International Journal of Public Opinion Research Vol.  No.  © World Association for Public Opinion Research
; all rights reserved
THREE WORLDS OF PUBLIC OPINION?
VALUES, VARIATION, AND THE EFFECT
ON SOCIAL POLICY
F. John Mehrtens III
A B ST RA C T
The variation among advanced capitalist democracies in terms of welfare programs has
been well documented. Given the theoretical relationship between public opinion and
public policy in democratic states, we should expect to see similar differences among
such countries in terms of their mass political preferences. Rooted in Esping-Andersen’s
familiar and influential ‘Three Worlds’ typology, this paper provides an analysis of this
relationship for  advanced capitalist democracies using International Social Survey
Program, World Values Survey, and Eurobarometer data. Results suggest () a noticeable
ideological rift between the more libertarian English-speaking countries and the more
socially-oriented states of Scandinavia, continental Europe, and Japan; () a strong fit
between differences in public opinion and variation in policy orientation when ideological
responses are considered separately from policy-oriented results, consistent with both Free
and Cantril () and Coughlin (); () the unexpected emergence of Sweden and Great
Britain as anomalous cases which exhibit patterns of policy preferences at odds with their
statuses as bulwarks of the Social Democratic and Liberal welfare types, respectively;
and () a relationship between public opinion and social policy roughly similar to that of
previously-established causal relationships, justifying the further examination of mass
political preferences as a source of the variation in policy among our sample countries.
By definition, democracy implies at least some sort of relationship between public
opinion and government policy. Since we know there is significant variation
between countries, even seemingly similar ones, in terms of their social policy
(levels of spending, range and availability of programs, etc.), should we not also
expect to see related differences in public opinion at the national level? This is
the primary question which motivates this study.
Specifically, this essay will analyze the attitudes of citizens from  advanced
capitalist democracies with respect to the proper role of government within the
larger society, focusing particularly on the social services spending designed to
The author would like to thank John Stephens, Marco Steenbergen, Gary Marx, Carol Mershon, Jim Stimson,
Cyndy Mehrtens, and the editor, staff, and reviewers of IJPOR for their invaluable advice and assistance.
The article was first submitted to IJPOR January , . The final version was received August , .

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
enhance standards of living, such as health care, education, pension, and unemployment programs. Viewed within the context of the larger global community,
these  states share a number of important characteristics: democracy, capitalism,
a half-century or more of political and economic stability, a high standard of
living, culture, and status in the global arena, to name a few of the most obvious.
Yet when one looks at this small and seemingly homogenous group of countries
in isolation, it does not take long before the policy differences become apparent.
Perhaps the most significant and most analyzed source of variation among these
countries is the extent to which they manage the free market’s self-regulating
mechanism in an attempt to combat some of the unpleasant external costs which
the market tends to ignore (if not cause). In other words, some countries are more
willing than others to let the ‘invisible hand’ govern itself, even though while
producing a socially desirable allocation of resources it simultaneously offers some
people little more than a nasty punch in the nose. More to the point, if the market
provided universal health care and education, as well as care for the elderly and
unemployed, it would be unnecessary for governments to do so, but the governments of all advanced capitalist democracies do indeed provide such services to
compensate for the market’s inadequacies, though programs vary a great deal from
country to country in terms of their funding level, range, duration, and eligibility
requirements. Given this well-documented diversity (Esping-Andersen’s 
The Three Worlds of Welfare Capitalism being the hallmark study), along with the
significance of public opinion to democratic governance, it seems worthwhile to
inquire as to whether or not these clear differences between countries’ public
policies are reflected in similarly clear differences in the attitudes of their citizens.
Following a review of the relevant literature and a brief methodological
discussion, this paper will proceed to accomplish three specific tasks. As a foundational first step, I will briefly justify my methodological decision to rely on EspingAndersen’s () decommodification index as a policy indicator later in the paper.
Second, we will undertake a direct comparison of cross-national values in order to
assess whether or not countries do indeed cluster together with respect to values as
they do in terms of social policy. Last we will address the issue of causality. Data
limitations render more advanced statistical tests implausible, but it is possible,
through simple correlation and regression analyses, to compare the strength of the
relationship between public opinion and policy against a number of variables
previously found to affect the variation among advanced capitalist democracies.
T HE O R E T I C A L O R I E N T A T I O N
PUBLIC OPINION AND PUBLIC POLICY
According to Webster’s Dictionary (Mish, , p. ), the term democracy,
based on its Greek roots, means ‘government by the people’ or ‘rule of the
THREE WORLDS OF PUBLIC OPINION

majority’. Direct democracy, which would be the most straightforward application of the concept, is fraught with logistical and theoretical complications which
render it impractical at best, dangerous at worst. As such, democracy as it is
practiced in modern nation-states is largely of the representative variety. Policy
makers are chosen by their fellow citizens to represent community interests in
the political process, or in other words, public opinion shapes governmental
policy. So goes the theory, at any rate, but to what extent does policy reflect
public preferences in the real world?
As Burstein () has noted, the causal relationship between public opinion
and public policy is an understudied area of inquiry in the social science literature.
More common are projects which attempt to decipher the basic nature of public
opinion (such as Free & Cantril, ) or evaluate public opinion as the dependent
variable (a few classic examples are McClosky & Zaller, ; Zaller, ; a couple
of very current examples would be Moene & Wallerstein, ; Iversen & Soskice,
). However, political science research has yielded a couple of important
findings in answer to the question of whether public opinion input does indeed
influence public policy output.
Page and Shapiro () provide a pioneering study in the effort to uncover
the causal dynamics at work by comparing changes in both public opinion and
public policy in the United States and find that public opinion often has at least a
proximal affect on policy. And while feedback between the two certainly seems to
be present, it is indeed public opinion which seems to cause policy more often
than the reverse. However, they caution against any grand illusions about the
representativeness of the American political system, noting that fairly simplistic
research design necessitates further study before such a claim could be made.
Stimson, MacKuen, and Erikson () develop the notion of ‘dynamic representation’ and, through the use of a more advanced statistical model, find that
each branch of American government, and indeed the government as a whole, is
highly responsive to changes in public opinion. Focusing on defense spending in
the United States, Wlezien (, ) addresses the feedback issue by conceptualizing public opinion as a ‘thermostat’, which registers responses to existing
government policies (or changes in them), so that government can in turn
respond by adjusting policy until it coincides with the public’s preferences.
Though some of the specifics of the complicated causal relationship remain
elusive, studies such as these provide an important foundation by convincingly
establishing the fact that public opinion does play a role in shaping public policy,
at least in terms of the American example, as democratic theory suggests.
VARIETIES OF CAPITALISM
The body of literature which seeks to make sense of the significant variation in
terms of government policy and social outcomes among seemingly homogeneous

