You are on page 1of 13

Cognitive Therapy and Research, Vol. 6, No. 3, 1982, pp.

287-300
The Attributional Style Questionnaire1
Chr i s t o phe r Pe t e r s o n and Amy Se mme l
University of Pennsylvania
Carl y o n Bae ye r
University of Saskatchewan
Ly n Y. Ab r a ms o n and Geral d I. Me t al s ky
State University of New York at Stony Brook
Mar t i n E. P. Se l i gman 2
University of Pennsylvania
Of current interest are the causal at t ri but i ons of f ered by depressives f o r the
good and bad event s in their lives. One i mport ant at t ri but i onal account o f
depression is the ref ormul at ed learned helplessness model , which proposes
t hat depressive s y mpt oms are associ at ed with an at t ri but i onal style in whi ch
uncont rol l abl e bad event s are at t ri but ed to internal (versus external),
stable (versus unstable), and gl obal (versus specific) causes. We describe
the A t t ri but i onal St yl e Questionnaire, which measures i ndi vi dual
di f f erences in the use o f these at t ri but i onal di mensi ons. We report means,
reliabilities, intercorrelations, and test-retest stabilities f o r a sampl e o f 130
undergraduates. Evi dence f o r the quest i onnai re' s validity is discussed. The
At t r i but i onal St yl e Quest i onnai re promi ses to be a reliable and valid
i nst rument .
'This work was supported by PHS grant MH-19604 to M. Seligman, NSF grant BNS76-
22943 to the Center for Advanced Study in the Behavioral Sciences, and Social Sciences
and Humanities Research Council of Canada grant 463-80-0003 to C. von Baeyer. C.
Peterson is now at Virgina Polytechnic Institute and State University. A Semmel is at the
University of Texas, and L. Abramson and G. Metalsky are at the University of
Wisconsin.
2Address all correspondence to M. E. P. Seligman, Psychology Department, University of
Pennsylvania, 3813-15 Walnut Street, Philadelphia, Pennsylvania 19104.
287
0147-5916/82/0900-0287503.00/0 1982 Plenum Publishing Corporation
288 Peterson, Semmel, von Baeyer, Abramson, Metaisky, and Seligman
Depression has recently been conceptualized as a cognitive disorder (Beck,
1967, 1976); in particular, a number of theorists have further suggested
that depressive symptoms might be profitably understood by taking into
account the causal attributions offered by depressives for the good and bad
events in their lives (e.g., Abramson & Sackeim, 1977; Abramson,
Seligman, & Teasdale, 1978; Golin, Sweeney, & Shaeffer, 1981; Gong-Guy
& Hammen, 1980; Harvey, 1981; Ickes & Leyden, 1978; Janoff-Bul man,
1979; Klein, Fencil-Morse, & Seligman, 1976; Kuiper, 1978; Miller &
Norman, 1981; Peterson, 1979; Peterson, Schwartz, & Seligman, 1981;
Rizley, 1978; Seligman, Abramson, Semmel, &von Baeyer, 1979; Wort man
& Dintzer, 1978). All of these attributional theories of depression propose
that depressives and nondepressives differ in their causal judgments and
that these differences are closely linked to the presence and extent of
depressive symptomatology. Some theories hypothesize that a particular
style of making attributions is a demonstrable risk factor for subsequent
depression (e.g., Golin et al., 1981).
One of the most important attributional accounts of depression is the
reformulated learned helplessness model (Abramson et al., 1978).
According to this model, depression is the result of experience with
uncontrollable aversive events. However, the nature of the depression
following uncontrollable events is governed by the causal attributions the
individual makes for them. If they are seen as caused by something about
the person ( i n t e r n a l a t t r i b u t i o n s ) , as opposed to something about the
situation ( e x t e r n a l a t t r i b u t i o n s ) , then the resulting depression is
hypothesized to involve loss of self-esteem. If the uncontrollable events are
attributed to nontransient fact ors ( s t a b l e a t t r i b u t i o n s ) , i n contrast to
transient ones ( u n s t a b l e a t t r i b u t i o n s ) , then the depressive symptoms are
expected to be long-lasting. Finally, if the uncontrollable events are
attributed to causes present in a variety of situations ( g l o b a l a t t r i b u t i o n s ) ,
as opposed to more circumscribed causes ( s p e c i f i c a t t r i b u t i o n s ) , then the
ensuing depression is proposed to be pervasive.
