Tilburg University
Worlds of Welfare, Worlds of Consent?
Gelissen, J.P.T.M.
Publication date: 2001
Document Version
Publisher's PDF, also known as Version of record Link to publication in Tilburg University Research Portal
Citation for published version (APA):
Gelissen, J. P. T. M. (2001). Worlds of Welfare, Worlds of Consent? Public Opinion on the Welfare State. Thela Thesis.
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal
Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
C (
2.
Worlds
of
Welfare,
Worlds
of
Consent?
Public Opinion on
the
Welfare
State
Proefschrift
ter verkrijging vandegraad van doctor
aande KatholiekeUniversiteitBrabant,
op gezag vande rector magnificus,
prof. dr. F.A. van der DuynSchouten,
in het openbaarteverdedigenten overstaan van
een door hetcollegevoor promotiesaangewezencommissie
in de aula vandeUniversiteit
op vrijdag 1 juni 2001 om 14.15 uur
door
Jean PetrusTheodorusMaria Gelissen,
geboren op 3 oktober 1972 teUbach overWorms
Acknowledgements
All is well that ends well...
About5yearsago,afterfinishingmy master-thesis, I was put in theluxuriousposition of being able to choose between a Ph.D.-student position at the Faculty
of
SocialSciences of Tilburg University or one at theInteruniversityCenterfor Social Science
Theoryand Methodology (ICS) atUtrecht University. At that time, I preferred the
former. Now that Ihavefinished my dissertation, Iamworking asaresearchfellow at the ICSin UtrechtandNijmegenafter all. So, in the end, all my ambitions have been
satisfied.
At that time, the choice between thetWO pOSitionS was very difficult, but I have never regretted taking on the project in Tilburg.
Above all, this is to
the credit of Wil Arts, my supervisor at the Department of Sociology at Tilburg University. Although he hadthe difficult jobof
reorganising the department,he alwaysfound thetime to
helps
youngPh.D. studentsearchingfor direction in his researchproject. Ienjoyedworking with himonseveralarticles, benefitting fromhisinsights, humor and
enthousiasm for our discipline. Most importantly, I am indebted to him for sharing his knowledge with me on howto handle myselfwithinthe scientificcommunity.
I amalso grateful to RuudMuffels forthe stimulating discussions I had with him
- especially during the final stages of this dissertation. Like Wil Arts, his door was
always open to me when advise was needed. Finally, I thank Olli Kangas (University
of
Turku, Finland), Jos Berghman (University of Leuven), Jacques Hagenaars andWim van Oorschot (both Tilburg University) foragreeing to be on the dissertation committee.
Many others have made my time at Tilburg University very enjoyable. My
roommate BrigitteWiddershoven - who happened to share the same 'nationalroots'
as myself - and fellow Ph.D. students Trudie Schils and Tamara van der Hoek
made the final months less tedious. Loek Halman was my brother-in-arms in the
incomprehensibility ofvalue-structures and modern micro-computers. Also, many
colleagues at theFacultyofSocialScienceswere alwaysgood company:WilcoEmons, Paqui Gallindo-Garre, Sandravan Abswoude, Andries van derArk,JeroenVermunt,
Pietervan Harberden, Ellie Roelofs, Pascale Peters, Ronald Dekker,AntonisKlidas, Johan Verweij, Ivan Nyclicek, Marloes vanEngen. Specialthanks go to Ruud Luijkx
for teaching me the research-skills I needed to start the project in the first place. I
appreciated the many useless discussions with Roel Rutten concerning our private
'Methodenstreit' duringcoffee-breaks and hiking-trips, and our common enjoyment
of the 'good things in life'. I am much obliged to Emanuel 'le sauvage' Aris and
Wicher Bersma for their support (among other things, with mastering IdrEX). Igor
'the player' Heinen and Werner Ubachs made sure that I didnot forgetto imbibe in
order to gain new inspiration. Most importantly, I thank them all for theirfriendship. On manyoccasions,TonHeinen (AiO-coordinator ofthe Faculty) made sure that financial aidwasavailabletopresentmyideasabroad. I alsothankthemembers of the secretarial pool (inparticular, MargreetMorren,JanePinas, ElsVerhoevenandAnita
Roestenberg) for theirassistance inmasteringthe bureaucracy ofauniversity. Sandy
Reijnhart (TransEd, Bloemendaal) had the difficult task
of
correcting the mistakesa Dutch native speaker inevitably makes. Of course, any errors left are my own
responsibility. Ialsothank Matthijs Kalmijn and PauldeGraafforgiving me the time to add thefinal touches to this book.
I amdeeplyindebted to myparentsfortheir love andsupport, not only in matters
of writingthisdissertation.Without theirnon-academic wisdom,which Ioftenneeded
to putthings into perspective, this studywould have notbeenfinished. Therefore, I
dedicate this book to them. Jacqueline andFrank Willtgens also gave methestrength
to keep pushing forward. They all made sure that, while visiting my home town in
Limburg, I was able to putsome distance between myself and my dissertation - not
an easy task,asevery Ph.D. studentknows.