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
advanced capitalist democracies is perhaps more developed than that on the
causal relevance of public opinion (though it has by no means answered all of its
own relevant questions). As mentioned previously, Esping-Andersen () provides us with the most widely-accepted typology for classifying these countries.
Beginning with the notion that social outcomes are more important than the
mere existence of programs or the amount spent on them, Esping-Andersen
focuses on the explicitly Polanyi-esque concept of decommodification, or the
degree to which social programs ‘emancipate individuals from market dependence’
(, p. ), as the crucial feature differentiating  advanced capitalist democracies from one another. Because basic human needs must be met on a consistent
basis, labor is different from the other factors of production in that its price cannot
realistically fall below subsistence level in the medium or long term. Decommodification is used as a measure of the degree to which a country’s residents are
able to secure basic needs free from exposure to the unpredictability of market
forces (supply and demand).
Measuring decommodification through an analysis of old-age pensions, sickness
benefits, and unemployment insurance, Esping-Andersen finds that countries
naturally separate into three distinct clusters. The Social Democratic model characteristic of the Nordic states, which promotes equality and social rights through a
relatively high degree of market intervention, provides a level of benefits that
effectively reduces the gap in standard of living between rich and poor. The Conservative (or Christian Democratic or corporatist) model of continental Europe
(Japan is placed here also), with an emphasis on social stability and a preservation
of hierarchical class or status relationships, provides substantial universal
benefits, but not at a level sufficient to achieve any significant redistributive
effect. Lastly, the Liberal model of the Anglo-American countries, epitomized by
a relatively heavy reliance on the market mechanism and therefore relatively little
protection from it, offers means-tested benefits, typically in paltry amounts and with
strict eligibility requirements, to a relatively few citizens. Esping-Andersen’s
investigation indicates that none of these models is epitomized in any one country
as an ideal type, but that three distinct clusters do indeed emerge.
Combining Esping-Andersen’s important analysis with the above discussion
about the relationship between policy and public opinion, we arrive at the question
which serves as the foundation for this analysis. Are there variations in public
opinion from country to country which mirror (and may in fact underlie) the
observed variations in managing the social effects of a capitalist economy? This
question has been addressed previously by both Coughlin () and Svallfors
(, forthcoming). Among Coughlin’s key findings is a discernable relationship between social policy and mass ideology, marked by a striking absence of
hardcore laissez-faire sentiment, in the eight countries he studies. Similarly,
Svallfors has also found some evidence of a fit between public opinion and welfare
type across eight countries, using a four-way classification scheme developed by
THREE WORLDS OF PUBLIC OPINION

Castles and Mitchell (). The problem with these studies, as addressed by
both of the authors themselves, is limited data availability, which, like a nagging
fly, perpetually plagues anyone who studies comparative public opinion (and
certainly goes a long way toward explaining the relative dearth of studies in this
area). Coughlin, writing over two decades ago, was largely forced to rely on the
pasting together of various surveys conducted independently in the eight countries
which made up his sample. The obvious problems are that interesting questions
asked in one country may not be asked in another, and, even if they are, the
translation of the questions may be different enough that the results are rendered
incomparable. Svallfors, writing just a few years ago, had quite a few more
options, but has confined his analysis to the in-depth coverage of a single survey, and
a relatively small number of questions, which in turn limited him to a relatively
small number of countries. The benefits of this approach are an enhanced level of
conceptual precision, but the costs are that the results of one survey may be
contradicted by another (see Mehrtens, ). While it may come with the price
of somewhat reduced specificity, this project will attempt to contribute to the
pioneering efforts of both of these researchers by following the same line of
inquiry, but looking at more countries, more questions, more surveys, and more
time, in an effort to simultaneously decipher a number of available surveys.
IDEOLOGICAL VS . O PERATIONAL ATTITUDES
Free and Cantril () identify a crucial split between ideological and operational attitudes among the American public. That is, Americans tend to give very
conservative (laissez-faire, anti-government, etc.) responses to broad ideological
questions, but then tend to respond in favor of specific policy programs, such as
health care, education, and the like, which helps explain why such programs
persist despite an enormous amount of anti-government (or at least minimal
government) rhetoric in American political discourse. As mentioned above,
Coughlin () applies this logic to his comparative analysis and achieves a similar
result. Specifically, he finds that public opinion toward specific spending
programs (operational attitudes) vary little across countries (consistently high
support), but that responses to broad, general questions (ideological attitudes) do
vary to a significant degree, and in accordance with the documented cross-national
variation in policy outcomes. As we shall see, this insight will be vital in our
efforts to make sense of the survey data analyzed in this study.
CAUSES OF VARIATION AMONG A DVANCED CAPITALIST D EMOCRACIES
The literature on welfare state development has identified a number of key variables that have contributed to the formation of distinct types of policy orientation.
For example, within the tradition of power resources theory (Korpi, ;

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
Stephens, ), both left government and unions have been found to positively
affect welfare state development through the political mobilization of previously
marginalized societal segments (particularly workers) that have much to gain
from government social activism. Similarly, numerous authors (Schmitter, ;
Katzenstein, ; Crouch, ; Hicks, ) have cited the presence of corporatist bargaining arrangements as a key to the emergence of more extensive
social programs, as the very definition of the term entails a level of cooperation
and compromise between business and labor in which the latter repudiates its more
radical tendencies (such as the abolition of private property) in exchange for
concessions such as health coverage and unemployment assistance (Przeworski,
). As an indicator of institutional legacy, Hicks () discusses early
program consolidation, arguing that those countries with a lengthy history of
governmental generosity can be expected to display similar characteristics in the
present. Huber, Ragin, and Stephens () find that constitutional structure
affects welfare state development by determining whether narrow majorities will
have the opportunity to veto populist policies.
HYPOTHESES
Based on the established findings discussed above, we are able to derive and test
three specific hypotheses. First, if we know that there is a relationship between
public opinion and governmental policy, and we know that advanced capitalist
democracies vary significantly in terms of their policies and outcomes, then we
should expect them to exhibit similar differences with respect to country-level
mass attitudes. Specifically, my hypothesis is that support for social programs
should be higher in Social Democratic than in Christian Democratic countries,
and Christian Democratic states should in turn bear the same relationship with
respect to Liberal countries. Next, within the context of testing our first hypothesis
we must incorporate Free and Cantril’s () and Coughlin’s () results
regarding operational and ideological attitudes into our comparative analysis.
Based on their work, we expect to find very little variation between countries on
questions of specific policy (which register widespread support), but a great deal
of variation on questions of broad ideology, and variation consistent with the
Three Worlds typology (Esping-Andersen, ). Last, it is expected that a brief
empirical examination will show that public opinion merits further attention as a
significant cause of welfare state variation.
S A M P L E , D A T A, A N D M E T H O D S
In an attempt to extend the efforts of Coughlin () and Svallfors (, forthcoming), and consistent with the large body of work in comparative politics on
advanced capitalist countries (such as Esping-Andersen, ), this study will focus
THREE WORLDS OF PUBLIC OPINION

