Professional Documents
Culture Documents
Garonna 1999
Garonna 1999
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
International Statistical Institute (ISI) is collaborating with JSTOR to digitize, preserve and extend access to
International Statistical Review / Revue Internationale de Statistique.
http://www.jstor.org
Summary
Societal change, which takes a variety of directions and forms and in no way can be assimilated or
reduced to a single dimension, is often accompanied by a perception of insufficient understanding and lack
of control. There is a frustrated need for mastering complexity and instability, separating the voluntary
from the involuntary, the intended from the unintended, opportunities from risks, getting to the real
causes and dominating the uncertain implications of social change. Social change catches us unprepared
and confused.
In this context statistics are generally considered a fundamental instrument of knowledge, but also
part of the problem! In the public debate and in the specialized literature, the ability to measure social
phenomena through current statistics and indicators is increasingly questioned. Data-it is claimed--are
lacking, particularly longitudinal data; their quality (accuracy, relevance, timeliness, comparability, etc.)
should be improved; indicators do not provide early warning signals, policy performance evaluation,
and a precise indication of outcomes. Statistics cannot be used as a reliable and timely basis for decision
making by individuals, organizations, governments, and for understanding these decisions. In some cases,
statistics have been accused of giving a misleading and false picture of reality: do we measure the real
extent of social exclusion and unemployment? Do we fully capture the quality of life and the degradation
of the environment? Mismeasurement has been deemed by some commentators as being responsible
for the wrong focus in inflation and stabilization policies, science and technology, unemployment and
poverty. The productivity paradox, the informal economy, failure to measure welfare and the quality of
urban life are instances where statistics do not seem to provide complete and satisfactory answers to the
demand for information and knowledge. Our paper illustrates how, quite independently of measurement
techniques and data production processes, the inadequacy of the conceptual framework may explain
mismeasurement in relation to complex (multidimensional) and dynamic social phenomena. It is then to
social theories, explanations and interpretations that statisticians need to turn, in order to come to grips
with the new challenges in social measurement.
We will develop this thesis looking at a few cases where measurement issues can be connected to both
theoretical and empirical difficulties. The statistical gap which reveals itself in the mismeasurement or
difficult measurement of social phenomena is closely interconnected with the social science gap. Only
close collaboration between statisticians and social scientists can bring about continuous advancement in
social science and quality improvement in social statistics.
Undeniably considerableprogress has been made at the nationaland internationallevel for the
developmentof social statistics:populationcensuses have become richerandmoreprecise;adminis-
trativefiles (such as populationregistersor welfaredata)are being exploitedfor statisticalpurposes;
sample surveys have become widespread(e.g. multipurposesample surveys);time budget and lon-
gitudinal surveys provide precious information;definitions and classifications are improved and
standardized;datacollection techniques(areasamplingor CASIC) and methodsfor datacorrection,
compilation,imputation,etc. have been upgraded;datapresentation,analysis and disseminationare
considerablymore developed now.
The 'gap in social statistics', in spite of the considerable investmentbeing made upon it, is
mainly due to an unprecedentedsurge in expectationsand demands;it is thereforethe by-product
of education,growthin civil society and in its considerationof the importanceof statistics;this gap
must thereforebe addressedby a preliminaryinvestigationof users' requirements,particularlythe
new features.
Five such featurescan be singled out as being the most relevantand intriguing,albeitnot the only
ones:
1. Performanceassessment of social policies has become both more needed and more difficult.
Social policy reform requires complex, informed, rational and forward-lookingdecision-
making.Tightbudgetspush towardscost-efficiencyandeffectiveness.Alternativeinstitutional
arrangementsare being experimentedwith and tested. 'Coherent'policy making (see OECD
1993), a cross cutting approachto indicators,accounting,and analysis, and the integrationof
differentsocial data archivesfigurehighly in the political agenda.
2. The increasingrole in social protectionof non-profit,and otherprivateorganizations,house-
holds and citizens, reliance on self-help, individualresponsibility and choice determine a
widespreaddemandfor guidance,counselingand publicinformation.Which school has better
facilities and teaching? In which areas and sectors is one more likely to find a job? Which
degrees give a higher probabilityof successful careers?How safe is a neighborhood?How
do we choose a hospitalor a specializedphysician?Should one postponehaving childrenfor
careerreasons?etc. While individualsandbusinesshave, in general,easy access to basic infor-
mation for their economic choices for consumption,investment,saving, etc. (e.g. short-term
indicators,CPI, interestand exchangerates),datafor social choices are less easily available.
