby inspecting theassociatedbètacoefficients inthe
first stepof theanalysis.Thisanalysisalsoallowsthebalancemodel tobe
tested byexamining thecontribution of theinteraction termofmasculinityand
femininity (note6)totheexplainedvarianceofthe regressionequation,by
examiningwhether theexplainedvariance ofthe regressionequation showsa
significant increase after inclusionof theinteraction terminthe second step
of theanalysis.However, the interaction termwasnotdetermined asa simple
product ofMx F,because thecorrelationbetweenMx FandM and Fwouldbe so
large thatmulticollinearitywouldemerge inthecorrelationmatrix, thusgiving
risetoestimationproblems including theweightsb., b,andb, (Winnubst,
Marcelissenand Kleber, 1982,p.478).instead,the interaction termwas
determined using themethodemployed byWinnubst etal. (1982)asfollows:

a+b.M+b,F+bjfM-Cj^) (F-c,), where:

M ismasculinity
 -90-

F is femininity
y isthedependent variable.
The constant factorsarechosen insuchaway that the interaction term
(H-c,)(F-c,)is independent ofM and F.The constant factorsc.andc,are the
regressionweightsofM and F,determined ina separate regression analysis with
theproduct ofM and Fasdependent variable.
For thisconservative regressionmethod p < .10was selected as the minimum
level of significance,which isinlinewithWinnubst etal. (1982).

RESULTS

Total sample
Relationships between thescales.ThePearson correlation coefficients showing
the relationshipsbetween themasculinity and femininity scales,the three
eatingbehaviour scalesand thescales foranxietyand self-concept are
presented inTable 1forthetotal sample.Themeansand standarddeviations of
these scalesarealso presented.
Both emotional eating andexternaleatingwere found tobe related to
feminine stereotype traitsbutnot tomasculine stereotype traits;and
restrained eating tobepositively related toboth feminine and masculine
stereotype traits.Emotional eatingwas found tobepositively related to
external eating, andboth typesofeatingbehaviourwerepositively related to
anxietyand negative self-concept andnegatively related topositive
self-concept. Further,endorsement of feminine stereotype traitswas positively
related toanxietyandnegative self-concept, andnegatively related topositive
self-concept.Whileendorsement ofmasculine stereotype traitswas positively
related topositive self-concept andnegatively related toanxietyand negative
self-concept.

Two subsamples
As correlation coefficientsofthe total sampleweregenerally low, the total
samplewasdivided randomly intotwo independent subsamplesand regression
analyseswerecarried out inthesesubsamples.Sample 1consisted of260women,
and sample 2,whichwasused tocross-validate sample 1,of280women.
 -91-

<*•> ÏN OJ

R. S

O r-l UI

>

V 0) c at
H r-t 0 C
h
•H
tfl >% > >
4J •H
•U
•H
a ca •H U 0} 4-t
> o 4J (U 4J •H •H 4 V V
o ca O 4J CO X 09 00

S •s
O
JS a o.
S ta
ä CU e
CS
0)
H CN en m \o r*. co S
*»
 -92-

Androqyny:theadditivemodel.Thebètacoefficients associatedwith the

main effectsofmasculinity and femininity (step1)onthe three typesof eating
behaviour,anxiety and self-concept arepresented inTable2.
Insample 1,therewasno support for theadditivemodel ofandrogyny, as
all regression equations revealed either amaineffect of femininity inthe case
ofemotional,external and restrained eatingbehaviour,ormaineffects opposite
indirection, ofbothmasculinity and femininity inthecase of anxiety,
positive self-concept andnegative self-concept. In sample 2,all resultswere
cross-validated, except thoseon restrained eating.Inthecaseof restrained
eating, some support for theadditivemodel ofandrogynywasobtained because
thebètacoefficients formasculinityand femininitywereboth inthe same
direction and significant inthis sample.
Androgyny: thebalancemodel.The interactioneffectsofmasculinity and
femininity (step2)associatedwith thethreetypesofeatingbehaviour, anxiety
and self-concept are shown inTable 2.insample 1,a significant increase in
explainedvariancewasobtained after inclusion of the interaction term of
masculinity and femininity foremotionaleating,positive self-concept and
negative self-concept.The interaction term inthe regression equations for
emotional eating and negative self-conceptwasnegativeand forpositive
self-concept, itwaspositive.This indicates thatwomenhaving abalance of
masculineand feminine stereotype traitshavea lower tendency toeat in
response tonegative emotions,andhave ahigher positive anda lower negative
self-concept.However, these resultswerenot cross-validated insample 2,as
none ofthe regression equations revealed a significant increase inexplained
variance after inclusion ofthe interaction termofmasculinity and femininity.
Anxietyand self-concept asmoderatingvariables.The regressionanalyses
indicated thatmasculinity and femininity (although the latterwas significant)
donot account fora substantial amountofvariance inthe caseofthe three
types ofeating behaviour,but thesevariablesexplaina substantial amount of
variance inthe case ofanxietyand self-concept.Thus the significant positive
contributionof femininitytoemotionalandexternaleatingmaybeduemainly to
anxietyandnegative self-concept associatedwith female stereotypes traits.To
investigate this,anxiety, positive andnegative self-conceptswere entered ina
stepwise incremental fashion inthe regressionequationwhichalready contained
masculinity, femininityand the interaction termas independentvariables.The
bètacoefficients of these regression equationsaregiven inTable 3.
 -93-

* «
«*
CN
CO
*«0 0
O")
-d-

c
« o
r r

4-1 CO <ü
CO ÖO t»
0) S
"^^^^ c >*
4J
Ui
01 •o o
rH a X cü
CO
co
O
5 «
o
r

10 oo
n e'
4J «H

o
f
S

B «O
U C
0) -H

o
r

<0
S
•2 2

0) <u
r-f
CO fa
O
CO
«
O
CO X X
B
•H
S fa S Ni
BS
 -94-

o o o o

m I-H o u*t oo N N 00
O i-H rH o o

3
r\ O
<U *H

co oj
Pu
^^ 00

>. -5

*£> r>* vo

ÖO
c
•H « n N co u"i Ov C7» <N

e C
O
y-i •H
o +J
O
B
e

>% CU >
•U •H

•H
>
11 •H 4J
dl •H •H n)

5
-ao 00

JT* s
X
5
X CO
O
CU
00
CÜCNJ

PM
 -95-

Inbothsample1andcross-validation sample 2,femininityno longer

contributed significantly toemotional andexternal eatingwhen anxiety,
positive andnegative self-conceptwere included intheanalysis.This indicated
thatthecontribution of femininity toemotional and external eating behaviour
isduemainlytoanxietyandnegative self-concept associatedwith female
stereotype traits.The resultson restrained eatingobtained insample 1,were
notcross-validated insample 2 (Table3).
Finally, insample 1thecontribution ofmasculinity toemotional eating
became significant after inclusionof othervariables inthe regression
analysis,and the samewasthecase forexternal eating insample 2.Although
these findingswereunexpected, they canbeeasilyexplained, because anxiety
andnegative self-conceptwhichbothcontributed significantly andpositively to
emotional and external eating,werenegatively related tomasculinity. By
removing theirvariance inmasculinity, the contribution ofmasculinity to
emotional and external eatingbehaviour was increased.However, inbothcases,
the resultswerenot replicated intheother sample.

DISCUSSION

The resultsof thepresent studyonwomenareinlinewith theviewthat the

degree towhichparticular typesofeatingbehaviour areexhibited is related to
sex-role orientation.Resultsonemotional andexternal eatingbehaviour were
consistent, and indicated amaineffect of femininity, butnotof masculinity.
The resultson restrained eatingwereequivocal,asfindings inone sub-sample
werenot cross-validated intheother independent subsample.This inconsistency
makes itdifficult todeterminewhether restriction of food intake is
particularly prevalent insex-typedwomen or inwomenhavingbothmasculine and
feminine stereotype traits.However, the results indicate that the supposition
thatnon-sex typedwomen strife for slimness requiresmodification, as inboth
subsamples, no supportwasobtained for restrained eatingprevailing inthe
subgroupofmasculine sex-typedwomen,because thiswould imply amaineffect of
masculinity only. Inaddition, the lowvarianceexplained by masculinity,
femininity and theothervariables in restrained eatingbehaviour indicates that
other factorsnot considered inthepresent studymaybemore important for this
type ofeating behaviour.
Thepositive relationshipsbetweena female sex-roleorientation, and
emotional and external eatingbehaviour are inclose agreementwith the
suppositions of Leonand Finn (1984)andHall andHavassy (1981),and indicate
 -96-

again that femininity isdetrimental towell-being. Bothemotional and external

eatingbehaviour havebeen shown topromoteovereating, andareconsidered to
beprimary causal factors forweight gainand obesity (e.g., Slochower, 1983 a ;
Lowe and Fisher, 1983;Schachter andRodin, 1974), andasisknown, overweight
ishazardous tophysical andpsychological health.Forexample, inthepresent
sample ofyoungwomen (maximum ageattimeofassessment 31years) overweight
was shown tobeassociatedwithanumber ofpathological conditions, especially
hypertension, back and jointproblemsandheadache,and a largeproportion of
those overweight hasvisited their familydoctor oranother physician inthe
yearpreceding the study (Deurenberg,VanPoppel andHautvast, 1984). In
addition, public attitudes toward theoverweight inWestern societyare
generally extremelynegative (DeJong, 1980), and thismay lead tothe
generation of lowself-esteem, negativemood statesandpoor self-image inthe
obese (Allon, 1982).
Inthepresent study, femininity accounted foronlya smallamount of
totalvariance inbothemotional andexternal eating behaviour.The
contribution ofanxietyandnegative self-conceptwasmuchhigher toboth types
of eatingbehaviour. Further,high relationshipswere foundbetween femininity,
andanxietyandnegative self-concept.When thevarianceofanxiety and
negative self-concept in femininity scoreswaspartialed out, femininity no
longer contributed significantly toemotional andexternal eating.This
indicates that the contribution of femininity toboth typesofeating behaviour
isduemainly toanxiety andnegative self-concept associatedwith female
stereotype traits.
Inthepresent study, support for thebalancemodelofandrogynywas
obtained inone sample foremotional eating, andpositiveand negative
self-concepts,but these findingswerenot cross-validated inthe other
subsample.These resultsare inlinewithTaylor andHall (1982), who intheir
reviewalso found littleevidence ofbalance androgynybut instead founda
potent effect ofmasculinity. This isincontrastwithpresent findings,
becausemasculinityhadonlya strongpositive effect onpositive self-concept,
and the relationshipswithall othervariableswere found tobeweak ornot
present atall.A possible explanation for thesecontradictory results is,that
thepresent resultswereobtained fromcommunity residents,whereasmost
studies onpsychological androgyny arecarried outoncollege students.Inthe
predominantly protestant and conservative community of Ede,masculinity may
have lessobviousutility forwomen than inauniversity community. However,
 -97-

even inthispopulation, nosupportwasobtained for thenotion that femininity

contributes towellbeing inwomen.

NOTES

1. Inpopulations ofwomen the relationshipsbetween scalesofmasculinity and

femininity and theRestraint-Scale ofHermanand Polivy (1975)has been
studiedbyHawkins II,Turell andJackson (1983).However, this scalemay
notmeasuredietingper se,butdisinhibition of cognitive restraint. This
findingwasobtainedwith theThreeFactorEatingQuestionnaire (TFEQ),
whichmeasures separately cognitive restraint of food intake (FactorI ) ,
disinhibition ofcognitive restraint (Factor II)andHunger (Factor III)
(StunkardandMessick, 1985).TheHermanand PolivyRestraint Scalewas
foundnot tobe related to thecognitive restraint scaleof theTFEQ (Factor
I), but thisscalewashighly related tothedisinhibition scale ofthe TFEQ
(Factor II).Inaddition,not Factor I,but Factor IIoftheTFEQwas
related toovereating (Shrager,Wadden,Miller, Stunkardand Stellar, 1983),
which indicates that theexcessive food intake found insubjects scoring
highontheRestraint Scaleinthe studiesofHermanand Polivy (e.g.,
Hermanand Polivy, 1980 )maynotbe related to restraint per se,but to
disinhibition of restraint.
Also, theDiet subscale of theEatingAttitudes Test (Garner andGarfinkel,
1979),whichwasused inthe studyofHawkins IIet al. (1983)inonlyone
of thepopulationsmeasured amoreextreme (unhealthy)typeofdieting than
considered inthepresent study.
2.According toTaylorandHall (1982), inmany reportsthismasculinity effect
hasbeencamouflaged inthedominant methodsofdata-analysis of relying on
pairwise comparison ofthe four sex-role subgroupmeans.Many reports have
not recognized that theobservation, thatandrogynous andmasculine subjects
scoreequallyhigh,and indifferent and feminine subjectsequally low,
impliesamaineffect ofmasculinity.
3.Unpublished dataset.
4.This restrained eating scalewhichmeasuresa "healthy" typeofdieting, is
similar in itemcontent tothecognitive restraint scaleoftheTFEQ (see
note 1) (VanStrien, Frijters,BergersandDefares, inpress), andwas found
tohavegoodpredictivevalidity forrestrictionof food intake (VanStrien,
Frijters,Van Staveren,DefaresandDeurenberg, inpress).
5.As theeffectsateach level areadjusted by taking intoaccount
within-level effectsandprior level effects,this formof analysis
corresponds tothe "classic ...analysisofvariance" (Feather, 1984,p.615)
6.As stated byTaylor andHall (1982)"... foranANOVA tableofanysize,
interaction effectsare represented asdeviations fromzeroafter the grand
meanandmaineffectshavebeen subtracted from thecellmeans.The
requirement that suchdeviationsmust sumtozero ineachcolumn and row
implies that inthe two-by-twoANOVA table, interactionmust always take one
particular form: thetwodiagonalsmust beequal inmagnitudebut opposite
insign from thetwodeviations intheother diagonal" (p.349).Thus,a
significant interaction effect implies that subjects inthe balanced
diagonal of theANOVA Table (lowmasculine, lowfeminine;highmasculine,
high feminine)differwith regard tothedependentvariable from subjects in
the sex-typed diagonal (male sex-typed; female sex-typed).
 -98-

CHAPTER 12

LIFEEVENTS,EMOTIONALEATINGANDCHANGEINBODYMASS INDEX

TatjanavanStrienwithHattiA.Rookus,GerardP.A. Bergers,JanE.R. Frijters

andPeterB.Defares*.

It isacommonly heldbelief thatstressful lifeeventscanbe important

aetiological factors inhumanobesity.Asearlyas1949,Shorvanand Richardson
(1949)cited casesof individuals,most ofthemwomen,whobecameobeseafter
severepsychic trauma (forexample, thebombing ofLondonduringthe Second
WorldWar, thedeathofaparent, financial reverse).Other studieshave also
provided evidenceofobesityassociatedwith traumaticevents (Atkinsonand
Rinquette, 1967;Meyer andTuchelt-Gallwitz,1967;Nutzinger, 1979).
Stress,definedasan imbalancebetweenenvironmental loadand theability
of theindividual tocope,createshightenedarousaloractivation.As this
stateofarousal inhibitsthephysiological correlatesofhunger (Cannon,
1915), most individuals respond tosucharousalwithlossofappetiteand
subsequentweight loss.Incontrast,others respond tosucharousalwith
excessive intakeof food,ifthistypeofemotional eatingoccurs frequently,
itmayleadultimatelytoweightgainorobesity.
Thepresent studydescribes resultsofastudyontheinteractioneffect
ofnegative lifeeventsandemotionaleatingonchange inbodymassindex (BMI;
2 2
weight/height (kg/ta)).itispredicted that lowemotionaleatersgainless
weight andhighemotionaleatersgainmoreweightafterexperiencing negative
lifeevents,than thosewhohavenotexperienced suchevents.

*Thisarticle isinpressintheInternational JournalofObesityand is

printedherewithkindpermission ofJohnLibbey&Company, limited.
 -99-

C
ZD OO
Q)
BJDp JU3UJSS8SSD pU0D8S Vi IM T J
3 o^ 0)
«3
o <u
Vi
Vi (U ui 4-1
a nj. U
C
(U
to öO
<D 0) co
J_i )-i n) a>
a>
3 ^3U (UC &
tn
V}
O 01
0 J3
^o
xi

J3

co
ON

ajDp juaiussassD 4SJ^

1-1 C
cO QJ
e
o CU
•I-I
>
4J ai
0 M-l
B •i-t

• | = OO

V)
 -100-

METHOD

Overview
Ina longitudinal studyonoverweight, a scaleonemotional eating and a
questionnaire on lifeeventsexperienced during thepreceding sixmonthswere
administered on twomeasurement dates.Inaddition, bodyweight of subjects was
measured on these twomeasurement dates,andalso sixmonthsbefore and after
bothdates.Bodyheight hadbeenmeasured earlier inthe same study (Fig.1 ) .
On thebasisof their scoreon the scaleonemotional eating and the
questionnaire on lifeevents, subjectswere classified aseitherhighor low
emotional eaters,whohad experienced eithernonegative lifeeventsor more
than threenegative lifeevents.The interaction effectwasassessed during six
periods, the sixmonthspreceding the twoassessment dates,the six months
following these twodates,and 18monthsand twoyearsafter the firstassess-
ment date.Short-term effectswere analysed for the total sample. Long-term
effectswere analysed fora subgroupwhose emotional eating behaviour and life
events statushadnotaltered during thecourse of the study (seeFig.1 ) .

Subjects and procedure

The subjectswere participants ina longitudinal studyonoverweight carried
out inthemunicipality of Ede,TheNetherlands.Detailsof the study popula-
tionhavebeendescribed elsewhere (Baecke,Burema, Frijters,Hautvast andVan
derWiel-Wetzels, 1983;Van Strien, 1985). Inthepresent study, information
wasobtained from 589menand 619women inthreeagegroups (19to21,24to26
and 29 to31yearsasat 1January, 1980).Women,pregnant for threemonthsor
longer onat least oneof theassessment dateswere excluded fromanalysis.
On eachmeasurement date of the longitudinal study, subjectswere invited
bymail tocomplete aquestionnaire.Several days later, theywerevisited at
home bya trained assistant who collected thequestionnaire and checked itfor
completeness. Shealsomeasured bodyweightwithout shoesand jacket to the
nearest 0.5 kg.Bodyheight of the subjectshad beenmeasured atamobile
researchunit earlier inthe longitudinal study, soitwaspossible to assess
thebodymass index of the subjects.
On twomeasurement dates (Spring 1982andAutumn 1983) (note 1 ) * ,the
questionnaire contained theDutch EatingBehaviour Questionnaire (DEBQ) (Van
Strien, Frijters,BergersandDefares,inpress)including,a scaleon

* seepage 106 for notes

 -101-

emotionaleating (note 2 ) ,andaquestionnaireaboutlifeeventsexperienced

duringthepreceding sixmonths.As thesetwomeasurementdatesarecentralto
thepresentstudy,theyarehenceforth referredtoasthefirstandthesecond
assessmentdate.

Instruments
Thescaleforemotionaleating.Thisscalecontains11items (forexample,do
youhaveadesiretoeatwhenyouareirritated;doyouhaveadesiretoeat
whenyouhavenothingtodo).Thescalehasbeenshowntohavehigh internal
consistencyand factorialvalidity (VanStrienetal., inpress). A respondent
indicatesonfive-pointLikertscales (never (1)...veryoften (5))howoften
heor she resortstoeachofthe11typesofeatingbehaviour.A scoreforthe
scaleisobtainedbydividingthesumofitemscoresbythetotalnumberof
itemsendorsed,ahigh scoreindicatesahighdegreeofemotionaleating.Thus,
thepossible score rangeforthescaleisfrom1to5.
DutchLifeEventsQuestionnaire.Lifeeventswereassessedbymeansofa
DutchLifeEventsQuestionnaire (Lennings,1980),whichcontainsquestions
derived fromtheSocialReadjustmentRatingQuestionnaire (HolmesandRahe,
1967), theQuestionnaireofPaykelandUhlenhuth (PaykelandUhlenhuth, 1972),
theRecentLifeChangeQuestionnaire (Rahe,1975), theLifeExperience Survey
(Sarason,JohnsonandSiegel,1978)andtheLifeEventsInventory (Tennantand
Andrews, 1978). Onbothassessmentdates,thesubjectswereasked tomark
eventswhichhadoccurred inthesixmonthsprior totheassessmentdate.

Lifeevent index
Inthepresentstudy,onlynegativeeventsweretakenintoconsideration,be-
causeundesirable lifechangeshavebeenshowntobebetterpredictorsof
stress reactionsthanneutralordesirablelifechanges (Sarasonetal.,1978;
TennantandAndrews,1978;Gersten,Langner,EisenbergandOrzak,1974;Vinokur
andSelzer,1975;Mueller,EdwardsandYarvis,1977;RossandMirowskyII,
1979).
Anadditive indexbasedonacumulativescoreofall 47negativelife
eventswasused,because itisknownthatweighted indices,forexampleasde-
rived fromtheSocialReadjustmentRatingQuestionnaire (HolmesandRahe, 1967),
donotpredictdependentvariablesanybetterthanindicesbasedonsimple
counts (Muelleretal., 1977;RossandMirowsky,1979;Lorrimor,Justice,McBee
andWeinman,1979;LeiandSkinner,1980;Zimmerman,1983). The classification
 -102-

of lifeeventsasbeingnegativewasmade independently bythe firstand the

third author,andwas found tobe inclose agreement.