Finally, I thank Jonneke for her love, solidarity, and, above all, her patience with a - at times - willful young scientist. Although she usually showed a great deal of
patience concerning my work on this dissertation, she made sure that I always kept on
Contents
Acknowledgements vii
1 Introduction
and Research Questions 11.1 Introduction . . . . 1.2 Researchquestions . . 1.2.1 Worlds
of
welfarecapitalism . . . 51.2.2 Popularsupport forthewelfare
state . . . 6
1.2.3 Notionsofsolidarity andchoices ofjusticeprinciples . . . 8
1.2.4 Publicsupportforhealthcare systems . . . 9
1.2.5 Responsibilityforold-age pensions . . . . . . . . . . . 11
1.3 Data . . . .1 2 1.3.1 InternationalSocialSurvey Program . . . 12
1.3.2 Eurobarometersurveyseries . . . 1 3 1.3.3 European Values Study . . . . . . . .1 4 1.4 Scope andlimitations ofthis
study . . . 14
1.4.1 Cross-national comparabilityofattitudes . . . .1 4 1.4.2 Selectionofcountries foranalysis . . . .1 5 1.4.3 Contextual-effects models and the smallNProblem . . . 16
1.5 Outlineofthe b o o k. . . .1 8 2 Three Worlds ofWelfare Capitalism or more? In Search of Ideal Types and Real
Forms 21
2.1 Introduction . . . .2 1 2.2 Research
questions . . . 23
2.3 Idealt y p e s. . . .2 4 2.4 Threeworldsof'welfare capitalism' ormore? ... 27
2.4.4 The Antipodes . . . .3 4
2.4.5 Gender... . . . .3 6
2.4.6 East-Asia . . . 37
2.5 Ideal and real typeS . . . · · · 38
2.6 Empirical robustness ofthe three-way-classification . ... .. 42
2.7 Conclusion anddiscussion . . . 46
3 Popular Support for Institutionalised Solidarity: A Comparison among European WelfareStates 51
3.1 Introductionand research questions . . . 51
3.2 Theories andfindings about welfarestate support . . . 53
3.2.1 Dimensions andlevels ofattitudes towards the welfarestate . . 53
3.2.2 Motivesto supportthe welfare state . . . 55
3.2.3 Socialpositionandsupport forthewelfarestate . . .... . . 57
3.2.4 Welfarestateregimes andsupport forthewelfarestate . . . 59
3.2.5 Other Contextual-level characteristics and support for the
wel-fare state . . . 60
3.3 Hypotheses . . . .6 0 3.4 Data, Operationalization and
Method . . . 63
3.4.1 Data . . . , , , . . . , , , . ,6 3 3.4.2 Operationalization . . . .6 4 3.4.3 Method . . . , , / . . . .6 9 70 3.5 Results . . . . 3.6 Summary and
discussion . . . 75
4 Welfare States, Solidarity and Justice Principles: Does the Typereally
matter? 79
79 4.1 I n t r o d u c t i o n. . . . 4.2 Research questions . . . 80
4.3 Conceptualframework . . . 82
4.4 Differencesinsolidarity and justice among welfare statesregimes . . . 84
4.5 Modelsand hypotheses . . . 88
4.5.1 Distributivejustice and solidarity: acausalmodel . . . 88
4.5.2 Distributive justice and solidarity: additional hypotheses . . . . 90
4.6 Data, Operationalizationand
Method . . . 93
5 Public Health Care in the Balance: Exploring Popular Support for
Health Care Systems in the EuropeanUnion 109
5.1 Introductionand research questions . . . 109
5.2 Solidarity anditsmotivationalbases . . . 111
5.3 Reasons forwelfare state
s u p p o r t. . . 113
5.3.1 Welfarestate
regimes . . . 113
5.3.2 Institutional characteristics ofthe nationalcare system . . . 1 1 5 5.3.3 Individual,socialand ideological position . . . 120
5.4 Hypotheses . , , . . . 121
5.5 Data, operationalizationandmethod . . . 125
5.5.1 Data . . . 125
5.5.2 Operationalization
. . . 126
5.5.3 Method . . . 132
5.6 Results . . . · · · 133
5.6.1 Attitudestowards public
health care . . . 133
5.6.2 Explainingdifferences inattitudes towardspublichealth care . 135 5.7 Conclusions and discussion . . . 139
6 Old-age Pensions: Individual or Collective Responsibility? An Investi-gation of Public Opinion within European WelfareStates 143 6.1 Introduction andresearch questions . . . 143
6.2 Theory and hypotheses . . . . . . . . . . . . . . . . . 145
6.2.1 Welfare state regimesandpublic opinionon pensions . . . . . 145
6.2.2 Socialpositionand opinions about old ageprovisions . . . 150
6.3 Data, operationalizationandmethod . . . 151
6.3.1 Data . . . 151
6.3.2 Operationalization . . . 152
6.3.3 Method , , , . . . 154
6.4 Results . . . 154
6.5 Conclusion anddiscussion . . . 166
7 Summary and Discussion 171 7.1 Summary . . 7.1.1 Introduction . . . . 7.2 Answers to the research
questions . . . 172
7.2.1 Worlds ofwelfare capitalism... 172
7.2.2 Popular support forthewelfare
state . . . 173
7.2.3 Notionsofsolidarityand choices ofjusticeprinciples . . . 175
7.2.4 Public support for healthcaresystems . . . . . . . . . . . . . . 176
7.3 Conclusion . . . , , . , . . . . 178
7.4 Discussion . . . 184
7.4.1 Scientificrelevance . . . 184
7.4.2 Societalrelevance... . . . 186
7.4.3 Someshortcomingsandprospectsforfutureresearch . . . 188
7.5 Concluding remarks . . . 191
References 193
List of
Tables
2.1 Anoverviewoftypologies ofwelfare states. . . . 30 2.2 Classification
of
countries according to 8 typologies. . . . . 402.3 Empiricalrobusmess of theThree-Worlds-Typology. ... .. .. 43
3.1 Factor loadings, correlations and fit-indices ofmeasurement models
for extensiveness andi n t e n s i t y. . . 66 3.2 Scale distributions of popular support for extensiveness andintensity
ofsocialpolicy. ... ... .. 70 3.3 Two-level modeloffactors affectingthe levelofsupport foranextensive
andintensive welfare state. . . . 72
4.1 Factor loadings and fit-indices for Confirmatory Factor Models of solidarityitems. . . . 96
4.2 Means (and standard deviations) of measures of preferred level of
solidarityandjustice,across c o u n t r i e s. . . 99
4.3 Two-level hierarchical linear model
of
factors affecting people's pre-ferredlevel ofsolidarity. . . . 1024.4 Two-levelmodeloffactors affecting preferenceforjusticeprinciples. . 104
5.1 Factor loadingsforitem s q 1 2 1c,q12 ld and q122formeasurements of the levelofsupportfor publichealth care. . . 127
5.2 Correlations betweenfactorscoresestimated with the Bartlett method
andwith multigroupanalysisin LISREL. . . . 127
5.3 Structural characteristics of the health and social care systems of the Europeanc o u n t r i e s. . . 129
5.4 Measuresof distribution
for
itemsq121c, q12ld and q122. . . . 1345.5 Means, standard deviations andvalid N on 'Support for publichealth
care'bycountry. . . . 135
5.6 Two-level modeloffactors affecting supportforhealth carewithwelfare
state regimesasmacro-level explanatoryvariables. . . . 136
5.7 Two-level model of factors affecting support for health care with structural characteristics of the care systemasmacro-level explanatory
variables.
. . . 138
6.1 Welfare stateregimes andthe relevance
of
modelsof
social security.. . 1486.2 Opinions about individual and collective responsibility for old age
pensions... 155
6.3 Multinomial logit regression ofthe public preference forindividual or
collective responsibilityfor pensions... . . . . . . . . . . . . . . . 158
6.4 Multinomial logit regression ofthe public preference for individual or
collective responsibilityfor pensions... . . . . . . . . . . . . . . . 163
7.1 Main findings concerningtherelationship between welfarestate regimes andpublic attitudes. . . . 180 7.2 Main findings concerning the effects
of
structural characteristics onpublic attitudes. . . 181
List of
Figures
1.1 Thestructureofcontextual explanation... 4
3.1 Relationsbetween support for extensiveness and intensiveness of
wel-fare statei n t e r v e n t i o n s. . . 54
4.1 TheoreticalModel oftheImpact of InstitutionalArrangements.. . .. 90 4.2 Determinants
of
People's Notions ofSolidarity and Choice ofJusticeP r i n c i p l e s. . . .9 3
5.1 Residential and community care services for older people in Europe
(Source: Eurostat, 1 9 9 9) . . . 119
1 Introduction
and
Research
Questions
1.1 Introduction
The scope
of
public consent to welfare policies constitutes an important topic in the political and social scientific debate about the welfare state. In all Western democracies, the idea that the preferences ofthe majority - how vaguely these may be in practice - should have at least somebearing on actual policydevelopments has become self-evident (Taylor-Gooby, 1995, p. 11). Furthermore, all welfare statesare confronted with severe challenges: the ageing population, family instability and
the labour market consequences ofglobalization and technological change. These
have led tothealmost universal claim that the welfare statehasbecomeunsustainable
andtherefore incapableofsatisfyingexpressed social needs(Esping-Andersen, 2000).