Thus, the reformulated learned helplessness model holds that
attributing uncontrollable bad events to internal, stable, and global factors
leads to depression. To the extent that individuals show characteristic
attributional tendencies, it is appropriate to speak of an attributional style.
The present paper describes a measure of attributional styles. This
instrument yields scores for individual differences in the tendencies to
attribute the causes of bad and good events to internal (versus external),
stable (versus unstable), and global (versus specific) factors. We describe
this Attributional Style Questionnaire (ASQ), its instructions, items, and
scoring, and present such dat a as item and scale means and standard
deviations, scale intercorrelations and reliabilities, and test-retest stabilities.
Attri buti onal Style Questionnaire 289
Some suggestions f or t he quest i onnai re' s use are also made. Finally, we
present val i di t y evi dence f or t he ASQ.
METHOD
Development of the Questionnaire
Several consi derat i ons led t o t he f or mat empl oyed. We want ed
quest i ons t hat woul d measur e t he degree t o which subjects used t he
at t r i but i onal di mensi ons of i nt ernal i t y, stability, and gl obal i t y as def i ned
by Abr ams on et al. (1978). However , it seemed a poor idea t o pr ovi de t he
subjects with possible causes, e. g. , ability, ef f or t , luck, task di ffi cul t y, and
so on, t hat we believed t o cor r espond t o t he di mensi ons of concer n, since
such an oper at i onal i zat i on has in t he past pr oven t o be pr obl emat i cal . First,
t here is no guar ant ee t hat an at t r i but i on r egar ded by an at t r i but i on t heori st
as, f or exampl e, unst abl e is so r egar ded by all subjects; some may believe
t hat l ow ef f or t is a stable charact eri st i c of t he i ndi vi dual , while ot hers may
percei ve it as unst abl e. Second, Abr ams on et al. (1978) argued t hat ability,
ef f or t , l uck, and task di f f i cul t y are t heoret i cal l y or t hogonal t o the gl obal -
specific di st i nct i on. Fal bo and Beck (1979) have shown t hat causal
at t r i but i ons assumed by Wei ner (1974) and ot her at t r i but i on t heori st s to
oper at i onal i ze i nt ernal -ext ernal and st abl e-unst abl e nei t her occur
pr eponder ant l y in t he free responses of subjects nor cluster as expect ed?
On t he ot her hand, Ross' s (1977) obser vat i on t hat t he codi ng of causal
at t r i but i ons i nt o abst r act cat egori es depends mor e on the (t heoret i cal l y
i rrel evant ) gr ammat i cal f or m of t he at t r i but i on t han on its act ual meani ng
argues agai nst t he use of compl et el y open- ended quest i ons. Elig and Frieze
(1979) r epor t ed t hat open- ended at t r i but i onal measures are not as reliable as
f i xed- f or mat pr ocedur es. Thus, a r easonabl e compr omi se, whi ch we
empl oyed, was t o ask subjects t o generat e a cause themselves f or each of
a number of events and t hen t o rat e t he cause al ong 7-poi nt scales
cor r espondi ng t o t he i nt ernal i t y, stability, and gl obal i t y di mensi ons. The
f or mat does not const r ai n or creat e t he causal at t ri but i ons made by t he
3Weiner (1980) recently argued t hat specific at t ri but i ons and their dimensional representa-
tions have distinguishable (al t hough related) effects on subsequent affect and behavi or.
Because t he concern here is with at t r i but i onal s t y l e , only the dimensional ratings are of
interest. However, Weiner' s (1980) argument is well t aken, and the possibility t hat t he
specific at t ri but i ons offered by depressives clarify t hei r sympt omat ol ogy in ways above and
beyond t hai provi ded by consi derat i on of the internality, stability, and globality ratings is
likely (cf. Pet erson et al., 1981).
290 Peterson, Semmel, von Baeyer, Abramson, Metaisky, and Seligman
subj ect but at t he same t i me allows simple and obj ect i ve quant i f i cat i on of
responses.