on the following  countries: Australia, Austria, Belgium, Canada, Denmark,
Finland, France, Germany, Great Britain, Ireland, Italy, Japan, the Netherlands,
Norway, New Zealand, Sweden, Switzerland, and the United States.
Mention of this group of countries necessitates an immediate turn to the issue
of data availability because, while it would be nice if the hypotheses generated
above could be examined in light of survey responses from each of these  countries, this is unfortunately not the case, and that needs to be made clear from the
outset. As with previous analyses of cross-national attitudes, this project suffers
to some extent from spotty data coverage, though the relatively recent emergence
of the International Social Survey Program, World Values Survey, and Eurobarometer data are an immeasurable improvement over what was previously
available and indeed have made this project possible.
The significance of these three surveys is that they represent coordinated
efforts to ask important and interesting questions of a fairly sizable sample in a
number of different countries at roughly the same time, and just as importantly,
in successive versions. The International Social Survey Program (or ISSP) was
organized in , is structured around specific social science themes (for
instance, the ‘Social Inequality’ and ‘Role of Government’ studies analyzed here),
and has grown to include  countries. The World Values Survey (or WVS) has
been conducted in three waves (, –, and –) and has covered
more than  societies in various combinations throughout those three waves.
The Eurobarometer has, since the early s, conducted twice-yearly topicallyoriented surveys of European countries (one problem with Eurobarometer from
the perspective of this project is that it includes only one country from the Liberal
cluster, the somewhat anomalous Great Britain, to compare against an armada of
Social Democratic and Christian Democratic counterparts—as the name of the
survey implies, this is, at least for my purposes, simply an unfortunate side-effect
of its geographical focus and in no way reduces the contribution of the survey as
a whole). Once again, the extent to which these three survey projects have facilitated
comparative studies of the type undertaken in this study cannot be overstated.1
An important caveat must be issued here about interpreting poll responses:
when attempting to draw broad conclusions from these surveys, with some questions
it is essential that one does so only in the context of existing national circumstances. In other words, some questions ask for relative rather than absolute
responses, which are necessarily shaped by the respondent’s own personal experiences within her country. For example, Question  in the  ISSP survey
1
In some cases the data files contain information on response rates, in other cases they do not. In still other
cases, response rates are completely unavailable. For the ISSP Role of Government surveys, response rates
(RR  according to the ‘Standards and Best Practices’ page of the website for the American Association of
Public Opinion Research, www.aapor.com) as available in the data files range from  percent (France, ) to
 percent (USA, ); for the ISSP Social Inequality surveys from  percent (France, ) to  percent
(USA, ). A compilation of the response rates contained in the data files is available from the author.

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
asks whether income differences in the respondent’s country are too large, and
Americans are more likely than Norwegians to answer in the affirmative. So can
we conclude from these results that Americans are more concerned than Norwegians
about inequality? Clearly we cannot because inequality is much higher in the US
than in Norway, so to some extent we should probably expect this result. But
when we look at Question  from Wave  of the World Values Survey, which
calls for a comparison based on a one-to-ten response scale where one indicates
that incomes should be made more equal and ten denotes that larger income
differences are needed as an incentive for effort, the results suggest that Americans
are more comfortable with inequality than are Norwegians, as intuition might imply.
Methodologically, this essay is rather crude, perhaps necessarily so given the
nature of the data. Much of what is presented is simply my own analysis of the
survey data. However, bivariate correlations and simple regression equations will
be utilized to test for both fit and causality between attitudes and policy.
A few notes need to be made here regarding data presentation. First, to maintain
theoretical consistency, and for ease of presentation, countries will be grouped
according to Esping-Andersen’s ‘Three Words’ typology in all relevant tables
within this paper (the reasons for this approach will be explained in the next part
of the paper). On that note, there has been some disagreement about which cluster
best fits a couple of countries. Probably the most questionable among the choices
I have made are the placement of Ireland into the Liberal group and the Netherlands
into the Social Democratic group. I have done so because this approach is the most
conservative from a research perspective, as it lessens the difference in mean
cluster scores and thus should make it harder to find the hypothesized pattern.
Second, rather than merging the second and third waves of the WVS and
presenting the most recent results from each country (which would provide a
sample size of ), poll results from the two waves will be listed separately. This
is because there is a noticeable shift in all countries toward a pro-government
stance between the two waves, rendering the results of the two surveys largely
incomparable—what looks like a relatively liberal aggregate response in Wave 
becomes a relatively conservative response in Wave  (for more on this, see
Mehrtens, ). Finally, in all of the tables I have listed each of the  sample
countries, regardless of whether survey data exist for all countries in that
particular table. While this may be a bit unusual, I have presented the results in
such a manner to provide an ongoing indicator to the reader of exactly which
data we are working with in comparison to what we would optimally like to have
available.
DECOMMODIFICATION AS AN INDICATOR OF POLICY V ARIATION
Before turning to the substantive results of this study, it is necessary to tackle one
last methodological issue, namely the exclusive reliance on Esping-Andersen’s
THREE WORLDS OF PUBLIC OPINION

() decommodification index as my preferred indicator of variation in social
policy among advanced capitalist democracies, both in this paper and throughout
the larger study of which it is but a part. While I do feel the need to justify this
approach, I take comfort in the facts that many respected and influential
researchers hold the decommodification index in such high regard, and it is
certainly among the most (if not the single most) cited work within the body of
literature on welfare state variation. To be fair, Esping-Andersen’s typology has
generated both criticism and alternative classification schemes (see Castles &
Mitchell, ; Sainsbury, ; Gornick, Meyers, & Ross, ; Hicks &
Kenworthy, ; Edwards, ), but as Goodin, Headey, Muffels, and Dirven
(, p. ) note, ‘not only does it remain the firm focus of most on-going
discussions in these areas, it has by now become a well-established landmark in
relation to which any subsequent work . . . must situate itself’. The point here is
not to unduly reify Esping-Andersen’s classification scheme, but rather to justify
its centrality within my own analysis.
Theoretically rooted in the influential work of Karl Polanyi (), decommodification, in Esping-Andersen’s own words, ‘occurs when a service is
rendered as a matter of right, and when a person can maintain a livelihood without
reliance on the market’ (, p. –). As mentioned earlier, actual decommodification scores used in Esping-Andersen’s analysis are operationalized as a
combination of pension, sickness, and unemployment benefits, according to
characteristics such as replacement rates, eligibility rules, and duration of coverage.
The variation in these characteristics from country to country provides evidence
of differing levels of decommodification, thus resulting in Esping-Andersen’s
familiar ‘Three Worlds’ typology. Decommodification scores for each of our
sample countries are listed by cluster in Table .
Perhaps most importantly, the use of decommodification figures as an indicator of policy variation is justified by the fact that it is very highly correlated with
the alternative measures one might logically use in its place. For example,
decommodification is very highly correlated with alternative policy indicators,
such as welfare effort (Hicks & Misra, ; Hicks, ) and economic freedom
(Fraser Institute, ). It is highly correlated with indicators of policy performance, such as reductions in poverty and inequality, as well as with social outcomes
such as post-tax and -transfer poverty and inequality (Hicks & Kenworthy, )
and social indicators such as those relating to health, literacy, and crime
(Mehrtens, forthcoming). In sum, many of the variables mentioned above
could be used as an indicator of variation among advanced capitalist democracies, but I have chosen Esping-Andersen’s decommodification scores because
there seems to be a national economic, political, and social orientation which
characterizes advanced capitalist democratic countries and which the decommodification index seems to capture, both in the numbers and through the
historically rooted theory.

