Content uploaded by Christian Welzel
Author content
All content in this area was uploaded by Christian Welzel on Oct 21, 2014
Content may be subject to copyright.
Political Culture and Democracy:
Analyzing Cross-Level Linkages
Ronald Inglehart and Christian Welzel
Abstract
Do individual-level attitudes play a significant role in sustaining democratic
institutions at the societal level? In a recent contribution to this journal, Seligson
claims that the strong aggregate-level correlations that Inglehart has found between
political culture and stable democracy are “spurious” because Seligson does not find
individual-level correlations between Inglehart’s indicators of political culture, and
overt support for democracy. Seligson’s analysis exemplifies precisely the sort of
cross-level fallacy that he attributes to Inglehart: he equates individual-level support
for democracy, with the presence of democratic institutions. Surprising as it may
seem, however, individual-level lip service to democracy is only weakly linked with
societal-level democracy.
At this point in history, democracy has a positive image
almost everywhere, but these favorable opinions are often superficial, and unless they
are accompanied by deeper-rooted orientations of tolerance, trust, and a participatory
outlook, the chances are poor that effective democracy will be present at the societal
level. On the other hand, as we demonstrate, the linkage between these deeper-rooted
orientations and effective democracy, is remarkably strong: a political culture of
tolerance, trust and participatory orientations seems to essential to effective
democracy.
This article is forthcoming in Comparative Politics.
1
NOTE ON CONTRIBUTORS
Ronald Inglehart is a Professor of Political Science and Program Director at the
Institute of Social Research, University of Michigan. Since 1988 he has directed the
World Values Surveys. Author of more than 170 publications, his recent books
include Modernization and Postmodernization: Cultural, Economic and Political
Change in 43 Societies (Princeton University Press, 1997); and (with Pippa Norris)
Gender Equality and Cultural Change (New York and Cambridge: Cambridge
University Press, 2003). Recent articles include (with Wayne Baker) “Modernization,
Cultural Change and the Persistence of Traditional Values.” American Sociological
Review. (February, 2000):19-51; “How Solid is Mass Support for Democracy—And
How Do We Measure It ?” PS: Political Science and Politics, January, 2003.
Christian Welzel is Associate Professor of Political Science at the International
University Bremen (IUB). He is a member of the Steering Committee of the World
Values Surveys Group. Author of more than 40 publications, his recent book is
Fluchtpunkt Humanentwicklung: Die Grundlagen der Demokratie und die Ursachen
ihrer Ausbreitung [Focus Human Development: The Foundations of Democracy and
the Causes of Its Expansion] (Opladen: Westdeutscher Verlag, 2002). His recent
article is (with Ronald Inglehart and Hans-Dieter Klingemann) “The Theory of
Human Development: A Cross-Cultural Analysis.” European Journal of Political
Research 42 (2), forthcoming.
2
Political Culture and Democracy:
Analyzing Cross-Level Linkages
Introduction
This article responds to Mitchell Seligson’s contribution in a recent issue of
this journal.
1
Seligson raises a classical and still controversial issue in comparative
politics: what role does political culture play in sustaining stable democratic
institutions? He examines this question in light of one of the central methodological
problems in cross-national research: the linkage between individual level and
aggregate level relationships.
Seligson starts by claiming that cross-national correlations that do not also
appear at the individual level within each nation are “spurious,” citing a passage to
this effect by Przeworski and Teune.
2
Although this claim has been widely accepted,
it is clearly groundless, as this article will demonstrate. Basing his argument on this
assertion, Seligson attempts to invalidate Inglehart’s findings that there are strong
aggregate-level correlations between political culture and stable democracy. Seligson
claims that the aggregate level findings are spurious because he does not find
individual level correlations between these political culture indicators and support for
democracy.
This article will demonstrate that—ironically enough-- Seligson’s conclusions
exemplify precisely the sort of cross-level fallacy that Robinson warned against.
3
The
central point of the Ecological Fallacy thesis is that strong aggregate-level
relationships are not necessarily reproduced at the individual level. When Robinson
was writing, districts with large percentages of African Americans (then located
mainly in the South) generally elected segregationist candidates; but, as Robinson
3
demonstrated, this relationship was not reproduced at the individual level—Blacks did
not vote for segregationist candidates. This did not mean that the aggregate level
relationship was somehow “spurious;” no one questions the fact that, districts with
large numbers of African Americans really did elect the worst sort of segregationists,
in a pattern of repression that endured for decades.
Seligson turns the argument the wrong way around, claiming that an
aggregate-level finding must be reproduced at the individual level – and if it isn’t, it is
somehow “spurious.” This claim is groundless, as Robinson demonstrated more than
50 years ago, and as more recent evidence presented in this article will confirm.
Moving farther in misinterpreting the Ecological Fallacy, Seligson equates
individual-level support for democracy, with the presence of democratic institutions.
Superficially, this seems plausible. But in fact, at this point in history, individual-
level lip service to democracy is only weakly linked with societal-level democracy.
Since the collapse of communism, democracy has attained a positive image in
virtually every country in the world. But these favorable opinions are often
superficial, and unless they are accompanied by deeper-rooted orientations of
tolerance, trust, and a participatory outlook, the chances are poor that effective
democracy will be present at the societal level. In striking contrast to Seligson’s
unproven cross-level assumption, mere lip service to democracy is not necessarily
linked with actual democracy at the societal level, as we will demonstrate: at this
point in history, it is almost as strong in authoritarian societies or unstable
democracies, as in stable democracies.
On the other hand, as this article will demonstrate, the linkage between a
deeper-rooted syndrome of “Self-expression” orientations, and effective democracy,
is remarkably strong. A controversial body of literature that goes back to Lipset
4
and
4
Almond and Verba
5
is basically correct: a specific type of political culture seems to
be an essential precondition to effective democracy.