Admi ni s t r at i on
The quest i onnai re is gr oup- admi ni st er ed, with t he fol l owi ng di rect i ons
appear i ng on t he first page of t he test bookl et :
Pl eas e t r y t o vi vi dl y i magi ne your s e l f in t he s i t uat i ons t ha t f ol l ow. I f s uc h a s i t uat i on
ha ppe ne d t o you, wha t woul d you feel woul d ha ve caus ed i t ? Whi l e event s ma y
ha ve ma n y caus es , we wa nt you t o pi ck onl y o n e - t h e maj or caus e i f t hi s event
ha ppe ne d t o you. Pl ease wri t e t hi s c a us e in t he bl a nk pr ovi ded af t er each event .
Next we want you t o a ns we r s o me que s t i ons a bout t he cause a nd a f i nal que s t i on
a b o u t t he situation. To s umma r i z e , we wa nt you t o:
1. Read each s i t uat i on a nd vi vi dl y i ma gi ne it h a p p e n i n g t o you.
2. Deci de wha t you feel woul d be t he maj or caus e o f t he s i t uat i on i f it ha ppe ne d
t o you.
3. Wr i t e one caus e in t he bl a nk pr ovi ded.
4. Ans we r t hr ee que s t i ons a bout t he cause.
5. Ans we r one que s t i on a b o u t t he situation.
6. Go on t o t he next s i t uat i on.
Fol l owi ng t hese i nst ruct i ons are 12 di f f er ent hypot het i cal events. Ha l f
are good events, while hal f are bad event s ? Addi t i onal l y, hal f t he events
are i nt er per sonal / af f i l i at i ve, while t he ot her hal f are achi evement - r el at ed.
Thi s l at t er di st i nct i on was made t o bui l d cross-si t uat i onal general i t y i nt o
t he measur e of this "st yl e, " as well as t o al l ow f or t he possibility t hat
at t r i but i onal style f or affi l i at i ve events is di f f er ent f r om at t r i but i onal style
f or achi evement events.
A pr el i mi nar y versi on of t he scale was pi l ot -t est ed in a sampl e of 145
i nt r oduct or y psychol ogy st udent s at t he Uni versi t y of Pennsyl vani a.
It em analysis reveal ed t hat 3 of t he 12 events pr oduced low vari ance in
ratings and l ow i t em-t ot al cor r el at i ons, due t o a t endency of most subj ect s
t o rat e t hese items near t he mi ddl e of each scale. These t hree event
4These des i gnat i ons were c onf i r me d by a s ki ng a s a mpl e o f unde r gr a dua t e s at t he Uni ver s i t y
o f Pe n n s y l v a n i a t o r at e t he event s a l ong sever al s e ma nt i c di f f er ent i al scal es t a ppi ng
e va l ua t i on. The " good" event s were cl earl y per cei ved as mo r e desi r abl e t h a n t he " ba d"
event s. Al t h o u g h depr es s i ve s y mp t o ms as as s es s ed wi t h t he Beck De pr e s s i on I nve nt or y
(Beck, 1967) i nf l uenced t hes e r at i ngs , t he di s t i nct i on bet ween " good" a n d " ba d" event s
was cons i s t ent l y ma i n t a i n e d r egar dl es s o f t he ext ent o f s y mp t o ma t o l o g y .
5Judges were abl e t o di s t i ngui s h t he af f i l i t at i ve event s f r o m t he a c hi e ve me nt event s , but it
is acknowl edged t h a t t her e is n o t al ways a pr eci se de ma r c a t i on bet ween s uc h cl asses o f
event s , or at l east bet ween t hos e us e d her e. Th u s , t he f ai l ur e o f our s ubj ect s t o r e s pond
di f f er ent i al l y t o af f i l i at i ve a nd a c hi e ve me nt i t ems (see bel ow) is not s ur pr i s i ng. For a
di s c us s i on o f domai n- s peci f i c a t t r i but i ona l me a s ur e s , see Le f c our t (1979).
Attributional Style Questionnaire 291
Tabl e I. Hypot het i cal Event s o f t he At t r i but i onat Style Quest i onnai r e
Out c ome Goal ar ea Event s a
Good Achi evement (3) You become very rich.
Good Af f i l i at i on
Bad Achi evement
Bad Af f i l i at i on
(10) You appl y f or a posi t i on t hat you
want very badl y (e. g. , i mpor t a nt j ob,
gr aduat e school admi ssi on) and you
get it.
(12) You get a raise.
(1) You meet a f r i end who compl i ment s
you on your appear ance.
(6) You do a pr oj ect t hat is hi ghl y
pr ai sed.