RESULTS

Classification of the subjects

Analyses formenandwomenwere carried out separately. Subjectswere
classified asbeingeither lowemotional orhighemotional eatersonthe basis
of amedian split of scoreson theemotional eating scale,administered onthe
first assessment date.Scores ranged from 1.00 to4.17 formenand from 1.00 to
4.50 forwomen.Cutoffpoints for lowandhigh emotional eatingwere 1.9 and
2.1 formen andwomen respectively.
Subjectswereplaced inthe "nonegative lifeevents"category, if they
had notexperienced any sucheventduring theassessment-period. Theywere
classified in thenegative lifeevents category iftheyhad experienced three
ormore negative lifeevents,corresponding totheupper quartile of
distribution ofvaluesof thenegative lifeevents index.Subjects inthe
middle twoquartileswere excluded fromanalysis.
The final result consistsofa cross-classification tableof the following
four categories:lowemotional eater/no negative lifeevents;low emotional
eater/negative lifeevents;high emotional eater/nonegative lifeevents;and
high emotional eater/negative lifeevents.

Short-term effects ofnegative lifeeventson change inbodymass inlowand

high emotional eaters
A two-wayanalysis ofvariancewas carried outover the six-months periods
preceding and following the twoassessment dates (classicapproachwith unequal
cell frequencies;Nie,Hull,Jenkins,Steinbrenner andBent, 1970;note 3 ) .It
waspredicted that theeffect ofnegative lifeeventsonchange inBMIwould be
larger inhighemotional eaters than inlowemotional eaters.Thus inthe
paper, the resultsareconsidered intermsofwhether theyare inagreement
with this supposition andwhether these interaction effects are significant.
Main effects arenot reported astheyarenotof relevance tothe present
hypothesis.
The resultswere not aspredicted inboth sexes intheperiods preceding
the first and second assessment dates,andno significant interaction effects
were obtained.
 -103-

Inthe sixmonths subsequent tothefirstassessmentdate, resultswere as

predicted inthesample ofmen (Table 1 ) .Lowemotionaleatersgained less
weight andhighemotional eatersgainedmoreweightafterexperiencing negative
lifeevents,thanthosewhohadnotexperienced suchevents.Ingeneral,there
wasasignificant interaction effectofemotionaleatingandnegative life
eventsonchange inBMI (F (1,326)-6.43; p-.Ol).Thisinteraction remained
significant afteradjustment forBMI,asobservedonthefirstassessmentdate
(F (1,325)-6.59; p-.Ol). However,intheperiodafter thesecond assessment
date, thisinteractioneffectwasnot replicated (F (1,390)- .43;notsignifi-
cant (ns)). Forwomen,nointeractioneffectswereobtained inthe sameperiods
(F (1,377)- .54;ns)andF (1,455)- .01;ns,respectively) (Table1 ) .

Table 1.Change inbodymass indexsixmonthsafterthe firstassessment date

(positive-weight gain;negative-weight loss,seefootnote 1)

Negative liveevents
Hen Women
Nonegative >3negative Nonegative >3negative
lifeevents lifeevents lifeevents lifeevents

Lowemotional eater
mean .01 -.19 -.11 .05
sd .79 .70 .74 .98
n 83 84 89 75

Highemotional eater
mean -.15 .07 .19 .21
sd .73 .72 .80 .73
n 63 100 55 162

Long-termeffectsof negative life eventsonchange inBMI inlowandhigh

emotional eaters
It isnot clearwhynointeraction effectswere found formenon the second
assessmentdate.Todeterminewhether the resultsof thefirstassessment date
couldbeattributable tochancevariationor togenuineeffects,long-term
effectsofnegative lifeeventsonchange inBMI inlowandhighemotional
eaterswere studied inthe18monthsand twoyears following the first
assessmentdate inasubsample of subjects classified inthesamecategoryof
emotionaleating/negative lifeeventsonbothassessmentdates.
Thedifferential long-term effectsareshown inTable2.Inthemen,
similar resultswere found intheperiodsof18monthsandtwoyearsafter the
 -104-

first assessment date.These resultswereaspredicted and similar to short-

termeffectsobtained onthefirstassessment date.

Table 2.Changes inbodymass index 18monthsand 2yearsafter the first

assessmentdate*

Men Women
Periodafter first No negative > 3negative No negative >3negative
assessment date li feevents life events li feevents life events

18months

Lowemotional eater
mean .24 - .08 - .10 .06
sd 1.00 .96 1.11 1.14
n 25 25 30 44

Highemotional eater
mean - .22 .24 - .32 .40
sd 1.15 1.20 .66 1.23
n 16 51 16 102

2 years

Lowemotional eater
mean .51 .13 .15 .23
sd 1.25 1.04 1.03 1.50
n 25 25 30 44

Highemotional eater
mean - .06 .60 - .01 .66
sd 1.04 1.19 .99 1.54
n 16 51 16 102

* Ina subsample ofmenandwomenwhohad thesameemotional eating/negative

lifeeventsclassification onboththe firstand secondassessment date.

Ingeneral,these interactioneffectswereborderline significant for the

18-monthsperiod after the first assessment date,and significant for the two-
year periodafter thisdate (F (1,113)- 3.16;p-.08)and F (1,113)- 4.98;
p-,03). Theeffects remained the sameafter adjustment forBMIasobserved on
the firstassessment date (F (1,112)- 3.21;p-.08andF (1,112)= 4.70;p-.03
respectively).
Forwomen, similar interactioneffectswerenot found (F (1,188)- 1.71;
ns, and F (1,188)=1.31;ns),although the resultswereaspredicted for high
emotional eaters (Table2 ) .
 -105-

DISCUSSION

A significant interaction effectwas found inthesixmonths following the

first assessment date formen.This suggests that the impact ofnegative life
eventsbecomesmanifest after acertaindelayor latentperiod.Although this
interaction effectwasnot foundon the secondassessment date,the long term
effects suggest that the significant effectobtained on the first assessment
date isprobably attributable toagenuine effect andnot tochance variation.
The effect twoyearsafter the first assessment datewas stronger than
thatobtained 18months after thisassessment date.This indicates that the
short-term effects ofnegative lifeeventsasexperienced on the second assess-
ment date strengthen the long-term effects,although theydonothavea signi-
ficant interaction effect (seeFigure1 ) .
Similar interaction effectswere not found inwomen.The long-term results
were aspredicted forhigh emotional eaters,but lowemotional eatersdid not
showthenormal reaction of lessweight gain in response tostressand there-
foreno interaction effectwas found.
Thus, thepresent hypothesiswas confirmed tosomeextent inmen,but not
inwomen. Itispossible thatdieting hasobscured the results inwomen, as
women tend tobemore conscious of their appearance andbodyweight. However,
thisexplanation doesnot seemplausible, assimilar resultswere obtained af-
terpost hoc adjustment fordieting, asassessed by theRestrained Eating Scale
of theDEBQ.Also, response setsmayhave obscured theclassification ofwomen
intohigh or lowemotional eaters,asthereare indications that the response
setsacquiescence and socialdesirability areprevalent inwomen inthe present
longitudinal study (VanStrien, 1985). Inmen, thereareno indications of
response sets inthis study (VanStrien, 1984 ),sotheyhavebeen classified
more according totheir realemotional eating status.Although the precise
effects of these response setson thewomen's emotional eating scoresarenot
clear, theymayhave confounded theeffect ofemotional eatingand life events
on change inBMI.
Thereare indications that stress-illness relations aremediated byvaria-
bles suchas the individual's ability tocopewith stressful lifeevents (Rahe,
1974), thedegree of social support during and justafter stressful circumstan-
ces (Cobb, 1976), and theperceived control over anevent (Johnson and Sarason,
1978). Further, subjective weights regarding theundesirability of life events
may also reflect individuals'vulnerability tolifeevents (Sarason et al.,
1978; Zimmerman, 1983).Unfortunately, thesevariables arenotaccounted for in
 -106-

thepresent study,andthisnaybe the reasonthatthehypothesizedeffectwas

not found forwomen,and formen inonlyoneofthetwosixmonthsperiods.

NOTES

1.As theBMI showsa rythmic fluctuationwiththeseasonoftheyear theabso-

lutebodymasschangesprecedingorfollowingthetwoassessmentdatesare
notcomparable. Instead,thedifferencebetweenthesubgroupsshouldbeta-
ken inconsiderationandcanbe compared.
2.Thisquestionnaire alsocontainedascaleon restrained eating.
3.Examinationofhomogeneityofvariancewhichunderliesanalysisofvariance
indicated thatperforminganalysisofvarianceonthe rawdatawas
appropriate inallanalysesonmen,butnot inallanalysesonwomen.Even
inthelatter case,however,analysesof rawdata seemedpreferabletothe
alternatives,astreatmentofthedatabyeithera square root,ora log
transformationwouldmake findingsdifficult tointerpret,making such
transformations counterproductive.
 -107-

CHAPTER 13

ONLONGITUDINAL VERSUSCROSS-SECTIONAL STUDIESOFOBESITY:POSSIBLE ARTEFACTS

Tatjanavan Strien*

Scalesonemotional eatingand restrained eatingdeveloped ina large-scale

researchproject onoverweightwere administered toa cross-sectional sample of
normalweight andoverweightwomen (VanStrien, Frijters,Roosen,Knuiman-Hijl
andDefares, inpress), and alsotoasample ofwomenwhohadparticipated ina
longitudinal studyonoverweight on the thirdassessment date (VanStrien,
1984 a ). inboth samples, relationships betweeneachof these twotypesof
eatingbehaviour,and alsobetween eatingbehaviour andbodymass index (BMI;
2
weight/height )were determined.
Inthe cross-sectional study, emotional eatingwas found tobe related to
BMI, butno relationshipwas foundbetweenBMIand restrained eating. In
addition,no relationshipwas foundbetweenemotional eatingand restrained
eating.However, inthe longitudinal study, bothemotional eating and
restrained eatingwere foundtobe related toBMI,andalsoa significant
relationshipwas observed between restrained eating andemotional eating.When
thedifferences between the corresponding correlation coefficients between
emotional eating, restrained eatingandBMI inthetwostudieswere tested, all
threepairsofcorrelations differed significantly (VanStrien, 1984 a ).
Thus thequestion arosewhether theseearlier resultscouldbe replicated in
another assessment date of the samelongitudinal studyand inanew
cross-sectional sample, and further,whether systematic differences in results
of these twostudies canbe explained.

Response setsasartefacts inlongitudinal studies

Longitudinal studiesareconsidered tobepreferable tocross-sectional studies
because repeatedmeasurements onthe same individuals enables the investigator
todrawconclusions about causal effectsand intra-individual changes
(NesselroadeandBaltes, 1979).

Thischapter hasbeenpublished inthe International Journal ofObesity (1985)

9, 323-333,and is reprinted herewith kindpermission ofJohnLibbey&
Company, Limited, copyright 1985.
 -108-

However,difficultiesmayarisewhenmethodsarebased on the individual's own

report ofhisbehaviour.Studieswhich relyheavilyon tests, questionnaires
and interviews fordataare confrontedwith response sets,suchas,
acquiescence and socialdesirability (Cronbach, 1959;Jackson andMessick,
1958; Bentler, Jackson andMessick, 1971;Paulhus, 1984). The response set,
acquiescence, that isa "tendency tobeagreeable ina rather submissive way,
that isadesire tosaywhat isexpected" (Rorer,1965;p 134)may result from
themotivation ofsubjects tobe supportive and compliant tothe perceived
experimental hypothesis (Orne, 1962).The response set, social desirability,
that isa tendency toconsciously dissemble aself-report bydenying socially
undesirable characteristics andadmitting sociallydesirable characteristics
may result from the subject'swillingness tomakeagood impression on the
investigator (Riecken, 1962).
Longitudinal studiesmaybeparticularly susceptible tothegeneration of
response sets.Repeatedmeasurementwhich isa characteristic of longitudinal
designmay sensitize subjects tothenatureofthetestsused (Windle, 1954).
Inaddition, intensive contact between subjectsand investigator asa
consequence of repeatedmeasurementsmay lead toagreaterdegree of compliance
and ahighermotivation togive sociallydesirable responses,that isto "look
good" to the investigator.Therefore, the response sets,acquiescence and
socialdesirability, maybemoreprevalent inlongitudinal than in
cross-sectional studies.
Response sets reducethevalidityof resultsobtainedwith aparticular
measuring instrument (PhilipsandClancey, 1970), especially sowhen the
responsesofonegroupare systematically eithermorepositiveormore negative
thananother groupwithin the same sample (Cunningham, Cunningham andGreen,
1977). Inthepresent longitudinal studyonoverweight and eating behaviour,
thismayhavebeen the case forgroupsofnormalweight andobese subjects;
acquiescence mayhavebeen thedominant response set inthenormalweight, and
social desirability intheobese.

Bodyweight and response sets

Obese individuals havebeendemonstrated tobemore concerned about self-
presentation thannormalweight subjects (Elman,Schroeder and Schwartz,1977;
Glass, Lavin,Henchy, Gordon,MayhewandDonohoe, 1969;Rodin and Slochower,
1974; Younger and Pliner, 1978)because oftheir deviant status ina culture
where slimnessisthedesired norm (Krantz, 1978).Asaneed tobeaccepted and
approved of socially isassociatedwith giving sociallydesirable responses
 -109-

(McGee, 1962 a ' ), itmaybeexpected that the responsesofobese individuals

aremore affectedby the response set, socialdesirability thanwere thoseof
normalweight individuals.Self-reportsofeatingbehaviour especiallymaybe
systematicallydistortedby impressionmanagement intheobese (Stunkardand
Messick, 1985).
More specifically, forquestionson restrained eatingandemotionale eating,
itmaybeproposed that it issociallydesirable forobese subjects toendorse
questionsaboutbeingonadiet (Ruderman, 1983), andnot toendorse those
abouteatingwhen feelingannoyedor lonely.Thus,itmaybeexpected that
obese subjectsoverreport dieting behaviour andunderreport emotional eating
when socialdesirability istheirdominant response set.Incontrast, normal
weight subjectsare lessmotivated togive sociallydesirable responses to
questionsondieting and emotional eating.This inclination, togetherwith the
fact thatnormalweight individualsare ingeneral lessconcerned thanthe
obeseabout self-presentation,makes ithighlyplausible thatnormalweight
subjectsaremoremotivated tosupport theperceivedexperimental hypothesis
than togivea sociallydesirable response.Thus itmaybeproposed that
acquiescence isthedominant response inthenormalweight.

Indicationsof theprevalence of response sets

Althoughno independent scales forthe relevant response sets (Bentleret al.,
1971;Paulhus,1984)are included inthepresent study,examinationofthe
patternofcorrelation coefficientson thescalesforemotional eatingand
restrained eatingmay indicate theprevalence ofparticular response setsin
groupsof subjects.Theprocedureusedwasanalogous tothatusedby Jackson
andMessick (1961),whoobservedhighcorrelationsbetween scaleswhichwere
similar insocialdesirability, and that thecorrelation became increasingly
smaller andnegative as theybecamemoredissimilar indesirability. Jackson
andMessick contended that: "Sincehighcorrelationsbetween ... scales
probably indicate the covariationofacquiescent responses to itemssimilar in
desirability, itseemslikely that thesecorrelationswillprove tobe
interprétableprimarily intermsofthese twostylisticdeterminants" (e.g.,
acquiescence and socialdesirability) (I.e.p775).
Thus,an indicationof socialdesirabilitybeing thedominant response set
intheobesemaybeobtained froma lowor zerocorrelationbetween emotional
eating and restrainedeating,becauseofthe relatively largediscrepancy in
socialdesirability of these twotypesofeatingbehaviour for thesesubjects.
Similarly, an indicationofacquiescence beingthedominant response set in
 -110-

normalweight subjectsmaybeobtained fromahighcorrelationbetween

emotional eatingand restrainedeating,becauseofthe relativelysmall
discrepancy indesirabilityofthesetwotypesofeatingbehaviour for these
subjects.
Indicationsofahighdegreeofprevalenceofaparticular response set in
subjectsparticipating inthelongitudinal studycomparedwith those
participating inthecross-sectional studymaybeobtained fromtestsof
differencesofthecorrelation coefficientsbetweenemotionaleatingand
restrained eating,andalsofromthemeanscoresonthetwoscales.

METHOD

Subjects
Studypopulation.Inthefirstpartof1980,astudywascarriedout inthe
municipalityofEde,theNetherlands,inwhichall inhabitantsinthreeage
groups (19-21,24-26and29-31yearsasat 1January1980)were invited to
participate.Theiraddresseswereobtained fromtheCivilRegistrationOffice
inEde.Complete informationwasobtained from3,936subjects,representing33%
ofall invited subjects.Detailsofthemunicipality, thestudypopulation,and
theprocedure followedhavebeenpresented elsewhere (Baecke,Burema,Frijters,
HautvastandVanderwiel-Wetzels, 1983). Someoftheparticipants inthis
longitudinal studywere selected toparticipate inalongitudinal studyon
overweight,andparticipants incross-sectional studieswere selected fromthe
remainingsubjects.

Longitudinal sample.Fromthetotal studypopulation,a sampleof1663

individuals,thedistributionof sex,age,and levelofeducation,ofwhomwas
comparablewith thestudypopulation,wasinvited inNovember 1980topartici-
pate inthepresent longitudinal study.Thisstudycomprisesatotalof seven
assessmentdates,twoperyear.Onthefirstassessmentdate (Spring 1981),
complete informationwasobtained from1533subjects representing92%ofall
those invited.Datadiscussed inthepresentpaperwereobtained forwomenon
the third (Spring1982)andsixth (Autumn1983)assessmentdates,because the
scalesoneatingbehaviourwereadministeredonthesetwoassessmentdates
only,andbecause inthecrosssectional study,datawereobtained forwomen
only.Women,pregnant forthreemonthsorlonger,andwomenwhohadmoved out
of thedistrict,wereexcluded fromanalysis.
 -in-

complete informationwasobtained onthethirdassessmentdate from744women

andon thesixthassessmentdate,from 570women (note 1 ) * .

Cross-sectional sample.A sample of 345women,thedistributionoftheageand

levelofeducationofwhomwascomparablewith thestudypopulationwas selected
toparticipate inthisstudy.

Procedure
Longitudinal study.Oneachassessmentdateofthisstudy, subjectswere invited
bymailtocomplete aquestionnaire athome.Severaldayslater,theywere
visited athomebyafield-workerwhocollectedandchecked thequestionnaire
for completeness.Shealsomeasuredbodyheightandbodyweightwithout shoes
and jacket tothenearest 1mmand 0.5kg respectively.Onallassessmentdates,
subjectswerevisitedbythe samefield-worker.
Thequestionnaire contained threeparts.Twopartswereadminsitered tothe
subjectsonallassessmentdates,andcontainedquestionson lifeevents,and
changes inbehaviouralandsocio-demographicaspects.Thethirdpartwasdiffe-
rentoneachassessmentdate,andon thethirdand sixthassessmentdate
contained thescalesoneatingbehaviour.

Cross-sectional study.Subjectsofthecross-sectional studywere invitedby

mail toparticipate.Severaldayslater,theywerevisitedathomebya
field-worker,whomeasuredbodyheightandweightwithout shoesand jacket to
thenearest 1mmand 0.5 kg respectively.The field-worker alsoleftaquestion-
naire tobe completed. Severalhourslater,itwascollectedandchecked for
completeness.

Scalesforemotional and restrainedeating

The scalesonemotionaleatingand restrainedeatingarederived fromtheDutch
EatingBehaviourQuestionnaire (DEBQ),andhavebeen showntohaveahigh
internal consistencyand factorialvalidity (VanStrien,Frijters,Bergers&
Defares,inpress). HieRestrainedEatingscaleconsistsofsevenitems,andthe
Emotional Eating scaleofeleven items.All itemshavethe response format:
never (1);seldom (2);sometimes (3);often (4);andveryoften (5). A high
score indicatesahighdegreeoftheeatingbehaviour inquestion.Table 1 shows
Englishtranslationsofthe items,and their item-rest correlationcoefficients.

* seepage 118 fornotes

 -112-

Table1.Itemsofthetwoeatingbehaviourscalesandtheiritem-rest
correlationcoefficientsforthethirdassessmentdateofthelongitudinal
study.