1. Misconceptions of the Ecological Fallacy
Seligson’s contribution is based on a misconception of the problem of cross-
level inferences. This is its crucial flaw. It also involves a minor problem that we
will briefly discuss before addressing the major one. First, Seligson equates
aggregating individual level responses with the individualistic fallacy, as if
aggregating such responses were inherently wrong. Actually, aggregating individual
level attitudes to the nation level is a perfectly legitimate procedure, and is essential to
any attempt to depict the features of national mass cultures. The individualistic fallacy
consists in making the incorrect assumption that an individual-level relationship also
has similar strength and direction at the aggregate level.
Seligson’s crucial misconception is that cross-national correlations are
“spurious” if they are not also present at the individual level within each nation.
Although this claim is often accepted, it is simply not true—and, in fact, it represents
precisely the type of fallacious cross-level inference that was exposed by the
Ecological Fallacy thesis. Deciding whether a relationship is genuine or spurious, on
the basis of whether this relationship occurs at another level of analysis, is exactly
what Robinson warned us not to do: it is an unwarranted cross-level inference.
Whether or not a relationship is spurious, can only be determined by evidence at the
same level of analysis. Thus, in the classic case, the question of whether or not
individual African-Americans were voting for segregationist candidates could only be
decided by individual-level evidence—not by state-level correlations. The
methodological axiom on which Seligson bases his analysis is a clear
5
misinterpretation of the level of analysis problem. To demonstrate this point, let’s
consider some examples.
In Robinson’s case, the fact that electoral units with high percentages of
African-Americans tended to elect segregationist representatives did not mean that
African-Americans were segregationists--the opposite was true. Conversely, the fact
that African-Americans were not segregationist did not mean that the district-level
linkage between racial composition and segregationist policies was “spurious.” The
correlation between race and electoral behavior reversed its sign when one moved
from the individual level to the aggregate level, and the findings at both levels of
analysis were genuine and important.
Similarly, in contemporary France the vote for the xenophobic National Front
tends to be highest in districts with high percentages of Islamic immigrants. This does
not mean that the immigrants are supporting the National Front. They are not. And
conversely, the fact that the immigrants are not voting for the FN does not mean that
the linkage between ethnicity and politics is “spurious:” the relatively high percentage
of immigrants has a major impact on the vote for the National Front, even though the
correlation between vote and immigrant status reverses its polarity from one level of
analysis to another.
Likewise, the fact that jobless Germans in the early 1930s did not show a
stronger tendency to vote for the Nazis than those Germans who still were employed
does not mean that there was no causal linkage between unemployment and the Nazi-
vote share. Sharply rising unemployment rates created a climate of anxiety that
affected all social groups, whether employed or not, increasing their readiness to vote
for the Nazis. Thus, the rise in unemployment levels from the late 1920s to the early
1930 was followed by a strong increase in the Nazi vote. The fact that the
6
unemployed were as likely to vote communist as Nazi, at the individual level, does
not mean that unemployment was unimportant.
As these examples demonstrate, it is perfectly possible—and is frequently the
case-- that an aggregate level linkage is not reflected at the individual level. This does
not mean that this linkage is somehow unreal or “spurious.” Quite the contrary,
aggregate-level linkages often have more impact on societal phenomena than those
found at the individual level. Assuming that rising unemployment has no impact on
support for extremist parties because there is no linkage between unemployment and
extremism at the individual level would be committing the “individualistic fallacy.”
6
Seligson, nevertheless, claims that the linkage that Inglehart found between
interpersonal trust and democratic institutions at the aggregate level is “spurious”
because he finds no linkage between trust and support for democracy at the individual
level. This conclusion is a classic case of the individualistic fallacy.
2. Outdated Measures of Political Culture and Democratic Institutions
Seligson’s article examines the individual-level correlations among a set of
indicators that Inglehart used in analysis of the 1981 World Values Surveys. Anyone
reading this article would probably assume that it also refers to Inglehart’s recent
work. Actually, Inglehart’s analysis of the 1990-91 surveys and his subsequent work
moves beyond the indicators tested in Seligson’s article (life satisfaction and
interpersonal trust), incorporating them into a broader set of indicators of political
culture. This fact is of relatively minor importance. The critique in this article applies
equally to Inglehart’s original findings or to the more recent work: societies with
relatively high levels of interpersonal trust and life satisfaction are significantly more
7
likely to have democratic institutions than societies with lower levels, and this linkage
is by no means “spurious.”
Nevertheless, to bring the argument up to date, it is worth noting that
Inglehart
7
, Inglehart and Baker
8
, as well as Welzel
9
and Welzel, Inglehart and
Klingemann
10
have identified a broader syndrome of “self-expression values” that
includes not only interpersonal trust and life satisfaction but also several other
attitudes that seem to play even more important roles in promoting democracy. As we
will demonstrate, the respective publics’ locations on this self-expression values
dimension, together with economic indicators, explain roughly 80 percent of the
variance in democratic institutions. The dependent variable in this analysis is an
indicator of democratic institutions that we refer to as “effective democracy.” Let us
briefly describe these measures before analyzing their linkages.
3. Improved Measures of Political Culture and Democratic Institutions
S
ELF-EXPRESSION VALUES: Self-expression values are a syndrome of mass
attitudes that tap a common underlying dimension, reflecting emphasis on freedom,
tolerance of diversity, and participation. This is true at both the individual level and
the aggregate level, as the factor loadings in Table 1 indicate. Self-expression values
are present in a political culture in so far as the public emphasizes “liberty and
participation,” “public self-expression,” “tolerance of diversity,” “interpersonal trust”
and “life satisfaction” (see the footnotes to Table 1 for the construction of these
variables). All these attitudes tap a common underlying dimension, showing positive
loadings on a self-expression values factor. This pattern applies at three different
levels of analysis: the individual level within nations, the pooled cross-national
individual level data, and the aggregate national level. The strength of the factor
8
loadings rises systematically as we move from the individual level within nations to
the aggregate cross-national level.