(9) Your spouse ( boyf r i end/ gi r l fri end)
has been t r eat i ng you mor e l ovi ngl y.
(2) You have been l ooki ng f or a j ob un-
successful l y f or s ome t i me.
(5) You give an i mpor t a nt t al k in f r ont
o f a gr oup and t he audi ence react s
negat i vel y.
(8) You can' t get all t he wor k done t hat
ot her s expect o f you.
(4) A f r i end comes t o you wi t h a
pr obl em and you don' t try t o hel p.
(7) You meet a f r i end who act s host i l el y
t owar d you.
(11) You go out on a dat e and it goes
badl y.
~Numbers in par ent heses r ef er t o t he or der o f t he event s in t he act ual
quest i onnai r e.
descri pt i ons were rewri t t en t o pr oduce t he versi on of t he scale t hat is
r epor t ed he r e ? The t wel ve event descri pt i ons are shown in Tabl e I, al ong
with t hei r desi gnat i ons as good or bad and af f i l i at i on or achi evement .
Fol l owi ng each event are paral l el quest i ons. First, t he subj ect is asked
t o "wri t e down t he o n e maj or cause" of t he event . Then t he subj ect is asked
t o rat e t he c a u s e al ong t he t hr ee at t r i but i onal di mensi ons. Al so, t he subj ect
is asked t o rat e t he i mpor t ance of t he si t uat i on described. 7 The wor di ng of
6The ver si on o f t he ques t i onnai r e appear i ng in t hi s paper is avai l abl e on r equest f r om
M. E. P. Sel i gman, Uni ver si t y o f Penns yl vani a, 3813-15 Wal nut St reet , Ps ychol ogy Depar t -
ment , Phi l adel phi a, P A 19104.
7The i mpor t a nc e r at i ngs were i ncl uded in l i ght o f t he possi bi l i t y t hat t he pr opos e d
r el at i onshi p o f at t r i but i onal st yl e and depr es s i on woul d occur onl y f or i mpor t a nt event s,
or mor e st r ongl y f or i mpor t a nt event s t han f or uni mpor t a nt event s. However , anal yses of
t he pr esent dat a reveal ed t hat t he i mpor t ance var i abl e di d not consi st ent l y medi at e t he
at t r i but i on- depr es s i on cor r el at i on (cf. Gong- Guy & Ha mme n , 1980).
292 Peterson, Semmel, yon Baeyer, Abramson, Metalsky, and Seligman
t he var i ous quest i ons reflects the specific event to be expl ai ned, but the
fol l owi ng exampl e i l l ust rat es t he nat ur e of these quest i ons:
You have been looking for a job unsuccessfully for some time.
1. Write down the one major cause
2. Is the cause of your unsuccessful job search due to something about you or
to something about other people or circumstances? (circle one number)
Totally due to
other people Totally due
or circumstances 1 2 3 4 5 6 7 to me
3. In the future when looking for a job, will this cause again be present? (circle one
number)
Will never again Will always
be present 1 2 3 4 5 6 7 be present
4. Is the cause something that just influences looking for a job or does it also in-
fluence other areas of your life? (circle one number)
Influences just Influences
this particular all situations
situation 1 2 3 4 5 6 7 in my life
5. How important would this situation be if it happened to you? (circle one number)
Not at all Extremely
important 1 2 3 4 5 6 7 important
Scoring
The three at t ri but i onal di mensi on rating scales associated with each event
description are scored in the directions of increasing internality, stability, and
globality. Composi t e scores are created simply by summi ng the appropri at e
i t ems and di vi di ng the sum by t he numbe r of i t ems i n t he composi t e. 8
The const ruct i on of the scale allows for the deri vat i on of 20 different
subscales based on different composites of items. At the finest level of analysis,
one can derive 12 subscales based on three items each (e.g., rated stability of the
at t ri but i ons for the three good-out come achievement-related events).
Collapsing across the achi evement -affi l i at i on distinction, one can obt ai n six
subscales based on six items each (e.g., rated stability of the at t ri but i ons for the
six good-out come events). Finally, one can combi ne the internality, stability,
and globality scales i nt o two composite at t ri but i onal style scores, one for good
and one for bad events, based on 18 items each.