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
TABLE  Scores for the three components of decommodification and combined decommodification scores
Liberal regimes
Australia
Canada
Ireland
New Zealand
Great Britain
United States
Mean score
Conservative regimes
Austria
Belgium
France
Germany
Italy
Japan
Switzerland
Mean score
Social democratic regimes
Denmark
Finland
Netherlands
Norway
Sweden
Mean score
Overall mean
Overall standard deviation
Old-age
pensions
Sickness
benefits
Unemployment
insurance
Composite
decommodification
score
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Note: The Netherlands is often placed in either the Social Democratic or Conservative groups. See Goodin et al.
() for a further discussion. Also, I have chosen to call the middle group ‘Conservative’ rather than ‘Christian
Democratic’ simply because of the presence of Japan, which shares an emphasis on social traditionalism, but
obviously not a Christian heritage.
Source: Esping-Andersen, , pp. –.
VA R I A T I O N I N P U B L I C O PI N I O N AN D PU B L I C PO LI C Y
Given that national policies and outcomes cluster as they do into Social Democratic, Christian Democratic, and Liberal groups, it seems logical to assume that
national attitudes might follow a similar pattern. That is, we might reasonably
expect citizens in Social Democratic countries to hold more favorable views toward
government spending programs than those in Christian Democratic countries,
and we would expect citizens in Liberal countries to be the most hostile toward
those same services.
THREE WORLDS OF PUBLIC OPINION

GENERAL R ESULTS
At first glance, the relationship between differences in public opinion and differences in national policy orientations appears to be rather weak. In response to a
substantial number of questions, our paradigmatic understanding of the differences
between advanced capitalist democracies seems to display at best an unimpressive
relationship to differences in country-level public opinion. In general, among the
countries which come across as the most consistently supportive of government
programs are some of our Conservative states. For example, respondents in these
countries are among the most likely to feel that income differences in their country are too large (ISSP, , Question ), incomes should be made more equal
(WVS, Waves  and , Question ), government is not doing enough to fight
poverty (Eurobarometer, , Question ), and society is unfair (Eurobarometer,
, Question ), as we see in Table . As further evidence, note the mean
scores for each cluster, which show that it is the Conservative group which
generally displays the most pro-government sentiment.
Simultaneously, some of our egalitarian Social Democratic countries produce
responses to the political right of some of our typically more anti-government
Liberal states on a number of questions, though this seems largely related to the
aforementioned issue of absolute vs. relative questions. Swedes, for example, are
more likely than Americans or Australians to feel that people should provide for
themselves rather than relying on the government (World Values Survey, Waves
 and , Question ) and if one were to focus only on the  version of the
ISSP (not to be confused with the first data column of Table , which provides
the most recent result for each country), it would appear that respondents in the
US, Great Britain, and New Zealand are more likely to see inequality as a problem
than are citizens in Norway (ISSP, , Question ).
What is quite clear, however, is that our Liberal countries consistently tend to
congregate at the conservative end of the distribution with respect to survey
responses, regardless of the question asked. For example, Australia, Canada, Great
Britain, and the United States are among the most likely to argue that larger income
differences are needed as an incentive for effort (WVS, Waves  and , Question
), and those polled in Australia, Canada, and the US are among the least likely to
favor an increase in government ownership at the expense of private ownership
(WVS, Waves  and , Question ). Respondents in Australia, Canada, New
Zealand, and the US are among the least likely to feel that the government should
reduce inequality (ISSP , Question , and , Question ). Similarly,
Australians, Canadians, and Americans are among the least likely to feel that government should provide a basic income (ISSP , Question  and , Question ). Other countries do occasionally provide responses similar to those of the
Liberal countries, but it is generally not the same country with any degree of regularity (with the possible and highly unexpected exception of Sweden). Likewise,
Liberal regimes
Australia
Canada
Ireland
New Zealand
Great Britain
United States
Mean score
Conservative
regimes
Austria
Belgium














Most recent results
Percent agree







Too much Should gov’t Should gov’t
provide a
reduce
inequality
basic
()
inequality?
income?
()
()
.
.
.
.
.
.
.
Wave 
Gov’t should
provide
for everyone
vs. people
should
provide for
themselves ()
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Wave Wave  Wave 
Means, -to- scales
Incomes should be
made more equal vs.
larger differences
needed as an
incentive ()
.
.
.
.
.
.
.
Wave 
.
.
.
.
Wave 
Private ownership
should be
increased vs.
gov’t ownership
should be
increased ()
TABLE  Public opinion on relevant issues in countries with different welfare state regimes










Percent agree
Is society
Public
authorities unfair
to you
not doing
()
enough
for the
poor
()

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH











.











.

.








.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.


.
















Note: See Appendix for question wordings.
Sources: () ISSP, , Question , , Question , and , Question . () ISSP, , Question , , Question , and , Question . () ISSP, ,
Question , and , Question . () WVS, Waves  and , Question . () WVS, Waves  and , Question . () WVS, Waves  and , Question . () Eurobarometer, , Question . () Eurobarometer, , Question .
Overall mean
Overall standard
deviation
France
Germany
Italy
Japan
Switzerland
Mean score
Social democratic
regimes
Denmark
Finland
Netherlands
Norway
Sweden
Mean score
THREE WORLDS OF PUBLIC OPINION


INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
Liberal countries may produce atypical responses from time to time, but they generally tend to cluster together at one end of the distribution (with the possible and
also unexpected exception of Great Britain). Once again, comparison of the mean
responses for each cluster may provide the clearest indication of the overall picture.
This discussion is not meant to imply that no pattern emerges, just that the
pattern is not as clear as anticipated. In general terms, the picture painted by the
data is of a fairly obvious split between our Liberal countries, on the one hand,
and our Social Democratic and Conservative countries, on the other, but then
with no discernible gap between Social Democratic and Christian Democratic
countries. This, perhaps not coincidentally, corresponds to an intermingling of
Esping-Andersen’s own decommodification scores between the two groups (see
Table ), and may provide evidence that the real-world differences between the
two clusters are rather subtle.
T WO EXCEPTIONAL CASES —G REAT B RITAIN AND S WEDEN
Looking below the surface, there seem to be two specific reasons for this mediocre
fit between variation in opinion and policy. The first is the emergence of two highly
anomalous cases—Sweden and Great Britain (also noted briefly by Svallfors, ;
Ginsburg, , p. , perhaps most accurately refers to Britain as ‘an odd mixture
of the “socialist” and “liberal” types’). The former is, of course, considered the
preeminent example of an active and generous welfare state, successfully combating
poverty and inequality through extensive government programs, the high taxes
necessary to fund them, and an extraordinary level of social cohesion, while the latter
is generally viewed, particularly in the popular media, as a lone European bulwark
trying to pull her continental counterparts in the direction of a U.S.-style capitalism
based on individualism, competition, and minimal governmental interference.
However, survey results show these characterizations to be grossly exaggerated,
if not downright erroneous, with respect to the mass attitudes of Swedish and
British citizens, as we see by again looking at Table . Sweden, for its part, as of
the most recent polling, is the country most likely to favor inequality as an incentive
for effort (WVS, Wave , Question ), among the least likely to argue that income
differences are too large (ISSP, , Question ), among the most likely to feel
that citizens, rather than government, should provide for themselves (WVS,
Wave , Question ), and among the least likely to see competition as harmful
(WVS, Wave , Question , not included in the table). Great Britain, on the
other hand, is among the countries most likely to argue that competition is harmful
(WVS, Wave , Question ) and that government should provide a basic
income (ISSP, , Question ). The British are also among the most likely to
view success as the result of luck and connections rather than hard work (WVS,
Wave , Question , not included in the table) and see the world in zero-sum
rather than positive-sum terms (WVS, Wave , Question , not included in
THREE WORLDS OF PUBLIC OPINION