These more recent challenges have led to considerable reforms
of
welfare statearrangements in mature as well asinimmature welfare states. Consequently, against
the background
of
these developments, critics ofthe welfare state have reconsidered the chancesof
survival and the adverse consequencesof
welfare state arrangements (see, for example, Zijderveld, 1999; Schmidtz & Goodin, 1998). The question ofthe extent to which the general public is, in spite of this criticism, still committed
to the solidary foundations of the welfare state has, since the 197Os, increasingly
become the subject matter of empirical research (Coughlin, 1980; Wilensky, 1975; Papadakis & Bean, 1993; Svallfors, 1997; Kluegel & Miyano, 1995; Peillon, 1996;
Taylor-Gooby, 1998). The study
of
popularcommitment to welfare state solidarity hasbeenparticularlyintensified in the wakeof
large-scaleprojects forthecollection ofsurvey data on opinions, values and attitudes. Ingeneral, ithasbecome increasingly recognised that this fundamental knowledge isvaluable,as theseorientations may be
essential in guidinghuman behaviour. However, as Svallforsstates "we are now rich ondata,whilequalifiedanalysesand interpretations lag considerablybehind" (1995a,
P. 7).
This study can be situated in the latter tradition
of
cross-national research on attitudes and opinions. Specifically, itisconcerned withadescription and explanation2 Chapter 1
of
public attitudes towards welfarestatesolidarity anddistributivejustice inWestern,Antipodean and South EastAsian welfare states. In addition, this study investigates
the relationship betweenpublic commitmenttowelfare Statesolidarityandfairness on
the one hand, andthedifferentwelfarestateregimes - as they havebeen conceived by Esping-Andersen (1990) and, lateron, extended andamended by his critics - on the other. Citizens ofdifferent welfarestates canbeexpected tobecommitted differently
to welfare state solidarity and to differ in their choices
of
justice principles. One ofthe main objectives is therefore to investigate theextent to which tile level of public commitment to welfare state solidarity and fairness are related to the institutional
context that constitutes the welfare state ofa country. Svallfors (1995b, pp. 118-119) argues that
"in
trying to explain national differences in attitudes a focus oninstitutions can beveryfruitful".According tohim, "comparativeresearchhasshown
that institutions have a substantial impact on things such as the income distribution,
the standard
of
living, social mobility, and voting behaviour. What comparativeattitude research should aim at is to study variations and similarities in attitudes
across national contexts, and explain, or at least interpret, these as the outcome of institutional arrangements. The attitudes we may register in oursurveys are, at best,
today's traces
of
yesterday's history. Theyare remnantsof
historical processes thathavebeen structured by nationalinstitutions. National differences inattitudes could be explained as the outcome ofthe lived experience and interpretations
of
nationalinstitution". One of the main objectives is therefore to answer Svallfors' call, and
to investigate the ways in which welfare state arrangements may matterforpeople's
commitmenttowelfarestatesolidarityanddistributivejustice.
Moreover, not only, as we have stated above,willcitizensof differentwelfarestates
showdifferentlevels
of
commitmenttowelfarestatesolidarity, butalso,withinwelfarestates, theseattitudes
will
differamongsocial groups. Thefirst objective ofthis studyis to examine the determinants
of
people's commitment to welfare state solidarityand their choices ofjustice principles. This study not only investigates whether or
not these variations reflect differences in institutional context created by differing welfare statearrangements, butalsowhether or not they reflectindividual differences.
In particular, we investigate the relevance ofself-interest and ideology for people's
commitment to welfare solidarity and fairness. All studies on this topiC have shown that people in different social locations and with differingsocio-political orientations
areindeedcommitteddifferentlytowelfarestatesolidarity and that theyhavedifferent preferences with respect to justice principles, (see, for example, Papadakis, 1993;
Introductionandresearch questions 3
1.2 Research questions
The central problematic dealt with in this study concerns the following questions:
To what extent are citizens of different welfare states and with different social
characteristics committed differently to welfare state solidarityand certain principles
ofsocial justice; how can these differences can be explained. More specifically, five
research questions have been formulated. The first one addresses the discussion on
the variety in and clustering
of
welfare state arrangements atthe national level. The remainingfour research questionsconsist oftwo parts. The firstpartaddresses howand why there maybe differences among welfare state regimes in their population's
endorsement
of
welfare state solidarityand particular justice principles. The secondpart raises the question of how and why individuals maydiffer in their commitment
towelfarestatesolidarity anddistributivejustice.
With respect to the differences among welfare states, and specifically among welfare state regimes, it has been proposed that public attitudes will tend to reflect varyingtraditions
of
characteristic welfare governance. This proposition is based on the assumptionthatthesewelfarestateregimes each have beeninfluenced bydifferenthistorical circumstances,politicalframeworks andsocial values(Taylor-Gooby, 1995).
Accordingto Ullrich (2000), this is one of themore promising lines
of
researchto explain national consensus and national particularity. However, thus far, only
a limited number of studies have systematically investigated, with varying results,
the impact
of
welfare state arrangements on people's commitment to welfare statesolidarity,apartfrom theirindividual-level determinants (Svallfors, 1997;Gundelach, 1994; Papadakis & Bean, 1993; Mau, 1997). This study is also focused onassessing
the impact of bothindividual-level (social characteristics and socio-political beliefs) and group-level variables (type ofwelfare state regime and structural characteristics
of the national health and pension system) on an outcome at the individual level,
namely people's commitment to welfare state solidarity and distributive justice.
An important
contribution of
this study to previous such research is that explicitpropositions concerning the impact ofthe institutional context are formulated and
specificinformation on group membershipis included into the analytical framework. Essentially, attitudes are related to aspects ofthe social context and to indicators of location inthesocial structure andofsocio-politicalbeliefs. InFigure 1.1, wepresent
the structure
of
explanation whichisfollowed inthis study.As far as therelevance
of
societalinstitutionsisconcerned,weexamine theextentto which the institutional conditions created by welfare state arrangements have an
impact on people'scommitmenttowelfare statesolidarity andtheirchoices
of
justice principles. Here, the point of departureisEsping-Andersen's well-known classification4 Chapter 1 SOCIETAL LEVEL
Determinants:
Institutional conditions created bywelfare statearrangements x
\
Outcome:
Structuralpositions
\
Public opinionCulturalconvictions
-onwelfare state
solidarity and fairness INDIVIDUAL LEVEL
Figure 1.1: The structure of contextual explanation.
differentmodelsofwelfare provision. However, this study will notbelimitedtowelfare
states which canbeplaced underthese headings. Esping-Andersen'sclassification of welfare states has not only been applauded, but it has also been heavily criticised,
whichhasresulted in importantextensions ofhis original classification. Basically, his
critics argue that more than just threetypesofwelfare provisionexist in therealworld,
andthatclassifications
of
welfarestates should also take thesealternativemodes intoaccount. Depending on the data available, we
will
investigate the extent to whichthese different styles
of
welfare provisionrelate to people's attitudes towards welfarestatesolidarity anddistributive justice.