[Table 1: about here]
The fact that self-expression values are more strongly structured at the
aggregate level than at the individual level, reflects a well-known phenomenon:
individual level survey data are affected by random measurement error that is
cancelled out through aggregation. As Blalock
11
observed some time ago, the
variation in individual-level attitudes consists of a systematic component and a
random component. Consequently, the correlation between two different attitudes
consists of a systematic term and a random term, in which the random term
diminishes the correlation—what Blalock called the “attenuation effect.” This
attenuation effect is relatively large at the individual level because, as Converse
12
first
observed, significant numbers of survey respondents give random answers, producing
a substantial amount of measurement error. In so far as the responses are random, this
weakens the correlations between them, making individual level correlations
relatively weak.
13
But when attitudes are averaged across nations, the random variations offset
each other: random negative and positive deviations from the national mean tend to
cancel each other out.
14
Following the law of large numbers, this “reduction of error”
becomes more pronounced as the number of individuals being aggregated rises.
Consequently, the random term becomes smaller, and the systematic correlation
larger, at higher levels of aggregation. Consequently, aggregation to the nation-level
does not produce “spurious” correlations. Quite the contrary, aggregation often
reveals systematic correlations that may be hidden by measurement error at the
individual level within nations. Hence, the syndrome of self-expression values is
9
much more pronounced at the aggregate national level than at the individual level
within nations (compare columns 1 and 3 in Table 1). As Erikson, MacKuen and
Stimson
15
argue, analysis at the aggregate often provides a more accurate view of the
underlying relationships than is available from individual-level analysis.
The strength of the correlations at the pooled individual level, falls between
these two extremes (compare column 2 with columns 1 and 3 in Table 1). At the
pooled individual level, the variation in an attitude is composed of the deviations from
the mean within nations, which may be random to a relatively large degree, and the
deviations from the mean between nations, which are largely systematic. Thus, at the
pooled individual level there is more systematic variation than at the individual level
within nations; and in turn, the pooled individual level entails more random variation
than the aggregate level. This is why the factor loadings of the self-expression values
syndrome increase from the individual level within nations to the pooled individual
level to the aggregate level.
In short, relatively weak correlations at the individual level do not indicate that
relatively strong aggregate level correlations are somehow false or “spurious.” Quite
the contrary, aggregate correlations may reveal linkages that are obscured by random
measurement errors at the individual level. Moreover, the aggregate level is precisely
the level at which democracy exists: democracy is an attribute of nations, not of
individuals. Hence, if we are interested in the impact of mass attitudes on democracy,
it is a society’s mass tendency in these attitudes that matters and not the individual-
level attitudinal structure, as Seligson assumes.
E
FFECTIVE DEMOCRACY: Since democratic institutions will be our dependent
variable, it is important to measure them with reliable indicators. In particular, it is
10
crucial to differentiate between mere formal democracy or “electoral democracy,” and
effective democracy.
Democracy is central to people’s lives because it establishes civil and political
rights that enable them to make free choices. Providing legal guarantees of these
rights creates formal democracy--which is a necessary component of democracy. But
formal rights alone are not sufficient: formal rights are effective only in so far as
elites respect these rights in their actual behavior. Law-abiding elite behavior, or “elite
integrity,” is an expression of the rule of law that, as Rose
16
and others have pointed
out, distinguishes effective democracy from formal democracy. Hence, our measure
of effective democracy combines formal democracy (i.e., freedom rights) and elite
integrity. We weight the scope of freedom rights by the extent to which elite integrity
is present, in order to measure effective democracy.
17
We measure freedom rights using the combined Freedom House scores for
civil and political rights.
18
The scores from Freedom House range from 1 to 7 on each
of the two scales, with 1 indicating the highest and 7 the lowest level of freedom (i.e.,
civil rights and political rights).
19
We reversed this scale so that higher figures
indicate a broader scope of freedom rights. The scores from Freedom House are
expert ratings of the extent to which certain individual rights are guaranteed.
20
We use
the most recent Freedom House scores from 1999-2000 in order to ensure that our
measure of freedom rights is subsequent in time to the political culture indicators that
we use as predictors of democracy.
The Freedom House scores are imperfect measures of freedom rights. They do
not take into account the extent to which given rights are respected in actual elite
behavior. To overcome this problem, we use the corruption perception indices
developed by Transparency International.
21
These scores are also expert ratings; they
11
judge how corrupt the political, bureaucratic and economic office holders of a country
are. One indication of the validity of these estimates is that they strongly correlate
with aggregate measures of the citizens’ perception of elite corruption in
representative surveys.
22
The Transparency scores range from 1 to 100, with 100 indicating the greatest
amount of corruption. Reversing these scores, one obtains a measure of law-abiding
elite behavior or elite integrity.
23
We operationalize effective democracy through
weighting freedom rights by elite integrity. Since elite integrity shall operate as a
weighting factor and not as a compensating factor, we standardize it to 1.0 as its
maximum, obtaining fractions from 0 to 1. Hence, to obtain effective democracy, we
multiply freedom rights (standardized to a maximum of 100) by fractions from 0 to 1
for elite integrity. This produces an index of effective democracy that has 100 as its
maximum. Since we use the most recent Transparency scores from 1999-2000, we
obtain a measure of effective democracy in 1999-2000:
Effective Democracy = Freedom Rights * Elite Integrity
(percentages) (fractions of 1.0)
Note that even if a country comes close to a maximum elite integrity of 1.0
(i.e., almost no elite corruption), the weighting procedure would not compensate for a
low level of freedom rights: When a regime reaches only five per cent of the possible
maximum in the freedom rights measure, a maximum elite integrity of 1.0 cannot do
more than reproducing these five per cent.
24
On the other hand, a freedom rights level
close to the maximum of 100 per cent can be severely devalued, if elite integrity is so
low that it reaches only a small fraction of 1.0. Hence, given freedom rights levels are
12
devalued to the degree that elite integrity is absent—reflecting that given
constitutional guarantees are made ineffective in proportion to elite corruption. High
levels of elite integrity cannot produce effective democracy, in absence of freedom
rights. High levels of freedom rights, on the other hand, produce formal democracy,
but formal democracy is only effective to the degree that elites base their activities on
rights instead of bribes.