RestrainedEating Item-rest
correlationcoefficient

1.Whenyouhaveputonweight,doyoueatlessthanyou
usuallydo? .71
2.Doyoutrytoeatlessatmealtimesthanyouwouldliketoeat? .68
3.Howoftendoyourefusefoodordrinkofferedbecauseyouare
concernedaboutyourweight? .76
4.Doyouwatchexactlywhatyoueat? .68
5.Doyoudeliberatelyeatfoodsthatareslimming? .71
6.Whenyouhaveeatentoomuch,doyoueatlessthanusualthe
followingday? .66
7.Doyoutakeintoaccountyourweightwithwhatyoueat? .83
Cronbach'salphacoefficient .90

EmotionalEating Item-rest
correlationcoefficient
1.Doyouhaveadesiretoeatwhenyouareboredorrestless 775
2.Doyouhaveadesiretoeatwhenyouhavenothingtodo? .60
3.Doyouhaveadesiretoeatwhenyouaredepressedor
discouraged? .79
4.Doyouhaveadesiretoeatwhenyouarecross? .65
5.Doyouhaveadesiretoeatwhenyouareexpectingsomething
unpleasanttohappen? .70
6.Doyougetadesiretoeatwhenyouareanxious,worriedor
tense? .70
7.Doyouhaveadesiretoeatwhenyouarefeelinglonely? .67
8.Doyouhaveadesiretoeatwhenthingsaregoingagainstyou
orwhenthingsgowrong? .81
9.Doyougetadesiretoeatwhenyouareirritated? .78
10.Whenyouareexcited,doyougetadesiretoeat? .65
11.Doyouhaveadesiretoeatwhensomebodyletsyoudown? .76
Cronbach'salphacoefficient .93

RESULTS

Differencebetweenthelongitudinalstudyandcross-sectionalstudy
Themeansandstandarddeviationsofscoresonthescalesforemotionaleating
andrestrainedeating,andBMIandageforthetwoassessmentdatesofthe
longitudinalstudy,andthecross-sectionalstudyaregiveninTable2.As
scoresforthetwoassessmentdatesofthelongitudinalstudyweresimilarto
eachotherandalsotoscoresonthecross-sectionalstudy,sample
characteristicsofthelongitudinalstudyandthecross-sectionalstudymaybe
consideredtobesimilar.
 -113-

Table2.Meansand standarddeviationsoftheRestrained Eatingand Emotional

EatingScales,bodymassindex (BMI)andage*,obtained forthetwo assessment
datesof thelongitudinal study,and for thecross-sectional study.

Longitudinal study Cross secti onal

study
Third assessment Sixth assessment
(n=744) (n-570) (n-345)

mean sd mean sd mean sd

Restrainec eating 2.45 .94 2 63 .84 2.51 .92
Emotional eating 2.21 .76 2 13 .74 2.27 .68
BMI 22.56 2.78 22 64 2 .91 22.67 3.29
Age 27.72 3.95 28 72 3 .95 29.55 4.49

*Asat1Januaryoftheyear ofassessment

Pearsoncorrelation coefficients showing the relationshipbetween the two

eatingbehaviour componentsandBMI,partialcorrelation coefficients adjusting
forconfounding effectsand resultsoftestsofdifferences (Fisher's rtoZ
transformation;Hays,1969)betweencorresponding pairsof correlation
coefficients arepresented inTable 3.Forthelongitudinal study,the
correlation coefficientsbetweenthetwoeatingbehaviour componentsandBMI
forthethirdassessmentdatedifferednot significantly fromthe corresponding
correlation coefficients forthesixthassessmentdate.However,twoof the
threecorrelation coefficients ofthecross-sectional studydiffered
significantly fromthecorresponding correlationcoefficients forboth
assessmentdatesofthelongitudinal study.Thus,itmaybeconcluded that
therearesystematic differencesbetween resultsofthesetwostudies.

Acquiescence andsocial desirability

Pearson's correlation coefficientsbetweenthetwoeating behaviour
components forthenormalweight (BMI<26),andoverweightwomen (BMI>26)in
thethreestudiesarepresented inTable4.Forallthreesetsofdata,a
significant relationshipwasfoundbetween restrained eatingand emotional
eating inthenormalweightwomen,butnosuch relationshipwasobtained inthe
overweightwomen.Thus,itmaybesuggested thatacquiescencewasthe dominant
response setinthenormalweightwomen,andsocialdesirability inthe
overweightwomen.
 -114-

ïO •r4
<D •U
a <Ö
f-H
>*
•o d>
kl
0
.0 1-
U-i 0
•
< 4r~? PO

9 cw
O 0
•H
4J

5>
01
co
01
u
A
01
00 £
c 4J

4J •o
c
«
o> cd
J3 C
0
3 «
co x e o

CQ « v^

s
CQ
ü
01
CQ
01
CM u
•o CM
e •v
c« e
01 a
00 .n
c AJ 00
•M e
 -115-
to e
CO
o >.
o
h •H
.u
1•u
U U (0

c
CD
a .-»
Ä to >©
u to h*
M tu H
-H M e
tO D) V

f tu s

5 1

a
ta

c
m
a
o
» >>
•O
9
u 4J
Si
M
•H rH
01 tu
C
» •H

3 •s
h •4J
H
O 00
2
e
o Ü %£
hJ

o t-t

«
w
«
c •H
e
ai
•H
'S 4J
O •H
f* M
»W C
«W 0
<D i-H
O
Cl m
£
u
c
•o
H
Lt e
o
a
iH
V e
u «
a

0 /•
co
H M
01 S o
0) M
A
 -116-

fi
cd
cu

o CD cd

» oe
*J n o
O -H
JS 4J
00 /-s cj

J= B
4j cy o>
xa "

£
00
•H 4J
O •o e
3 u cu
•H a il
•H J= co •W
4 J CO cd
2 0) T3
a
u ^ ca
o < co
s v ^ cd

B T> T3
cd CU CU
IV W rH
T4 «H
e cd cd
0) 00 4-t 4J
s CU 0
•H
4J B »
cd O «
CU
•o •o
^ -^^^
ITI rH
""
iH
<u
4J
JC CU O O
cd B 00
S •H •H B •c
H
o B • cd B . CU O B • cd B V V
•H cd
•H 6.13
•H
*J cd o n cd o
01 • u V • »
U a• u4J cd
CU
Q
o. o.
Ä
«
S =
O 4-1
O S co co S co
e CU
CU
I* CO co S CO
Hl
H O CO
m as o> « « ««
 -117-

A test of thedifference ofthe correlation coefficients between emotional

eating and restrained eating fornormalweightwomen inthe three studies,
revealed that this relationshipwas stronger inbothassessment datesof the
longitudinal study than inthecross-sectional study (t- 2.88;p < .01),and
t - 2.75;p < .01).Thusthe response set,acquiescence,mayhave beenmore
prevalent inthenormalweightwomen inthe longitudinal study than inthe
cross-sectional study.
The testsofdifferences (t-tests)between themeansonemotional eating and
restrained eating scores fornormalweight andoverweightwomen inthe three
data setsarepresented inTable 5.Inthenormalweightwomen,no significant
differenceswere obtained. Intheobesewomen,nosignificant differences were
found inemotional eating,but thoseparticipating inthelongitudinal study
scored significantly higher ontheRestrained Eating scaleonboth assessment
dates than those participating inthecross-sectional study.Thus,the
restrained eating score ofoverweight women of the longitudinal studymayhave
beenaffected moreby the response set socialdesirability thanthe same score
foroverweightwomen inthe cross-sectional study.

SUMMARY AND CONCLUSION

Most results of the cross-sectional studydiffered significantly from those of

the longitudinal study, indicating that theremaybe systematic differences
between the results inthese twostudies.Therearealso indications that these
differences maybedue to increased prevalence of the response set,
acquiescence, inthenormalweight, and socialdesirability, inthe overweight
participants inthe longitudinal study.However, it shouldbe stressed that
thisexplanation canonlybea supposition at this stage and requires
confirmationwith independent response-set marker scaleswhichwere notused in
thepresent study. Itwould bepreferable tocheck self-reports against
objective external criteria, because substantive traits cannot alwaysbe
separated from responsebiasbymeansof independent response-set marker scales
(McGrae and Costa, 1983). Forexample, analternative explanation for thehigh
scoreson restrained eating intheoverweightwomenof thelongitudinal study
maybeahigher levelof restriction infood intake asa result of increased
weight consciousness, because of thehalf-yearlymeasurement ofbodyweight in
the longitudinal study.Thisexplanation doesnot seemveryplausible.As obese
women arenotorious forunderestimating theamount of food theyeat (Beaudoin
andMayer, 1954), itwould be surprising iftheir responses to questionnaires
 -118-

oneatingbehaviourdidnot reflect thistendency.However,thiscanonlybe

confirmedbychecking self-reportsagainst objectiveexternal criteria.A high
scoreonasocialdesirability scale indicatesahighsensitivity tosocial
approval inthe individual.Inself-reportingofdieting,ahighsensitivityto
socialapprovalmay result incompliancewiththenormativepressure of
slimnessand thusactualdieting,butalsoin reportofdieting, irrespective
oftheactual level in restrictionof food intake (Blanchard andFrost,1983;p
261). Thusbymeansofaself-report scaleon socialdesirability inthe
assessmentof restrainedeating, theconfoundingeffectofactual dieting
cannotbeeliminated completely.
Often, it isverydifficult toverifythisagainst external criteria.Also,
substantivetraitsarenotalwaysentangled tosuchagreat extentwith
response sets inallaspectsofself-reportsand inall contexts.Thus itis
recommended thatusebemadeof independent response-setmarker scales insuch
studiessothat response setscanbe identified,andthevariance attributable
tostylistic tendenciespartiallyadjustedstatistically (Gadourek, 1972), or
byuseofthe ipsativeprocessto reduce response-setbias (Cunninghamet al.,
1977). Recently,a shortsocialdesirability inventorywhich isphrased in
dietary terms (Worsley,Baghurst &Leitch,1984)hasbeendevelopedwhichwill
beofuse instudiesonoverweightandeatingbehaviour.Response-setmarker
scalesare recommended especially inlongitudinal field studies.Repeated
measurement sensitize subjectstothenatureofthetestsused.This,together
withclose contactbetween subjectsand investigator,maylead togreater
compliance,and inturntothe response set,acquiescence,insomesubjects,
while inother subjectsahighermotivation forsocialapprovalmaylead in
turntothe response set,socialdesirability.

NOTES

1.Thedrop-out of subjectson thesixthassessmentdate isduepartlyto

missingdata.On thisassessmentdate,subjectscouldgiveanot relevant
response toall itemswhichwerepresented inaconditional format (thatis,
doyoueatwhenyou feelboredor restless).Subjectsgiving sucha response
tomore thanone itemwere excluded fromanalysis.
 -119-

CHAPTER 14

GENERAL DISCUSSION

Themain aimof this studywastoassesswhich theory seems themostvalid with

reference totheaetiology ofobesity, psychosomatic theory,externality theory
or the theoryof restrained eating, and todeterminewhichvariables are
related toemotional, external and restrained eatingbehaviour.This ultimately
maylead tomore insight inwhat thebestmethod of treating overeating and
overweight mightbe.Theconstruction ofaquestionnaire for theassessment of
these typesofeating behaviour,whichwould enablepsychometric studiestobe
carried outon the relationships between theseeatingbehaviour components and
othervariables, comprised a further important goal.

TheDutch EatingBehaviour Questionnaire

In theDutch EatingBehaviour Questionnaire (DEBQ)whichwasconstructed for
thepurpose of this study (Chapter 8 ) ,the scales for restrained, emotional and
external eatingbehaviourwere shown tohavehigh internal consistency,
factorial validity anddimensional stability.The scales for restrained and
external eatingbehaviourwere found tobeunidimensional.The scale for
emotional eatingwas showntocomprise twodimensions,onedealingwith eating
in response todiffuse emotions and theotherdealingwitheating in response
toclearly labelled emotions.However, thedifference between these two
dimensionswas found tobe lessdistinct intheoverweight rather than inthe
normalweight subjects.This isinaccordancewith findingsof Slochower
(1983 ).Sheobserved thatvarious typesofemotionsproduce different typesof
eating responses:onlydiffuse emotional states trigger theovereating response
and clearly labelled emotionsdonotaffect eatingbehaviour.However, shealso
found indications thatobese individuals perceive thoseemotions,whichare
clearly labelled inmostnormalweight individuals,asdiffuse.This finding
demonstrated thatoverweight individuals alsoovereat inresponse to so-called
clearly labelled emotions,and thismayexplainwhy thedifference between the
twodimensions of the itemson theemotional eating scalewere found tobe less
distinct intheoverweight. Though scaleshavebeen constructed dealing
separatelywith clearly labelled anddiffuseemotions,this issuewasnot
further examined inthe current project. Instead, the two-dimensional emotional
eating scalewasused, forageneralassessment ofemotional eating.
 -120-

•ttiepredictivevalidity of the restrained eating scaleof theDEBQwas

studied inapopulation ofwomen (Chapter 9). Theultimate criterionof res-
trained eating, that is,thedegree towhich awoman eats less than shedesi-
res,was studied indirectly, using estimates of thedeviation from required
energy intake.Itwasnecessary touse thistechnique because of the impossibi-
lityof conducting a studyofbothactual food consumption and restriction of
food intake.However, theuse of themagnitude ofdeviation fromenergy re-
quirement asthecriterion of restrained eatingmayhave resulted inanunder-
estimate of the scale'sability topredict deviation ofenergy intake from
desired intake for the following two reasons.Firstly, energy requirement may
give a conservative estimate of the food intakedesired byawomenwhowould
like toeatmore than isrequired tomaintain thepresentbodyweight. Second-
ly,awomanmayeat lessthan shedesires,andyet stilleatno less than
required for themaintenance ofherbodyweight,when thisbodyweight hasbeen
reached following successfulweight lossdue todieting (note1 ) * .
Thus, theobserved relationship between restrained eating scoresand the
magnitude ofdeviation fromenergy requirement suggests that the scale predicts
eating less thanwhat thenorms forweight saythesewomen should be eating.
Thispossibly results ina conservative estimate of the scale'sability topre-
dict eating less than isdesired.A further indicationof the scale's ability
topredict restrictionof food intakewasobtained fromthehighnegative rela-
tionships observed between scoresonthe scaleand the intake of fatand sugar.
However, it should be stressed that these conclusions arebased on the self-re-
portsof thewomen studied, and such reportsof food intake are particularly
notorious for their inaccuracy inestimating actual food intake (Block, 1982).
Thoughnotdesigned assuch, the study inChapter 12maybe considered asa
validation studyof theemotional eating scaleof theDEBQ. Itwas found that
inmale subjects lowemotional eatersgained lessweight andhigh emotional
eaters gainedmoreweight after experiencing negative lifeevents,than those
whohadnotexperienced suchevents.This is inlinewith theclinical observa-
tionthat some individuals, socalled 'emotional eaters' respond to stresswith
anexcessive intake of food,whilemost individuals respond tosuch circumstan-
ceswitha lossof appetite and subsequent weight loss.Thiscanbe considered
asan indication that theemotional eating scale issuccessful in predicting
weight change inpersonsdesignated asloworhighemotional eaters, according
tothe scoreson this scale.However, similar resultswerenot found amongst

* seepage 135 for notes

 -121-

women. Inthe long teemhighemotional eatersgainedmoreweight afterexpe-

riencingnegative lifeevents than thosewhohad notexperienced suchevents,
while lowemotional eatersdid not show the 'normal'response oflessweight
gain following stressful lifeevents.A possible explanation for this finding
maybe that theemotional eating scalehasbeen lesssuccessful in correctly
classifyingwomen asloworhighemotional eaters.Itmaybeproposed that
socially it islessdesirable forobese, rather thannormalweight,women to
endorse questionsoneatingwhen feeling annoyed or lonely.As a consequence,
overweightwomenmayunderreport emotional eatingmorecommonly thannormal
weightwomen, and this reduces thevalidity obtained using theemotional eating
scale.Aswillbeelaborated below, thereare indications that such response
setshavebeenprevalent toaheightened extentamongst thewomen studied in
Chapter 12,due toprolonged contactwith the investigator, inturna result of
the longitudinal design ofthisstudy.Though theprecise effectsof these res-
ponse setsonthe resultsarenotclear,theymayhaveconfounded the influen-
cesofemotional eatingand lifeeventsonchange inBMI.
Endorsement of itemsonexternal eating isprobablyalso less socially
desirable forobese than fornormal-weight women. It islikely that the reverse
istrue for itemson restrained eating.However, theprecise extent towhich
sociallydesirable tendenciesaffect scoreson theeatingbehaviour scales in
both longitudinal and cross-sectional populations isnot known. Itwould be
preferable to investigate thisusing objective external criteria,as substanti-
ve traitscannot alwaysbe separated from response biasbymeansof independent
response-set marker scales (McCraeand Costa, 1983).A high scoreona social
desirability scale indicates ahigh sensitivity tosocial approval on thepart
of the individual.Inself-reporting ofdieting ahigh sensitivity tosocial
approval may result incompliancewith thenormative pressure of slimnessand,
therefore,withactualdieting, andalso inthe reportingofdieting, irrespec-
tiveof theactual level of restrictionof food intake (Blanchard and Frost,
1983). Similarly, inself-reporting ofexternal eatingbehaviour, ahigh sensi-
tivity to socialapprovalmaybe indicative ofageneralised sensitivity toex-
ternal stimuli ofwhich a sensitivity toexternal food stimuli is just another
aspect. Itmayalsobe indicative ofanunderreporting ofexternal eating,be-
cause it isnot sociallydesirable toadmit toeating in response to external
food cues.Thus,it isdifficult todifferentiate between response setsand
substantive traitsbymeansof sociallydesirability scales.
Thisproject produced nodata concerning thevalidity of the external
eating scale,and onlya start hasbeenmadewith theconstructvalidity of the
 -122-

scales, through theassessment oftheir relationshipwithothervariables.

These associationswillbe reported inseparate sections.Nevertheless, the
studyof thethreeeatingbehaviour scalespermitted anexploration of the
validity ofthemain theories inthedevelopment andmaintenance of obesity.
Firstly, however, itneeds tobe considered thatdataobtained from female
participants ofthe longitudinal studymayhavebeenaffected toa heightened
extent by the response-sets acquiescence and social desirability.

The response-sets acquiescence and social desirability

Results obtained forwomen participating inthelongitudinal studyon
overweightwere found todiffer systematically from thoseobtained forwomen in
thecross-sectional study.Nodifferencesbetween thetwotypesofstudieswere
foundwhen applied tomale subjects (VanStrien, 1984 ).As isshown inChapter
13, thedifferencesbetween thewomenmaywell beattributable toa greater
prevalence inthe longitudinal studyof the response-sets,acquiescence and
social desirability. The supposition isput forward that the intensive contact
between subject and investigator, togetherwith the sensitisation of the
subjects tothequestionnaires due to repeatedmeasurements,mayhave led toan
increasedwillingness toco-operate,and, asaconsequence, to social
acquiescence inthewomen ofnormalweight.However, an increased concern for
self-presentation and, thus,socialdesirability asa response-set may have
beenmore prevalent inobesewomen.As aconsequence, inthe longitudinal study
responses obtained fromoverweight womenmaybe systematically either more
positive ormore negative than thoseobtained fromnormal-weightwomen. This
reduces thevalidity ofthe resultsobtained withaparticular measurement
instrument.Theeffect ofthis is,that the resultsobtained forwomen inthe
longitudinal study,as reported in the final studyofChapter 8,and the
studiesofChapters 11and 12,shouldbe interpretedwith caution, especially
where the resultsdiffer fromthoseobtained inthecross-sectional studyor
fromgenerally accepted theoreticalnotions.
As alreadymentioned above,nodifferences between the resultsof the
longitudinal and cross-sectional studieswere foundwith respect tomen.
Therefore, there isno reason tosuspect that response setshavebeen prevalent
toa largeextent inthe resultsobtained from themaleparticipants of the
longitudinal study.Sexdifferences inresponse setshavenotbeenexamined in
thepresent project.However, thehigher prevalence of response setsmay be
explained by the following lineof reason.
 -123-