the table). In response to some survey questions, those polled in Sweden and
Great Britain respond more as one might expect given their more general
national orientations. However, it would not be an overstatement to suggest,
based on the poll results analyzed here, that the British public generally holds
attitudes which are more receptive to governmental activism and more hostile
toward the free market than those of their Swedish counterparts.
This discussion of Sweden and Great Britain needs to be viewed in light of the
issue of relative and absolute questions discussed earlier. That is, Swedish citizens
might be less likely than British citizens to support increased governmental
activity because their government is already quite active in, not to mention
successful at, managing the economy for the sake of larger social goals. However,
I maintain that the larger point about the anomalous nature of these cases
remains valid. Comparing Sweden’s responses with the cluster means, Sweden
often produces results most similar to the Liberal group, while Great Britain is
consistently far more likely than any of the other Liberal countries to display
interventionist tendencies. This raises one further question. If it is just about
relative vs. absolute questions, why does Sweden so often look different from the
other Social Democratic countries, and why does Great Britain appear even
more strikingly different from its Liberal counterparts? Norway and Denmark
both have active governments which successfully manage social problems
through market intervention, and the US and Australia both have relatively
uninvolved governments which let social outcomes be determined by market
forces, but in all of these cases, mass attitudes bear at least a general resemblance
to governmental orientation. It is only Sweden and Great Britain which consistently emerge as anomalous. Clearly the dynamic process by which public opinion
is translated into public policy in these two countries deserves further attention.
IDEOLOGICAL AND O PERATIONAL ATTITUDES
The second and more important reason for an apparently unimposing fit between
public opinion and public policy concerns the nature of the questions themselves.
As the work of Free and Cantril () suggests, and as Coughlin () indeed
finds, we might expect our preliminary finding of a merely modest fit between
variation in attitudes and variation in policies to change if we consider the distinction between ideological and operational attitudes. That is, we should find similar
(and generally high) levels of support, across our sample of countries, for specific
public spending programs, but we should find mixed levels of support for broader
ideological questions, with variation occurring in accordance with each country’s
general policy orientation, as evidenced by a much stronger relationship between
differences in mass attitudes and decommodification scores.
Examination of simple bivariate correlations allows us to reach some basic
conclusions. Tables  and  provide evidence of the operational–ideological

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
dichotomy, thereby reinforcing Coughlin’s () results and establishing the
validity of the Three Worlds paradigm in the context of cross-national public
opinion. As we can see from the first four correlations presented in Table  (the
aggregated country-level data are presented in Table ), which are based on data
from the ISSP’s Role of Government polls, the relationship between decommodification scores and four specific policy-oriented questions, which ask whether
the respondent would like their government to increase spending on health,
education, pensions, and unemployment, respectively, even if doing so might
TABLE  Bivariate correlations between Esping-Andersen’s () decommodification scores and country level responses to various policy and ideology
questions
Increase spending for . . .
health, . . .
education, . . .
pensions, . . .
unemployment, . . .
even if it requires a tax increase?
Government’s responsibility to . . .
provide health care for the sick?
provide decent living for the
elderly?
provide decent living for the
unemployed?
reduce differences in income?
Government doing enough for
people in poverty?
Income Distribution Index,
measuring optimal pay
differences between various
occupations
Success is the result of hard work,
rather than luck and connections
People in need because of laziness
or social injustice?
Is it just that money buys better
health?
Is it just that money buys a better
education?
Bivariate
correlations
n
Sources
−.
−.**
.
.




ISSP, , Q. 
ISSP, , Q. 
ISSP, , Q. 
ISSP, , Q. 
.
.**


ISSP, , Q. 
ISSP, , Q. 
.***

ISSP, , Q. 
.*

.**

ISSP , Q. ; ,
Q. ; , Q. 
WVS, Wave , Q. 
−.*

Svallfors, , p. 
−.***

.*

WVS, Waves  and ,
Q. 
WVS, Wave , Q. 
−.**

ISSP, , Q. 
−.**

ISSP, , Q. 
*** p = .; ** p = . level; * p = . level (one-tailed tests).
Note: Country-level responses to these questions are provided in Table .
Source: Decommodification scores obtained from Esping-Andersen, , p. . See Appendix for question
wordings.
THREE WORLDS OF PUBLIC OPINION

require a tax increase (ISSP, , Questions , , , and ), is unimpressive
to say the least. The correlations for three of the four propositions are statistically insignificant, and the fourth, regarding education, is highly significant, but
negative, contrary to expectations. This is a noteworthy finding because this is
the same battery of questions analyzed by Iversen and Soskice (), in which
they argue that differences in vocational training (general vs. specific) shape
policy variation among advanced capitalist democracies, through a simple rationalitybased effect on citizen preferences.
Turning to the second four correlations in Table , however, we see that more
broadly ideological questions concerning similar issue areas are asked within the
same survey, but produce a much different outcome. Specifically, respondents
were asked about the role of the government in providing for the sick, the elderly,
the unemployed, (ISSP, , Questions , , ) and in reducing inequality
(ISSP, , Question , , Question, , and , Question ) and
aggregate results for three of the four questions are significantly and positively
correlated with decommodification scores (the fourth, regarding health care, just
misses significance at the . level, seemingly due to the low French response, but
still provides a much better fit than the related item in the specifically policyoriented battery of questions).
Merely rephrasing the questions in a more broadly ideological, rather than
policy-specific, manner drastically alters the goodness of fit between variations in
mass attitudes and public policy. To be fair, the results diverge somewhat from
what we should expect based on Coughlin’s () results, which, as mentioned
before, show that specific policy questions draw universally high support, but
then broader ideological questions draw high levels of support in some countries,
and lower levels of support in others, with the result being a pattern which corresponds closely to policy variation. What we see here, particularly for the questions
on health care and pensions, is fairly high levels of support in all countries (all at
or above  percent for both questions), but the variation that exists is more
highly correlated with the decommodification index (the questions regarding
unemployment benefits and reducing inequality do more closely correspond
with Coughlin’s results). One might also argue that the caveat about higher taxes
in the first battery of questions merely introduces a distortion to the respondent
but that these are both series of questions which are inherently policy-oriented,
rather than ideological in nature. I will, of course, leave this issue to the judgment
of the readers, though I would argue that the second battery of questions (‘is it
the responsibility of the government to . . .’) is a step towards broader ideology as
compared to the first (‘would you like to see more or less . . .’), and as we will see
momentarily, the fit with policy variation gets even better as we move to
questions which are even more removed from specific policy.
The remaining correlations in Table  are between decommodification and six
other interesting and relevant questions of broad ideology, each of which also
Belgium
Liberal regimes
Australia
Canada
Ireland
New Zealand
Great Britain
United States
Mean score
Conservative
regimes
Austria
Percent agree
Percent agree
























































Health Education Pensions Jobless Provide Provide Provide Reduce
benefits
for
for
for
inequality
sick
elderly jobless
Government’s
responsibility to . . .
Increase spending
for . . . , even if it
requires a tax
increase
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.