As Figure 1 illustrates, it is much more difficult for nations to obtain high
scores on effective democracy than on freedom rights. Freedom rights translate into
effective democracy in a curvilinear way: a relatively large variation in the lower four
fifths of the freedom rights scale translates into a relatively small variation in effective
democracy; while a small variation in the top fifth of the freedom rights scale
translates into large variation in effective democracy. This reflects the fact that
freedom rights are a necessary condition to create effective democracy: only nations
scoring high in freedom rights can attain high scores on effective democracy. But
freedom rights are not a sufficient condition for effective democracy: not all nations
scoring high in freedom rights also score high in effective democracy. Whether or not
we include elite integrity in our operationalization of democracy makes a crucial
difference: including it clearly provides a more realistic measure of democracy.
25
[Figure 1: about here]
We now have comprehensive and meaningful measures, self-expression
values and effective democracy, allowing us to examine the linkage between political
culture and democratic institutions on a valid basis.
4. The Linkage between Political Culture and Democratic Institutions
13
Inglehart and his collaborators’ analyses of the relationship between political
culture and democracy do not imply that the linkage between effective democracy and
self-expression values is present because individuals who emphasize self-expression
necessarily endorse democracy more than individuals with little emphasis on self-
expression. This is the type of cross-level fallacy to which Seligson inadvertently falls
victim, arguing that the societal-level correlation between democracy and political
culture is spurious unless it is reflected in individual-level correlations between these
political culture indicators and support for democracy. Seligson assumes that
individual-level lip service to democracy can be equated with the emergence and
survival of democratic institutions at the societal level—precisely the sort of cross-
level inference that Robinson warned against. The assumption that individual-level
endorsement of democracy can be equated with societal-level democracy, is
fallacious. At this point in history, overt support for democracy has become
extremely widespread, and the citizens of Albania or Azerbaijan are as likely to
express a favorable opinion of democracy as are the citizens of Sweden or
Switzerland. But these favorable opinions are often superficial, and unless they are
accompanied by deeper-rooted orientations of tolerance, trust, and a participatory
outlook, the chances are poor that effective democracy will be present at the societal
level.
These deeper-rooted orientations, such as those tapped by self-expression
values, have their impact at the societal level in promoting effective democracy. In
order to demonstrate a linkage between political culture and democratic institutions,
individual-level attitudes must be aggregated to the nation level, since democracy is
an attribute of nations, not of individuals. Thus, one can only test the hypothesis that a
given political culture is conducive to democratic institutions, at the societal level--
14
which is the level at which Inglehart and his collaborators have investigated the
relationship: no cross-level assumption is involved. The ecological fallacy (and the
individualistic fallacy) are based on unwarranted assumptions that a phenomenon that
exists at one level, also exists at another level. Inglehart and his collaborators have
made no such assumption. Democracy is a societal-level variable, not an attribute of
individuals; consequently, the hypothesis that self-expression values are conducive to
democracy must be tested at the societal level.
[Figure 2: about here]
The aggregate level linkage between political culture and democratic
institutions is remarkably strong, as Figure 2 demonstrates. A society’s prevailing
attitudes on the self-expression values dimension in about 1990 (see Appendix),
explain fully 75 per cent of the cross-national variation in effective democracy in
1999-2000.
26
This effect does not simply reflect other influences, such as economic
development. The effect of self-expression values remains robust when one controls
for economic development, experience with democracy and even support for
democracy, as the regression analyses in Table 2 shows.
[Table 2: about here]
Comparing model 1 with model 5, economic development
27
adds about 6 per
cent to the effect of self-expression values on effective democracy. Economic
development also captures part of the impact of self-expression values, diminishing
their effect from beta=.86 in model 1 to beta=.51 in model 5. Considered conversely,
however, the inclusion of self-expression values diminishes the effect of economic
development from beta=.84 in model 2 to beta=.43 in model 5, adding 10 per cent of
explained variance to what economic development alone explains. This indicates that,
although self-expression values and economic development are strongly correlated
15
with each other, they are not completely exchangeable, since both add a significant
amount of explained variance to the effect of the other.
By contrast, the length of time a society has experienced under democratic
institutions
28
adds very little to the effect of self-expression values on effective
democracy (2 per cent to be precise, see models 1 and 6). Moreover, taking into
account a society’s experience with democracy only slightly diminishes the effect of
self-expression values on effective democracy (the beta-coefficient shrinks from .86
in model 1 to .73 in model 6). Conversely, however, experience with democracy’s
impact on effective democracy shrinks from beta=.75 in model 3 to beta=.18 in model
6, controlling for self-expression values. This implies that self-expression values do
not result from the presence of pre-existing democratic institutions. If this were the
case, the length of the society’s experience with democracy would capture significant
parts of the effect of self-expression values—but it does not.
In sharp contradiction to Seligson’s unproven cross-level inferences, is the
finding that overt support for democracy
29
adds nothing to the effect of self-
expression values on effective democracy (compare explained variances in models 1
and 7). Accordingly, support for democracy only captures a negligible part of the
effect of self-expression values on effective democracy (beta shrinks from .86 in
model 1 to .83 in model 7). But conversely the effect of support for democracy on
effective democracy literally vanishes (shrinking from beta=.60 in model 4 to an
insignificant beta=.07 in model 7)--once we control for self-expression values. It
may seem surprising that overt support for democracy has so little impact on the
presence of effective democracy at the institutional level, but it is important to bear in
mind that, since the collapse of communism, lip service to democracy has become
almost universal, being given favorable ratings by over 90 percent of the publics of
16
most countries. It does not tap the qualities of tolerance, self-expression, trust, well
being and participatory orientations that are crucial to the functioning of democracy.