8This scoring procedure has the effect of according each item equal status, since (as
will be seen) they tended to have comparable means and standard deviations. However,
other researchers are cautioned against summing items when they are greatly discrepant with
regard to means and/or standard deviations, since this unweighted procedure would
give some items more weight than others in the composite. In the case of such discrepancies,
it is recommended that the items be normalized (i.e., subtract the item mean for the
sample from the item score and divide by the item standard deviation for the sample),
Attributional Style Questionnaire 293
Subjects
The questionnaire was completed during class by 130 undergraduates (50
males, 80 females) enrolled in an abnormal psychology course at the State
University of New York at St ony Brook. Five weeks later, the questionnaire
was again completed by 100 of these students.
RES ULTS
Sex Differences. When the analyses report ed below were comput ed
separately for males and females, no differences were found. Thus, the dat a
were pool ed across this distinction.
Means and Standard Deviations. Tabl e II presents the item and scale
means and st andard deviations. As can be seen, these values were more
comparabl e within the bad items and within the good items t han they were
between bad and good items. Good events tended to be explained more
internally, stably, and globally t han bad events (p's < .0001), hardly a
surprising finding (cf. Weiner, 1974), at least in a mostly nondepressed
popul at i on.
Table II. Item and Scale Means and Standard Deviations: Under-
graduate Sample (N = 130)
Item means (standard deviations)
Event (see Table I) Internality Stability Globatity
Rich 4.30(2.04) 4.62(1.89) 5.15(1.84)
Position 5.67(1.26) 5.71 (.99) 5.40(1.25)
Raise 5.93(1.06) 5.49(1.08) 5.43(1.37)
Compliment 5.39(1.33) 5.05(1.07) 4.76(1.59)
Praise 5.31(1.59) 5.78 (.93) 5.03(1.51)
Loving 4.94(1.39) 5.49(1.01) 4.89(1.68)
Dimensions for 5.26 (.79) 5.36 (.68) 5.11 (.80)
good events
Composite style for 5.25 (.62)
good events
Unsuccessful 4.31(1.71) 4.27(1.41) 3.31(1.91)
Negative talk 4.57(1.64) 4.11 (1.24) 3.86(1.69)
Can' t finish work 4.57(1.63) 4.67(1.16) 4.49(1.47)
Friend' s problem 4.41(1.88) 3.51(1.37) 3.98(1.90)
Hostility 3.80(1.63) 3.87(1.19) 3.53(1.72)
Bad date 4.06(1.13) 4.39(1.10) 4.05(1.59)
Dimensions for 4.29 (.84) 4.14 (.71) 3.87(1.07)
bad events
Composite style for 4.12 (.64)
bad events
294 Peterson, Semmel, von Baeyer, Abramson, Metalsky, and Seligman
Tabl e I I I . Di me ns i on Rel i abi l i t i es a n d I nt er cor r el at i ons ( N = 130)
Di me ns i on 1 2 3 4 5 6
Good
1. I nt er nal i t y (.50) b
2. St abi l i t y . 62 a (.58)
3. Gl obal i t y .38 a . 59 a (.44)
Bad
4. I nt er nal i t y .11 .01 - .03 (.46)
5. St abi l i t y - . 1 7 - . 0 7 .03 . 18"
6. Gl obal i t y - . 1 5 .04 .24 a . 28"
(.59)
.45 ~ (.69)
~p < .05.
~Fi gures in pa r e nt he s e s ar e rel i abi l i t i es e s t i ma t e d by Cr o n b a c h ' s (1951)
coef f i ci ent al pha.
Internal Consistency. The internal reliability of each subscale was
estimated using Cronbach' s (1951) coefficient alpha. Respectable al pha
coefficients of .75 and .72 were obt ai ned f or the composite attributional style
scales for good events and bad events, respectively. These scales were based on
18 ratings each.
The six-item subscales reflecting separate attributional dimensions
achieved a mean reliability of .54 (range f r om .44 t o .69); the exact coefficients
are report ed in Tabl e III. At the finest level of analysis, three-item subscales
were derived. These subscales did not attain sufficient reliability to make t hem
useful in future' research (mean al pha = .38; range .21 to .53).