Ithasbeen found thatwomenareexpected tobemore sensitive tothe needs

of others (e.g.,Bakan, 1966;ParsonsandBales, 1955), and that theyaremore
socially oriented (Carlson, 1971)anddemonstratemore social conformity (e.g.
Allen andCrutchfield, 1963;Tuddenham, 1958)thanmen.While these
suppositions remain tobe further investigated, itispossible that this
general tendency 'tobeagreeable'canbeextended to research situations,and
that instudieswhich relyon tests,questionnaires or interviews fordata, the
response-set acquiescence ismore prevalent inwomen than inmen.
However, ithasalsobeendemonstrated thatwomenhaveahigher need for
approval thanmen (e.g.,Hoffman, 1972).As theneed tobeaccepted and
approved of socially isassociated with sociallydesirable responses (HcGee,
1962 a ' ),itmaybeexpected that the responsesofwomenaremore affected by
the response-set socialdesirability than thoseofmen.Aswillbe elaborated
below, thismaybeespecially true forobesewomen,particularly in
self-reports of eating behaviour.
Public attitudes towards theobese inour societyareextremely negative
(DeJong, 1980). Thisderogation mayhave resulted inpart from the presumption
that fatpersons are responsible for their owncondition.Fatness isbelieved
tobe theoutcome of "immoral self indulgence" (Maddox,Back andLiederman,
1968), and obesepeople canscarcely avoid interactionswhereweight and eating
behaviour are topicsofdiscussion orcovert determinants of theevaluation of
others (Wooley,Wooley andDyrenforth, 1979).As a result of these negative
attitudes,obese individualsmaybehighly self-conscious andvery concerned
with their appearance toothers.Ithasbeendemonstrated invarious studies
that obese individuals aremore concerned about their self-presentation than
normal-weight subjects (Elman,Schroeder and Schwartz,1977;Glass, Lavin,
Henchy,Gordon,MayhewandDonohoe, 1979;Rodinand Slochower, 1974;Younger
and Pliner, 1978). Self reportsofeatingbehaviourmaybeespecially distorted
by impression management intheobese (Stunkard andMessick, 1985). However,
deceptionwith respect toeatingbehaviour seemsmore likelytooccur inobese
women than inobesemen.
Overweight isviewed more seriously inwomen than inmen (Harris,1983)by
both societyandbythewomen themselves (Allon, 1982), partlybecause being
attractive ismore important forwomen than formen.Traditionally, beauty has
beenamajor asset forwomen inobtaining a suitablemarriage partner and
achieving upward socialmobility (MaccobyandJacklin, 1974), and, according to
a studyofPlayboy centrefolds andMissAmerica pageant contestants (Garner,
Garfinkel, Schwartz and Thompson, 1980), beautyhas increasingly been equated
 -124-

with slimnessforthepast 20years.Not surprisingly,womenaremore likely to

be concerned about theirweight thanmen (DwyerandMayer, 1970), and dieting
hasbecomeaway of life formanywomen.At the same time,however,bodyweight
control ishampered byanumber ofelements central tothe traditional female
sex role,suchasthe roleof food intraditional women'sworkand the
encouragement ofdependent attitudes inwomen.As observed byHall and Havassy
(1981), thepositionofahomemaker who iseasily tempted bynice and
appetising food is roughlyanalogous tothat ofanalcoholic bartender.And
dependency, aswillbeelaborated ina separate section,may function to
"...reinforcenon-effective responses ininterpersonal situations requiring
assertive behaviours" (Leonand Finn,1984,p.323), forexample, emotional
eating.Theopposing pressures toeatornot toeatmakeseatingbehaviour a
highly laden subject formanywomen,and thismay explain the fact that women
frequentlyunderestimate their food intake,whereasmen rarelydo so (Stunkard
andKoch, 1964).Thisunderestimation of food intake inwomenmaybe either
unconscious, resulting froma lack ofawareness ofhavingeaten toomuch,
caused bya reductionof cognitive dissonance, or conscious, dissembling
overeating asa result ofawillingness tomake agood impression on the
investigator, orboth.Inthe lightof thegenerally accepted notion that it is
unladylike forawoman toeat largeamountsof food (Leonand Finn, 1984)and
society's pressure tobe thin,overweightwomenmay find it especially
difficult toadmit that theyhaveovereaten,both tothemselvesandothers.
Incontrast towomen, formen theamount eaten isoften seenasa measure
ofmasculinity, power and strength (Leonand Finn, 1984), and obesemen have
even been found toboast about theamount of food theyate (Hendelson,Weinberg
and Stunkard, 1961). This sexdifference inthe "denial"of food intake may
also result fromvariations intheopportunities available for concealing
overeating.While menhave little opportunity toeatunobserved between meals
(sowhen theyovereat, theymust overeat inpublic),womenhavemany
opportunities toeatunobserved byothers inthe solitude of their kitchen
(Stunkard and Koch, 1964,p.81).Thus,womenhaveboth theoccasion and reason
todeny that theyhaveovereaten, and thismayexplainwhy the responses of
obesewomen inthe realmof researchoneatingbehaviour maybemore affected
by the response-set socialdesirability thanthoseofobesemen.
 -125-

The three theoriesof overeating

The theoryof restrained eatingbehaviour.According tothetheoryof

restrained eating, restriction of food intake isaccompanied by psychological
side-effects,whichmaypromote overeating orweight gain.This supposition is
based on theobservation that restrained normal-weight individuals show
"counterregulatory"eatingbehaviour orexcessive food intakewhen their
cognitive control over food intake isundermined bydisinhibitors, suchasa
preload, alcohol,anxietyordepression (HermanandMack,1975;Hermanand
Polivy, 1975;PolivyandHerman, 1976a, ).Inthese studies,however, subjects
weredesignated as restrainedeatersbymeansof theRestraint Scale,whichmay
notmeasuredieting per se,butdisinhibition ofcognitive restraint. This
findingwasobtained usinganewquestionnaire toassess restrained eating, the
Three Factor EatingQuestionnaire (TFEQ) (StunkardandNessick, 1985). The TFEQ
separatelymeasures cognitive restraint of food intake (FactorI),
disinhibitionofcognitive restraint (Factor II)andhunger (Factor III). It
hasbeen shown thatwhile theHermanand PolivyRestraint Scalewasnot related
to thecognitive restraint scaleoftheTFEQ (Factor I),itwasclosely related
tothedisinhibition scaleof theTFEQ (Factor II) (StunkardandHessick,1985;
Weissenburger,Rush,Giles&Stunkard, submitted). Inaddition, itwas found
that Factor II,rather thanFactor I,oftheTFEQwas related toovereating
(Shrager,Wadden,Miller,Stunkard and Stellar, 1983).Thus,itcanbe
tentatively concluded that theexcessive food intake found insubjects scoring
highontheRestraint Scale inthe studiesofHermanand Polivywasnot related
to restraint per se,but todisinhibition of restraint.Thismeans thatnot all
dieters,butonlya sub-populationhavinga tendency towards disinhibition,
showexcessive food intakewhen "comingoff"adiet.
Examination ofthe restrained eating scaleof theDEBQ revealed thatmost
of itsitemswere similar to thoseof thecognitive restraint scale (Factor I)
of theTFEQ.This isnot surprising asbothcontained itemstaken fromPudel's
Fragenbogen für LatenteAdipositas (Pudel,MetzdorffandOetting, 1975)inthe
initial itempool. Inaddition, the restrained eating scalewas found tohave
good predictive validity for the restrictionof food intake.Therefore,it
appears that the restrained eating scalediffers fromtheRestraint Scale, in
that it seems tomeasuredieting toamuch larger extent.
Inthe studyofwomen inChapter 10nosignificant relationshipswere found
between the scales for restrained eatingandeither of thescales for emotional
or external eatingbehaviour.These findingsarenot inaccordancewith the
 -126-

theoryof restrained eating,which suggeststhat dieting isa causalagent for

emotional and external eatingbehaviour, the so-called "obese"eatingpatterns.
Bather, indicationswere found that "obese"eatingpatternsprecede restrained
eatingbehaviour -nodifferencewas found inthedegree ofemotional or
external eatingbehaviour betweenwomen of current overweight andwomen of
prior overweight, butofnormalweight atthe timeof the studydue to
successful dieting, the "obese"and "latentobese"women.This finding isin
linewithbothpsychosomatic andexternality theory,and indicates that "obese"
eating patterns arepotential causesofoverweight andof dieting.
Lowe (1984)also found indications that "obese"eatingpatternsmay precede
dieting. Inhis studythedegree towhich subjectswere currently suppressing
theirweightwascalculated by subtracting the current percentage overweight
from theprecentage overweight at theirgreatestweight. Itwas found that
high-restrained normalshad suppressed theirweightmore than low-restrained
normals.Inaddition, attheir greatest previousweight high restrained normals
hadbeenmore than fourtimesmore overweight thanunrestrained normals.Also
their currentweightwashigher than thatofunrestrained normal-weight
subjects.According toLowe,these findings suggest that "... high restraint
normalsbehave like theobesebecause theyhave retained characteristics
associatedwithprior obesity rather thanbecause of their current level of
cognitive restraint" (I.e.p.247).
Shrager et al's (1983)finding that cognitive restraintwasnot related to
overeatingdoesnot contradict the suggestion that "obese"eating patterns
precede restrained eating.However, it isdifficult toassesswhat implication
their finding thatdisinhibitionwas related toovereating hason the current
argument. Inspection of thedisinhibition scaleoftheTFEQ (Factor II) reveals
that itcontains twosortsof items:those referring to restraint breaking and
bingeing (e.g.,while onadiet, ifIeat a food that isnotallowed, Ioften
then splurge and eat otherhighcalorie foods;doyougooneatingbinges even
thoughyouarenothungry)and those referring toemotional and external eating
behaviour (e.g.,when Ifeel lonely, Iconsolemyself byeating;when Ismell a
sizzling steak or seea juicypieceofmeat, Ifind itverydifficult tokeep
fromeating, even if Ihave just finished ameal).Thus,the relationship
betweendisinhibition and overeatingmayhavebeencaused bythedualeffect of
disinhibition per seand "obese"eatingpatterns,asahigh score on
disinhibition can result fromahigh subscore oneither of these two
dimensions.Because of this interpretative difficulty, thequestionas to
whether theovereating of thosehavinga tendency towardsdisinhibition in the
 -127-

studyofShrageretal. (1983)iscausedmainlyby "obese"eatingpatterns,

suchasemotionalandexternaleatingbehaviour,orbytendencies towards
restraint breaking andbingeingmust beleft open.
Itshouldbenoted that the resultsof the study inChapter 10 should be
interpretedwith caution. Firstly, preliminary scaleshavebeenused inthis
studytomeasure thethreetypesofeatingbehaviour, and inspectionofthe
itemsof the scales inthis study (seeTable 1,Chapter 10)and thoseof the
final scales (seeTable 1,Chapter 8) reveals that the scales for restrained
andexternal eatingbehaviour inparticular containdifferent items inthe
final scales.Secondly, someof the resultsof the study inChapter 10havenot
been replicated bythe femaleparticipants of the longitudinal study (Chapters
8 and 11).Inthis study significant relationshipswere found between
restrained eating, andemotional andexternal eating.Thediffering resultsmay
have beendue totheuse ofdifferent scalesfor theassessment of the three
typesofeatingbehaviour.Theymayalsohavebeencaused bydifferences inthe
sample characteristics of thetwostudies:thewomen studied inChapter 10had
ameanBMI ofapproximately 25,while thoseexamined inthe longitudinal study
had ameanBMI ofapproximately 22.5.However,asargued byVan Strien (in
pressand inChapter 13),it seemsmore feasible toattribute this discrepancy
inthe results toaheightened prevalence of response setsamongst the female
participantsof thelongitudinal study,astwocross-sectional studies,which
used sample characteristics and scales tomeasure emotional and restrained
eating identical tothoseused inthe longitudinal study, revealed no
significant relationships between thetwotypesofeatingbehaviour.The
discoveryofa significant association between restrainedandemotional eating
doesnotnecessarily contradict the suggestion that "obese"eating patterns
precededieting, rather thanbeing causedbydieting — "obese"eating patterns
maycauseweight problems andmakedieting necessary.The suggestionmade here
thatthe restrained eating theoryappears invalidmaypossiblybe premature.
More research isneeded,preferably utilising the final scalesof theDEBQto
reachamore specific conclusion.Asalso suggested byLowe (1984)future
research should, inaddition, take intoaccount theprior and current weight
statusof the subjectsunder study.

Thepsychosomatic and externality theories

As already stated, the supposition that "obese"eating patterns precede

dieting, rather thanbeing causedbydieting, isinagreementwith both the
 -128-

psychosomatic and externality theories.The significant relationships found

betweenBMI and emotional and external eatingbehaviour, andalsobetween the
two typesof eatingbehaviour inthe studycontained inChapter 10,alsolent
some support tothe twotheories.Slochower (1983a), oneof theadvocates of
psychosomatic theory, hasproposed thatemotionality and food cues operate
conjointly toelicit eatingbehaviour:astateofhighuncontrollable anxiety
mayenhance reactions toexternal cues.According toexternality theory,a
close relationshipbetween both typesofeatingbehaviour isalsoassumed if,
as statedby Schachter andRodin (1974)ahighdegree ofemotionalityis
considered tobeamanifestation of the general trait of externality.
However, on thebasisof thepatternof relationships foundbetween the two
typesof eatingbehaviour andvariouspersonality traits inthe study contained
inChapter 10,theauthor is inclined tostress the importance of emotional
eating asa causal factor inthegenesisof overweight. This inclination is
based on the fact thatno significant relationshipswere foundbetween the
scale forexternal eatingand personality scales,which reflected aspects of
externality or emotionality, inthe study inChapter 10.This result isnot in
accordance with the stimulus-binding concept ofexternality theory. Similar
findingshave alsobeen reported byNisbett andTemoshok (1976)and Isbitsky
andWhite (1981). Incontrast, the resultswere inaccordancewith the argument
ofpsychodynamic theory that intrapsychic conflicts arecentral in inducing the
eatingpatterns ofemotional eaters- significant relationshipswere found
between emotional eating andanumber ofpersonality traitswhich reflected
aspects ofemotional instability.
Inthe studyofwomen contained inChapter 11 significant relationships
were also found betweenemotional and external eating.Moreover,both typesof
eatingbehaviour were found tobe related toanxietyandnegative self-concept.
At first sight the significant relationship found inthis study between
external eating andemotional eating (r-.27 (n-540),p-.OO)weakens the current
argument, as this findingdoesnot seemtocontradict the stimulus-binding
concept of theexternality theory, that is,thatahighdegreeof emotionality
is justoneother aspect of thegeneral trait ofexternality. However, ina
post hocanalysis, this relationshipwasmarkedly reduced after adjustments
were made for the effect of emotional eating (r-.12 (n-538)p-.OO). By
contrast, the significant relationship betweenemotional eatingand anxiety
(r-.34 (n-540)p-.OO)wasnotmuch reduced aftermakingadjustment for the
effects ofexternal eating (r-.25 (n-538)p-.OO). These findings suggest that
onlya small part of theexplained variance between external eating and anxiety
 -129-

isdue toexternal eating itself.Emotional eating accotantsfor the largest

partofthevariance, indicating that the relationshipbetweenanxiety and
emotional eating isthemost significant.Asa result, ithasbeen suggested
that thepsychosomatic theoryappears tobe themostvalid.However, it should
be kept inmind that thisconclusionwas reached inpartusing preliminary
scalestomeasure the three typesofeatingbehaviour (e.g., the study in
Chapter 10).More research iscertainlyneeded inorder to reachadefinite
conclusion regarding thevalidityofthethree theoriesof overeating.

Emotional eating

The finding that inwomen emotional eatingwaspositively associated with

personality traits reflecting feelingsof inadequacy» lowself-esteem, low
sociability, emotional instability and social anxiety isconsistentwith the
contention ofthepsychosomatic theory thatovereating isassociated with
emotional andpersonalitydisturbances.Bruch (e.g.,1981)has suggested that
theseproblemshavebeencausedbya lackofadequate parental response to cues
originating inthechild ininfancy. Iftheparental response is continuously
inappropriate,be itneglectful orover-apprehensive, thechild isprevented
fromdeveloping a sense of competency andautonomy. Intheareaof feeding, it
willnot learntodifferentiate between feeling hungryor sated, or as to
whether it suffers fromsomeotherdiscomfort. Inaddition,many other
developmental aspectsmaybedisturbed. Thepositive relationship found between
emotional eatingand fearof failure inthe study contained inChapter 10
supported thiscontention. Fearof failure-characterised asa debilitating
anxiety in relativelyunstructured situations-hasbeenconsidered asbeing
associatedwith adeficient inner cognitive andaffective structure,and has
been conceptualised asbeginning in infancy,a result ofalack of adequate
parental response toexpressions of insecurityoraneed forconfirmation of
attainment ofprescribed standardsby thechild (Hermans, 1971).The positive
relationshipbetween emotional eating and fearof failurepoints toa general
deficit and failure inself-awareness, and supports the suggestion of Slochower
(1983a)thatuncontrollable emotional statesmay result inovereating. Precise-
lyunstructured situationsmayprovokeuncontrollable anxiety,whichmaybe
associatedwith emotional eating inthosehaving adeficient inner cognitive
andaffective structuredue tospecificparent-child interactions early in
life.
 -130-

Remarkably,ahigh fearof failurehasalsobeen found ingirlsbeingat

risk for thedevelopmentofanorexianervosa (Weeda-Hannak, 1984).According to
Bruch (1981), bothanorexianervosapatientsand individualstending toovereat
havemany features incommon, thoughat firstsighttheymayseemtobedirect
opposites.Bothgroupssuffer fromdisturbances inbody imageandbody concept,
and,preoccupied astheyarewithnon-eatingoreating,bothhave incommon an
inabilitytocorrectlyidentifyhungerordistinguishitfromotherbodilyor
emotional tensions.Inaddition,bothgroupshavea "paralyzingsenseof
ineffectivenesswhichpervadesall their thinkingandactions" (I.e.p.216).
Inanorexianervosapatients,however,thesenseofhelplessness iscamouflaged
byexcellent performanceandhighachievementmotivation (Weeda-Nannak, 1984),
whichextendstoeveryareaof life,beingmostnotable inthe refusalof food.
Bycontrast,overeating isthemanifest expressionofalack ofinnercontrol
andwillpower.The studycontained inChapter 10showedthatemotional eating
appeared tobenegativelyassociatedwithachievementmotivation (PMT-P)and
socialascendency (AS).
One findingwhichneedsfurtherconsideration isthatemotionaleatingwas
shown tobemoreprevalent inwomenthaninmen (seeTable 1,Chapter 8 ) .This
sexdifference inemotionaleatingcanbeexplainedbythefollowing lineof
reason.
Oneofthecausesofparental "imperviousness"tothechild'sneedsand
emotionsmaybedissatisfactionwith thesexofthechild (Bruch,1964,p.
270), and researchfindingssuggest thatgirlsarenot somuchwanted asboys
(e.g.,Pohlman,1969;GordonandGordon,1960;Gillman, 1968).Moreover,since
girlsarevalued less forthemselves,beinga "goodgirl"mayassumemore
importance for them.For fearofloosingtheloveofothers,agirlismore
motivated toshowconforminganddependentbehaviour,and lessmotivated to
expressangeroraggressivenessevenwhenmotivated tojustifiableanger.
Dependencyandpassivityalso fit inwith theideal feminine sex-role
stereotype;lestawomanbe thought "masculine",shemustavoid independence,
aggressiveness,competence anddominance.However,dependencyhas itsprice.A
passivedependentattitude islikelytoevokeanxiety, sincetheperson feels
powerlessandhelpless inthe faceoflifeevents (Sherman, 1971). Indeed,
womenhavebeen found tobemoreanxiousandemotionallyunstable (e.g.,
Oetzel, 1966), tohaveatendencytoworrymore (BradburnandCaplovitz,
1965), and tobemore frequentlydepressed thanmen (e.g.,Warheit,Hölzerand
Schwab, 1973). Ithasbeen suggested thatdepression isanorganismic response
to suppressedhostilityandaggression,theoutcomeofabeliefthataction is
 -131-

futile (Seligman, 1975). Likewise,under-orovereating maybe another

organismic response totheperception ofhavingno control over events,given
specific learning experienceswith regard to food.Aswomen tend toperceive
themselvesasbeing lesspowerful incontrolling events thanmen (Olsen, 1969),
thedevelopment of "organismic responses" ingeneral and eatingdisorders in
particular islikely tobemore common inwomen than inmen.
The studyofwomendescribed inChapter 11demonstrated theexistence of a
positive relationship betweenendorsement of feminine stereotype traits and
emotional and external eatingbehaviour. Inaddition, itwas found that the
contribution of femininity toemotional and external eatingwasduemainly to
theanxiety andnegative self-concept associated with female stereotype traits.
Therefore, insupportofthe foregoing argument, itseems that the dependency
andpassivity associated with female stereotype traitsevoke low self-concept
and high anxiety.Moreover, feminine sex-typedwomen seemtobeespecially at
riskwith respect tothedevelopment ofeating disorders.

Successful dietingandweight loss

Some indications astowhich traitsare related toahighdegree of restrained

eatingand successfulweight losshavebeenobtained inthe studyofwomen
contained inChapter 10.Inthis studypersonality differences between "obese"
and "latent obese"womenwereassessed. Bothgroupscouldbeconsidered as
having been confronted withweight problems. Inaddition, bothgroups have a
high tendency toeat in response tonegative emotions and external foodcues.
Incontrast tothe "obese"women,however, the "latent obese"hadby definition
a normal BMI asa result of strict dieting.
Itwas found that "latentobese"womenascomparedwiththe "obese",
functioned better insocial situations,weremore outgoing inattitude,had a
higher self-esteem, higher masculinity of interests and emotions, lower fear of
failure andmore internal control.Thehighermasculinity, lower fear of
failure andhigher internal controlpoint toabetter cognitive and affective
functioning onthepartof the "latentobese".Thishealthier psychic
constellation islikely tobeassociatedwithbetter social functioning and
higher self-esteem, andmaybean important causal factor insuccessful dieting
andweight loss.This supposition isinlinewith the findings of other
studies, inwhich thedegree of internal control hasbeen reported tobean
important predictor oftheeffectivenessofweight reductionprogrammes and
successful weight loss (e.g.,BalchandRoss,1975;Salzer, 1982).A similar
 -132-

hypothesis hasalsobeenput forward byKarpovitz and Zeis (1975),who claimed

thatmental and emotional functioning isoneof the factorsassociated witha
willingness and ability towork onweight control.It isevenmore likely, that
there isa continuous interaction betweenweight lossandpsychological and
social functioning.A womanwho successfully reducesherweight no longer
anticipates negative evaluations ofherappearance byothers, consequently
feelsmore at ease in social situations,and, accordingly,has ahigher
self-esteem.