To illustrate the findings from Table 2 more clearly, Figure 3 displays the
partial plots, showing the effects of self-expression values on effective democracy;
and the effects of overt support for democracy on effective democracy— controlling
for the effects of the other independent variable in both cases. These partial plots
make strikingly clear that the impact of self-expression values on effective democracy
is unaffected by the fact that we are controlling for overt support for democracy: it
continues to show a strong relationship with effective democracy. But, by contrast,
the effect of overt support for democracy on effective democracy disappears when we
control for levels of self-expression values.
[Figure 3: about here]
These findings indicate that the impact of a pro-democratic political culture on
effective democracy does not operate through its impact on public support for
democracy. Figure 4 suggests why this is so: Public support for democracy can be
very strong among publics that show low levels of tolerance, trust, participatory
orientations and the other components of self-expression values. Let us examine this
figure more closely.
[Figure 4: about here]
Strong self-expression values seem to be a sufficient condition to create a
minimum amount of support for democracy: once we move above the level of self-
expression values found in Japan, there are about fifty or more percent of “solid
democrats” within each population. On the other hand, strong self-expression values
are by no means a necessary condition to create a certain proportion of solid
democrats. This is evident from the fact that among nations with weak emphasis on
17
self-expression there can be very low as well as very high proportions of solid
democrats (consider, for instance, Albania and Hungary in Figure 4). These
observations indicate that overt support for democracy is sometimes inflated by
superficial lip service that is not necessarily linked to deeper-rooted democratic
values.
Yet, we should refrain from such cross-level inferences without testing them.
Let us examine at the individual level what motivates people to express overt support
for democracy. Bratton and Mattes
30
conducted such an analysis using data from the
Afrobarometer. They found that individual support for democracy is determined far
more by instrumental motives than by normative commitments to the values that are
inherent to democracy. This finding is perfectly replicated in our analysis of the
World Values Surveys, as Table 3 shows. Although there is a linkage between support
for democracy and self-expression values at the individual level (see model 2), what
people think about the performance of democracy in running the economy and
maintaining law and order is a much better predictor of their overt support for
democracy (compare the explained variances of models 1 and 2). To be sure, people
with strong emphasis on self-expression almost always prefer democracy to
autocracy, but there is a large number of people who support democracy for reasons
of expected performance—even if their emphasis on self-expression is weak. Hence,
overt support for democracy is a poor indicator of intrinsic support, since overt
support is inflated by instrumentally motivated lip service.
[Table 3: about here]
Conclusion
18
In analysis of data from the 1981 World Values Surveys, Inglehart found that
societies with relatively high levels of interpersonal trust and life satisfaction were
much likelier to have democratic institutions than societies with relatively low levels
of trust and well being. This is a reliable finding that has been replicated in
subsequent waves of surveys covering much larger numbers of countries. It is by no
means “spurious,” as Seligson claims. Seligson attempts to refute this societal-level
correlation by demonstrating that there are only weak correlations between trust and
life satisfaction, on one hand, and lip service to democracy at the individual level.
This attempted refutation depends on the implicit assumption that individual-level
endorsement of democracy can be equated with democratic institutions at the societal
level. This is an unwarranted cross-level assumption—and one that proves to be
false. Initially, it may seem plausible to assume that countries with widespread lip
service to democracy are more democratic than those where it is less widespread, but
it is empirically untrue—because at this point in history, democracy has a favorable
image almost everywhere.
Seligson’s assumption that overt support for democracy at the individual level
is a reliable measure of democratic institutions at the societal level, is mistaken—and
represents an example of the individualistic fallacy. Today, lip service to democracy
is widespread-- but it does not necessarily reflect a deep commitment to crucial
democratic norms. On the other hand, the evidence indicates that a political culture
that emphasizes self-expression, tolerance, trust, life satisfaction and participatory
orientations, plays a crucial role in effective democracy. This linkage is remarkably
strong and it persists when we control for levels of economic development and for
how long a society has experienced democratic institutions. A political culture of
19
tolerance, trust, and the other components of self-expression values, seems to be
essential to the flourishing of democratic institutions.
As shown in recent analyses by Welzel as well as Welzel, Inglehart and
Klingemann
31
, effective democracy is an evolutionary phenomenon. It emerges from
a broader process of human development, in which economic development tends to
promote rising self-expression values that in turn tend to fuel effective democracy. In
conclusion, effective democratic institutions are rather a consequence than a
precondition of a democratic mass culture.
20
Appendix
National aggregates of self-expression values have been calculated running the
factor analysis shown in Table 1 across the time-pooled aggregated data set of the
World Values Surveys, including 137 “nation per wave” units. The time-pooled data
matrix provides aggregates of self-expression values from the 2nd WVS (about 1990)
for 34 countries, including: Argentina, Austria, Belarus, Belgium, Brazil, Bulgaria,
Canada, Chile, China, Denmark, Germany (East), Germany (West), Finland, France,
Great Britain, Hungary, Iceland, India, Ireland, Italy, Japan, Latvia, Mexico,
Netherlands, Nigeria, Norway, Portugal, Russia, South Korea, Slovenia, Spain,
Sweden, Turkey, U.S.A.
For another 29 countries, missing self-expression values in the 2nd WVS have
been estimated from existing self-expression values in the 3rd WVS (about 1995). For
estimation, we used the following regression equation (which explains 91% of the
variance across 21 countries): “S
ELFEXVAL1990 = .124 + .841 * SELFEXVAL1995.”
Estimates based on this equation have been assigned to the following countries:
Albania, Armenia, Australia, Azerbaijan, Bangladesh, Bosnia-Herzegovina,
Colombia, Croatia, Czech Republic, Dominican Republic, Estonia, Georgia, Ghana,
Lithuania, Macedonia, Moldova, New Zealand, Pakistan, Peru, Philippines, Romania,
South Africa, Slovakia, Switzerland, Taiwan, Ukraine, Uruguay, Venezuela,
Yugoslavia.