Consistency Across Goal Areas (achievement-affiliation). Ratings of
internality, stability, and gtobality for achievement events were significantly
correlated with the respective ratings for affiliation events (separately for good
and bad outcomes). These correlations had a mean of .37 and a range f r om
.23 to .59 (all p' s < .05). The corr.elations mat ch or exceed the reliabilities
of the respective subscales. Thus, there is no evidence for the discriminability
of achievement f r om affiliative goal areas. While this failure t o distinguish
achievement f r om affiliation items may be a fault of the scale, it may well
reflect an actual failure of discrimination by the subjects. Particularly in an
American college sample, affiliation may be viewed in economi c and achieve-
ment terms, and attributions about affiliation may overlap greatly with
attributions about achievement.
We counsel the researcher not t o bot her maki ng a distinction between
these items unless there is a specific interest in compari ng correlations of
achievement and affiliation subscales t o external criteria t hat distinctly pertain
to each of these goal areas. I f onl y one type of external criterion is used,
prediction f r om the ASQ is likely t o be i mproved by using the composites
collapsed across goal areas rat her t han the separate subscales.
At t r i b u t i o n a l S t y l e Qu e s t i o n n a i r e 295
Consistency Across Outcomes (good-bad). We have found that
composite attributional style scores based on all of the items for bad events and
on all of the items for good events are more strongly related to depression than
are the individual attributional dimensions (Seligman et al., 1979). Moreover,
these attributional style composites are unrelated to each other, with a correla-
tion of .02 in the present sample. Separate correlations for each attributional
dimension appear in Table III. The lack of correlation between corresponding
ratings for good and bad events underscores the importance of dividing the
data along these lines. High internality for good events does not necessarily
imply high internality for bad events, and conceptions that confound the
two (e.g., Rotter, 1966) are not maximizing the power of the concept.
Intercorrelations of Dimensions (internal-stable-global). In Table III,
the intercorrelations of the individual attributional dimensions are
presented, together with their reliabilities, means, and standard deviations.
For the good events, the individual attributional dimensions were intercor-
related at a level near that of their reliabilities, suggesting that the present
questionnaire did not succeed in distinguishing them. The distinctiveness
of the three dimensions was more adequate for the bad events.
Why is there less discrimination among internality, stability, and
globality for good events? Perhaps people make fewer distinctions among
good events since they may not spend as much time ruminating over them as
they do over bad events, and may attend more to the causes of bad
events (cf. Langer, 1978; Peirce, 1955, pp. 9-12; Ryle, 1949).
If a researcher expects differential relationships between these
dimensions and some other variable, as, for example, the helplessness
reformulation does in according the internality dimension the role of
determining self-esteem, or stability for bad events the role of determining
time course of helplessness effects, then the dimensions should be separately
employed. Otherwise, though, the researcher should use the composite
Ta b l e I V. Test-Retest Correlations ( N = 100)
r
Attributional dimension for good events
Internality .58"
Stability .65 ~
Globality .59"
Composite .70 a
Attributional dimension for bad events
Internality .64"
Stability .69 a
Globality .57 ~
Composite .64 a
1 9 < .001.
296 Peterson, Semmei, von Baeyer, Abramson, Metaisky, and Seligman
scores, particularly for good events. The use of fewer subscales possessing
higher reliability will also facilitate data analysis.
S t a b i l i t y . Table IV summarizes the 5-week test-retest correlations
of the attributional dimensions and the composites. These are respectably
high, as we had hoped since the scores are hypothesized to represent a
"style." It is worth observing that not all measures of putative cognitive
"styles" prove to be so reliable, either internally or across time (cf.
Goldstein & Blackman, 1978; Streufert & Streufert, 1978), so this finding
should be underscored. 9
DISCUSSION
The present paper has reviewed some of the psychometric properties
of the Attributional Style Questionnaire. Compared to other scales in their
initial stages of development, this questionnaire seems satisfactory in terms
of the internal consistency and stability of its composites. The discrimina-
tion between the individual dimensions was not particularly precise,
especially for good events. We have speculated that this may reflect the
rough-grained nature of actual attributions about good events.
We are now engaged in alternative means of assessing attributional
style, so that the method variance associated with such assessments can
be estimated. In addition to this multimethod approach, we are at-
tempting to assess the degree to which attributional style is actually a
style. Work in personality research and attitude measurement implies that
some people have consistent styles, while other people are inconsistent.
We are also exploring the relationship of this scale to other conceptually
related scales.
In closing, we wish to discuss briefly some data bearing on the
validity of the ASQ. A number of lines of evidence, described by Peterson
and Seligman (Note 1), show that the ASQ yields scores that are related
as expected to a variety of other variables.