Implications

For research.Oneoftheseriousmethodological problemsassociatedwith this

study concerned thepossibility ofaheightened prevalence of response sets
amongst the femaleparticipants of the longitudinal study,whichmayhave
confounded thevalidity of the findings.Another problem pertains to the
different versions of theeatingbehaviour questionnaire utilised inthe study
described inChapter 10andtheother studies.Becauseof these methodological
problems,the resultsof the longitudinal studycouldnotbe unequivocally
interpreted, especially incaseswhere thesediffered from thoseobtained in
the cross-sectional study inChapter 10,orgenerallyaccepted theoretical
notions.Moreover, the resultspermitted only tentative conclusions tobemade
with regard tothemainobject ofthe study, theassessment ofthe comparative
validity of the three theoriesofovereating. Forthepurposeof research, it
canbe suggested that the studyonwomen inChapter 10be replicated, using the
finalversion of theDEBQ for theassessment ofeatingbehaviour. Inorder to
reachadefinite conclusion regarding the contention of the theory of
restrained eating that "obese"eatingpatternsare causedbydieting, more
attention should bepaid totheweight history of the subjectsunder study,
preferably bymeansofaweight historyquestionnaire. Inaddition toweight
categories, also the currentweight statusof the subjects should be taken into
account. Inorder todetermine whether counterregulatory eatingbehaviour is
caused by "obese"eatingpatterns, suchasemotional and external eating
behaviour, orby tendencies toward restraint breaking andbingeing, orboth,
separate scales shouldbeused for themeasurement of thetwodimensions.A
scalehas recentlybeendeveloped measuring exclusively restraintbreaking and
bingeing, which facilitates themeasurement ofdisinhibition per se (Witteman,
VanDusseldorp, Frijters andVan Strien, submitted). Inaddition, the supposi-
tion of theauthor thatadeficient inner cognitive and affective structure,
 -133-

and theperception ofhaving nocontrol over events,arecausal factors inthe

development ofeatingdisordersneeds further confirmation.Useof the Rotter
I-Escales,theHerman's PMT,scales forsocialanxiety, self-esteem, masculi-
nityand femininity, and soon,arehighly recommended in future studies.
However, theauthordoesnot favour aone-stage approach inthe assessment
ofpersonality structure and obesity for the following reasons.Firstly, past
research taughtus that conclusions regarding theexistence ofan "obese perso-
nality structure"cannotbederived frompsychometric studiesonthe relation-
shipbetweenweight statusand personality characteristics (e.g.,Leon and
Roth, 1977;McReynolds, 1983;Diehland Paul, 1985).Secondly, ashas hopefully
become clear fromtheabove remarks,aone-to-one assessment of the relation-
shipbetweenpersonality andobesity constitutes anoversimplification of the
problem. Instead,atwo-stageapproach is recommended, inwhich eating
behaviour isconceived ofasbeing themediator betweenpersonality traits and
bodyweight.Noaccountwastakenofthephysicalactivityofthe subjects in
thisstudy.A studyofBaecke,VanStaverenandBurema (1983)using another
sample taken fromtheMunicipality of Ede,comparable regarding ageand social
status tothe longitudinal sampleused inthis study, revealed that the female
populationwasveryhomogeneouswith respect tophysical activity. Accordingly,
it isnotexpected thatphysicalactivity scoreswouldprovidemuch additional
information for thepurpose of this studyusing the samepopulation. However,
as thiscouldwelldiffer amongst otherpopulations, infuture studies
attention should bepaid tothephysical activityof the subjects analysed.
Finally, investigators shouldbeaware ofthe fact that longitudinal
studiesareapt toevoke response sets.To improve the identification and
control of response sets,theuseof independent response-set marker scales is
recommended. However, itwouldbepreferable tocheck self-report against
objective external criteria, as substantive traitscannot alwaysbe separated
from response setsbymeansof independent response-setmarker scales.

Forpractice.The structure ofan individual's eatingbehaviour can in

principle beassessed inavalidmannerusing theDEBQ.Taking additionally
intoaccount theperson'sweight historyand current bodyweight, it becomes
possible todecidewhichparticular typeofweight reductionprogramme can best
2
be recommended .
A highdegree ofemotional eatingpoints toadeficient inner cognitive and
affective structure.For this reason,high scorersontheemotional eating
scalemaybenefit best from therapy,which focusesonpsychological problems
 -134-

rather thanweight.During thetherapeutic process, feelings of inadequacy and

lack of control over events shouldbecounteracted, and treatment should focus
onevoking awareness inthepatient of impulses, feelingsandneeds originating
withinhimorherself.Inmanycaseshighemotional eatingmaybe associated
withhigh levelsofexternal eating,but this sensitivity tofood cuesmay
disappear inthe therapeutic process.However, ifapatient insistsonworking
onweight control,onemay consider sendinghimorher toan independent weight
control programme, based, forexample, on stimulus-control procedures. During
the therapeutic process itself theproblemofoverweight shouldnotbe directly
attacked, since overeating isprobablyonlya symptom ofunderlying emotional
problems; removal of the symptom probably doesnot last long ifthe underlying
cause isleftunsolved (Slochower,personal communication; Bruch, 1977;Wolman,
1982).
A high scorewith respect toexternal eating,unsupported byahigh score
regarding emotional eating,points toa sensitivity toexternal food cues.
Though thepatientmayhave lowself-esteem andhigh social anxiety, this is
more likely tobe the result ofnegative social pressures concerning
overweight, thanunderlying emotional problems.However, this should be checked
before deciding ona stimulus-control programme.
Restrained eating scores shouldnever beconsidered alone, inpersons
demonstrating ahighdegree ofemotional orexternal eatingbehaviour, the
strict restriction of food intake isnot likely to result inlasting weight
loss,unless theunderlying psychic problemsare solved, or the sensitivity to
food stimuli istreated. For this reason, restrained eating scores should
alwaysbeconsidered inconjunctionwith theother scalesof theDEBQ.The
weight history and the currentweight statusof thepatient should also be
taken intoaccount,and, inaddition, thetendency towards restraint breaking
or bingeing (forexample, bymeansof the Loss ofControl Scale for Bulimic
EatingBehaviour (Wittemanet al. submitted).Though the restrained eating
scale hasbeendesigned sothat it refers to "healthydieting", the fact that
someof those having ahigh scoreonthis scale,alsohave tendencies towards
restraint breaking, bingeing oranorexia nervosa, cannot bedisregarded. For
example,underweight incombination with severe dietingmaypoint to anorexia
nervosa,while ahighdegreeofweight fluctuation incombination with severe
dieting and tendencies towards restraint breaking may indicate bulimia nervosa.
Finally, apersonwith ahigh restraint score,but showinga lowlevel of
emotional and external eating,may requiremore accurate information concerning
nutrition and caloric balance.However, if thispersonhasalways been
 -135-

overweight,heor shemaydobetter toaccept hisorher heavybuild, instead

of continuously starvinghimorherself.With respect tothis itshould be
noted that amoreaccepting socialandmedical attitude towards overweight
would alsobehelpful inalleviatingmanyofthepsychological problems faced
by theoverweight, suchassocialanxietyand low self-esteem.

NOTES

1.ithasbeen suggested that individualswith longdietinghistoriesneed less

energy compared with their equalweight-height peerswhohave notengaged in
similar dieting activities (Garrow, 1978).A studybyRavussin, Burnand,
SchutzandJéquier (1985)has shown that thisdecrease inoverall energy
expenditureduring caloric restrictionwasdue tothree factors:thelossof
leanbodymass (whichexplains the lower restingmetabolic rate), the lower
thermogeneticeffect of foodwhich isrelated tolowerenergy intake,and
the reduced cost ofphysical activity, a resultof thelowerbodyweight. On
thebasisof these findings,the investigators concluded that "...there
seems little reason toevoke additionalmechanisms toexplaindecline in
energy expenditure during dieting".
2. It should bekept inmind that thisprocedure innowayaccounts for
physiological factors inweight gainorobesity, sothat inthecase of
doubt hormonal,metabolic disturbances and such like shouldbe separately
assessed. However, theremaybean interaction ofphysiological disturbances
andpsychological problems.
 -137-

SUMMAKY

Inthis study, three theories onthedevelopmentandmaintenance ofhuman

obesity are investigated. These theoriesare thepsychosomatic theory, the
externality theoryand the theoryof restrained eating.The psychosomatic
theory focusesonemotional factors,andattributes overeating to confusion
between internal arousal statesaccompanying emotional statesandphysiological
statesofhunger and satiety. Individuals having the tendency toeat in respon-
se toemotional statesare considered tobeunadjusted and tosuffer fromun-
stable emotionality. Externality theory focusesonexternal food cues,and
attributes overeating toahyper-responsiveness tofood-related cues inthe
environment togetherwithunresponsiveness tointernal cuesofhunger or
satiety.This tendency isconsidered tobeamanifestation of thegeneral trait
of externality. The theoryof restrained eating focuseson sideeffectsof
dieting, that is,thepossiblebreakdown of restrictive control sothat sup-
pressed eating behaviour isdisinhibitedandexcessive food intakeoccurs.
Psychosomatic theoryemphasizes internal instigation ofeating and externa-
lity theory, focusesonexternal instigation ofeating.Boththeories contend
thatdieting results fromovereating andweight gain,whereas according to the
theoryof restrained eating,dietingmaylead toovereating andweight gain.As
these theories differ inassumptionswhy individuals overeat, it isdifficult
todetermine howovereating or overweight canbeadequately treated.Thus, the
principal aimof this studywas totestanumber ofhypothesesevoking from
these theories.Thiswasdoneby reviewing the literature on these theories
(Part Iof thisdissertation)and subsequently bycarryingouta seriesof psy-
chometric studiesonthe relationships between thethree typesofeating beha-
viour central to these theories (emotional,external and restrained eating
behaviour)andvariables, suchaspersonality traitsandbodymass (Part IIof
thisdissertation).

Part I.literature review

The processesdescribed byeachof the three theories,and theprinciples of

weightmanagement asadvocated bythesetheorieshavebeen reviewed. However,
no conclusion could bedrawn concerning thevalidity of these theories because
thewhole rangeofhypothesesderived fromeach theoryhasasyetnot been
confirmed byempirical studies.Most clinical studiesgive support tothe
psychosomatic theory, inthatmost individualsovereatwhen feeling emotionally
 -138-

upset,but these findingscouldnotbe replicated incontrolled laboratory

studies.However,with the introduction of thedistinction between the concepts
of clearlylabelled anddiffuseemotions,overweight individualswere shown to
overeat inresponse tocertain typesof stressors incontrolled studies.Also,
overeatinghasbeen shown toexertananxiety-reducing function.
Thedistinction between internal andexternal cues inexternality theorywas
found tobe toosimple anexplanation of thedifference ineating behaviour
between normal and obese subjectsbecause individualswhoare hyperresponsive
toexternal stimuli orunresponsive tointernal stimuli are found inallweight
categories.Thediscovery ofnormalweight individualswithobese eating
patterns, referred toas "latentobese",and theobservation thatbodyweight
isinfluenced byphysiological variables,offersadditional support tothe fact
that the relationshipbetweenexternal responsiveness and food intake and body
weight ismediated by cognitive andphysiological variables.
The theoryof restrained eatingwascorraborated instudieson restrained
non-obese individuals,butnot instudieson restrained obese individuals.This
ambiguity infindingsmayhave resulted fromdifficulties inmeasuring
cognitive restraint butmayalso reflect inconsistencies inthetheoryof
restrained eating.

Part II.Psychometric studies

A number ofpsychometric studieswas carriedout toassess the relationships
between the three typesofeatingbehaviour andvariables suchas personality
traitsandbodymass.Asno satisfactory scaleswere available,the Dutch
Eating Behaviour Questionnaire (DEBQ)comprising scales foreachof the three
typesofeatingbehaviourwasdeveloped. Each scalewas showntohave high
internal consistency, factorialvalidity anddimensional stability.The scales
for restrained and external eatingwere found tobeunidimensional.The scale
foremotional eating contained twodimensions,onedealingwitheating in
response toclearly labelled emotions and theotherwith eating in response to
diffuse emotions.This two-dimensional scalewasused inthegeneral assessment
of emotional eating,but separate scaleswere also constructed for clearly
labelled anddiffuse emotions (Chapter8 ) .
When testedonapopulation ofwomen, theRestrained eating scaleof the
DEBQwas found tohavemoderate togoodpredictivevalidity for restriction of
food intake.Highnegative relationshipswereobserved between scoreson the
scaleand intakeof fatand sugar.The relationshipbetween restrained eating
scoresandmagnitude of thedeviation fromenergy requirement (an indirect
 -139-

measureofdiscrepancy betweenactual anddesired intakeof energy),was also

high (Chapter9 ) .
Thevalidity ofthe threepsychological theoriesofovereatingwas investi-
gated ina sampleofwomenbyexamining the relationshipsbetween the three
eatingbehaviour scales,personality scalesandbodymass index (BMI;
2
weight/height ).Toascertainwhichpersonality traitswere related to
successfulweight loss,differences ineatingbehaviour andpersonality were
investigated between overweightwomen,andwomenofnormalweightwho had
previously beenoverweight: "obese"and "latentobese"women (Chapter10).
No significant relationshipwas found between the scale for restrained
eating, and that foremotional or that forexternal eatingbehaviour.These
findingsdonotoccurwith thetheoryof restrained eating,which suggests that
dieting isa causal agent ofemotional andexternal eatingbehaviour, that is
"obese"eating patterns.Incontrast, indicationswere found that "obese"
eatingpatternsprecede restrainedeatingbehaviour, becausenodifference was
found inthedegree ofemotional orexternal eatingbehaviour between the obese
and the latent obesewomen.Thisfinding isinlinewithbothpsychosomatic and
externality theory.However,on thebasisof thepattern of relationships
between the scales foremotional and external eatingbehaviour theauthor is
inclined tostress the importance ofemotional eatingasa causal factor in the
genesis ofoverweight: nosignificant relationshipswere foundbetween the
scale forexternal eatingand personality scales reflecting aspectsof
externality or emotionality inthis study.Such findingsarenot in accordance
with the contention ofexternality theory that sensitivity tofood cues isonly
amanifestation ofexternalityasageneralpersonality trait.Results support
instead psychosomatic theory, that intrapsychic conflicts induce theeating
patternsofemotional eaters,because significant relationshipswere found
between emotional eatingandanumber ofpersonality scales reflecting aspects
of emotional instability.
Inspectionofpersonalitydifferencesbetween the "latent obese"and the
"obese"inChapter 10 indicated that sometraitswere related toahigh degree
of restrained eatingand successfulweight loss.As compared with the "obese",
"latent obese"women functionedmore effectively insocial situations,were
moreoutgoing inattitude,hadhigher self-esteem,ahigherdegreeofmasculi-
nity, less fear of failure andmore internal control.Highermasculinity, less
fear of failure andhigher internal control indicate abetter cognitive and
affective functioning of the "latent obese",and thismaybean important
causal factor for successful dieting andweight loss.Alternatively, better
 -140-

psychic functioning mayalsohavebeen the resultof successfulweight loss,

because oftheanticipation ofamorepositive evaluation of their appearance
by others.A thirdpossibility isa continuous interactionofweight loss,and
psychological and social functioning.
Psychic functioningwas found tobe thecentral factor inthedevelopment of
eating disorders inChapter 11.Intwo independent subsamples ofwomen,positi-
ve relationshipswere found between emotional and external eating behaviour,
and female stereotype traits.Thiscouldbeattributed mainly toanxiety and
negative self-esteem associated with female stereotypical traits.Specific
learning experiences early inlifewith regard to food,dependency/ lackof
assertiveness andpassivity (infemale stereotypes)andassociated anxiety and
negative self-concept may result inthedevelopment ofeatingdisorders asan
"organismic" response totheperception ofhavingno control over events.
A longitudinal studyoneffectsofnegative lifeeventson change inbody
mass indexinmenandwomen classified asloworhighemotionaleater is repor-
ted inChapter 12.Inmen,highemotional eaterswere found togainmore
weight, and lowemotional eaters togainlessweight after experiencing negati-
ve life events that thosewhohadnotexperienced suchevents.This is in line
with the clinical observation that some individuals, referred toas emotional
eaters, respond to stresswith excessive intakeof food,whilemost individuals
respond to suchcircumstanceswith lossofappetite and subsequent weight loss.
Similar resultswere not found inwomen: Inlong-termhighemotional eaters
gained moreweight after experiencing negative lifeeventsthanthosewho had
not experienced suchevents,but lowemotional eatersdid not show the "normal"
response of lessweight gainafter stressful life events.A possible explana-
tionmaybe that theemotional eating scalehasbeenless successful inclassi-
fyingwomen asloworhighemotional eaters.As isshown inChapter 13,there
are indications inthe femaleparticipants inthe longitudinal study of the
response sets,acquiescence and socialdesirability. Intensive contact between
subject and investigator, andalso sensitization of subjects tothe question-
naires asa result of repeatedmeasurement may have led to increased willing-
ness toco-operateand asa consequence, tosocialacquiescence inthewomen of
normalweight.However, increased concern for self-presentation and thus social
desirability asa response setmayhavebeenmoreprevalent inobesewomen.
Thus, thevalidity of resultsobtained withaparticular measurement instrument
mayhavebeen reduced, and thismayhavebeen thecasewith the results of the
female participantsof thislongitudinal study inChapter12.
 -141-

Inthegeneraldiscussion ofthe resultsofthestudies, it isconcluded

thatpsychosomatic theoryseems tobe themostpromising theory regarding the
maintenance and thedevelopment ofhumanobesity.However,because of the
possibleprevalence of response sets inthe femaleparticipants inthe
longitudinal study thiscanonlybeatentative conclusion. Implications of the
research findingsfor future researchand forclinicalpracticeare discussed.
The structureofan individual's eatingbehaviour can inprinciple be.assessed
with theDEBQ, and therefore thisquestionnaire maybeuseful indetermining
which particularweight reductionprogrammewouldbemost appropriate.
 -142-

SAMENVATTING

Centraal indit proefschrift staandriepsychologische theorieën over het

ontstaan enhethandhavenvanovergewicht:depsychosomatische theorie,de
externaliteitstheorie ende theorievan lijngerichtheid. Elkvandeze theorieën
verschilt op fundamentele puntenvandeandereenookde behandelingsstrategie
isbinnen elk theoretischuitgangspuntweer anders.Doelvanditonderzoek was
derhalve na tegaanwelkevande theorieënhetmeest geldig lijkt.Dit gebeurde
optweewijzen:a. literatuurstudie (Deel Ivanditproefschrift),b.het
uitvoerenvaneen reekspsychometrische studiesnaarverbanden tussende
eetgedragingen diehetmeest kenmerkend zijnvoor elkvandedrie theorieën
(emotioneel eten, "extern"etenen lijngericht eten)en persoonlijkheidskenmer-
ken,matevanovergewicht endergelijke (Deel IIvanditproefschrift).