For still another 10 countries, missing self-expression values in the 2nd WVS
have been estimated from existing self-expression values in the 4th WVS (about
2000). For estimation, we used the following regression equation (which explains
92% of the variance across 28 countries): “S
ELFEXVAL1990 = .047 + .858 *
21
SELFEXVAL2000.” Estimates based on this equation have been assigned to the
following countries: Egypt, El Salvador, Greece, Iran, Jordan, Luxembourg, Malta,
Poland, Uganda, Zimbabwe. In case of Jordan, New Zealand and Pakistan, aggregates
for self-expression values have been calculated excluding “tolerance of diversity” (see
fn. 1 in Table 1 for operationalization), since the relevant questions were not asked
there.
22
Table 1: The Dimension of Self-Expression Values
Levels of Analysis:
Variables:
Individual level within
nations (mean
loadings)
Individual level across
nations (pooled data)
Aggregate cross-
national level
Strong self-expression values
reflect strong emphasis on the
following attitudes/behavior:
- Tolerance of diversity
1
.47 .68 .82
- Public Self-expression
2
.45 .65 .87
- Liberty and Participation
3
.54 .59 .82
- Interpersonal Trust
4
.34 .47 .64
- Life Satisfaction
5
.13 .44 .76
Weak self-expression values
reflect weak emphasis on these
attitudes/behavior.
Explained variance 23% 29% 54%
Number of cases 137 national surveys 158,803
individuals
137
nation per wave units
Notes: Entries are factor loadings. Explorative principal components analysis (extraction of factors with
‘Eigenvalues’ above 1 adviced), no rotation. Source: European/World Values Surveys I-IV.
1
“Not mentioned” for “disliked neighbors” coded “1” and dichotomized against 0; scores added for
neighbors with AIDS (V58) and homosexual neighbors (V60).
Aggregate data are national averages on this 0-2 scale.
2
“Have done” for “signing petitions (V118) coded “1” and dichotomized against “0.”
Aggregate data are national percentages have done.
3
Respondents’ first and second priorities for “giving people more say in important government
decisions” and “protecting freedom of speech” (V106-107) added to a four-point index, assigning 3
points for both items on first and second rank, 2 points for one of these items on first rank, 1 point for
one of these items on second rank and 0 for none of these items on first or second rank.
Aggregate data are national averages on this 0-3 scale.
4
Respondents believing “most people can be trusted” (V27) dichotomized as “1” against “0.”
Aggregate data are national percentages of people trusting.
5
10-point rating scale for life satisfaction (V65).
Aggregate data are national averages on this 1-10 scale.
23
Figure 1: Freedom Rights and Effective Democracy
00
10
20
30
40
50
60
70
80
90
100
110
00 10 20 30 40 50 60 70 80 90 100 110
Freedom Rights 1999-2000
Effective Democracy 1999-2000
Small variance in high values of
Freedom Rights
translates into
large variance in Effective Democracy.
24
Figure 2: Political Culture and Democratic Institutions
Albania
Argentina
Australia
Austria
Azerbaijan
Belarus
Belgium
Brazil
Bulgaria
Canada
Chile
China
Colombia
Croatia
Czech R.
Denmark
Egypt
El Salv.
Estonia
Finland
France
Ghana
Greece
Hungary
Iceland
India
Ireland
Italy
Japan
Jordan
Latvia
Lithuania
Luxemb.
Macedon.
Mexico
Moldova
Netherld.
New Zeald.
Nigeria
Norway
Peru
Philipp.
Poland
Portugal
Romania
Russia
S-Africa
S-Korea
Slovakia
Slovenia
Spain
Sweden
Switzerld.
Taiwan
Turkey
Uganda
G.B.
Ukraine
Uruguay
U.S.A.
Venez.
Germany (W.)
Yugosl.
Zimbabwe
-05
05
15
25
35
45
55
65
75
85
95
105
-1.5 -1.3 -1.1 -0.9 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1
Strength of Self-Expression Values (1990)
Degree of Effective Democracy (1999-2000)
y = 38.94 + 29.40 * x
R sq. = 0.75
25
Table 2: The Effect of Self-Expression Values on Effective Democracy Controlling for Rival Predictors
Dependent Variable: Effective Democracy 1999-2000
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
Model 7
Predictors:
B (SE)
Beta
B (SE)
Beta
B (SE)
Beta
B (SE)
Beta
B (SE)
Beta
B (SE)
Beta
B (SE)
Beta
Self-
Expression
Values
27.15***
(1.94)
.86
15.95***
(3.13)
.51
23.07***
(3.01)
.73
24.88***
(3.03)
.83
GDP per
capita
1995
.003***
(.000)
.84 .001***
(.000)
.43
Experience
with
democracy
until 1995
.50***
(.05)
.75 .12*
(.06)
.18
Support for
Democracy
1995-98
.91***
(.19)
.60 .10
(.15)
.07
Constant
43.61***
(1.80)
13.22***
(2.13)
23.06***
(3.22)
-2.25
(9.55)
28.98***
(3.76)
38.65***
(3.09)
37.82***
(7.69)
Adjusted
R
2
.74
.70
.55
.34
.80
.76
.74
N
68
99
69
44
68
67
44
26
Figure 3: Partial Effects of Self-Expression and Democratic Support
on Effective Democracy
Yugoslavia
Venezuela
Uruguay
U.S.A.
Ukraine
Turkey
Taiwan
Switzerld.
Sweden
Spain
South Africa
Slovenia
Slovakia
Russia
Romania
Peru
Norway
Nigeria
New Zeald.
Moldova
Mexico
Macedonia
Lithuania
Latvia
South Korea
Japan
India
Hungary
Germany (W.)
Georgia
Germany (E.)
Finland
Estonia
Czech R.
Croatia
Chile
Bulgaria
Brazil
Belarus
Azerbaij.
Australia
Armenia
Argentina
Albania
-35
-25
-15
-5
5
15
25
35
-30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30
Percentage Solid Democrats
(residuals unexplained by self-expression values)
Effective Democracy
(residuals unexplained by self-expression values)
R squared: .01
Yugoslavia
Venezuela
Uruguay
U.S.A.