1. As predicted by the learned helplessness reformulation, a style in
which internal, stable, and global attributions are offered for bad events
is associated with depressive symptoms in college students, adults, out-
patients, and inpatients; to a lesser degree, the opposite style for attributing
good events is also associated with depression (Seligman et al., 1979).
In these studies, depressive symptoms have been variously measured by self-
report questionnaires and by formal diagnosis. The individual dimensions
~In t he pr es ent s ampl e, at t r i t i on bet ween t he t wo t es t i ng s es s i ons was 23070. Howe ve r , it
is unl i kel y t ha t t hi s at t r i t i on i nf l uenced t he r esul t s, si nce in a n o t h e r s ampl e o f under -
gr a dua t e s , wi t h at t r i t i on o f 4. 5% over a 6-week i nt er val , t est - r et est st abi l i t i es were
c ompa r a bl e .
Attributional Style Questionnaire 297
(and of course the composites) are consistently correlated with the extent of
depressive symptomatology. Yet to be investigated in a depressed
population are the specific roles assigned the individual attributional
dimensions by the helplessness reformulation.
2. In a cross-lagged panel design, Golin et al. (1981) found that ASQ
scores predict which college students would develop depressive symptoms
1 month later.
3. Similarly, we have found that ASQ scores are associated with the
development of depressive symptoms following poor performance by
college students on a midterm examination. This finding has been replicated
several times.
4. In several studies, we have shown that ASQ scores correlate
positively with actual attributions made by subjects for specific events, such
as rejection in a dating situation, poor performance at laboratory tasks,
and the occurrence of stressful life events.
5. When "naturally" occurring attributions are extracted from therapy
transcripts and rated blindly along our three attributional dimensions,
high correlations with the therapist's ratings of depression are observed.
6. When subjects in a learned helplessness laboratory paradigm
(e.g., Hiroto & Seligman, 1975) are divided into high and low groups
based on stability scores for bad events, only those in the high group
showed helplessness deficits 3 days after experience with uncontrollable
events. This finding remains even when internality and globality scores are
used as covariates. Thus, the specific role hypothesized for the stability
dimension is supported in a helplessness paradigm.
7. Similarly, when learned helplessness laboratory subjects are divided
into high and low groups based on globality scores for bad events,
only those in the high group showed helplessness deficits at a task highly
dissimilar to the pretreatment task. Again, this finding remains even when
the other ASQ scores are held constant.
Overall, then, we conclude that the ASQ has considerable construct,
criterion, and content validity. Its reliability is satisfactory. Further work is
needed that addresses the reliability and validity of the individual
dimensions. On the whole, the Attributional Style Questionnaire promises
to be a useful means for assessing habitual tendencies in the attributions
of causes.
REFERENCE NOTE
I. Peterson, C., & Seligman, M. E. P. Helplessness and attributional style in depression.
Paper presented at the Heidelberg Symposium on t he Development of Met acogni t i on,
July 15, 1980.
298 Peterson, Semmel, von Baeyer, Abramson, Metalsky, and Seligman
R E F E R E N C E S
Abr amson, L. Y., & Sackeim, H. A. A paradox in depression: Uncont rol l abi l i t y and self-
blame. PsychologicalBulletin, 1977, 84, 838-851.
Abr amson, L. Y., Seligman, M. E. P. , & Teasdale, J. D. Learned helplessness in humans:
Critique and reformul at i on. Journal o f Abnormal Psychology, 1978, 87, 49-74.
Beck, A. T. Depression: Clinical, experimental, and theoretical aspects. New York:
Harper & Row, 1967.
Beck, A. T. Cognitive therapy and the emotional disorders. New York: Int ernat i onal
Universities Press, 1976.
Cronbach, L. J. Coefficient al pha and t he internal structure of tests. Psychometrika, 1951,
16, 297-334.
Elig, T. W., & Frieze, I. H. Measuring causal at t ri but i ons for success and failure. Journal
of Personality and Social Psychology, 1979, 37, 621-634.
Falbo, T., & Beck, R. C. Naive psychology and the at t ri but i onal model of achievement.
Journal o f Personality, 1979, 47, 185-195.
Goldstein, K. M., & Bl ackman, S. Cognitive style: Five approaches and research. New
York: Wiley, 1978.