Deel I. literatuurstudie

Depsychosomatische theorie.Een "normale" reactieopnegatieveemotiesof

stress isvermindering vaneetlust:stress roeptdezelfde verzadigingsver-
schijnselen opalsopnemingvanvoedsel;de samentrekkingen vandemaag houden
openhetbloedsuikergehalte gaatomhoog.Er zijnechtermensendiena stress-
ervaringen juist reagerenmethet etenvanvoedsel.Deze,zogenaamde emotionele
eters,hebbenvolgensdepsychosomatische theorieniet leren discrimineren
tussenhongerprikkels en fysiologischeprikkelsbijemotiesalsangst, boosheid
ofvrees,onder andere tengevolgevandewijzevanopvoeden door demoeder die
het kind ook teetengafwanneerhetomandere redenendanhonger huilde.Veel-
vuldigemotioneel etenkan leiden totgewichtstoeneming enuiteindelijk tot
overgewicht.Volgensdeze theorieën functionerenpersonenmet eendergelijk
eetgedragpsychischnietgoedenbevinden zijzichdikwijls ineensituatievan
stress.
Deexternaliteitstheorie.Deexternaliteitstheorie heeftmet de
psychosomatische theoriegemeendatovereten kanontstaan door het onjuist
interpreterenvan sensatiesvanhongerenverzadiging. Integenstelling totde
psychosomatische theorie, schrijft deexternaliteitstheorie overetenniet toe
aaneengebrek aanonderscheidend vermogen tussen fysiologische symptomen bij
emotiesen fysiologische symptomen bijhonger enverzadiging,maar aan een
algemene gevoeligheid voor externe prikkels,waarvan de gevoeligheid voor
voedselprikkels, zoalsgeuren smaakvanvoedsel, slechtseenonderdeel is.Er
wordt vanuit gegaandatexternaliteit eenpersoonlijkheidskenmerk is.
 -143-

Detheorievan lijngericht eetgedrag.Volgensde theorievan lijngericht

eetgedrag heeft iederpersooneeneigen "natuurlijk gewicht", eenvast gewicht
datdoor het lichaam ineenhomeostatische regelkring instandwordt gehouden.
Oppogingen totverlagingvandit "natuurlijke gewicht"door lijnen reageert
het lichaammet fysiologische defensiemechanismen, bijvoorbeeld honger.Er is
dan sprakevaneenconflict tusseneetlust envoedselonthouding, tussendewil
omteetenendewil omaftevallen.Wordtde cognitieve controleoverhet
eetgedrag doorbrokendoor bijvoorbeeld het etenvan "verboden"voedsel,dan
geven lijnersveelalaanhuneetlust toeenkomthetbij sommige lijnersvoor
dat zegeweldigehoeveelhedenvoedsel consumeren (counter-regulatie).Naast
"verboden"voedsel kunnenooknegatieve emotiesenexternevoedselprikkels de
cognitieve controleopheteetgedragbij lijnersdoorbreken.Het voortdurend
negerenvan gevoelensvanhonger kanbovendien leiden toteen blijvend
onvermogen om interne fysiologische sensatiesvanhonger enverzadiging opde
juistewijze te interpreteren.Volgensde theorievan lijngericht eetgedrag is
lijneneenmogelijk oorzakelijke factorvanemotioneel enexterneten-ookwel
"obees"eetgedrag genoemd. Indit opzichtverschilt de theorievan lijngericht
eetgedragvan zoweldepsychosomatische alsdeextemaliteitstheorie;deze
theorieën zien lijngedrag alseenmogelijk antwoord opovergewicht datop zijn
beurtdoor "obees"eetgedrag isveroorzaakt.
Omdat dedrie theorieënelkeeneigen zienswijzehebbenophetontstaan van
overeten enovergewicht, isookdebehandelingsstrategie binnenelk theoretisch
uitgangspunt weer anders.Indevisievandepsychosomatische theorie zalhet
emotionele etenpasverdwijnen alsonderliggende emotionele problemen zijnop-
gelost (inpsychotherapie).Indebehandelingvolgensde extemaliteitstheorie
staateengedragstherapeutische benadering gericht opstimulus-contrôle cen-
traal. Indebenaderingvolgensde theorievanlijngericht eetgedrag ligthet
accent ophet (opnieuw)aanlerenvan sensatiesvanhonger enverzadiging, en
het lerenaccepterenvanhet eigen "natuurlijke" gewicht.
Deuitkomstenvandebelangrijkste onderzoeken ophetgebiedvandedrie
theorieën gevengeenuitsluitselwelkevande theorieënhetmeest ondersteund
wordt.Geenderhypothesenvoortkomend uitdezedrie theorieënwerd inonder-
zoekopconsistentewijzebevestigd. Bevestiging voor de psychosomatische
theoriewerdwelgevonden inklinischpsychologisch onderzoek -demeeste per-
sonen rapporteerdenveel teetenals zij ineen toestandvannegatieve emoties
verkeerden -maar eendergelijk eetgedrag konnietaltijdworden aangetoond in
experimentele studies.Nogmaar sindskort,nade invoeringvanhet onderscheid
tussendebegrippen "duidelijk omschreven"en "diffuse"emoties,bleekook in
 -144-

laboratorium-situatiesdatdikkemensenveel eten instress-situaties waarin

diffuse emotiesopgewektwerden endatdit eetgedrag eenangst-reducerende
functie heeft.
Ook dedichotomie interneversusexterneoriëntatie inde externaliteits-
theorie bleek tesimpel voor hetverklarenvanverschillen ineetgedragvan
personenmet eennormaalgewicht tegenoverdatvanpersonenmet overgewicht; in
elkegewichtsklasse bleken sommige personen overgevoeligvoor externe prikkels
enongevoeligvoor interneprikkels tezijn.Erblekenook personen tebestaan
met eennormaal lichaamsgewicht diewél "obees"eetgedragvertoonden.Deze per-
sonenworden "latent obesen"genoemd.Dezebevindingènhet inzichtdat iemands
uiteindelijke lichaamsgewicht ookwordtbepaalddoor fysiologische factoren,
heeft duidelijk gemaaktdatergeen rechtstreeksverband bestaat tussenmate
vanexternaliteit enerzijdsende consumptie vanvoedsel en lichaamsgewicht
anderzijds.Deze relatiewordtmedebepaalddoor cognitieve factoren (b.v.het
besluit te lijnen)en fysiologische factoren.
De theorievan lijngericht eetgedrag, tenslotte,werdwel bevestigd in stu-
diesmet lijnende proefpersonenmeteennormaal lichaamsgewicht, maar niet in
studiesmet lijnendeproefpersonenmet overgewicht.Er zijnaanwijzingen dat
deze tegenstrijdigheid in resultaten toegeschrevenmoetworden aanproblemen
met hetmeetinstrument voor hetbepalenvan lijngerichtheid, de "Restraint
Scale".Daarnaast ishetmogelijkdatde theorievan lijngericht eetgedrag
slechtsvantoepassing isvoor eenbepaalde soortvanpersonen.

Deel II.Eigenempirisch onderzoek

Het tweededeelvanditproefschrift bevat zespsychometrische studiesnaar

emotioneel,externen lijngericht eetgedrag.Allereerstmoest ereenmeetin-
strument voordezeeetgedragingenworden ontwikkeld, aangezien een bevredigend
instrument niet beschikbaarwasbijhetbeginvanditonderzoek.Hoofdstuk 8
beschrijft de constructie vandit instrument, deDutchEating BehaviourQues-
tionnaire (DEBQ)ofweldeNederlandseVragenlijst voor Eetgedrag (NVE)met
schalenvoor emotioneel,externen lijngericht eetgedrag.Hetbleek dat deze
schalen eenhogematevan interne consistentie, factoriëlevaliditeit endimen-
sionele stabiliteitvertoonden.Verderblekende schalenvoor lijngericht en
extern eetgedraguni-dimensioneel,maarde schaalvoor emotioneel eten bi-di-
mensioneel te zijn.De eerstedimensiehadbetrekking opetenbij duidelijk
omschrevenemotiesende tweedevooral opetenbijdiffuse emoties.Hoewel er
ook aparte schalen zijngeconstrueerdvoorhetmetenvanelkvandeze twee
 -145-

dimensies, isinditproefschrift de twee-ditnensioneleschaal gebruiktvoor het

metenvanhetetenbijnegatieveemoties.
Hoofdstuk 9"bevateen studieoverdepredictievevaliditeit vande schaal
voor lijngericht eetgedrag ineen steekproefvanvrouwen.Hetbleekdat deze
schaal een redelijke totgoedepredictieve validiteit heeftvoor hetmetenvan
beperking vanvoedselinneming:erwerdenhogeverbandengevonden tussen scores
opdeze schaal endegerapporteerde consumptie vanvetten en suikers.Daarnaast
wasook hetverband tussen scoresopdeze schaal enmatevandeviatievanener-
giebehoefte goed.Dit iseen indirectemaatvoor lijngerichtheid, d.w.z.de
matewaarin iemandminder eetdanhijof zijwel zouwillen eten.
Kernvanditproefschrift isHoofdstuk 10.Ditbevat eenonderzoek ineen
steekproefvanvrouwennaardegeldigheid vandedriepsychologische theorieën
vanovereten enovergewicht.Tevenswerd inditonderzoek nagegaan welke
persoonlijkheidskenmerken samenhangenmet succesvol gewichtsverlies.Dit ge-
beurdedoor eenanalysevandeverbanden tussendedrie eetgedragingen,
persoonlijkheidskenmerken enmatevanovergewicht (bodymass index (BMI);
2
gewicht/lengte )endaarnaastdoorna tegaanopwelkepunten (eetgedrag en
persoonlijkheidskenmerken) dikkevrouwenverschillen vanvrouwen dieooit te
dik zijngeweestmaarophetmomentvanhetonderzoek eennormaal lichaamsge-
wicht haddendoor succesvol lijngedrag: "obese"en "latentobese"vrouwen.
Hetbleekdat scoresopde schaalvoor lijngericht eetgedragniet samenhin-
genmet die opde schalenvoor emotioneel enexterneetgedrag.Dit resultaat
komtniet overeenmetdetheorievan lijngericht eetgedrag,waarin lijngedrag
gezienwordt alsmogelijke oorzaakvoor emotioneel-enextern-, "obees",
eetgedrag.Wélwaren eraanwijzingen dat "obees"eetgedragvooraf gaat aan
lijnen, aangezien "obese"en "latentobese"vrouwen inongeveer gelijke mate
eengeneigdheid totemotioneel en "extern"etenbleken te rapporteren.Dit
resultaat isinovereenstemming met zoweldepsychosomatische alsdeexternali-
teitstheorie. Echter,hetpatroonvanverbanden tussende schalenvoor emotio-
neel enextern etenenerzijdsenverschillende persoonlijkheidsschalen ander-
zijdsdoetvermoedendathet etenbijnegatieve emotieswel eensde belangrijk-
steoorzakelijke factorvoor overgewicht zoukunnen zijn.Ditvermoeden isge-
stoeld ophet feitdat er indit onderzoek geenverbandenwerden gevondentus-
sende schaalvoor externeetgedrag enpersoonlijkheidsschalen voormatevan
externaliteit enemotionaliteit.Dezebevinding isintegenspraak met de ziens-
wijzebinnendeexternaliteitstheorie dateengevoeligheidvoorexterne voed-
selprikkels slechtséénvandeuitingen isvaneenalgemene gevoeligheid voor
externeprikkels,vanexternaliteit alspersoonlijkheidskenmerk dus.Daaren-
 -146-

tegenwarende resultatenwel inovereenstemmingmet het gezichtspunt vande

psychosomatische theoriedat een slechtpsychisch functioneren een centraal
element isbijhet tot stand komenvanhet specifieke eetgedragvan emotionele
eters,aangezien er significante verbandenwaren tussende schaalvoor emotio-
neel eetgedrag enpersoonlijkheidsschalen voor emotionele instabiliteit, angst
ennegatief zelfbeeld.
Deverschillen inpersoonlijkheidsstructuren tussen "obese"en "latent
obese"vrouwen inhet onderzoek inHoofdstuk 10,gevenenkele aanwijzingen om-
trentde soortvanpersoonlijkheidskenmerken diesamenhangenmeteenhogemate
van lijngerichtheiden succesvol gewichtsverlies.Hetbleekdat invergelijking
met "obese"vrouwen "latentobese"vrouwenbeter functioneren insociale situa-
ties, sociaal actiever zijn,eenhoger gevoelvaneigenwaarde hebben,eneen
meermasculine instelling, eenhogerematevaninterne controleeneen lagere
matevan faalangstvertonen.Dehogerematevanmasculiniteit, de lagere faal-
angst endehogerematevan interne controlewijzenopeenbeter cognitief en
affectief functionerenvan "latentobese"vrouwen.Mogelijk zijndit belang-
rijkeoorzakelijke factorenvoor succesvol lijngedrag engewichtsverlies.Het
kanechter nietuitgeslotenwordendat beterpsychisch functioneren een gevolg
isvan succesvol gewichtsverlies,aangezien iemand nagewichtsverlies inminde-
rematedanvoorheennegatieve opmerkingenvananderen overhetuiterlijk moet
ondergaan.Tenslotte isookeenwisselwerking tussengewichtsverlies en psy-
chischen sociaal functionerenmogelijk.
Ook resultaten inHoofdstuk 11wijzenophetbelangvanhetpsychisch func-
tioneren bijproblemenmet eetgedrag. Intweeonafhankelijke steekproeven van
vrouwenwerd ereenpositiefverband gevonden tussenenerzijds emotioneel en
extern etenenanderzijdsvrouwelijke stereotype eigenschappen,verder kwam
naarvorendatditverbandvooral toegeschrevenmoetwordenaanangst ennega-
tiefzelfbeeld inhetvrouwelijke sexe-stereotiep.Dezegegevenswijzen erop
dat, gegevenbepaalde leerervaringen indevroege jeugd,eetproblemen mogelijk
ontstaanals "organische reactie"ophetgevoeldeeigensituatieniet inde
hand tehebben,door gebrek aanonafhankelijk enassertief gedrag,angst eneen
negatief zelfbeeld.
Hoofdstuk 12beschrijft resultatenvaneen longitudinaal onderzoek naar ef-
fectenvannegatievelevensgebeurtenissen opverandering inlichaamsgewicht bij
mannenenvrouwendieals lageofalshogeemotionele eterwarengeclassifi-
ceerd.Bijmannenwerd gevondendathoge emotioneleetersmeer enlageemotio-
neleetersminder ingewicht toenamennadat zijnegatieve gebeurtenissen hadden
meegemaakt,danwanneer zijdergelijke gebeurtenissen niethadden meegemaakt.
 -147-

Dit resultaat isinovereenstemmingmetdeobservatie inde klinische praktijk

dat sommigemensen,deemotioneleeters,opstress reagerenmetexcessief eet-
gedrag, terwijlverminderingvaneetlust (engewichtsverlies)de "normale"
reactie opzo'n situatie is.Bijvrouwenwerdendergelijke resultaten evenwel
niet gevonden:oplange termijnblekenhogeemotionele eterswel zwaarder te
wordenals zijnegatieve gebeurtenissen haddenmeegemaakt,maar bij lageemo-
tionele eterswerd de "normale"reactievangewichtsverliesnanegatievege-
beurtenissen niet aangetroffenbijvrouwen.Eenmogelijkeverklaringvoor dit
resultaat is,datde schaalvooremotioneeletenopminder succesvolle wijze
vrouwenals lageofhogeemotionele eterheeft gekenmerkt. InHoofdstuk 13komt
naarvorendathet antwoordgedrag vandevrouwen inde longitudinale studie
mogelijk voor eenbelangrijk deelbepaald isdoorde antwoordtendenties
"acquiescence" ("ja"-zegtendentie)en socialewenselijkheid. Het intensieve
contact tussen onderzoeker enonderzochte endegroterematevan test-ervaring
vandeproefpersonen inhet longitudinale onderzoek doordeherhaalde metingen,
heeftmogelijk bijdevrouwenvannormaal gewicht geleid toteenverhoogdebe-
reidwilligheid bijhetonderzoek behulpzaam te zijnendaarmeeeenverhoogde
"ja"-zeg tendentie (socialacquiescence).Daarentegenwarenvrouwenmetoverge-
wicht erwaarschijnlijk meer opgericht goedvoordedag tekomenen eerder
geneigd sociaalwenselijke antwoorden tegeven.Uiteraardwordt devaliditeit
vande resultatenmeteenbepaaldmeetinstrumentdoordeze antwoordtendenties
verlaagd.Mogelijk isdithet geval geweestbijde resultatenbijvrouwen in
het longitudinale onderzoek inHoofdstuk 12 (bijmannen zijndergelijke ant-
woordtendenties nietaangetroffen).
Het onderzoeksgedeeltewerdafgeslotenmet eenalgemenediscussie vande zes
studies.Geconcludeerdwordt,datdepsychosomatische theoriedemeeste steun
verdient,maar hierbijdientbenadrukt tewordendatdezeconclusie slechts
voorlopig kan zijnvanwegedemogelijke aanwezigheid vanantwoordtendenties in
hetantwoordgedrag vandevrouwenuithet longitudinale onderzoek (deonderzoe-
ken inHoofdstuk 8, 11en 12).Voor eendefinitieve conclusie ismeer onderzoek
vereist enenkele suggestiesvoorverder onderzoekwordengegeven.Tenslotte
wordt aandacht besteed aanmogelijke consequentiesvande resultatenvandit
onderzoek voordeklinische praktijk:deNederlandseVragenlijst voor Eetgedrag
maakt hetvaststellenvan iemandseetpatroonmogelijk enditvormt een goede
basisvoor het bepalenvandemeest geschiktebehandelingsvormvan een
specifiek eetprobleem.
 -149-

Abramson,E.E.&Stinson,S.G. (1977).Boredomandeatinginobeseandnonobese
individuals.AddictiveBehaviors,2,181-185.
Abramson,E.E.&Wunderlich,R.A. (1972).Anxiety,fearandeating :A testof
thepsychosomaticconceptofobesity.JournalofAbnormalPsychology,79,
317-321.
Allen,V.L.&Crutchfield,R.S. (1963).Generalisationof experimentally
reinforced conformity.JournalofAbnormalandSocialPsychology,67,
326-333.
Allon,N. (1982).The stigmaofwomen ineverydaylife.In:Wolman,B.B.
(Ed.).Psychological aspectsofobesity:ahandbook.NewYork:VanNostrand
ReinholdCompany.
Andriessen,J.H.T.H. (1972). Interneofexternebeheersing.Nederlands
TijdschriftvoordePsychologie,27,173-198.
Atkinson,R.M. &Ringuette,E.L. (1967).A surveyofbiographical and
psychological features inextraordinary fatness.PsychosomaticMedicine,
29,21-133.
Baecke,J.A.H. (1982).Determinantsofbodyfatnessinyoungadultsliving ina
Dutchcommunity.DoctoralDissertation,Agricultural University.
Baecke,J.A.H., Burema,J., Frijters,J.E.R.,Hautvast,J.G.A.J.&vander
Wiel-Wetzels,W.A.M. (1983).Obesity inyoungDutchadults:I,
sociodemographicvariablesandbodymassindex.International Journalof
Obesity,7,1-12.
Baecke,J.A.H., Burema,J., Frijters,J.E.R.,Hautvast,J.G.A.J.&Vander
Wiel-Wetzels,W.A.M. (1983).Obesity inyoungadults:II,daily life-style
andbodymassindex.InternationalJournalofObesity,7,13-24.
Baecke,J.A.H.,VanStaveren,W.A.&Burema,J. (1983).Foodconsumption,
habitualphysicalactivityandbodyfatnessinyoungDutchadults.American
JournalofClinicalNutrition,37,278-286.
Bakan,D. (1966).Thedualityofhumanexistence. Chicago:RandMcNally.
Balch,P.&Ross,W.A. (1975).Predicting successinweight reductionasa
functionoflocusofcontrol:anunidimensionaland multidimensional
approach.JournalofConsultingandClinicalPsychology,43,119.
Beaudoin,R.&Mayer,J. (1954).Food intakesofobeseandnon-obesewomen.
JournaloftheAmericanDieteticAssociation,29,29-33.
 -150-

Beck,J.B.,Ward-Hull,C.J.&McLear,p.H. (1976).Variables relatedtowomen's

somaticpreferencesofthemaleand femalebody.JournalofPersonalityand
SocialPsychology, 34,1200-1210.
Bern,S. (1974). Themeasurement ofpsychologicalandrogyny.Journal of
ConsultingandClinicalPsychology, 12,155-162.
Bern,S. (1975). Sex roleadaptability:one consequenceofpsychological
androgyny.JournalofPersonalityandSocialPsychology, 31,634-643.
Bern,S.L. (1977).On theutilityofalternativeprocedures forassessing
psychological androgyny.JournalofConsultingandClinical Psychology,45,
196-205.
Bern,S.&Lenney, E. (1976). Sex typingand theavoidanceof cross-sex
behavior.JournalofPersonalityandSocialPsychology, 33,48-54.
Bentler,P.M., Jackson,D.N.&Messick,S. (1971). Identificationofcontent
and style:a two-dimensionalinterpretationofacquiescence.Psychological
Bulletin,76,86-204.
Bernard, L.C. (1980).Multivariate analysesofnewsex role formulationsand
personality.JournalofPersonalityandSocialPsychology, 38,323-336.
Blanchard, F.A. &Frost,R.O. (1983).Two factorsof restraint:concernfor
dietingandweight fluctuation.Behaviour ResearchandTherapy,21,
259-267.
Block,G. (1982).A reviewofthevalidationofdietaryassessmentmethods.
American JournalofEpidemiology, 115,492-505.
Borgers,H.&Roskam, E. (1967). Factor rotationtocriterionpattern (ROTA
01).A programdevelopedattheDepartment ofMathematicalandStatistical
Psychology,Nijmegen,universityNijmegen.
Bradburn,N.&Caplovitz,D. (1965). Reportsonhappiness:apilot studyof
behaviour related tomentalhealth.Chicago:Aldine.
Brand,J. &Clotz,B. (1982). ZumBegriffderExtemalitat:eineGegenüberstel-
lungvon SchachtersAdipositastheorie,Rotters Kontrollerwartungskonzept
undAttributionstheorie.PsychologischeBeiträge,24,181-198.
Bray,G.A. (1976). Theobesepatient.Philadelphia:W.B.Saunders.
Brehony, K.A. &Geller,E.S. (1981). Relationshipsbetweenpsychological
androgyny,socialconformity,andperceived locusofcontrol.Psychologyof
WomenQuarterly, 6,204-217.
Bruch,H. (1957).The importance ofoverweight.NewYork:Norton.
Bruch,H. (1961).Transformationoforal impulses ineatingdisorders:A
conceptual approach.PsychiatricQuarterly, 35,458-481.
 -151-

Bruch,H. (1964).Psychological aspects inovereating and obesity.