3. People increasinglydemanddisaggregatedatarelatingto theirterritory,communityand envi-
ronment.Informationmust be susceptibleof being linked throughgeo-referencingwith other
informationdrawnfrom differentsources.
4. Developmentof telecommunications,global technologies and world news stimulatedemand
for comparablecross-countryindicatorsand analysis. This requiresstandardizationof con-
cepts, classificationsand definitionsat the nationaland internationallevel.
5. Democraciestoday strugglefor understandingthe mood and analyzingin depththe malaise of
A1U A2 U ... U An = A,
Ain A = , i # j.
It is evident thatmultiple breakdownsare possible. But not all of them are useful in detectingthe
phenomenonA breakdownthat identifiesa single constituentaspect, which we will indicate as A1,
for example, is manifestly erroneousalthoughconvenientin practicalterms. In this case, A1 = A,
but this contradictsour assumptionthat A is not observable.So, in this type of situation,although
we can make use of an indicatorthatwill give us a "perfect"measurementof the constituentaspect
this indicatorwould "indicate"itself, ratherthan the initial phenomenon,in other words, A1; and
A, is only a part of A. So, if social change regardsexclusively its complement this change goes
undetected.
The logical contradictionpointed out previously can be overcome by breakingdown the phe-
nomenoninto at least two measurableconstituentaspects or dimensions:A1 and A2. However,such
a breakdownmight not be "fine"enough. In this case, once again, social change may not emerge,
by virtue of a sort of "aggregationeffect". One way to surmountthis hurdle is to adopt a more
sophisticatedbreakdown:A1, A2,..., An, of the phenomenon-considered by n > 2.
Up to this point we have conductedthe discussionpostulatinga staticcontext.However,the world
is not static, so it is wiser to view the social phenomenonused as an example as a process evolving
over time. We indicate this fact utilizing the following notation:A', t = 1, 2,.... Introduction
of the dynamic aspect furthercomplicates the picture. Suppose we have identified, at time t, an
"optimum"or "ideal"breakdown:A', A:, ..., A', which allows us to capturethe changes affecting
the phenomenonconsideredin theirentirety.This is not to say that.attime t + k, this breakdownwill
still be optimum.For example, at time t, A' U A U -.. U A' = A', at time t +k, A U A U .--U A' C
At+k. In other words, social change will have renderedthe optimum breakdownrelative to time
t obsolete, in the sense that this breakdown,correspondingto time t + k, no longer "covers"the
phenomenonin its entirety.It is clear that,once again, social change cannotbe completely detected.
Ourargumentillustrateshow, quiteindependentlyof measurementtechniquesand dataproduction
processes, the inadequacyof the conceptualframeworkmay explain mismeasurementin relationto
complex (multidimensional)anddynamicsocial phenomena.It is thento social theories,explanations
and interpretationsthat statisticiansneed to turn,in orderto come to grips with the new challenges
in social measurement.
We will develop this argumentlooking at a few cases wheremeasurementissues can be connected
to both theoreticaland empiricaldifficulties.
unemploymentby age, sex and duration,the distinctionbetweenjob losers, job leavers and persons
seeking their first job, the disaggregationsby educationaland skill levels, by territorialarea, by
household typology and position within the household. One group of indicatorshas linked the job
searchwith the conditions of supply:minimumacceptablesalary,geographicavailability,the search
for temporaiyor atypicaljobs, the intensityof the job search(numberand types of search actions).
The labor force surveyprovides numerousrelevantindicatorsin this respect,albeit sometimes with
limited statisticalsignificance.The sensitivityof the unemploymentrateto variationsin these supply
conditions has proven to be significant and growing. This means that the labor marketfunctions
as such, and the circumstancesof purely Keynesian(or involuntary)unemployment,depending,in
other words, exclusively on effective demand,are limited.