Ukraine
Turkey
Taiwan
Switzerld.
Sweden
Spain
South Africa
Slovenia
Slovakia
Russia
Romania
Peru
Norway
Nigeria
New Zeald.
Moldova
Mexico
Macedonia
Lithuania
Latvia
South Korea
Japan
India
Hungary
Germany (W.)
Georgia
Germany (E.)
Finland
Estonia
Czech R.
Croatia
Chile
Bulgaria
Brazil
Belarus
Azerbaij.
Australia
Armenia
Argentina
Albania
-50
-40
-30
-20
-10
0
10
20
30
40
50
60
-1.8 -1.5 -1.3 -1.0 -0.8 -0.5 -0.3 0.0 0.3 0.5 0.8 1.0 1.3 1.5 1.8
Self-Expression Values
(residuals unexplained by % solid democrats)
Effective Democracy
(residuals unexplained by % solid democrats)
R squared: .62
27
Figure 4: Support for Democracy and Self-Expression Values
Yugoslavia
Venezuela
Uruguay
U.S.A.
Ukraine
Turkey
Taiwan
Switzerld.
Sweden
Spain
South Africa
Slovenia
Slovakia
Russia
Romania
Peru
Norway
Nigeria
New Zealand
Moldova
Mexico
Macedonia
Lithuania
Latvia
South Korea
Japan
India
Hungary
Germany (W.)
Georgia
Germany (E.)
Finland
Estonia
Dominican R.
Czech R.
Croatia
Chile
Bulgaria
Brazil
Bosnia
Belarus
Bangladesh
Azerbaijan
Australia
Armenia
Argentina
Albania
10
20
30
40
50
60
70
80
90
-1,5 -1,3 -1,0 -0,8 -0,5 -0,3 0,0 0,3 0,5 0,8 1,0 1,3 1,5 1,8 2,0 2,3
Mass-Emphasis on Self-Expression 1990
Percentage Solid Democrats 1995-98
y = 49.34 + 12.09 * x
R sq. = .35
28
Table 3: Predicting Support for Democracy at the Individual Level
Dependent Variable: Support for Democracy 1995-98
Instrumental Support Model
Intrinsic Support Model
Combined Model
Predictors:
B (SE)
Partial R
B (SE)
Partial R
B (SE)
Partial R
Democracies have bad economy
a)
–.86 (.02)
–.20***
–.84 (.02)
–.19***
Democracies are indecisive
b)
–.23 (.02)
–.06***
–.21 (.02)
–.05***
Democracies are bad in maintaining order
c)
–.65 (.02)
–.16***
–.59 (.02)
–.14***
Self-Expression Values
d)
.66 (.02)
.18***
.42 (.02)
.13***
Constant
–.56 (.06)
4.55 (.05)
–.16 (.07)
Adjusted R
2
.24
.11
.25
N
56,894
59,450
50,531
a) V160 “In democracies, the economic system runs badly.” Answers recoded (4: strongly agree, 3: agree, 2: disagree, 1: strongly disagree).
b) V161 “Democracies are indecisive and have too much quibbling.” For coding, see a).
c) V162 “Democracies aren’t good at maintaining order.” For coding, see a).
d) Pooled individual-level factor scores for variables specified in Table 1 (center column).
All effects significant at the .001-level. Effects obtained after introducing controls for cultural zones, using dummies for each of the nine cultural
zones specified by Inglehart and Baker (see footnote 8). Effects of cultural zone dummies not documented for reasons of space restriction.
Source: European/World Values Surveys III (1995-98).
29
1
Mitchell Seligson, “The Renaissance of Political Culture or the Renaissance of the
Ecological Fallacy,” Comparative Politics 34 (April 2002), 273-292.
2
Adam Przeworski and Henry Teune, The Logic of Comparative Social Inquiry (New
York: Wiley, 1970), ch. 3.
3
William S. Robinson, “Ecological Correlations and the Behavior of Individuals,”
American Sociological Review 15 (1950), 351-357.
4
Seymour Martin Lipset, “Some Social Requisites of Democracy: Economic
Development and Political Legitimacy,” American Political Science Review 53
(1959), 69-105.
5
Gabriel Almond and Sidney Verba, The Civic Culture: Political Attitudes in Five
Western Democracies (Princeton: Princeton University Press, 1963).
6
Hayward R. Alker Jr., “A Typology of Ecological Fallacies,” in Mattei Dogan and
Stein Rokkan, eds., Quantitative Ecological Analysis in the Social Sciences
(Cambridge: MIT Press, 1969), 69-86.
7
Ronald Inglehart, Modernization and Postmodernization: Cultural, Economic and
Political Change in 43 Societies (Princeton: Princeton University Press, 1997).
8
Ronald Inglehart and Wayne E. Baker, “Modernization, Cultural Change and the
Persistence of Traditional Values.” American Sociological Review 65 (February
2000), 19-51.
9
Christian Welzel, Fluchtpunkt Humanentwicklung: Die Grundlagen der Demokratie
und die Ursachen ihrer Ausbreitung [Focus Human Development: The Foundations
of Democracy and the Causes of its Expansion] (Opladen: Westdeutscher Verlag,
2002).
10
Christian Welzel, Ronald Inglehart and Hans-Dieter Klingemann, “The Theory of
Human Development: A Cross-Cultural Analysis,” European Journal of Political
Research 42 (April 2003), forthcoming.
11
Hubert M. Blalock Jr., Causal Inferences in Nonexperimental Research (New York:
Seminar Press, 1964).