Better
Better
health education
Percent agree
-point
scale
-point -point
scale
scale
-to-
scale
Is it just
that money
buys . . .
Need
Success
Gov’t Income
Dist. due to hard due to
doing
work vs. laziness
enough Index
luck and or social
for the
connections injustice
poor?
TABLE  Country-level responses to survey items used in correlation analysis in Table 

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH




.









.









.






.












.









.




.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.




.








.




Note: For opinions on government’s reponsibility to reduce inequality, the most recent result for each country is shown. The Income Distribution Index measures acceptance
within countries of income disparities between various occupations.
Sources: shown in Table ; see Appendix for question wordings.
Norway
Sweden
Mean score
Overall mean
Overall standard
deviation




.






.




Italy
Japan
Switzerland
Mean score
Social democratic
regimes
Denmark
Finland
Netherlands




France
Germany
THREE WORLDS OF PUBLIC OPINION


INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
displays a strong relationship. The last five of these questions, in particular, all
pertain in some way to the notion of social justice, and all are significantly correlated with variation in policy orientation. Focusing specifically on the last two
items in Table , it is particularly instructive that Great Britain and Sweden
exhibit aggregate responses more as one might expect, confirming that it is these
questions of broad ideology which really matter. These results imply that those
countries which provide more for their citizens in terms of social policy are
populated by citizens who have a more expansive conception of social justice, as
opposed to their counterparts living under less-generous governments. While
the causal relationship between preferences and policy is obviously complicated,
it seems clear that people are, at least to some extent, getting what they want
from government, in comparative context. At the aggregate level, those who
favor a more egalitarian society get a more active government which provides
more extensive services and lower levels of poverty, inequality, and other social
problems. Those who hold less egalitarian views get smaller government, lessgenerous social programs, and social outcomes left relatively unchecked. It is
thus fair to say that broad ideological attitudes do in fact vary in accordance with
the documented variation in policy orientation.
CONCLUSIONS
In sum, the preceding analysis provides us with three clear conclusions. First,
there is a sharp division in public preferences between Liberal countries, on the
one hand, and Social Democratic and Conservative countries on the other,
consistent with the decommodification numbers provided by Esping-Andersen
(), if not as clearly consistent with the underlying theory. Second, Sweden
and Britain emerge as anomalous cases, each consistently producing aggregate
responses uncharacteristic of their roles as bulwarks of the Social Democratic
and Liberal models, respectively. Finally, and most significantly, the operational–
ideological dichotomy works at the cross-national level, the result being a strong
fit between cross-national differences in broad ideology and social policy (and an
insignificant fit between policy and more specific policy preferences). This finding
necessitates a cursory investigation of the possible causal link between the ideology
and policy, and it is to this task that we now turn our attention.
CA U SA L IT Y
Working with a sample of only  countries and then having relatively spotty
data coverage leaves us in a position where the usual multivariate statistical
procedures are rendered problematic. However, having established a statistically
significant relationship between aggregate political attitudes and policy orientation, we can weigh public opinion alongside variables already known to shape

THREE WORLDS OF PUBLIC OPINION
welfare state variation. To do so, I have constructed two index variables using
factor analysis to simplify the analysis and presentation. The first combines
questions regarding the government’s responsibility in providing health care,
pensions, and unemployment assistance (ISSP, , Questions , , and ,
respectively), while the second combines questions asking whether it is just that
money buys better health and education (ISSP, , Questions , and ).
For the record, these are the two clusters of questions which will be the focus
of the larger study of which this essay provides a foundation (discussed below
in the concluding section). The factor loadings for each are available from the
author.
As we see in Table , the relationship between public opinion (measured as
the two indexed variables) and decommodification is roughly similar in strength
to the relationships between decommodification and left government (measured
as cumulative years of left government as of , based on data from Huber,
Ragin, & Stephens, ), union density (unionized workforce as a percentage of
total workforce via Huber et al., , and Visser, ), constitutional structure
(index available from Huber et al., ), and early consolidation (measured as
the average number of years that pension, sickness, and unemployment benefits
have been both binding and extensive in each country via Hicks, , p. ),
TABLE  Relative strength of public opinion, compared to variables previously
found to have causal significance, as a source of welfare state variation
Bivariate correlations with decommodification
Bivariate
correlations
n
Sources
Index of public support
for government provision
of health care, pensions,
and unemployment assistancea
Index of justice-based
opiniona
Left government
Union density
.*

ISSP, , Q. , ,
and 
−.*

.**
.*


Constitutional structure
−.**

.*
.**


ISSP, , Q. 
and 
Huber et al., 
Huber et al., ;
data from Visser,
 and OECD
Huber et al., ;
Lijphart, 
Hicks, , ch. 
Hicks, , ch. 
Early consolidationb
Neo-corporatism
** p = .; * p = . level (one-tailed tests).
a
Combined measures using factor analysis; factor loadings available from the author.
b
Dummy variable measuring whether or not program consolidation had been achieved by 
Source: Decommodification scores obtained from Esping-Andersen, , p. . See Appendix for question
wordings.

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
especially when one factors in differences in sample size, though not as highly
correlated as the relationship between decommodification and neo-corporatism
(specifically tripartite neo-corporatism from Hicks, , p. ).
Further, we can perform simple bivariate regressions, measuring the strength
of our two public opinion index variables against our other predictors of decommodification one at a time. The results, presented in the top half of Table ,
indicate that the first indexed variable performs only moderately, as it is only a
significant predictor of decommodification when paired with unionization and
TABLE  Public opinion as a predictor of decommodification weighed in
two-variable models against factors known to affect welfare state variation
OLS regression results with decommodification as dependent variable
Public opinion
index 
Left-government
.
(.)
.*
(.)
Union density
.*
(.)
.
(.)
Constitutional
structure
Early consolidation
.
(.)
.*
(.)
.
(.)
−.
(−.)
.
(.)
Constant
Adjusted R-Square
n
.***
.

.**
.

.***
.

.**
.

.***
(.)
.***
.

Public opinion
index 
Left-government
−.**
(−.)
.**
(.)
−.**
(−.)
−.**
(−.)
−.**
(−.)
−.
(−.)
Neo-corporatism
Union density
.**
(.)
Constitutional
structure
Early consolidation
−.***
(−.)
.*
(.)
Neo-corporatism
Constant
Adjusted R-Square
n
.***
.

.***
.

.***
.

.***
.

.***
(.)
.***
.