Golin, S., Sweeney, P. D., & Shaeffer, D. E. The causality of causal at t ri but i ons in
depression: A cross-lagged panel correlational analysis. Journal of Abnormal
Psychology, 1981,90, 14-22.
Gong-Guy, E., & Hammen, C. Causal perceptions of stressful events in depressed and
nondepressed outpatients. Journal of Abnormal Psychology, 1980, 89, 662-669.
Harvey, D. M. Depression and at t ri but i onal style: Int erpret at i ons of i mpor t ant personal
events. Journal of Abnormal Psyehol ogy, 1981,90, 134-142.
Hi rot o, D. S., & Seligman, M. E. P. Generality of learned helplessness in man. Journal of
Personality and Social Psychology, 1975, 31, 311-327.
Ickes, W. J. , & Leyden, M. A. At t ri but i onal styles. In J. H. Harvey, W. J. Ickes, & R. F.
Kidd (Eds.), New directions in attribution research (Vol. 2). Hillsdale, New Jersey:
Erl baum, 1978.
Janof f - Bul man, R. Characterological versus behavi oral self-blame: Inquiries i nt o depression
and rape. Journal of Personality and Social Psychology, 1979, 37, 1798-1809.
Klein, D. C., Fencil-Morse, E., & Seligman, M. E. P. Learned helplessness, depression,
and t he at t ri but i on of failure. Journal of Personality and Social Psychology, 1976,
33, 508-516.
Kuiper, N. A. Depression and causal at t ri but i ons for success and failure. Journal of
Personality and Social Psychology, 1978, 36, 236-246.
Langer, E. J. Rethinking t he role of t hought in social i nt eract i on. In J. H. Harvey, W. J.
Ickes, & R. F. Kidd (Eds.), New directions in attribution research (Vol. 2).
Hillsdale, New Jersey: Erl baum, 1978.
Lefcourt, H. M. Locus of cont rol for specific goals. In L. C. Perl mut er & R. A. Mont y
(Eds.), Choice and perceived control. Hillsdale, New Jersey: Erl baum, 1979.
Miller, I. W., & Nor man, W. H. Effects of at t ri but i ons for success on the alleviation
of learned helplessness and depression. Journal of Abnormal Psychology, 1981, 90,
113-124.
Peirce, C. S. Philosophical writings of Peirce (J. Buchler, Ed.). New York: Dover, 1955.
Pet erson, C. Uncont rol l abi l i t y and self-blame in depression: Investigation of t he paradox
in a college popul at i on. Journal o f Abnormal Psychology, 1979, 88, 620-624.
Pet erson, C., Schwartz, S. M. , & Seligman, M. E. P. Self-blame and depressive symptoms.
Journal of Personality and Social Psychology, 1981, 41, 253-259.
Rizley, R. Depression and di st ort i on in the at t ri but i on of causality. Journal of Abnormal
Psychology, 1978, 87, 32-48.
Ross, L. The intuitive psychologist and his shortcomings: Distortions in t he at t ri but i on
process. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol.
10). New York: Academic Press, 1977.
Attributional Style Questionnaire 299
Rotter, J. B. Generalized expectancies for i nt ernal versus external cont rol of rei nforcement .
PsychologicaIMonographs, 1966, 80(1, Whol e No. 609).
Ryle, G. The concept of mind. London: Hut chi nson, 1949.
Seligman, M. E. P. , Abr amson, L. Y., Semmel, A., & yon Baeyer, C. Depressive at-
t ri but i onal style. Journal of Abnormal Psychology, 1979, 88, 242-247.
Streufert, S., & Streufert, S. C. Behavior in the complex environment. Washi ngt on, D. C. :
Wi nst on, 1978.
Weiner, B. Achievement motivation and attribution theory. Morri st own, New Jersey:
General Learning Press, 1974.
Weiner, B. A cognitive (at t ri but i on)-emot i on-act i on model of mot i vat ed behavi or: An
analysis of j udgment s of help-giving. Journal of Personality and Social Psychology,
1980, 39, 186-200.
Wor t man, C. B., & Dintzer, L. Is an at t r i but i onal analysis of t he learned helplessness
phenomenon viable?: A critique of t he Abramson-Sel i gman-Teasdal e reformul at i on.
Journal of A bnormal Psychology, 1978, 87, 75-90.