However, real welfare states are only empirical approximations of ideal-typical
constructs. Therefore, they may have limited relevance with respect to better
un-derstanding people's commitmentto welfare state solidarity and fairness. Structural
aspects
of
specific sectors of the national system of social protection may show astronger association with attitudes towards welfare solidarity, as theseparticular in-stitutional arrangements relate more concretely to the acceptance
of
specific stylesof
welfare provision (Ullrich, 2000). Therefore, not only the contextual effect ofbelonging to a certain type
of
welfare state regimeis investigated here, but also theextent to which structural characteristics
of
specific sectors of the national systemof
social protection relate to people's commitment to welfare state solidarity. ForIntroductionand researchquestions 5
(health) care systemsis dependenton structural characteristicsofthese systems. The contextual approach whichwefollow inthisstudy requires both the
examina-tion
of
these differencesand similarities acrosswelfare states, but also withinwelfarestates. As Figure 1.1 illustrates, the extent to which individuals are supportive of
theirwelfare system and prefer certain principles of justice are seen as the result of their location in the social structure and their socio-political beliefs, controlling for
contextual effects.
1.2.1 Worlds of welfare capitalism
In the last decades, interest in comparing the welfare states
of
modern societies hasgrown fast. To grasp the differences and similarities which exist among the welfare
state arrangements
of
different countries, the useof
theoretical models has become both a widely accepted, and sometimes strongly disputed activity (see, forexample, Baldwin, 1996). Goodin et al. (1999, p. 37) argue in favour of the classification endeavour ofwelfare states, saying that the institutions ofthe welfare state are theresult of many political tugs-of-war over a long period of time. As they point out,
in the course of all this bickering over the design and redesign
of
these institutions,however,patterns have emergedand clusters havebeenformed. Tosomeextent, the
characteristics
of
welfare states within these clusters reflect certain intentions, ideasand values. This is partially because there are only a limited number of ways of
pursuinganygiven social objective.Those who are - fromapolicy-making perspective
- involved in the shapingandreshapingof institutionalarrangements, will inparticular cling to old intellectual routines to further serve their intentions and principles.
Moreover, theseroutines are real in thesensethatthere is, tosomeextent,an internal
'regime logic' that dictates whatinstitutional options canfit together coherently and work together well.
This ideathat qualitatively different regimelogics have crystallised inreal welfare
states has gained momentum since the appearance of Esping-Andersen's
ground-breaking book The 77:ree IForlds ofwelfare Capitalism (1990). In this book, Esping-Andersen combines Marshall's (1950) definitionofsocial citizenship with Titmuss's
(1958; 1974) classification
of
threedifferentprinciples fortheorganization ofwelfare provision: residual, industrial-achievement and institutional-redistributive welfarestate models. By distinguishing qualitatively different ways of welfare provision,
Esping-Andersen aims toexemplify how specific constellations
of
politicalpower have led to the development ofhistoricallydifferentwelfarestateregimes. For this purposehe relates socialcitizenshipandwelfare (Boje, 1996, p. 19). Esping-Andersendefines social citizenship bythenumberofsocialrightsattributedtoindividuals inthewelfare
6 Chapter 1
variations in the substance
of
social citizenship and welfare state stratification, we findqualitativelydifferent arrangementsamong the state, themarket and the family. These cluster intoa liberal, conservative and social-democratic regime-type(Esping-Andersen, 1990, pp. 22 & 26),whichare influencedbytheirhistorical roots, political
systems and social values.
It is not surprising that Esping-Andersen's claim of 'three worlds of welfare capitalism' has been contested by other students ofthe welfare state. Although his
effort was much acclaimed,agreatmany alternativeendeavours have beenundertaken
to categorise real welfare states into different welfare state regimes. Although the
basic division into 'liberal'/'social-democratic'/'conservative' has, by now, become
customary, opinions differ about whether these three regime-types are sufficient to
classifywelfarestateswhich werenotincludedinEsping-Andersen's original account. In addition, as the act oftypologiSing iS a matter
of
deciding that some features areimportant ina certain respectandothers not (cf. Baldwin, 1996), disputehas arisen
about the proper classification ofrealwelfare states. In the end,theexplanatory value
of
Esping-Andersen's typologyWas questioned. The first researchquestion relates tothe discussion concerningtherobusmess oftheEsping-Andersenianworkingtypology. It readsasfollows:
1. Are there families of real welfare states or are all welfare states rather unique specimens? If the former is the case, are there three or more ideal-typical worlds of welfare capitalism?
To answer this question, wefirst review Esping-Andersen's original classification; wethen considerthevariouscriticisms which havebeenleviedagainst the typolOgiCal
approach, in general, and the theoretical underpinnings ofEsping-Andersen's
typol-ogy, inparticular. In addition,weoffera surveyofwelfarestateclassifications which
havebeen proposed to extend and amend Esping-Andersen's original classification. Wealso offer acompilation ofrealwelfare statesand their classification according to the varioustypologies. Finally, we also discuss thevariousattempts which have been undertaken to test the goodness-of-fit
of
Esping-Andersen's classification.1.2.2 Popular support for the welfare state
Thesecond researchquestionaskswhether ornotpublic opiniontowards thewelfare statetendstoreflectthe distinctive corporatist, social-democratic andliberal-leaning
Introductionand researchquestions 7
security and justice. The second research question is limited to the public's
prefer-enceswith respect tothe goal-dimensions of government intervention. Roller divides
thisintoextensiveness (whether individuals actually hold thestateresponsible for the provisionofsocialprotection) and intensity (thedegree towhich governmentshould intervene in a certain area ofsocial protection). Both kinds ofpreferences are put underthe generalheading ofthepublic'spreferenceforinstitutionalised solidarity, as
welfarestatesolidarityisachieved herebymeansofstateintervention. Apart from an examination oftheextent towhichthedifferentstyles
of
governmentintervention areechoed in public opinion, severalcontextual factors are thoughtto affect the level of support forinstitutionalised solidarity:the levelofsocialprotection, incomeinequality
and taxpolicies. Theresearchquestion isasfollows:
2. To what extent do citizens of different welfare states support institutionalised solidarity? To what extent can differences among individuals and societies in this support be explained by differences among welfare state regimes, social protection
level, income inequality, and tax regime, and, at the individual level, by social
position and socio-political beliefs?