*** p = .; ** p = .; * p = . level (one-tailed tests). Standardized coefficients in parentheses.
Sources: See Table . Public opinion index  is the Index of public support for government provision of health
care, pensions, and unemployment assistance in Table , and Public opinion index  is the Index of justice-based
public opinion there.
THREE WORLDS OF PUBLIC OPINION

the dummy variable for early consolidation. However, a look at the standardized
coefficients, combined with a reminder about the tiny sample sizes, indicates that
this variable performs well enough to at least be included in the discussion about
determinants of welfare state variation.
The case is much stronger for the second indexed variable, which is a significant
predictor of decommodfication scores at the . level in all cases except when
paired with neo-corporatism (which is statistically, and perhaps theoretically, a
virtual proxy for decommodification). Further, a relatively high portion of the
variance in decommodification scores is achieved through most of these simple
bivariate models. The fact that the second indexed variable performs better than
the first is consistent with the previously discussed notion that it is broader ideology
which is truly important in helping to shape policy variation among advanced
capitalist democracies.
It is important not to overstate any causal claims given the temporal relationship between the public opinion and decommodification data. However, given
the strength of the statistical relationship between public opinion and social policy and the strong causal link between public preferences and policy outcomes in
democratic theory, it is clear that mass political preferences warrant increased
attention as a significant cause of welfare state development. It certainly seems to be
the case that, at least to some extent, citizens are getting what they want from
their own governments, relative to each other, in terms of social policy, which
implies that a thorough understanding of the variation among advanced capitalist
democracies needs to incorporate the role of political ideology.
C O N C L US I O N
Based on the above evidence, four clear findings emerge in response to the hypotheses derived earlier in this study. First, there is a significant gap in mass attitudes
between Liberal countries, on the one hand, and Social Democratic and Conservative societies on the other. Second, we find a relatively weak relationship between
policy-related public opinion and the differences in social policy among countries,
but we find a much stronger fit between broad ideology and policy variation, consistent with the distinction between ideological and operational attitudes offered by
Free and Cantril (). Third, and in addition to our hypothesis-driven results, we
unexpectedly find that Sweden and Great Britain emerge as clearly anomalous cases,
and therefore complicate our analysis, in that they each seem to display the mass attitudes one would expect of the other—social democratic Sweden often comes across
as relatively laissez-faire, even conservative, in terms of its aggregate poll responses,
while liberal Great Britain often emerges as rather egalitarian and pro-government.
Finally, and perhaps most importantly, crude correlation and regression analyses
indicate that public opinion, particularly of the broader ideological variety, deserves
consideration as an important determinant of welfare state variation.

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
As for future research, three possibilities come to mind. First, having established a relationship between public opinion and public policy within this data, it
is necessary to obtain a fuller understanding of the forces which shape public
opinion and the cross-national differences in it which we have observed here.
Specifically, we need to develop a more accurate model of the social forces which
shape attitudes at the individual level. Much recent work has focused on a sort of
basic economistic approach, arguing that political preferences are formed by a simple
cost–benefit analysis, whereby policies which financially benefit a particular
individual are supported, while those that do not are opposed. This rationalitybased perspective seems flawed, however, in that it is both anti-historical and
anti-sociological and therefore wrongly assumes that all similarly situated individuals will reach the same political decisions, regardless of social context, and
particularly with regard to cross-national variation in that social context. Within
the confines of this research project, this is the line of inquiry I will be following.
Second, the emergence of Britain and Sweden as anomalous cases raises interesting
questions. Specifically, if Britain has such egalitarian and pro-government views,
while Sweden exhibits such conservative and laissez-faire attitudes, how have they
evolved into exemplars of the Liberal and Social Democratic models, as has been so
well documented? There seems to be a mismatch between public opinion and public
policy in these two countries which cries out for further examination.
The third suggestion is simply a continuation of the current line of inquiry as
more data become available. Just as Coughlin () provided a first step, Svallfors
(, forthcoming) provides a second, and this project hopes to provide a third,
many more steps need to be taken before we can feel comfortable that we have a
sound understanding of the comparative nature of public support for governmental
spending programs. Given the important theoretical link between public opinion
and public policy, it is crucial that we definitively ascertain the degree to which
citizen preferences are reflected in the work of our elected representatives.
A P P E N D I X : S U R V E Y Q UE ST I O N S
ISSP, —SOCIAL INEQUALITY III
Question : Differences in income in this country are too large—percentage answering
‘strongly agree’ or ‘agree’ (most recent result for each country).
Question : It should be or should not be the responsibility of the government to reduce
the differences in income between people with high incomes and those with low
incomes—percentage answering ‘strongly agree’ or ‘agree’ (most recent result for each
country).
Question : Is it just or unjust—right or wrong—that people with higher incomes can
buy better health care than people with lower incomes?—Percentage responding ‘very
just, definitely right’ or ‘somewhat just, right’.
THREE WORLDS OF PUBLIC OPINION

Question : Is it just or unjust—right or wrong—that people with higher incomes can
buy better education for their children than people with lower incomes?—Percentage
responding ‘very just, definitely right’ or ‘somewhat just, right’.
ISSP, —R OLE OF GOVERNMENT III
Question : Please show whether you would like to see more or less government spending for health. Remember that if you say ‘much more’, it might require a tax increase to
pay for it—percentage answering ‘spend much more’ or ‘spend more’.
Question : Please show whether you would like to see more or less government spending for education. Remember that if you say ‘much more’, it might require a tax increase
to pay for it—percentage answering ‘spend much more’ or ‘spend more’.
Question : Please show whether you would like to see more or less government
spending for old age pensions. Remember that if you say ‘much more’, it might
require a tax increase to pay for it—percentage answering ‘spend much more’ or
‘spend more’.
Question : Please show whether you would like to see more or less government
spending for unemployment benefits. Remember that if you say ‘much more’, it might
require a tax increase to pay for it—percentage answering ‘spend much more’ or
‘spend more’.
Question : It should be or should not be the responsibility of government to provide
health care for the sick—percentage answering ‘definitely should be’ or ‘probably
should be’.
Question : It should be or should not be the responsibility of government to provide a
decent living for the elderly—percentage answering ‘definitely should be’ or ‘probably
should be’.
Question : It should be or should not be the responsibility of government to provide a
decent living for the unemployed—percentage answering ‘definitely should be’ or ‘probably should be’.
ISSP, —S OCIAL I NEQUALITY II
Question : Differences in income in this country are too large—percentage answering
‘strongly agree’ or ‘agree’ (most recent result for each country).
Question : It is the responsibility of the government to reduce the differences in income
between people with high incomes and those with low incomes—percentage answering
‘strongly agree’ or ‘agree’ (most recent result for each country).
Question : The government should provide everyone with a guaranteed basic income—
percentage answering ‘strongly agree’ or ‘agree’ (most recent result for each country, not
asked in  version).

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
ISSP, —S OCIAL INEQUALITY I
Question : Differences in income in this country are too large—percentage answering
‘strongly agree’ or ‘agree’ (most recent result for each country).
Question : It is the responsibility of the government to reduce the differences in income
between people with high incomes and those with low incomes—percentage answering
‘strongly agree’ or ‘agree’ (most recent result for each country).
Question : The government should provide everyone with a guaranteed basic income—
percentage answering ‘strongly agree’ or ‘agree’ (most recent result for each country, not
asked in  version).
WORLD VALUES SURVEY, WAVES  AND :
Question : A one-to-ten scale, where low numbers indicate that ‘incomes should be
made more equal’ and high numbers indicate that ‘we need larger income differences as
incentives for individual effort’—mean response.
Question : A one-to-ten scale, where low numbers indicate that ‘private ownership of
business and industry should be increased’ and high numbers indicate that ‘government
ownership of business and industry should be increased’—mean response.
Question : A one-to-ten scale where low numbers indicate that ‘the government
should take more responsibility to ensure that everyone is provided for’ and high numbers indicate that ‘people should take more responsibility to provide for themselves’—
mean response.
Question : A one-to-ten scale where low numbers indicate that ‘in the long run, hard
work usually brings a better life’ and high numbers indicate that ‘hard work does not generally bring success—it’s more a matter of luck and connections’—mean response.
Question : Why, in your opinion, are there people in this country who live in need?
Here are two opinions. Which comes closest to your view? () ‘they are poor because of
laziness and lack of will power’, or () ‘they are poor because society treats them
unfairly’—mean response.
Question : Do you think that what the government is doing for people in poverty in
this country is about the right amount, too, much, or too little? () ‘too much’, () ‘about
the right amount’, or () ‘too little’—mean response
EUROBAROMETER .—POVERTY AND S OCIAL E XCLUSION
Question : Do you think the public authorities do too much for people who are poor or
excluded, do all that they should, or do not enough?—Percentage answering ‘not
enough’.
Question : Taking everything into account, do you yourself have the feeling that
society is unfair to you?—Percentage responding ‘yes’.
THREE WORLDS OF PUBLIC OPINION