12
Philip E. Converse, “The Nature of Belief Systems among Mass Publics,” in David
E. Apter, ed., Ideology and Discontent (New York: Free Press, 1964), pp. 206-261.
13
Note that the individual-level data are measured in ordinal or dichotomous scales
(that transform into continuous scales at the aggregate level). Considering this scale
30
level, the Pearson product-moment correlations tend to underestimate the “real”
correlations. Using tetrachoric correlations alternatively, provides somewhat stronger
correlations at the individual-level (not documented here). But still, these correlations
are considerably weaker than those at the aggregate-level. See Karl G. Jöreskog,
“New Developments in LISREL: Analysis of Ordinal Variables Using Polychoric
Correlations and Weighted Least Squares.” Quality & Quantity 24 (1990), 387-404.
14
Benjamin Page and Robert Y. Shapiro, “The Rational Public and Democracy,” in
G. E. Marcus and R. L. Hanson, eds., Reconsidering the Democratic Public
(Pennsylvania: Pennsylvania State University Press, 1993), p. 43.
15
Robert S. Erikson, Michael B. MacKuen, and James A. Stimson, The Macro Polity
(New York: Cambridge University Press, 2002).
16
Richard Rose, “A Divergent Europe,” Journal of Democracy 12 (January 2001),
93-106.
17
This conception of effective democracy has been introduced by Welzel, p. 155-160.
18
The Freedom House scores can be obtained from the Freedom House homepage:
http://www.freedomhouse.org
. For a description of the estimation process and scale
construction, see Freedom House, ed., Freedom in the World (Lanham: University
Press of America, 1996), pp. 530-535.
19
Zachary Elkins, “Gradiations of Democracy? Empirical Tests of Alternative
Conceptualizations,” American Journal of Political Science 44 (April), 287-294,
provides convincing theoretical reasons, plus empirical evidence, that continuous
measures of democracy are superior to dichotomous classifications of democracies vs.
non-democracies.
20
Ted R. Gurr and Keith Jaggers, “Tracking Democracy’s Third Wave with the Polity
III Data,” Journal of Peace Research 32 (2 1995), 469-482, demonstrate that the
Freedom House scores correlate strongly with alternative measures of democracy. For
a cross-validation of the Freedom House scores in relation to alternative indicators,
see Kenneth Bollen and Pamela Paxton, “Subjective Measures of Liberal
Democracy,” Comparative Political Studies 33 (2, 2000), 58-86.
21
Data and methodological report can be obtained from Transparency International’s
homepage: http://www.transparency.org
.
22
Rose, p. 93-106.
31
23
Seymour Martin Lipset and Gabriel S. Lenz, “Corruption, Culture and Markets,” in
Lawrence E. Harrison and Samuel P. Huntington, eds., Culture Matters: How Values
Shape Human Progress (New York: Basic Books, 2000), 112-124.
24
Thus, uncorrupt authoritarian regimes do not wind up receiving the same effective
democracy score as slightly corrupt democratic regimes. Democratic regimes must be
extremely corrupt in order to slump down to the same effective democracy score as an
uncorrupt authoritarian regime.
25
The curvilinear relationship is not simply pre-defined by the way we construct
effective democracy. If, for instance, high levels of freedom rights would tend to
produce high rates of elite integrity, there would be a linear rather than a curvilinear
relationship.
26
This relationship is not tautological. Conceptually, self-expression values and
effective democracy measure clearly distinguished phenomena; empirically, the data
are taken from completely different sources.
27
Measured in 1995 per capita GDP in purchasing power parities. Data taken from
World Bank, ed., World Development Indicators (Washington D. C., 1998).
28
This variable measures the number of years that a country has spent under a
democratic constitution. These years haven been counted from the beginning of a
nation’s independence (or from 1850 onward in case of countries that have not
independent before 1850) until 1995. Countries that emerged from the dissolution of
the Soviet Union and Yugoslavia have been coded like their former mother country as
long as they belonged to it. A year has been counted as one under a democratic
constitution, if a country obtained at least +7 points on the “Autocracy-Democracy”
index from Gurr and Jaggers (see footnote 201). This index is based on an analysis of
constitutions considering the extent of restrictions on executive power and the voters’
opportunities to influence politics. Gurr and Jaggers classify countries as “coherent
democracies,” if they reach +7 or more points on their –10 to +10 index. Data and
methodological description can be obtained from the homepage of the “Polity 98”
project: http://www.bsos.umd.edu/cidcm/polity
. We used these data here because they
reach farther back in time than the scores from Freedom House and are therefore more
adequate to measure the endurance of the democratic tradition.
29
Democracy-scale according to Hans-Dieter Klingemann, “Mapping Political
Support in the 1990s: A Global Analysis,” in Pippa Norris, ed., Critical Citizens:
32
Global Support for Democratic Governance (New York: Oxford University Press,
1999), 31-56. In the first step, we added up respondent’s support of the statements
„Having a democratic political system“ (V157) and „Democracy may have problems
but it’s better than any other form of government“(V163). Support for these
statements could be expressed in four categories: „very good“ (code 3), „fairly good“
(code 2), „fairly bad“ (code 1) and „very bad“ (code 0) in case of V157 and „agree
strongly“ (code 3), „agree“ (code 2), „disagree“ (code 1) and „disagree strongly“
(code 0) in case of V163. People’s support for these statements has been added up to a
0-to-6 scale, with 6 representing the highest support for democracy. In the second
step, we added up people’s support of the statements „Having a strong leader who
does not have to bother with parliament and elections“ (V154) and „Having the army
rule“ (V156). Analogous to the first step, this creates a 0-to-6 scale of support for
autocracy. In the third step, we subtracted the “support for autocracy” scale from the
“support for democracy” scale to create an overall index of “autocratic versus
democratic support,” ranging from –6 (maximum autocratic support) to +6 (maximum
democratic support). In the fourth step, we calculated for each country the percentage
of people scoring on at least +4 on this index (since from +4 onward you are closer to
the maximum democratic support (+6) than to the neutral point (0)). Hence, we obtain
the percentage of “solid democrats” for each country.
30
Michael Bratton and Robert Mattes, “Support for Democracy in Africa: Intrinsic or
Instrumental?” British Journal of Political Science 31 (2001), 447-474.
31
See footnotes 9 and 10.
33