To answer this question, people's preferences for an extensive and intensive welfare state are compared among France, Belgium, The Netherlands, Denmark,
West Germany, Ireland, Great Britain, Italy, Spain, Portugal, and Greece. Using
the technique ofmultilevel-analysis, we estimate the effects
of
belonging to acertain welfare stateregime. The effectsof
welfare effort, income inequality, and tax regime arealsoestimated.In addition,weestimatetheeffects ofanindividual'ssocialpositionand his or her socio-political beliefs,asthesecharacteristics canbe expectedtoexplain
within-countryvariationsinlevelsofsupport.
Inprevious research, cross-national variations and similarities in public
commit-ment to the welfare state have been studied extensively (see, for example Svallfors,
1997; Taylor-Gooby, 1995; Papadakis & Bean, 1993). However, most studies are restricted to ananalysis of one or only a few countries. Moreover, they usually do not
explicitlyinclude contextualconditionsalongsideof individualsocialcharacteristics in
theiranalytical framework, although the impactofwelfare statearrangements isoften
emphasised. Consequently, this presumedassociation hasnotreallybeenempirically
established (however, see Svallfors, 1997; Roller, 1995). One major contribution
of
this study is thatit
extends the previous research on popular commitment tothe welfare state, in the sense that it analyses a relativelylarge number
of
countries simultaneously, while adding comparativemeasuresofcontextual and individual char-acteristics. In this way, we estimatethe impact of each typeof
welfare state regime,welfarestate effort, incomeinequality and the tax-regime on people's preferences for
8 Chapter 1
1.2.3 Notions of solidarity and choices of justice principles
Thethird researchquestion is concerned with the problem
of
whether ornotclassi-fications ofwelfare states matters forpeople's notions
of
solidarity and their choicesof
justice principles. This research question is an extension ofthe previousresearchquestion. Specifically,weinvestigate,first ofall, people's notions
of
solidarityintermsof
the preferred broadnessof
government intervention to provide social protection which are intended to benefit citizens. Secondly, we examine the cross-national differences in preferences for the equality, need, and equity principleof
distributivejustice (Deutsch, 1975). But whyfocus on thesetwo dimensions? As Goodin et al.
(1999, pp. 22-23)argue,welfarestates can, toa certain extent,beranked onthebasis
of
certain "external standardsof
assessment" such as thepromotionof
economic ef-ficiency, socialequality, social integration and stability, autonomy, and the reductionof
poverty. These values have - in one way or another and to a greater or lesserextent - traditionally been served
by various welfare state arrangements. Moreover,
they emphasise that abroad consensus existsconcerningthesemoralvalues across all
welfare regimes. These values are the moral embodiments through which allwelfare
stateregimes,
of
whatever type,arelegitimised bytheircitizens.Thisbroad consensusindicates that in the pursuit ofthese values certain similarities among welfare states
may exist. However, the fact that welfare states quite often substantiallydiffer in the allocation and distribution
of
welfare is a consequence of differences among them concerning which particular egalitarian justice principles theyemphasise and which specific notions of solidarity they embrace. Variations in both the levelof
supportforgovernment responsibility in guaranteeing certainsocialrights andpreferences for
certain justiceprinciples willbe related to the type
of
welfarestateregime acountry'swelfare state belongs to. In this way, we investigate whether certain moral under-pinnings ofwelfare state regimes are echoed in the public's attitudes. The research
question isasfollows:
3. To what extent do citizens of different welfare states have different notions
of solidarity and do they choose different justice principles? To what extent can differences among individuals and societies in these notions and choices be explained by differences among welfare state regimes, social position and
socio-political beliefs?
We compare popular notions ofsolidarity and choices ofjustice principles among Canada,theUnitedStates,GreatBritain, Ireland,France, West Germany,Denmark,
Norway, Sweden,Austria, Belgium,TheNetherlands,Greece,Portugal, Spain,Italy, Australia, New Zealand, Japan and The Philippines. After classifying these welfare
states into liberal, conservative, social-democratic, Mediterranean, Radical or
Introductionand researchquestions 9
Moreover, we also estimate the effects an individual's social position and his or her
socio-politicalbeliefs.
As far as the explanation of cross-national differences in people's
notions of
solidarity are concerned, we mainly follow Taylor-Gooby (1998) and Kluegel and
Miyano (1995). These studies compared people's notions
of
solidarity in terms oftheirconsent to governmentresponsibilityfordifferentaspectsofwelfare provision. In
both studies, onlyalimitednumberofcountries couldbecompared, as data onpublic
attitudes towards government intervention were not available for all welfare states.
Specifically, Taylor-Gooby (1985) only compared attitudes among West Germany, Great Britain, Italyand Sweden. Based on this limited number
of
countries, Taylor-Gooby concluded that"public
attitudes, though in some respects still distinctively 'national', seemobstinatelytoresist conforming tothe dominant policythemes of the welfarestatestheyinhabit". Kluegel andMiyano (1995) comparedpublicsupport forgovernment interventiontoreduce inequalities andthe effects
of
justice beliefs on the endorsementof
governmentinterventionamong GreatBritain,Japan,West Germany, The Netherlands, and theUnited States. Theyfound thatJapan ischaracterised by aespeciallyhighlevel
of
support and theUnitedStates, as is alsofound inotherstudies(forexample Papadakis & Bean, 1993; Svallfors, 1997) by an especiallylow level of support.
Onemajorcontribution ofourapproach is thereplication and extensionofprevious
research byTaylor-Gooby(1998), Svallfors (1997), andPapadakis and Bean (1993).
This is a replication in the sense that the type
of
welfare state regime is explicitlyincluded to assessthe explanatorypower ofthe regime-typologyforpeople's notions of solidarityandtheirchoices ofjustice principles. It is anextension in thesense that
this study includes significantlymore countries. Itenables us, firstly,to include more different welfare state regimes than the original tripartite classification by Esping-Andersen (1990), and secondly, to take into account that, also within welfare state
regimes, significantvariationmay exist among welfarestates inthepublic'sconsent to governmentintervention. Moreover, this study contributes to the literature about the relationship between welfarestate arrangements andpublic attitudes by investigating whether particular egalitarian justice principles, emphasisedby welfare state regimes,
are reflected in thepreferencesofcitizens forcertain justice principles.
1.2.4 Public support for health care systems
The latter two research questions relate to the public's general commitment to the
welfarestate, encompassingthewhole gamutofwelfare provisions. However, findings
may differ when attitudes towards specific policyareas areconsidered. Here, we have
chosen to examine more closely attitudes toWards tWO policy areas which lie at the
10 Chapter 1
The followingtworesearchquestions deal withthecross-national variationsin public
supportforgovernmentprovision
of
thesetwopolicyareas.The fourthresearch questionis concerned with the explanationofcross-national variations in levels
of
public support for national health care systems. This will bedonebystudying theeffects of the type
of
welfarestateregimetowhich welfarestatesconsidered can be assigned, the structural characteristics ofthe national health care systems, and the individual social and demographic characteristics. The research
questionisformulatedasfollows:
4. To what extent are citizens of different welfare states supportive of their health
care systems? To what extent can differences among inditiduals and societies
in their coni,nimie,it to natioiial health care systems be explained by dWerences among welfare state regimes, structural characteristics of the health care systems, social position and socio-political beliefs?