Psychosoroatics,5,269-274.
Bruch,H. (1973).Eatingdisorders.NewYork:BasicBooks.
Bruch,H. (1977).Anorexianervosaand itstreatment.Journal ofPediatric
Psychology, 2,110-112.
Bruch,H. (1981).Developmental considerations ofanorexianervosa andobesity.
CanadianJournalofPsychiatry, 26,212-217.
Cabanac,M.&Duclaux,R. (1970).Obesity:Absence ofsatietyaverson to
sucrose.Science,168,496-497.
Cabanac,M.,Duclaux,R.&Spector,N.H. (1971). Sensory feedback in regulation
ofbodyweight:Isthereaponderostat?Nature,229,125-127.
Cannon,W.B. (1915).Bodilychanges inpain,hunger, fearand rage.NewYork:
Appleton, 2nd ed.
Carlson,A.J. (1916).Thecontrolofhunger inhealthanddisease.Chicago:
universityofChicagoPress.
Carlson,R. (1971).Sexdifferences inego functioning:exploratory studiesof
agencyandcommunion.JournalofConsulting andClinical Psychology,37,
267-277.
Castelnuovo-Tedesco,P.&Schiebel,D. (1975). Studiesof superobesity: I.
Psychological characteristicsof superobesepatients.International Journal
ofPsychiatry inMedicine,6,465-480.
Chemin, K. (1981).Theobsession.NewYork:Harper Row.
Clarey,J.H.&Sanford,A. (1982).Femalecareerpreference and androgyny.
VocationalGuidanceQuarterly, 30,258-264.
Cobb,S. (1976).Social supportasamoderator oflife stress.Psychosomatic
Medicine, 38,300-314.
Committee onCaloricRequirements. (1950).CaloricRequirements.FAQ
Nutritional studiesNo.5,June.
Cronbach, L.J. (1959). Further evidenceon response setsandtestdesign.
Educational andPsychologicalMeasurements,10,3-31.
Crowne,D.P.&Marlowe,D. (1964).Theapprovalmotive.NewYork:Wiley.
Cunningham,W.H.,Cunningham, L.C.M.&Green,R.T. (1977).The ipsative process
to reduce response setbias.PublicOpinionQuarterly, 41,379-384.
DeGraaf,A. (1984). KonstruktievaneenverbeterdeversievandeGroninger
Androgynie Schaal (GRAS).Heyman'sBulletinPsychologische Instituten, R.U.
Groningen,HB-84-710-EX.
 -152-

DeJong,w. (1980).Thestigmaofobesity:theconsequencesofnaive assumptions

concerning thecausesofphysicaldeviance.Journal ofHealthand Social
Behavior, 21,75-87.
Deurenberg, P.,van Poppel,G.&Hautvast,J.G.A.J. (1984).Morbiditeiten
subjectiefwelbevindenbijjongevolwassenenmetovergewicht.Nederlands
TijdschriftvoorGeneeskunde,128,940-944.
Deutsch, C.J. &Gilbert,L.A. (1976).Sex-rolestereotypes:Effectson
perceptionof selfandothersandonpersonal adjustment.Journal of
ConsultingPsychology, 23,373-379.
Diehl, J.H.&Paul,T. (1985).Relativgewichtund Persönlichkeit:Analysenmit
dem Freiburger Persönlichkeitsinventar.AktuelleErnährungsmedizin in
KlinikundPraxis,10,14-23.
Drewnowski,A.,Riskey,D.&Desor,J.A.(1982). Feeling fatyet unconcerned:
Self-reported overweight and theRestraint Scale.Appetite, 3,273-279.
DutchExpertCommittee onEnergyandNutrientRequirements. (1983).Nederlandse
Voedingsmiddelen Tabel,aanbevolenhoeveelhedenenergieenvoedingstoffen.
'sGravenhage:VoorlichtingsbureauvoordeVoeding.
Dwyer,J.T. &Mayer,J. (1970).Potentialdieters:whoare they?Journal ofthe
American DieteticAssociation, 56,510-514.
Elman,D.,Schroeder,H.E.&Schwartz,M.F. (1977).Reciprocal social influence
ofobeseandnormal-weight persons.Journal ofAbnormal Psychology,86,
408-413.
Fallon,A.E.&Rozin,P. (1985). Sexdifferences inperceptionofbodyshape.
Journal ofAbnormal Psychology, 94,102-105.
Feather,N.T. (1984).Masculinity, femininity,psychological androgynyand the
structureofvalues.Journal ofPersonalityandSocialPsychology,47,
604-620.
Freed, S.C. (1947).Psychic factors indevelopment and treatment ofobesity.
Journalof theAmerican MedicalAssociation, 133,369-373.
Gadourek, I. (1972).Derek L.Philips'research toend the research.Mensen
Maatschappij, 47,119-128.
Garner,D.M. &Garfinkel,P.E. (1979).TheEatingAttitudesTest:An indexof
the symptoms ofanorexianervosa.PsychologicalMedicine,9, 273-279.
Garner,D.M.,Garfinkel,P.E.,Schwartz,D.&Thompson,M. (1980).Cultural
expectationsof thinness inwomen.Psychological Reports,47,483-491.
Garrow,J. (1978).The regulationofenergyexpenditure.In:G.A. Bray (Ed.).
Recent advances inobesity research,vol.2,London:Newman.
 -153-

Gaudreau,P. (1977).FactoranalysisoftheBernSex-Role Inventory.Journal of

Consulting andClinical Psychology, 45,299-302.
Gersten,J.C., Langner,T.S.,Eisenberg,J.G.&Orzak,L. (1974).Child
behaviourandlifeevents;undesirable changeorchangeper se,In:B.S.
DohrenwendandB.P.Dohrenwend (eds.),Stressful LifeEvents:TheirNature
and Effects.NewYork:JohnWileyandSons.
Gillman,R.D. (1968).Thedreamsofpregnantwomenandmaternal adaptation.
American Journal ofOrthopsychiatry, 38,688-692.
Glass,D.C.,Lavin,D.E.,Henchy,T.,Gordon,A.,Mayhew,P.&Donohoe, P.
(1969).Obesityandpersuability.Journal ofPersonality, 37,407-414.
Goldman,R., Jaffa,M.&Schachter,S. (1968).YomKippur,Air France,dormito-
ryfood,and theeatingbehavior ofobeseandnormalpersons.Journalof
PersonalityandSocialPsychology, 10,117-123.
Gordon,R.E.&Gordon,K.K. (1960).The split-level trap.UnitedStates:
BernardGeis.
Groffen,w.H.&VanKreveld,D. (1972).Schalenvoorprominentie,stijging en
goedkeuring.NederlandsTijdschriftvoordePsychologie, 27,199-214.
Guilford, J.P. &Zimmerman,w. (1949).TheGuilford-Zimmerman Temperament
Survey.BeverlyHills,California:SheridanSupply Company.
Hall,S.M.&Havassy,B. (1981).Theobesewomen:Causes,correlatesand
treatment. Professional Psychology, 12,163-170.
Hamburger,W.W. (1951).Emotionalaspectsofobesity.Medical ClinicsofNorth
America, 35,483-499.
Harris, (1983).Eatinghabits,restraint, knowledgeandattitudes toward
obesity. InternationalJournalofObesity, 7,271-286.
Hartigan,K.J., Baker-Strauch,D.&Morris,G.W. (1982).Perceptionsofthe
causesofobesityand responsivenesstotreatment.Journal ofCounseling
Psychology, 29,478-485.
Hautvast,J.G.A.J. (1975).Ontwikkelingvaneensysteemomgegevensvan
voedingsenquêtesmetbehulpvandecomputer teverwerken.Voeding,35,
356-360.
Hawkins II,R.C.andClement, P.F. (1980).Development and construct validation
ofaself-reportmeasureofbingeeatingtendencies.AddictiveBehaviors,
5,219-226.
Hawkins II,R.C.,Turell,S.andJackson,L.J. (1983).Desirableand
undesirablemasculine and feminine traitsin relation tostudent'sdietary
tendenciesandbody imagedissatisfaction. SexRoles,9,705-724.
Hays,W.L. (1969).Statistics,p.530-532.London:Holt,Rinehart &Winston.
 -154-

Hecht,H.B. (1955).Obesity inwomen:A psychiatric study. Psychiatric

Quarterly, 29,203-231.
Heilbrun,A.B. (1976).Measurementofmasculine and feminine sex role
identitiesasindependentdimensions.JournalofConsultingand Clinical
Psychology, 44,183-190.
Herman,C.P.&Mack,D. (1975).Restrained andunrestrained eating.Journal of
Personality, 43,647-660.
Herman,C.P.&Polivy,J. (1975).Anxiety, restraintandeatingbehavior.
Journal ofAbnormal Psychology, 84,666-672.
Herman,C.P.&Polivy,J. (1980 a ).Restrained eating, in:Stunkard,A.J. (Ed.):
Obesity.Philadelphia,London,Toronto:Saunders,208-225.
Herman,C.P.&Polivy,J. (1980 ).Stress inducedeatingand eating-induced
stress (reduction?):A response toRobbinsandFray.Appetite, 1,135-139.
Herman,C.P.&Polivy,J. (1982).Weight changeanddietaryconcern inthe
overweight:Are they reallyindependent?.Appetite:Journal for Intake
Research, 3,280-281.
Herman,C.P.&Polivy,J. (1984).A boundarymodel forthe regulationof
eating. In:Stunkard,A.J.&Steller,E. (Eds.):Eatingand itsdisorders.
NewYork:RavenPress.
Herman,C.P., Polivy,J., Pliner,P., Threlkeld, J.&Munie,D. (1978).
Distractability indietersandnondieters:An alternativeviewof
"externality".JournalofPersonalityand Social Psychology, 36,536-548.
Herman,C.P.,Polivy,J. &Silver,R. (1979).Effectsofanobserver on eating
behavior:The inductionof "sensible"eating.Journal ofPersonality,47,
85-99.
Hermans,H.J.M. (1968).Handleiding bijdePrestatieMotivatieTest. Amsterdam:
Swets&Zeitlinger.
Hermans,H.J.M. (1971).Prestatiemotief enfaalangst ingezinenonderwijs.
Amsterdam: Swets&Zeitlinger.
Hibscher,J. andHerman,C.P. (1977).Obesity,dietingand theexpression of
"obese"characteristics.JournalofComparative and Physiological
Psychology, 91,374-380.
Hoffman (1972).Earlychildhood experiences andwomen'sachievementmotives.
Journalof Social Issues,28,129-155.
Holland,J., Masling, J. &Copley,D. (1970).Mental illness inlower class
normalobeseandhyperobesewomen.PsychosomaticMedicine,32,351-357.
Holmes,T.H.&Rahe,R.H. (1967).The SocialReadjustment Rating Scale.Journal
Psychosomatic Research, 11,213-218.
 -155-

Isbitsky,J.R.&White,D.R. (1981). Externalityand locusofcontrol inobese

children.JournalofPsychology, 107,163-172.
Jackson,D.N.&Messick,S. (1958). Contentandstyle inpersonality
assessment.PsychologicalBulletin, 55,243-252.
Jackson,O.N.&Messick,S. (1961).Acquiescenceanddesirabilityas response
determinantsontheHMPI.EducationalandPsychologicalMeasurement,21,
771-790.
Janowitz,H.D. (1967). Roleofthegastrointestinal tract inthe regulationof
food intake.In:Cole,C F . (Ed.):Handbook ofphysiology:alimentorycanal
I.wahingtonDC:American Physiological Society.
Johnson,J.H.&Sarason,I.G. (1978). Lifestress,depressionandanxiety:
Internal-external controlasamoderatevariable.JournalofPsychosomatic
Research, 22,205-208.
Johnson,W.G.,Lake,L.&Mahan,M . (1983).Restrainedeating:Measuringan
elusive construct.AddictiveBehaviors,8,413-418.
Jones,W.H.,Chernovetz,M.E.&Hansson,R.O. (1978). Theenigma ofandrogyny:
Differential implication formalesand females?JournalofConsultingand
Clinical Psychology, 46,298-313.
Jung, F. (1973).AppetitverhaltenvonVorschulkindernunter experimenteller
Kontrolle.PraxisderKinderpsychologieundKinderpsychiatrie,22,167-171.
Jung, F. (1976).Untersuchungen zurNahrungsaufnahmevonKindernundNeugebore-
nenunterbesonderer BerücksichtigungpsychischerAktivierung.Doctoral
Dissertation,Göttingen.
Kalucy,R.S.&Crisp,A.H. (1974). Somepsychological and social implications
ofmassiveobesity:astudyofsomepsychosocial accompaniments ofmajor
fat lossoccurringwithoutdietary restrictionsinmassivelyobese
patients.JournalofPsychosomaticResearch,18,465-473.
Kaplan,H.I.&Kaplan,H.S. (1957).Thepsychosomatic conceptofobesity.
JournalofNervousandMentalDisease,125,181-201.
Karpovitz,D.H.&Zeis,F.R. (1975). Personalityandbehaviordifferencesof
obeseandnonobeseadolescents.JournalofConsultingand Clinical
Psychology, 43,886-891.
Kelly,J.A., Caudill,M.S.,Hathorn,S.&O'Brien,C.G. (1977). Socially
undesirable sex-correlated characteristics.Implications forandrogynyand
adjustment.JournalofConsultingandClinicalPsychology, 45,1185-1186.
Kelly,J.A. &Worell,J. (1977).Newinformationsofsex rolesandandrogyny:a
critical review.JournalofConsultingandClinical Psychology,45,
1101-1115.
 -156-

Kimlicka,T.,Cross,H.&Tarnai,J. (1983).A comparisonofandrogynous,

feminine,masculine,andundifferentiatedwomenonself-esteem, body
satisfaction,and sexual satisfaction.PsychologyofWomenQuarterly,7,
291-295.
Kirschenbaum,D.S.&Tomarken,A.J. (1982). Someantecedentsof regulatoryeat-
ingby restrainedandunrestrainedeaters.JournalofAbnormal Psychology,
91, 326-336.
Krantz,D.S. (1978).Thesocialcontextofobesityresearch:Another
perspectiveon itsplace inthefieldof socialpsychology.Personalityand
SocialPsychologyBulletin,4,177-184.
LaboratoriumvoorToegepastePsychologie (L.T.P.). (1974).GuilfordLTP
Temperament Survey.Lisse:SwetsenZeitlinger.
Lei, H.,&Skinner,H.A. (1980).A psychometric studyoflifeeventsand social
readjustment.JournalofPsychosomaticResearch, 24,57-65.
Lennings,C.I. (1980).Devragenlijstmetgebeurtenissendieeenmenskunnen
overkomen.Wageningen,TheNetherlands:Internalpublication,Department of
HumanNutrition,Agriculturaluniversity.
Leon,G.R.&Chamberlain, K. (1973). Comparisonofdailyeatinghabitsandemo-
tionalstatesofoverweightpersonssuccessfulorunsuccessful inmaintaining
aweight loss.JournalofConsulting &Clinical Psychology, 41,108-115.
Leon,G.R. &Finn,S. (1984). Sex-roledifferencesand thedevelopmentof
eatingdisorders.In:CathySpatzwidom (Ed.).SexRolesandPsychopathe—
logy.NewYork:PlenumPublishingCorp.
Leon,G.R. &Roth,Z. (1977).Obesity:Psychological causes,correlationsand
speculations.PsychologicalBulletin,84,117-139.
Loro,A.D.,Fisher,E.B.&Levenkron,J.C. (1979). Comparisonofestablished
and innovativeweight, reduction,treatmentprocedures.JournalofApplied
BehavioralAnalysis,12,141-155.
Lorrimor,R.J., Justice,B.,HcBee,G.W. &Weinman,M. (1979).Weightingevents
inlifeevent research.JournalofHealthandSocialBehavior,20,306-307.
Lowe,M.R.&Fisher,E.B. (1983). Emotional reactivity,emotional eating,and
obesity:A naturalistic study.JournalofBehavioralMedicine,6,135-149.
Lowe,M.R. (1984). Dietaryconcern,weight fluctuationandweight status:
furtherexplorationsoftheRestraint Scale.BehaviorResearchand Therapy,
22,243-248.
Luteyn, F., Starren,J. &VanDijk,H. (1975).HandleidingNederlandse Persoon-
lijkheidsvragenlijst. Lisse:Swetsenzeitlinger.
 -157-

Maccoby,E.E.&Jacklin,C.N. (1974).Thepsychologyof sexdifferences.

Stanford:StanfordUniversityPress.
Haddox,G.L.,Back,K.W.&Liederman,V.R. (1968).Overweight associal
deviance anddisability.JournalofHealthandSocialBehavior, 9,287-298.
HcGee,R.K. (1962 a ).The relationshipbetween response styleand personality
variables:II.Thepredictionof independentconformitybehaviour.Journal
ofAbnormal SocialPsychology,65,347-351.
HcGee,R.K. (1962 ).Response styleasapersonalityvariable:Bywhat
criterion?PsychologicalBulletin, 59,284-295.
NcCrae,R.R. &Costa,P.T. (1983).Socialdesirability scales:More substance
thanstyle.JournalofConsultingandClinicalPsychology, 51,882-888.
HcKenna,R.J. (1972). Someeffectsofanxietyleveland foodcuesontheeating
behavior ofobeseandnormal subjects:acomparisonof theSchachterianand
psychosomatic concepts.JournalofPersonality&SocialPsychology,22,
311-319.
HcReynolds,W.T.(1983).Towardapsychologyofobesity, reviewof researchon
the roleofpersonalityandlevelofadjustment,internationalJournal of
EatingDisorders,2,37-57.
Mendelson,H.,Weinberg,N.&Stunkard,A.J. (1961).Obesity inmen:A clinical
studyof25cases.Annual InternationalMedicine, 54,660-671.
Meyer,J.E.&Pudel,V.E.81972).Experimental studieson food intake inobese
andnormalweight subjects.JournalofPsychosomaticResearch, 16,305-308.
Meyer,J.E.&Pudel,V. (1977).Experimental feeding inman:A behavioral
approach toobesity.PsychosomaticMedicine, 39,153-157.
Meyer,J.E.&Tuchelt-Gallwitz,A . (1967). Psychiatrisch-Psychologische
UntersuchungenanWeiblichen Fettsüchtigen.Zeitschrift für
PsychosomatischeMedizinundPsychoanalyse,13,73-107.
Meyers,A.W., Stunkard,A.J.&Coll,M. (1980). Foodaccessibilityandfood
choice.ArchivesofGeneral Psychiatry, 37,1133-1135.
Milien,R.,Anderson,J. &Mills,M. (1976).Effectsofvisualpresentationof
caloricvalueson foodbuyingbynormalandobesepersons.Perceptualand
Motor Skills,42,155-162.
Milien,R.S.&Fisher,E.B. (1979).Theeffectsofcue salienceandpriortrai-
ningonthebehaviorofjuvenile-andadult-onsetobeseindividuals.
AddictiveBehaviors,4,1-10.
Mueller,D.P.,Edwards,D.W. &Yarvis,R.M. (1977).Stressful lifeeventsand
psychiatric symptomatology: Changeorundesirability?JournalofHealthand
SocialBehavior, 18,307-317.
 -158-

Nesselroade,J.R.&Baltes,P.B. (1979).Longitudinal researchinthestudyof

behavior anddevelopment.NewYork:AcademiePress.
Nie, N.H.,Hull,C.H.,Jenkins,J.G.,Steinbrenner,K.&Bent,D.H. (1970).
SPSS (Statistical Package fortheSocial Sciences).NewYork:McGraw-Hill.
Nisbett,R.E. (1968 a ).Taste,deprivationandweightdeterminantsofeatingbe-
havior.Journal ofPersonalityandSocialPsychology,10,107-116.
Nisbett,R.E. (1968 ).Determinantsoffoodintakeinobesity.Science,159,
1254-1255.
Nisbett,R.E. (1972).Hunger,obesityandthevertromedialhypothalamus.
Psychological Review, 79,433-458.
Nisbett,R.E.&Temoshok,L. (1976). Isthereanexternal cognitive style?
JournalofPersonalityandSocialPsychology, 33,36-47.
Nutzinger,D.O. (1979).ÜberdieBedeutungdepressiver ZuständebeiAdipösen.
PsychiatriaClinica,12,173-179.
O'Neil,P.H., Currey,H.S.,Hirsch,A.A.,Malcolm,R.J., Sexauer,J.D.,Riddle,
F.E.andTaylor,C.I. (1979).Development andvalidationoftheeating
behavior inventory.Journal ofBehavioralAssessment, 1,123-132.
Oetzel,R.M. (1966).Annotatedbibliography. In:E.E.Maccoby (Ed.).The
development ofsexdifferences.Stanford,California:Stanforduniversity
Press.
Olsen,D. (1969).Themeasurement of familypowerbyself-report and
behaviouralmethods.JournalofMarriageand theFamily, 31,549.
Orne,M.T. (1962).Onthesocialpsychologyofthepsychological experiment:
Withparticular referencetodemandcharacteristicsand their implications.
American Psychologist, 17,776-783.
Parsons,T.&Bales,R.F. (1955).Family,socialization,and the interaction
process.NewYork:FreePressofGlencoe.
Paulhus,D.L. (1984).Two-componentmodelsof sociallydesirable responding.
JournalofPersonalityand SocialPsychology, 46,598-609.
Paykel,E.S.&Uhlenhuth,E.H. (1972).Ratingthemagnitudeoflifestress.
CanadianPsychiatricAssociation Journal,17,93-100.
Pedhazur, E.J. &Tetenbaum,T.J. (1979).BernSexRole Inventory:A theoretical
andmethodological critique.Journal ofPersonalityandSocial Psychology,
37,996-1016.
Philips,D.L.&Clancey,K.J. (1970).Responsebiasinfield studiesofmental
illness.American SociologicalReview, 35,503-515.
Pliner,P. (1973 a ).Effectofcuesalienceon thebehavior ofobeseand normal
subjects.Journal ofAbnormal Psychology, 82,226-232.
 -159-