The datacould allow indicatorswhich highlightstructuraltypes of relationshipsto be constructed;
for example, persons willing to work on certain specific conditions depending on a shortage of
information,on the house rentalmarket,or searchcosts; or labordemandindicatorsconditionedby
the rigidity of hiring/firingpractices,by the difficultiesof finding labor,by the mismatchbetween
requestedandavailablequalifications.Importantdiscussionsandformulationsareunderwayto define
"discouragement"indicators,involving personswho do not seek a job because they doubtthey will
find one, involuntarypart-timelabor,and offers of a second or an informaljob.
Thereis a greatrequestfor dataon durationandflows, whichpermitus to understandthepersistence
(hysteresis) of unemployment,inter-companyand inter-sectorialmovementsand theirimplications
for salaries and careers(the Lilien index, for example).
Laborstatisticianshave sought to separateout the inactive,i.e. those who are not partof the labor
force, and the persons who are in the educationalor professionaltrainingsystem or whose status
alternates(cf. OECD, 1992). Operatorsin the so-called "thirdsector",those engaged in "socially
useful", unpaidcommunityor voluntaryactivitiesare hardto classify.
Registrationof vacancies, with the indispensableaccompanimentof informationon durationcon-
tractualand salaryconditions,has become particularlydifficult,primarilybecauseof the increasingly
marginalrole played by public placementoffices and the lack of statisticalsurveys on the subject.
Given this varietyof indicators,the problembecomes one of choosing the correctone or groupof
correct ones. At this point the lack of theoreticaland conceptualframes of referenceable to guide
these choices andfurnishassistancein the decision-makingprocessesby publicand privateoperators
and individualcitizens becomes clear.The prevailingtrendhas been to substitutethe unemployment
rate with a set of indicators,based on the considerationthat a single index is unableto satisfy the
need for informationand to adequatelyrepresenta complex andmulti-facetedsituation.This policy,
assumed by the pioneering AmericanBureauof LaborStatistics, has been followed by numerous
statisticalinstitutesand by Eurostat,althoughsometimes with a dose of abstractempiricism.
The true obstacle is the still wide divergenceexisting between analysts and labor scholars and
official statisticians.The formeroften tend to neglect conceptualquestionsconnected to statistical
production,only to then lamentthe scarcityand imprecisionof data.The latterfeel sometimes they
are able to act alone, taking chargeof survey techniquesand limiting themselves to a conventional
harmonizationof definitionsand classifications.
Reassessing unemploymentstatistics may, in this context, representan importantoccasion to
bridge this gap and work together.Under pressurefrom the EuropeanParliament,public opinion
and the governmentsof the member-states,the EuropeanUnion has initiatedan ambitiousprogram
to work out harmonized indicators, to define a common structureon labor force surveys, and
to develop the Europeanpanel survey. An equally ambitious programis underwayto formulate
conceptual and methodological definitions in preparationfor the 1998 meeting of the Conference
of Labor Statisticians.Withoutthe active involvementof the scientific community,labor scholars,
statisticians,economists, sociologists and institutionalscholars,etc., we will be unable to close the
gap between unemploymenttheory and measurementpractice.This will create heavy risks for the
multidimensionalnatureof phenomenasuch as the state of health, and often do not allow a full
understandingof structuralchange. It is well known for instance that a chronic state implies that
clinical parametersdo not change significantly;butthis does not meanthatthe benefitof alleviatinga
chronicdisease cannotbe measured(see Levorato,Rozzini & Trabucchi,1990).The scarcesensitivity
of clinical parametershas stimulatedthe searchfor new healthindicators,more accuratelyreflecting
the healthconditionsof the elderly.In particular,functionalmeasureshave been developed,aimed at
directlydetectingthe ability to lead daily activitiesandrespondto stimuliand conditions(Williams,
1989, Guralnik,1989). Howeverwhen the healthstate does not dependonly on physical efficiency,
even these indicatorsmay be inadequate.
Other relevant examples may be drawn from the field of education and training.In this case,
the objectives of policy are often defined in terms of knowledge and competence or in terms of
employabilityand labor marketleverage, going beyond specific professionalskills and operational
abilities. Therefore,measuringthe effectivenessof trainingor educationrequiresa complex process
of conceptual refinementand testing. A furthercomplicationis the need to consider the impact of
the environmentand otherinterviewingfactors;isolating the incidence of a specific policy measure
and controllingfor side or unintendedeffects may be quite an impossible task.