To answer this question, the 1996-level
of
public support forthe national health care systems is compared among Denmark, West Germany, Greece, Italy, Spain,France, Ireland, The Netherlands, Portugal, Great Britain, Finland, Sweden, and Austria.
Only a limited number
of
studies have examined public attitudes towardshealth care and their relationships to institutional characteristics
of
welfare states(Pescosolido, Boyer, & Tsui, 1985; Ardig6, 1995; Elola, 1996). Pescosolido et al.
(1985) found thatevaluations
of
welfare policies in health varied across 8 countrieswith patterned differences between welfare laggards and welfare leaders. Ardig6
(1995) foundthatcitizens
of
sevenEuropean countries andtheUnitedStatesconsid-ered good medical care 'veryimportant' and itsprovision an 'essential'responsibility
of
the government. He also showed that the way health care services are financedaffects the degree
of
responsibility attributed to the government for providing goodmedicalcare. Finally,Elola (1996) found thatpublicsatisfaction with thehealth care
system is lower in countries with a National Health System, than in countries with
socialsecuritybased healthsystems.
The analyses in this study follow the lead of these previous studies, but extend them in significant ways. For onething, more countries are included. This enables
us to assess whetherarelationship existsbetween the amount
of
emphasisplaced onuniversalityandcollectiveresponsibility inthe institutional arrangements ofa specific
type
of
welfare state regime and the levelof
support for the national health caresystem. Moreover, this study adds to previous research by examiningthe impact of
several structural characteristics ofthe national care system while, at the same time,
Introductionand research questions 11
1.2.5 Responsibility for old-age pensions
The fifth and final research question tackles the explanation ofvariations in public
consent to government intervention in another social policy area: old-age pensions. Researchoncross-national variations inattitudes about old-agepensionsis relatively
scarce. The informed studies that do exist are mainly descriptive. In a survey of
Elite opinion on retirement pensions in Denmark, France, Germany, Greece, The
Netherlands, and the United Kingdom, George and Taylor-Gooby (1996) found
only a few variations in opinions, which neither fell into any pattern nor correlated
with pension typologies. Walker and Maltby (1997) and Walker (1993) presented an overview of the opinions of EU-citizens concerning intergenerational relationships,
pensions and living standards ofthe elderly, employment and older workers and health and social care for the elderly. Among these many issues, one
of
theirmore notable findings was that the pension contract (i.e. those working accept the
responsibility
of
financially contributing to the provisionof
pensions for the elderly), - often seen as the basic expressionof
intergenerational solidarity between those in employment and older people - is ingood health. Consensus on this topiC is foundeverywhere. However, Walker and Maltby (1997, pp. 53-59) also found that most
national pensions systems donot succeedin providingpensions which are evaluated
as adequatebytheirbeneficiaries. Eventually, thisraisesthe question of howpensions
shouldbefinanced and who shouldberesponsible for the provisionofpensions.
In this study, we focus on the latter tWO iSSUeS. We
will
examine the extent ofdifferences and similarities inpublicattitudesabout the provision
of
old-agepensions.Specifically,weexpectthatseveral contextual factors affectpopularviewsconcerning the responsibilities forthe provision
of
pensions: the typeof
welfare stateregime andthe features of thenationalpension systems. We also assess theeffects
of
individuallevel social characteristics. In other words, the final research question is stated as follows:
5. To what extent do citizens of different welfare states endorse different ways of
providing old-age pensions? To what extent can differences among individuals
and societies in these preferences be explained by differences among welfare state
regimes, structural characteristics of the national pension system, social position and socio-political beliefs?
To answer this question, people's views on who should be responsible for the
provision of pensions are compared in 1992 and 1996 among France, Belgium, The Netherlands, West Germany,Italy,Denmark, Ireland,United Kingdom,Greece, Spain,Portugal,Finland, Sweden andAustria. We estimatetheimpactofwelfarestate
12 Chapter 1
controlling for country-differences, the effects
of
social characteristicsof
individualsare assessed.
In our analyses, we explicitly focus on the collectivist and market selectivist features which can be distinguished indifferent welfare state regimes. According to
Esping-Andersen (1990, p. 80), these features are especiallyimportantwhere old-age pensions are concerned. This is because pensions constitute "acentral link between
individualism and solidarity". The impact of the regime-classification on public
attitudesistestedagainst theimpactofcharacteristics ofthenational pensionsystem.
Again, the approach taken in this study
will
augmentpreviousresearchby comparinga relativelylarge number ofcountries, andin particularby simultaneously including
contextual conditionsand individuallevel social characteristics. In this way, we try to
achieve a betterunderstanding ofthe foundations
of
public attitudes concerning theresponsibility fortheprovision
of
pensions.1.3 Data
As we noted in Section 1.1 comparative research on values, attitudes and opinions
has become rich in data during the last two decades. Themost important data-sets
for comparative purposes now come from projects such as the International Social
Survey Project, the EuropeaniWorld Values Study, the International Social Justice Project and the Eurobarometer Suruey Series. The contextual nature of our study calls for datawhich consist of two levels
of
units -hierarchically arranged- where individualsare the primary or micro-level units and a grouping of the individuals constitutes the second macro-level (Ringdal, 1992, p. 235 ). All data-sets used in this study
meetthis requirement; theyarehierarchically structuredasopinions
of
citizenswithin countries. Inthis study we draw on data fromtheInternationalSocial Survey Program,theEurobarometersurvey series and theEuropeanValuesStudy. Inthe following, we offerashort description ofthesecross-national survey-projects.
1.3.1 International Social Survey Program
The InternationalSocial SurveyProgram (ISSP) isanongoing, annual programme of cross-national collaboration, in which the collaborating organizationsjointly develop
topical modules dealing with important areas of social science. These modules are fielded as fifteen-minute supplements to the regular national surveys (or a special
survey ifnecessary) and include extensive common cores of background variables. Since 1984, the ISSPhas growntoinclude 31 nations, the founding four-Germany,
the UnitedStates,GreatBritain,andAustralia - plus Austria, Italy, Ireland,Hungary,
The Netherlands, Israel, Norway, the Philippines, New Zealand, Russia, Japan,
Introduction and researchquestions 13
France, Portugal, Slovakia, Latvia, Chile, Bangladesh, Denmark and South Africa.
In addition,
East Germany was added to the German sample upon reunification.The ISSP researchers concentrate in particular on developing questions that are 1)
meaningful and relevant to all countries and 2) can be expressed in an equivalent manner in all relevantlanguages.ThequestionnaireisfirstlydraftedinBritishEnglish and then translatedinto otherlanguages using standard back translation procedures.