REFERENCES
Burstein, P. (). Bringing the public back in. Social Forces, , –.
Castles, F. G., & Mitchell, D. (). Worlds of welfare and families of nations. In
F. G. Castles (Ed.), Families of nations. Dartmouth: Aldershot.
Coughlin, R. M. (). Ideology, public opinion, and welfare policy. Berkeley: Institute of
International Studies.
Crouch, C. (). Industrial relations and European state traditions. Oxford: Clarendon
Press.
Edwards, E. (, April). Revisiting Esping-Andersen’s ‘Three Worlds’ using cluster analysis.
Paper presented at the Annual Meeting of the Midwest Political Science Association,
Chicago, IL.
Esping-Andersen, G. (). The three worlds of welfare capitalism. Princeton: Princeton
University Press.
Eurobarometer. (). Poverty and social exclusion (Version .) [Data file]. Compiled
at the Central Archive of the University of Cologne.
Fraser Institute. (). The economic freedom of the world:  annual report. Vancouver:
Fraser Institute.
Free, L. A., & Cantril, H. (). The political beliefs of Americans. New York: Simon &
Schuster.
Ginsburg, N. (). Divisions of welfare. London: Sage.
Goodin, R. E., Headey, B., Muffels, R., & Dirven, H.-J. (). The real worlds of welfare
capitalism. Cambridge: Cambridge University Press.
Gornick, J. C., Meyers, M. K., & Ross, K. E. (). Supporting the employment of
mothers: Policy variation across fourteen welfare states. Journal of European Social
Policy, (), –.
Hicks, A. (). Social democracy and welfare capitalism. Ithaca: Cornell University
Press.
Hicks, A., & Kenworthy, L. (). Varieties of welfare capitalism. Socio-Economic
Review, (), –.
Hicks, A., & Misra, J. (). Political resources and the growth of welfare in affluent
capitalist democracies, –. American Journal of Sociology, , –.
Huber, E., Ragin, C., & Stephens, J. D. (). Social democracy, christian democracy,
constitutional structure, and the welfare state. American Journal of Sociology, , –.
Huber, E., Ragin, C., & Stephens, J. D. (). Comparative welfare states data set.
Northwestern University and the University of North Carolina-Chapel Hill.
International Social Survey Program. (). Role of government I [Data file]. Compiled
at the Central Archive of the University of Cologne.
International Social Survey Program. (). Social inequality I [Data file]. Compiled at
the Central Archive of the University of Cologne.
International Social Survey Program. (). Role of government II [Data file]. Compiled
at the Central Archive of the University of Cologne.
International Social Survey Program. (). Social inequality II [Data file]. Compiled at
the Central Archive of the University of Cologne.
International Social Survey Program. (). Role of government III [Data file]. Compiled
at the Central Archive of the University of Cologne.

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH
International Social Survey Program. (). Social inequality III [Data file]. Compiled
at the Central Archive of the University of Cologne.
Iversen, T., & Soskice, D. (). An asset theory of social policy preferences. American
Political Science Review, , –.
Katzenstein, P. J. (). Small states in world markets. Ithaca: Cornell University Press.
Korpi, W. (). The democratic class struggle. London: Routledge and Kegan Paul.
Lijphart, A. (). Democracies: Patterns of majoritarian and consensus government in
twenty-one countries. New Haven: Yale University Press.
McClosky, H., & Zaller, J. (). The American ethos. Cambridge: Harvard University
Press.
Mehrtens, F. J., III. (). Shared values? Public opinion and social policy among advanced
capitalist democracies. Paper presented at the Southern Political Science Association
annual conference, January, New Orleans.
Mehrtens, F. J., III. (forthcoming). Laissez-faire vs. regulated capitalism: A cross-national
analysis of social costs. Sociological Spectrum.
Mish, F. C. (Ed.). (). Webster’s ninth new collegiate dictionary. Springfield, MA:
Merriam-Webster.
Moene, K. O., & Wallerstein, M. (). Inequality, social insurance, and redistribution.
American Political Science Review, , –.
Page, B., & Shapiro, R. Y. (). Effects of public opinion on policy. American Political
Science Review, , –.
Polanyi, K. (). The great transformation. Boston: Beacon Press.
Przeworski, A. (). Capitalism and social democracy. Cambridge: Cambridge University
Press.
Sainsbury, D. (Ed.) (). Gendering welfare states. London: Sage Publications.
Schmitter, P. C. (). Interest intermediation and regime governability in contemporary
Western Europe and North America. In S. Berger (Ed.), Organizing interests in Western
Europe. Cambridge: Cambridge University Press.
Stephens, J. D. (). The transition from capitalism to socialism. London: Macmillan.
Stimson, J. A., MacKuen, M. B., & Erikson, R. S. (). Dynamic representation.
American Political Science Review, , –.
Svallfors, S. (). Worlds of welfare and attitudes to redistribution. European Sociological Review, , –.
Svallfors, S. (forthcoming). Welfare regimes and welfare opinions: A comparison of
eight Western countries. In J. Vogel (Ed.), The European welfare mix: Institutional
configuration and distributional outcomes. Amsterdam: Kluwer Press.
Visser, J. (). Unionization trends: The OECD countries union membership file. Paper
prepared at University of Amsterdam, Centre for Research of European Societies and
Labor Relations (CESAR).
Wlezien, C. (). The public as thermostat: Dynamics of preferences for spending.
American Journal of Political Science, , –
Wlezien, C. (). Dynamics of representation: The case of US spending on defense.
British Journal of Political Science, , –.
World Values Survey. (). World Values Survey, –; –; and
– [Data file]. Compiled by the Institute for Social Research, Ann Arbor, and
THREE WORLDS OF PUBLIC OPINION

distributed by the Inter-University Consortium for Political and Social Research
(Study #).
Zaller, J. R. (). The nature and origins of mass opinion. Cambridge: Cambridge
University Press.
B IO GR A PH I CA L NO T E
F. John Mehrtens III is a Ph.D. candidate in the Department of Political Science at the
University of North Carolina–Chapel Hill, where he is completing his dissertation under
the direction of John Stephens. He is also currently an instructor in the Department of
Political Science at Miami University (Oxford, Ohio) and holds a Master’s degree in
Political Science from the University of Nebraska–Lincoln.
Address correspondence to F. John Mehrtens III,  Crystal Drive, Fairfield, OH
, USA, E-mail [email protected]