Pliner,F. (1973 ).Effect ofexternal cues inthethinking behavior of obese

and normal subjects.Journal ofAbnormal Psychology, 82,233-238.
Pliner,P. (1974).On thegeneralisabilityoftheexternality hypothesis. In:
S.Schachter &J.Rodin.Obese humans and rats.Washington, D.C.:Erlbaum/-
Halsted.
Pliner, P., Heyer,P. &Blankstein, K. (1974).Responsiveness to affective
stimuli byobese andnormal individuals.Journal ofAbnormal Psychology,
83, 74-80.
Plutchik, R. (1976).Emotions andattitudes related tobeing overweight.
Journal ofClinicalPsychology, 32,21-33.
Pohlman,E.H. (1969).Psychology ofbirthplanning.Cambridge, Massachusetts:
Shenkman.
Polivy, J. (1976).Perception ofcaloriesand regulationof intake in
restrained andunrestrained subjects.Addictive Behaviors, 1,237-243.
Polivy, J. &Herman,C.P. (1976 a ).Clinical depression andweight change:A
complex relation.Journal ofAbnormal Psychology, 85,338-340.
Polivy, J. &Herman,C.P. (1976 ).Effectsofalcohol oneating behavior:
Influenceofmood andperceived intoxication. Journal ofAbnormal
Psychology, 85,601-606.
Polivy, J. &Herman, C.P. (1976°)Theeffectsofalcohol oneating behavior:
Disinhibition or sedation?Addictive Behaviors, 1,121-126.
Polivy, J. &Herman, C.P. (1983).Breaking thediethabit: thenatural weight
alternative.NewYork:Basic Books.
Polivy, J., Herman, C.P. &Warsh, S. (1978).Internaland external components
of emotionality in restrained andunrestrained eaters.Journal ofAbnormal
Psychology, 87,497-504.
Polivy, J., Herman, C.P.,Younger,J.C.&Erskine,B. (1979).Effectsofa
model oneatingbehaviour:The induction ofa restrained eating style.
Journal ofPersonality, 47,100-114.
Preventie-project (1976).TheDutch-Self-Partner Scale.Interim rapport.
NijmeegsuniversitairHuisartsen Instituut.
Pudel,V. (1978).Zur Psychogenese undTherapie derAdipositas. Berlin:
Springer-Verlag.
Pudel,V.,Metzdorff, M.&Oetting,M. (1975).ZurPersönlichkeit Adipöser in
psychologischen TestsunterBerücksichtigung latent Fettsüchtiger.
Zeitschrift für Psychosomatische Medizinund Psychoanalyse, 21,345-361.
 -160-

Rahe,R. (1974).Thepathwaybetween subjects'recent lifechangesand their

near future illness reports:representative resultsand methodological
issues, In:B.S.Dohrenwend andB.P.Dohrenwend (Eds.).Stressful life
events:theirnature andeffects.NewYork:JohnWileyand Sons.
Rahe,R.H. (1975).Epidemiological studiesof life change and illness,
International Journal Psychiatry inMedicine,6, 133-146.
Rand,C.S.W. (1982).Psychoanalytic treatment ofobesity. In:Wolman,B.B.
(Ed.): Psychological aspectsofobesity:ahandbook.NewYork, Cincinnati,
Toronto,London,Melbourne:VNR, 177-191.
Rand, C.S.W.&Stunkard,A.J. (1978).Obesityandpsychoanalysis.American
Journal ofPsychiatry, 135,547-551.
Ravussin, E.,Burnand, B.,Schutz,Y. &Jequier, E. (1985).Energy expenditure
before andduring energy restriction inobesepatients.TheAmerican
Journal ofClinicalNutrition, 41,753-759.
Reznick, H.&Balch,P. (1977).Theeffectsof anxietyand response costmani-
pulations on theeatingbehavior ofobese andnormal-weight subjects.
Addictive Behaviors, 2,219-225.
Riecken, H.W. (1962).A program for research onexperiments in social
psychology. In:N.F.Washburne (Ed.).Decision,values and groups,Vol.11.
NewYork:Pergamon Press,25-41.
Robbins,T.W. &Fray, P.J. (1980).Stress-induced eating: Fact, fictionor
misunderstanding.Appetite, 1,103-133.
Rodin,J. (1973).Effectsofdistraction onperformance of obese and normal
subjects.Journal ofComparative and Physiological Psychology, 83,68-75.
Rodin,J. (1975).Causesand consequences of timeperception differences in
overweight and normalweight people.Journal ofPersonality and Social
Psychology, 31,898-904.
Rodin,J. (1978).Has thedistinction between internalversus external control
of feeding outlived itsusefulness? In:Bray,G.A.: Recent advances in
obesity research: II;Proceedings ofthe2nd International Congress on
Obesity.London:JohnLibbey &Co.,75-85.
Rodin,J. (1981 a ).Current statusof the internal-external hypothesis for
obesity.American Psychologist, 36,361-372.
Rodin,J. (1981 ).Psychological factors inobesity. In:Björntorp, P.,
Cairella, M. &Howard,A.N. (Eds.):Recent advances inobesity research,
vol. III.London:JohnLibbey &Co.,106-123.
 -161-

Rodin,J., Einen,D.&Schachter, S. (1974).Emotionality and obesity. In:

S.Schachter &J.Rodin (Eds.)Obesehumansand rats.Washington,D.C.:
Erlbaum/Halsted.
Rodin,J., Herman,C.P. &Schachter, S. (1974).Obesity andvarious testsof
external sensitivity. In:S.Schachter &J.Rodin (Eds.)Obese humansand
rats.Washington, D.C.: Erlbaum/Halsted.
Rodin,J., Silberstein, L.&Striegel-Moore,R. (1985).Womenandweight:a
normative discontent.Nebraska Symposium onMotivation.Lincoln,NE:
University ofNebraska Press.
Rodin,J. &Singer,J.L. (1976).Eye-shift, thought andobesity.Journal of
Personality, 44,595-610.
Rodin, J. &Slochower, J. (1974).Fat chance fora favor:Obese-normal
differences incompliance and incidental learning.Journal of Personality
and Social Psychology, 29,557-565.
Rodin,J. &Slochower, J. (1976).Externality inthenon-obese:Theeffects on
environmental responsiveness onweight.Journal of Personality and Social
Psychology, 33,338-344.
Rodin,J., Slochower,J. &Fleming, B. (1977).Effectsofdegree of obesity,
ageof onset, andweight losson responsiveness tosensoryand external
stimuli.Journal ofComparative and Physiological Psychology, 91,586-597.
Rodin, J. &Striegel-Moore,R.H. (1984).Predicting attitudes toward body
weight and food intake inwomen.Paperpresented at the 14thcongress of
EuropeanAssociation ofBehavior Therapy inBrussels,Belgium, September.
Rookus,M.A.,Burema,J., Deurenberg, P. &Hautvast,J.G.A.J. (Inpreparation).
Obesity inyoung adults III.Thenatural historyofbodyweight development
and fluctuation: a fouryear follow-up study.
Rorer,L.G. (1965).Thegreat response stylemyth.Psychological Bulletin,63,
129-156.
Roskam, E.and Borgers,H. (1969).Programma-Bulletin no.14:ROTA 01,ROTA 03.
University Nijmegen,Nijmegen:Publications of theDepartment of
Mathematical and Statistical Psychology.
Ross, C E . &Mirowsky, J. II (1979).A comparison of life-event weighting
schemes:Change,undesirability andeffect-proportional indices.Journal of
Health and Social Behavior, 20,166-177.
Rotter,J.B. (1966).Generalised expectancies for internal versus external
control of reinforcement. Psychological Monographs, 80,1-28.
Ruderman,A.J. (1983).The restraint scale:A psychometric investigation.
Behavior Research andTherapy, 21,253-258.
 -162-

Rudennan,A.J.&Christensen,H. (1983).Restraint theoryand itsapplicability

tooverweight individuals.JournalofAbnormal Psychology, 92,210-215.
Ruderman,A.J.&Wilson,G.T. (1979).Weight, restraint,cognitionsand
counterregulation.BehaviorResearchandTherapy, 17,581-590.
Salzer,E.B. (1982). Theweight locusofcontrol (WLOC)scale:A specific
measure forobesityresearch.JournalofPersonalityAssessment,46,
620-628.
Sarason, I.G., JohnsonJ.H. &Siegel,J.H. (1978).Assessingthe impact oflife
changes:developmentofthe lifeexperiencessurvey.Journalof Consulting
andClinicalPsychology, 46,932-946.
Schachter,S. (1964). The interactionofcognitiveandphysiologicaldetermi-
nantsofemotional state.In:Berkowitz,L. (Ed.):Advances inexperimental
psychology.Vol 1.NewYork:AcademicPress.
Schachter,S. (1967). Cognitiveeffectsonbodily functioning:studiesof
obesityandeating. In:Glass,D.C. (Ed.):Neurophysiologyandemotion.New
York:RockefellerUniversityPressandRussellSageFoundation.
Schachter,S.,Goldman,R.&Gordon,A. (1968). Effectsof fear,food
deprivationandobesityoneating.JournalofPersonalityand Social
Psychology, 10,91-97.
Schachter,S.&Gross,L.P. (1968).Manipulated timeandeatingbehavior.
JournalofPersonalityandSocialPsychology, 10,98-106.
Schachter,S.&Rodin,J. (1974).Obesehumansand rats.Washington D . C :
Erlbaum/Haltsted.
Schachter,S.&Singer,J.E. (1962). Cognitive,socialandphysiological
determinantsofemotional state.PsychologicalReview,69,379-399.
Seidell,J.C.,Bakx,J.C.,deBoer,E.,Deurenberg,P.&Hautvast,J.G.A.J.
(1985). Fatdistributionofoverweightpersonsin relationtomorbidityand
subjectivehealth.International JournalofObesity, 9,363-374.
Seligman, J. (1975). Helplessness:ondepression,developmentanddeath.New
York:W.H.Freedman &Co.
Sherman,J.A. (1971).On thepsychologyofwomen:asurveyof empirical
studies.Springfield, Illinois:CharlesC.Thomas,Publisher.
Shorvan,H.J. &Richardson,J.S. (1949). Suddenobesityandpsychological
trauma.BritishMedicalJournal,29,951-956.
Shrager,E.E.,Wadden,T.A., Miller,D.,Stunkard,A.J.&Stellar,E. (1983).
Compensatory intra-meal responsesofobesewomen toreduction inthesize
of foodunits.Abstract Society forNeuroscience,201,Abstracts62.8.
 -163-

Slochower,J.A. (1976).Emotionallabelingandovereating inobeseandnormal

weight individuals.PsychosomaticMedicine, 38,131-139.
Slochower,J.A. (1983a). Excessiveeating:the roleofemotionsand
environment.NewYork:HumanSciencesPress,inc.
Slochower,J.A. (1983 ).Life stress,weightandcue salience.In:Slochower,
J.A.: Excessiveeating.NewYork:HumanSciencesPress,inc.
Slochower,J.A. &Kaplan,S.P. (1980).Anxiety,perceived controlandeating in
obeseandnormalweightpersons.Appetite,75-83.
Slochower,J.A. &Kaplan,S.P. (1983).Effectsofcuesalienceandweight on
responsivenesstouncontrollableanxiety. In:Slochower,J.A. Excessive
eating.NewYork:HumanSciencesPress,inc.
Slochower,J.A., Kaplan,S.P. &Mann,L. (1981).Theeffectsoflife stressand
weight onmoodandeating.Appetite,115-125.
Spence,J.T.,Helmreich,R.&Stapp,J. (1975).Ratingsof selfandpeerson
sex roleattributesand their relationtoself-esteemandconceptionof
masculinityand femininity.JournalofPersonalityand Social Psychology,
32,29-39.
Spencer,J.A. &Fremouw,W.J. (1979).Bingeeatingasafunctionof restraint
andweight classification.JournalofAbnormal Psychology,8,262-267.
Spielberger,C D . ,Gorsuch,R.L.&Lushene,R.E. (1970).Manual for the
state-traitanxiety inventory.PaloAlto,California:ConsultingPsycholo-
gistPress.
Stuart,R.B.&Guire,K. (1978).Somecorrelatesofmaintenanceofweight loss
throughbehavior-modification. International JournalofObesity,2,
225-235.
Stuart,R.B.&Davis,B. (1972).Slimchance inafatworld:behavioral control
ofobesity.Champaign, Illinois:Professional ed.ResearchPress.
Stunkard,A.J. (1959 a ). Eatingpatternsandobesity.PsychiatricQuarterly, 33,
284-292.
Stunkard,A.J. (1959 ).Obesityandthedenialofhunger.Psychosomatic
Medicine,21,281-289.
Stunkard,A.J. (1980).Psychoanalysisandpsychotherapy.In:Stunkard,A.J.
(Ed.):Obesity.Philadelphia,London,Toronto:W.B.Saunders company,
355-368.
Stunkard,A.J. (1981). "Restrained Eating":What itisandanewscale to
measure it.In:L.A. Cioffi (Ed.)Thebodyweight regulatory system:normal
anddisturbedmechanisms.NewYork:RavenPress,243-251.
 -164-

Stunkard,A.J.&Albaum,J.M. (1981).Theaccuracyofself-reportedweights.
TheAmerican Journal ofClinicalNutrition, 34,1593-1599.
Stunkard,A.J.&Fox,S. (1971).The relationshipofgastricmotilityand
hunger:asummaryof theevidence.PsychosomaticMedicine, 33,123-134.
Stunkard,A.J.&Koch,C. (1964).The interpretation ofgastricmotility I:
Apparent biasinthe reportsofhungerbyobesepersons.Archivesof
General Psychiatry,11,74-82.
Stunkard,A.J.&Messick,S. (1985).TheThree-FactorEatingQuestionnaire to
measuredietary restraint,disinhibitionandhunger.Journalof
Psychosomatic Research, 29,71-83.
Stunkard,A.J.&Permick,S.B. (1979).Behaviormodification inthe treatment
ofobesity:Theproblemofmaintainingweight loss.ArchivesofGeneral
Psychiatry, 36,801-806.
Taylor,M.C. &Hall,J.A. (1982).Psychological androgyny:Theories,methods
and conclusions.PsychologicalBulletin, 92,347-366.
Tennant,C.&Andrews,G. (1978).Thepathogenicqualityoflifeevent stress
inneurotic impairment.ArchivesofGeneral Psychiatry, 35,859-863.
Tomarken,A.J.&Kirschenbaum, D.S. (1984).Effectsofplansfor further meals
oncounterregulatory eatingby restrained andunrestrained eaters.Journal
ofAbnormal Psychology,93,458-472.
Tuddenham,R.D. (1958).The influenceofadistoredgroupnormupon individual
judgement.JournalofPsychology,46,227-241.
VanderPloeg,H.M. (1979).Relationshipof thestate-trait anxiety toacademic
performance inDutchmedical students.PsychologicalReports,45,223-227.
Vander Ploeg,H.M., Defares,P.B.&Spielberger,C D . (1979).Zelf-beoorde-
lingsvragenlijst.STAI-versieDY-1enDY-2.Lisse:SwetsenZeitlinger.
VanderPloeg,H.M., Defares,P.B.&Spielberger,C D . (1980).Eennederlands-
taligebewerkingvande Spielberger state-trait anxiety inventory.De
zelf-beoordelingsvragenlijst.DePsycholoog,15,460-467.
VanderPloeg,H.M., Defares,P.M. &Spielberger,C D . (1981). Handleidingbij
de zelfanalyse-vragenlijst (Z.A.V.).Lisse:SwetsenZeitlinger.
Van Staveren,W.A.,Deurenberg,P., Burema,J., DeGroot,L.C.P.M.&Hautvast,
J.G.A.J. (submitted).Seasonalvariationintheenergybalanceofagroup
ofyoungadultDutchwomen consuming self-selecteddiets.
VanStrien,T. (1984a).Weight related susceptibilitytodemand characteristics
inalongitudinal studyoneatingbehaviour.Wageningen, theNetherlands:
Internalpublication,Department ofHumanNutrition,Agricultural
University, reportno. 0-1186.
 -165-

VanStrien,T. (1984b).Dedunnehuidvandikkemensen.Verwachtingskenmerken
bijonderzoeknaarovergewicht.DePsycholoog, 19,157-166.
VanStrien,T. (1985).Onlongitudinalversuscross-sectional studyofobesity:
possibleartefacts.International JournalofObesity,9,323-333.
VanStrien,T.,Frijters,J.E.R.,Bergers,G.P.A.&Defares,P.B. (inpress).
TheDutchEatingBehaviourQuestionnaire (DEBQ):assessmentof restrained
eating,emotional eatingandexternaleating. InternationalJournalof
EatingDisorders.
VanStrien,T.,Frijters,J.E.R.,Roosen,R.G.F.M.,Knuiman-Hijl,W.J.H.&
Defares,P.B. (inpress).Eatingbehaviour,personalitytraitsandbody
mass inwomen.AddictiveBehaviors.
VanStrien,T.,Frijters,J.E.R.,Van Staveren,W.A.,Defares,P.B.&
Deurenberg,P. (inpress).Thepredictive validityoftheDutchRestrained
EatingScale.International JournalofEatingDisorders.
Vinokur,A. &Selzer,M.L. (1975).Desirableversusundesirable lifeevents:
their relationshiptostressandmental illness.JournalofPersonalityand
SocialPsychology, 32,329-337.
Wardle,J. (1980).Dietary restraintandbingeeating.BehaviorAnalysisand
Modification, 4,201-209.
Warheit,G.,Hölzer,C.&Schwab,J. (1973).Ananalysisofsocialclassand
racialdifferences indepressive symptomatology: acommunity study.Journal
ofHealthandSocialBehavior,14,291-299.
Weeda-Mannak,W. (1984).Anorexianervosa:towardsanearlyidentification.
Hilversum:BenedictusPress.
Weiss,A.R. (1977).Characteristicsofsuccessfulweight reducers:abrief
reviewofpredictivevariables.AddictiveBehaviors,2,193-201.
Weissenburger,J., Rush,A.J.,Giles,D.E.&Stunkard,A.J. (submitted).Weight
change indepression.
Wiggins,J.S.&Holzmuller,A. (1978).Psychological androgynyand interperso-
nalbehavior.JournalofConsultingandClinicalPsychology, 46,40-52.
Willems,L.F.M.,Tuender-DeHaan,H.A.Defares,P.B. (1973).Een schaalom
socialeangsttemeten.NederlandsTijdschriftvoordePsychologie,28,
415-422.
Windle, C. (1954).Test-retesteffect onpersonalityquestionnaires.
Educational andPsychologicalMeasurements,14,617-633.
Winnubst,J.A.M.,Marcelissen,F.H.G.sKleber,R.J. (1982).Effectsof social
support inthestressor-strain relationship:aDutchsample.Social
ScienceandMedicine,16,475-482.
 -166-

Witteman,J.C.H., VanDusseldorp,H,Frijters,J.E.R.&VanStrien,T.
(submitted).TheLossofControlScale forBulimic EatingBehavior.
Wollersheim,J.P. (1970).Effectivenessofgrouptherapybasedupon learning
principles inthetreatmentofoverweightwomen.Journal ofAbnormal
Psychology, 76,462-474.
Wolman,B.B. (1982). interactionalpsychotherapyofobesity. In:Holman,B.B.
(Ed.).Psychological aspectsofobesity:ahandbook.NewYork,Cincinnati,
Toronto,London,Melbourne:VNR.
Woody, E.Z.,Costanzo,P.R., Liefer,H.&Conger,J. (1981).Theeffectsof
tasteandcaloricperceptionsontheeatingbehavior of restrained and
unrestrained subjects.CognitiveTherapyandResearch, 5,381-390.
Wooley,S.C.,Wooley,O.W.&pyrenforth,S.R. (1979).Theoretical,practical
and social issues inbehavioral treatmentsofobesity.Journal ofApplied
BehaviorAnalysis,12,3-25.
Wooley, S.C.,Wooley,O.W., Bartoshiek,L.M., Catenae,M.J.C., Feistl,R.,
Gutezert,G.W.R.,McFarland,D.J.,Oetting,M,Pudel,V.E.,Rodin,J. &
Simeons,F.J. (1976).Psychologicalaspectsof feedinggroup.In:
Silverstone,T. (Ed.):Appetite.DahlemKonferenzen,Berlin:Abalon.
Wooley,O.W., Wooley,S.C.&Dunham,R.B. (1972).Cancaloriesbeperceived and
do theyaffecthunger inobeseandnon-obesehumans.Journalof Comparative
and Physiological Psychology,80,250-258.
Worsley,A.,Baghurst, K.I.&Leiten,D.R. (1984).Socialdesirability response
biasanddietary inventory responses.Humannutrition:applied nutrition,
38A,29-35.
Yaremko,R.H.,Fisher,H.L.&Price,J.M. (1975).Pavloviangalvanic skin
responseconditioning inoverweight andnormalweightwomen,journalof
Abnormal Psychology, 84,429-432.
Younger,J.C.&Pliner,P. (1978).Obese-normaldifferences inthe
self-monitoringofexpressivebehaviour.Journal ofResearch in
Personality, 10,112-115.
Zimmerman,M. (1983).Weightedversusunweighted lifeevent scores:istherea
difference.Journal ofHumanStress,9, 30-37.
 -167-

CURRICULUMVITRE

De schrijfstervanditproefschriftwerd op12september 1954teGroningen

geboren. In1973behaalde zijhetdiplomaGymnasium-BètaaanhetWillem
Lodewijk Gymnasium teGroningen. Indatzelfde jaarving zijaanmet haar studie
psychologie aandeRijksuniversiteit teGroningen en inaugustus 1980behaalde
zijhetdoctoraalexamen.Van september 1980 tot september 1982was zij als
wetenschappelijk medewerkster verbondenaandevakgroep
Persoonlijkheidspsycholgie enPsychodiagnostiek vandeRijksuniversiteit te
Leiden. Inseptember 1982werd zijalswetenschappelijk ambtenaar aangesteld
bijdevakgroepHumaneVoeding,waar zij,met financiële steunvanhet
Praeventiefonds, tot september 1985het inditproefschrift beschreven
onderzoek verrichtte.Opditmoment iszijalsdocent vrouwenstudies verbonden
aandevakgroepKlinische PsychologievandeKatholiekeuniversiteit Nijmegen.