In this study, we use the 1996 module on the role
of
government which coveredattitudes towards a) civil liberties, b) education and parenting, c) welfare and social
equality and d) the economy (for more detailed information about the ISSP, see
http://www.issp.org) .
1.3.2 Eurobarometer suruey series
TheEurobarometersurveyseries isaprogrammeofcross-nationaland cross-temporal
comparativeresearch, which is designed to provide regular monitoring ofthe social andpoliticalattitudes of thepublic intheEuropeanUnion. TheEurobarometerpublic opinionsurveys have beenconductedonbehalf of the European Commissionsince the
early 1970s. Currently,the target-population for sampling includesthepopulation of
anyEUmember country, aged15years and over,resident in any oftheMemberStates.
The regularsample size instandard Eurobarometer surveys is 1000 respondents per
countryexcepttheUnited Kingdom (N=1000 inGreatBritainandN=300in Northern
Ireland). Since 1995, the survey series have covered all European Union Member
countries: France, Belgium, TheNetherlands, West Germany, EastGermany, Italy, Luxembourg, Denmark, Ireland, Great Britain, Northern Ireland, Greece, Spain, Portugal,Finland,Sweden andAustria. InallMemberStates,fieldworkisconducted
on thebasis
of
equivalent basicbilingual (French and English) questionnaires whichare translated into theotherlanguages.Thebasicsampling design in allMemberStates
is a multi-stage, random probability one, and selected respondents are interviewed face-to-face in their homes. In this study, we use data from Eurobarometer 37.1 (conducted in April and May, 1992), 44.0 (conducted in October and November, 1995) and 44.3 (conducted in February andApril, 1996). These include questions to
measurethepublic'sgeneral endorsement of the welfarestate (Eurobarometer 37.1),
preferences regarding the provision
of
pensions (Eurobarometer 37.1 and 44.0) andsupport for the national health care system (Eurobarometer 44.3). See Saris and
Kaase (1997) for more information. Information about the Eurobarometer survey
14 Chapter 1
1.3.3 European Values Study
The European Values Study is a large-scale, cross-national and longitudinal survey researchprogrammeon basichumanvalues,initiated bytheEuropean ValueSystems
Study Group (EVSSG) in the late 1970s. The major aim of the EVS is to design
and conduct anempirical study ofthe moral and social values underlying European
socialandpolitical institutionsand governing conduct. Twowaves werefielded in 26
countries in 1981 and in 1990. Here, we will use data from the most recent third
wave duringwhichthefieldworkwasconducted in 1999 or 2000 throughoutEurope. Each national survey consisted of a large representative sample
of
citizens aged 18or over. The observational method used wasface-to-face interviewsbased on largely standardised questionnaires.The currentresearchpopulation consists of thefollowing
34countrieswhichparticipated in the EVS 1999:Austria, Belarus, Belgium, Bulgaria, Canada, Croatia, Czech Republic, Denmark, Estonia, France, Germany, Greece,
Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Norway,
The Netherlands, Poland, Portugal, Romania, Russia, Slovak Republic, Slovenia, Spain, Sweden, Turkey, Ukraine, United Kingdom and the USA. The 1999-survey, used in this study, included, for the first time, questions about people's preferences
for
certain justiceprinciples (formore informationabouttheEuropean Values Study,see the EVS-homepageathttp://cwis.kub.nl/fsw_2/evs/index.htm).
1.4 Scopeand limitations of this study
Thepresent studyis among the firstto systematically investigate the ways in which
the specifics
of
welfare state regimes might affect public commitment to welfarestate solidarity and distributive justice. Moreover, this study examines the ways in which other contextual factors such as welfare state effort, methods of financing
the welfare state, income inequality and structural characteristics of the national pension and healthcare system are associated withthese public attitudes. However,
the comparative and contextual approach which we use in this study raises some
important methodological issues which we
will
discuss in the following sections. In particular, these issues relate to 1) the cross-national comparabilityof
attitudes andopinions, 2) the selection ofcountries for analysis and 3) the problem
of
estimating statistical contextual-effects modelsbasedon smallsample sizes.1.4.1 Cross-national comparability of attitudes
An importantguideline incross-national researchon values, attitudesand opinions is
that scores onascale can becomparedacrosspopulations only whentheirequivalence
Introduction andresearchquestions 15
about the welfare state, surprisingly little attention is paid to this issue, although
several students
of
public opinion have underscored the importance of establishing equivalence ofscales (see, for example Kangas, 1997; Svallfors, 1997; 1999). Thisispartlybecause, often, only single-item instrumentsare availabletoindicateconsent to a specific social policy aspect. This makes the construction
of
multi-item scalesand, consequently, the empirical testing ofequivalence of scales across populations nearly impossible. Often, the researcher simply has to trust that the cross-national questionnaire was designed meticulousnessly and that it takes into account different
meanings and connotations of various concepts. Therefore, equivalence is only implicitly assumed.
However, ifwefollowtheguideline given by vandeVijver andLeung (1997), the latter approachishighly unsatisfactory. It does notestablish equivalence ofmeasures
empirically. In this study, we will - depending on the available data - follow this
guideline and test the equivalence
of
scales acrosspopulations. To achieve this, we apply Confirmatory Factor Analysis (CFA) in several instances. In order to makevalid cross-national comparisons, it is necessary to demonstrate that the manifest variables (the items used) are similarly related to the same latent belief, attitude or
value in the countries being studied (Petterson, 1995).Withintheframework of CFA
models,twoimportanthypotheses can be tested(Halman & Moor, 1993):First of all,
the hypothesis that the same structure applies (i.e. only thesame manifest variables
are related to the same belief, attitude or value) and, secondly, the hypothesis of
invariance
of
factor loadingpaths. Thisexpressesthestrength of the relations betweenthe manifestvariables and the latent belief, value or
attitude:
Using this approach,we extend previous research on public endorsement ofthe welfare state, as we try to establish empirically cross-national comparability
of
concepts whenever possible, insteadof
simply assuming that theyarecomparable.1.4.2 Selection of countries for analysis
The primary objective ofthis study is to assess the extent to which differentwelfare
regimes relate tothe public's commitmentto welfarestate solidarity and distributive
justice. Therefore,thechoice
of
countriestostudy - and consequently the comparisonswe made - weremainly dictated by Esping-Andersen's typology anditslater extensions.
Apart from that, the choice
of
countries was largely opportunistic and data-driven.For example, cross-national surveys such as the Eurobarometer-survey series are
necessarilyrestricted tothemember states ofthe EuropeanUnion,makingitpossible
16 Chapter 1
of
public attitudes towards aspects of the welfare state, important comparisons withother countries, such as the United States, could not be made due to the scope of data collection.Thismeansthat conclusionsreachedby comparativeanalyses are very
much dependent onthecountries considered,whichnecessitatescarefulinterpretation oftheresults.