Difficultiesmay become insurmountableif the conceptualworkis carriedout only at the empirical
level or worse as a matterof conventionalaccordsamongnomenclaturistsandstatisticians.As a matter
of fact, the distinctionbetween outcome and output,which is proving expedientand stimulatingin
specific fields of statisticalanalysis,mightlead to an excessive simplificationof the conceptof output,
and endangera separationof social policy analysis from the work carriedout in the economics of
services productionandexchange.When,for instance,Vittadinistatesthat"outputis the standardized
productof individualservices, measurableat a given time in termsof quantityand quality (e.g. the
numberof blood tests, the numberof studentsbeing examined,the efficiency of specific machinery
in taking medical tests, etc.)", he seems to adopt a narrowand limited definition of output, and
risks confusing the output of a given service with the productionprocess of the service itself. In
the nationalaccounts literature(see Hills, 1977 in particular),a "service"is defined as a change in
the state of a person or a good belonging to an economic unit, induced by the activity of another
economic unit; thereforethe outputmust be measuredby identifying the extent of such a change,
and not by looking at the activity of the provider."Theprocess of producinga service is the activity
which affects the person or good belonging to some economic unit, whereasthe outputitself is the
change in the condition of the personor good affected"(Hill, 1977, p. 318).
To a largeextent outputand outcomeoverlapand interact;they both identifya changeof state and
conditionof the benefit-holder;they differhoweverin theirdrawingattentionto a differenttime span
for the assessmentof the impactof a measure.Outputis evaluatedby comparingthe initial state with
the one immediatelyafterthe service was provided."Servicesare consumedas they are producedin
the sense that the change in the conditionof the consumerunit must occur simultaneouslywith the
productionof the change by the producer:they are one and the same change"(Hill, 1977, p. 337).
Therefore, output is measuredat the time in which the activity ends. The outcome, instead, is the
long-termresults;it can only be determinedafter some time following the completionof providing
the service.
But here is where theory must come to the rescue if progress is to be made. When and where
do we collect the relevant information?And what is most relevant?Without a well-established
theoreticalfoundation,the searchfor outcomes,and effectiveness,become unsustainable.When the
longitudinaldimension is introducedinto statisticalinformationsystems, and when those systems
need to dialogue and integrate,only a solid theoreticalperspectiveenables a viable and effective
investmentprogramin measurementand datacollection.
Data inadequacies,particularlylongitudinaldata and lack of integrationof survey and adminis-
trativefiles, are often pointedout as being a majorcause of difficultyfor social theoryadvancement
and the assessment of social policy performance.However without social policy research and in-
vestmentin social analysis it would be practicallyimpossibleto fill the statisticalgaps. Paraphrasing
Thomas Jefferson we might say that we may measure a lot of aspects across different fields at a
given time, or a limited numberof focused (and theory-driven)aspects though time; but we cannot
measure everything everywhere always! Social theory must guide a bold and realistic statistical
policy reengineeringin the social field.
References
Barbagli,M. (1984). Sotto lo stesso tetto, Bologna II Mulino.
Blalock, H. M. (1968). The Measurementproblem.In Methodologyin Social Research,Eds. H.M. Blalock & A. B. Blalock
pp. 5-27, New York:McGraw-Hill.
Ciampi, C. A. (1996). Sfidaalla disoccupazione.Bari:EditoriLaterza.
Crozier,M. (1995). La crise de l'intelligence. Essai sur l'impuissancedes ilites a'se re4former.
Paris:InterEditions.
di Orio, F. (1988). Statisticamedica. Roma:La Nuova ItaliaScientifica.
Fukuyama,F. (1995). Trust.TheSocial Virtuesand the Creationof Prosperity.New York:A Free Press PaperbacksBook.
Galbraith,J. K. (1996). The Good Society. TheHumanAgenda.Boston: HoughtonMifflinCompany.
Garonna,P. (1994). StatisticsFacing the Concernsof a ChangingSociety. StatisticalJournalof the UnitedNations ECE. 11.
Garonna,P (1997). Misuree teorie del mercatodel lavoro: il ruolo della statistica ufficiale.ConvegnoSIS '97, La Statistica
per le Imprese,Torino,Italia,April 1997.
Garonna,P. & Geretto,P. (1996). OfficialStatisticsand ScientificResearch:an EvolvingRelationship.Conferencein Honour
of the Bi-centennialof Stefano Franscini'sBirth,Ascona, Switzerland,November 1996.