Themostimportant criterion for includingacountry was thatitcouldbeclassified
according to an extended, senary version
of
Esping-Andersen's typologyof
welfarestates. However, the comparison of only six countries - one for each regime type
-would be highly unsatisfactory because differences will always exist among different countries (cf. Svallfors, 1997, p. 286). Therefore,weaimedatincluding atleast two
countries foreach regime type. The availabilityofrelevantattitudinal dataforced us, at times, to include borderline cases -for exampleThe Netherlands or Great Britain - which are not that easy to classify. This makes the interpretation
of
results moredifficult. However, if one uses a typology, hybrid cases
will
irrevocably emerge. Nospecific case can everperfectly embody a particular idealtype (Goodin et al., 1999,
p. 56).
Byomitting countries simplybecause they cannot be classified beforehand,the researcher is not able to understand the peculiarities
of
these specific cases.Eventually, ifwe accept that there are limits to the empirical power
of
discernmentofclassifications, inclusionofborderline cases iswarranted, butshould be done with caution.
In addition, countries were selected on thebasis that similar and equal numbers
of
indicators (items) beavailableto constructthe dependentvariables whichmeasurepeople's attitudes about welfarestate solidarity andtheirchoicesofjustice principles.
Therefore, we strove for a similar basic construction ofthe dependent variables in orderto assessthe cross-national comparability oftheconstructs.
1.4.3 Contextual-effects models and the small N Problem
Inthis study, weusecontextual-effects models, which generally canbe defined as "an
efforttoexplain individual-level dependent variables using combinations
of
individualand group-level independent variables" (Blalock, 1984). A general characteristic of these models is that they allow for the impact
of
macro-level characteristics on anindividual actor,controlling fortheimpact
of
individual-levelsocialcharacteristics. Inthis study, bothindividuals and countries are the units
of
analysis. In thelatter case,where the line
of
approach is the comparison ofattitudes among countries and theassessmentofthe impact oftheinstitutionalandstructural context, this studyanalyses
significantlymorecountries than were included inprevious investigations. However,
as has beenclarifiedabove, thestudyis still limited bythenumber
of
countries whichcan be used to test the relevance of the Esping-Andersenian classification for the
Introduction and researchquestions 17
must takeinto account whatiscommonly known as 'the small N problem'.
As Goldthorpe (2000, p. 49) explains, the small N problem occurs whenever countries or other macrosocial entities are taken as the units
of
analysis. In those instances, thenumber availableforstudyislikely tobequitelimited. When individualsare the units, populations can be sampled which give Ns
of
several hundreds orthousands. However, when countries are the units N, in themost idealcase, cannot
rise much above one hundred, even ifall available cases are taken. Often, there are
far feweravailablecases. Goldthorpe further points out that inapplying techniques of
multivariateanalysis, seriousdifficulties then tend tobeencountered when thesample
size atthemacro-level is notmuch greater thanthetotalnumber
of
variablesinvolved. Statistically, this means that there are too few degreesof
freedom, that modelsbecome'overdetermined', that intercorrelations among independent variablescannot
be adequatelydealt with andthat results may notberobust. Substantively, itmeans
that competing explanations ofthedependent variable may not be open to anydecisive
evaluation. As Goldthorpe (2000, p. 52) correctly argues, this is essentially not a
problem ofthemethod used, but
"a
problemofinsufficient informationrelative to thecomplexity ofthe macrosociological questions that we seekto address". Eventually, both aqualitative and a quantitative approach may suffer from the smallNproblem. As we mentioned inSection 1.4.2, the number
of
countriesincluded in this study is generally larger than in previous studies. However, given the scopeof
welfare statetypologies and the available survey-data, the number
of
countries to be analysed is still relatively limited. Therefore, it is necessary to evaluate theconsequences which the smallNproblem might have for this study.Inthisstudy, we useastatistical modelling-approach. Twotechniques
ofmultivari-ate analysis will be usedto assesstheimpact ofregime-types, structuralcharacteristics
ofthe national care and pension system and individual-levelsocial characteristics. In most analyses, a random coefficient model, the most general type ofa hierarchical linear or multilevel model (Bryk & Raudenbusch, 1992; Snijders & Bosker, 1999),
will
be estimated. According to Snijders and Bosker (1999, p. 43), the random coefficient model should be used if the groupsare regarded asasample from a (real orhypothetical)population andthe researcher wishes to draw conclusionspertaining to this population, or iftheresearcherwishes totesteffects
of
group-level variables. Theadvantage
of
using these models is that it explicitly takes into account the clusteredstructure of the data. Specifically, these models enable us to estimate the effects of contextual variables on people's commitment to welfare state solidarity, controlling foreffectsofindividual-levelcharacteristics.Theadvantage,compared toanOrdinary
Least Squares (OLS) regression approach - in which effects ofcontextual variables
18 Chapter 1
often leads to a serious risk
of
committing type I errors (asserting on the basis of the observations that there is an effect, whereas in the population there is no such effect). Because the country-level part ofthe model takes into account the correctnumberofobservations (i.e. countries),theprobability
of
committing a type I error is significantlyreduced2. Anotheradvantageofthese models isthatexplainedvariances can be calculated separately for both the individual and the contextual-level. Thismakes it possibleto assess theexplanatory value
of
contextual factors, apart from the explanatory value ofindividual-level characteristics.Therandom coefficient model is used incases wherethe dependent variable can
be seen as continuous. However, whenthe dependent variable is categorical, we use
the multinomial logit model to asses the impact
of
individual and contextual-level explanatory variables. In these models, contextual factors are disaggregated to theindividual-level. Aswe discussed above, this approach leads to the problem that the
sample size ofcontextual factorsisdramatically exaggerated. It istherefore necessary
to realise that the data are clustered by countries. Therefore, we specify that the
observations are independent across groups (clusters), but not necessarily within
groups. This procedure affects the estimated standard errors, but not the estimated coefficients. This significantlyreducesthe probability
of
committing a type I error inthe evaluation
of
contextual-leveleffects.1.5 Outline ofthe book
Chapters 2 through 6 present the results of this study. The research questions formulated in Section 1.2aredealt with inthesefivechapters. Eachchapteraddresses
one ofthe research questions. The chaptersare writtenas separatearticles, of which
Chapters 2,3,5 and 6 have been previously published, and Chapter 4 has been
submitted toan international journalfor publication :
· Chapter 2 is an extended version of an article which was published inMe,ts & Maatschappij (Arts & Gelissen, 19998). The augmented version of this chapter
hasbeen submittedfor publication toan internationaljournal.
· Chapter 3 is an extended version of an article published in the International Journal of Social Welfare (Gelissen, 200Ob). An earlier version, in Dutch, was
publishedinSociale Wetenschappen (Gelissen, 1999).
· Chapter 5 appeared as an article published in the International Journal of Social IFefare(Gevers, Gelissen, Arts,& Muffels, 2000).