Glock, C. Y (1965). Y a-t-il un reveil religieux aux Etats-Unis?.In Mithodes de la sociologie, Eds. R. Boudon & P. E
Lazarsfeld.Vol I. Paris-LaHaye: Mouton.
Guralnik,J. M., Branch,L.G., Cummings,S. R. et al. (1989). Physical PerformanceMeasuresin Aging research.Journal of
Gerontology,44, 141-146.
Hill, T. P. (1977). On Goods and Services. TheReviewof Incomeand Wealth,December.
Hill, T. P. (1997). Tangible,Intangiblesand Services: A New Taxonomyfor the Clas-sificationof Output.CSLS Conference
on Service CentreProductivityand the ProductivityParadox,Ottawa,Canada,April 1997.
Hirschman,A. O. (1977). The Passions and the Interests:Political Arguments.forCapitalismbefore its Triumph.Princeton:
PrincetonUniversityPress.
Huntington,S. P. (1993). The Clash of Civilizations?ForeignAffairs,72, 51-83.
Lazarsfeld,P.E (1965). Le vocabulairedes sciences sociales. In Mithodes de la sociologie. Eds. R. Boudon& P.E.Lazarsfeld.
Vol I. Paris-LaHaye: Mouton.
Levorato,A., Rozzini, R. & Trabucchi,M. (1994). I costi della vecchiaia. Bologna: 11Mulino.
OECD (1992). EmploymentOutlook,Paris.
OECD (1993). Social Policies Ministerial,Paris.
Opit, L. J. (1993). The Measurementof HealthService Outcomes.Oxfordtextbookof health care, 10, OLJLondra.
Sabbadini,L. L. (1996). Trasformazionifamiliarie rilevazionistatistiche.RassegnaItaliana di Sociologia, XXXVII, n.2.
Saraceno, C. (1994). Unfamilismo ambivalente:le politiche della famiglia in Italia. InternationalConference"Mutamenti
della famiglia nei paesi occidentali",Bologna, Italia,October 1994.
Vittadini,G. (1997). Una metodologiastatisticaper la performanceevalutationdei servizi alla persona di pubblica utilitii,
Convegno SIS '97, La Statisticaper le Imprese,Torino,Italia,April 1997.
Williams,T. F. (1989). CurrentStatusof Biomedicaland Behavioralresearchin Aging. In Health an Aging Society.
Eds. S. Andreopulos& J.R. Hogness. New York:ChurchillLivingstone. Care.for
Resumi
Le changementsociale, qui prendsdirectionset formesdiverseset qui ne peuten aucunefagonetre assimil6ou r6dui' une
seule dimension,est souventaccompagn6parla perceptiond'une comprehensioninsuffisanteet d'une manquede contr6le.II
y a un besoin frustr6de maitriserla complexit6et l'instabilit6,touten s6parantle volontairede l'involontaire,l'intentionneldu
non voulu, les opportunit6sdes risques,pourarriveraux causes r6elleset dominerles implicationsuncertainesdu changement
sociale. Le changementsociale nous prendsau d6pourvuet confus.
Dans ce contexte, la statistiqueest g6n6ralementconsideree un instrumentfondamentalde la connaissance,mais aussi
une partie meme du probleme! Dans les d6batspublics, ainsi que dans la literaturespecialisee, la capacit6de mesurerles
ph6nomenesau moyen de la statistiquecouranteet les indicateursvient de plus en plus mise en doute. Les donnees--on
dit-manquent, en particulierles donnees longitudinelles;leurqualit6(precision,pertinence,opportunit6,comparabilit6,etc.)
doit &tream61ioree;les indicateursne fournissentpas des signaux d'alerte precoce, ni l'6valuationde l'accQmplissement
des politiques, ni une indicationprecise des resultats.Donc, on ne peut pas se servir de la statistiquecomme base croyable
et opportunepour le processus decisionnel des individus, des organisationset de gouvernements,ni pour comprendeces
decisions. En quelques ca, la statistiquea 6t6 accusee de donnerune image trompeuseet fausse de la r6alit: est-ce que nous
measuronsl'extension r6elle de l'exclusion sociale et du ch6mage?Est-ce que nous capturonsentierementla qualit6de la vie