Postmaterialism , religiosity and ethnocentrism : Interactive effects on political preferences

Relationships between value orientations and political attitudes are usually analyzed as linear and additive associations. Since values are commonly conceived as lacking independence of each other, particularly in politics where they usually appear in the shape of ‘ideological packages’, the paper examines how values interact when generating their political effects. We investigate the interactive effects between postmaterialism, religiosity and ethnocentrism when they are required to explain ideological and party preferences. The outlined problems are examined using the Dutch Parliamentary Election Studies (DPES) data. Results show that political expression of some values (e.g., ethnocentrism) is dependent on the level of the other values (e.g., postmaterialism).

etc. (e.g., Caricati, 2007).Operationally, this usually means that in regression equations their effects are specified as linear and additive, regardless of whether one focuses on a specific value orientation, or on a more comprehensive set of value orientations.
Theories of values, however, argue that they are not independent; they usually form clusters or 'packages'.According to Schwartz (1992), values form a 'quasi-circumplex structure', where the position of value types around the circle corresponds to their degree of congruence or conflict.Probability of manifesting a value is associated with the probability of activating values in its neighborhood.They, in other words, are not independent of one another.This seems to be the case especially in the political sphere, where values come in the form of ideological systems, built from a limited set of values in a more or less consistent manner (Kerlinger, 1984;Maio, Olson, Bernard, & Luke, 2003).
Linear and additive associations between value orientations and various variables which they are thought to explain may be only one part of the narrative.Attitudinal or behavioral expression of one value may depend on the levels of other relevant values and orientations.This paper addresses the ways in which political value orientations interact in order to generate effects on political attitudes and orientations.For instance, there may be specific affinities between particular values.Postmaterialists, being concerned with freedom of self-expression and cultural diversity (e.g., Inglehart 1977Inglehart , 1990)), may be inclined not only to be more tolerant towards ethnic minorities (which is implied by the linear association), but demonstrate less variance in ethnic attitudes that materialists do.The argument can also be construed from a reverse angle: ethnocentric orientation may constrain variance in political attitudes much more than the opposite value, i.e., ethnic tolerance.The association between postmaterialism (PM) and political preferences, therefore, is not necessarily constant across different levels of ethnocentrism.
This paper addresses the interplay between three general value or ideological orientations: religiosity, PM and nationalist attitudes, on the basis of a detailed study of Dutch survey data.The Netherlands represents a particularly suitable case for this purpose.Religious-secular cleavage has remained salient here.Postmaterialist orientation is relatively widespread (Inglehart & Abramson, 1994).Finally, nationalist or ethnocentric parties gained prominence in the past dozen years.

Theoretical framework and general hypotheses
The substantive focus is on the interactive effects of certain politically salient value orientations onto political attitudes and preferences.Value orientations in question include materialist-postmaterialist orientation, religiosity, and ethnocentrism.They belong to the core set of values that define the contemporary world, and thereby shape the political process.The following section describes the outlined value orientations, and their expected relationships with the examined political variables, including the interactive effects.
According to the classics in political sociology, social cleavages produce political divisions (Lipset & Rokkan, 1967).The link between these two, however, is established by corresponding value or ideological orientations. 1Despite the long-term process of secularization, the religious cleavage remains one of the key factors structuring political competition in most European countries, including the Netherlands (e.g., Todosijević, Mijić, & Hristić, 2014).
Postmaterialism, according to the originators of the concept (Inglehart, 1977(Inglehart, , 1990)), represents a new cleavage separating post WWII generations, raised in material and political security from the previous more materialistically oriented generations (Inglehart, 1990).Political outlook of these generations is often subsumed under the label "new politics", and is related with ecological and other left-liberal parties.
Ethnocentrism, and related concepts such as nationalism, chauvinism, nativism that demand exclusionary or otherwise negatively directed policies towards various ethnically/nationally defined groups, is also a constant feature of (not only) European politics.In the classical social cleavage model, these orientations reflect the division between center (dominant group) and periphery.With the rise of anti-immigrant issues across Europe this political dimension, while dormant in decades immediately after the WWII, gained in significance at the turn of the century.
Political variables, the dependent variables in this study, include general ideological orientation (left-right self-placement), and party preferences.Leftright semantics has dominated European political discourse for more than two centuries.The advantage of the left-right schema is its ability to represent party stands on many issues simultaneously (Dalton, 1988;Fuchs & Klingemann, 1990;Knutsen, 1995).Hence, in the survey research on political attitudes and behavior, left-right self-identification, as used in the present research, crystallized as a standard summary assessment of individual general ideological orientation.Since political parties are crucial actors in contemporary democratic politics, any value orientations that claim to be politically relevant are expected to be associated with party preferences.
Existing arguments about the relationships between values are typically founded upon the idea of ideological consistency (Converse, 1964;Rosenberg, 1960).Thus, the emphasis on individual freedom and self-expression, that characterizes postmaterialist orientation, can be construed as being cognitively and emotionally inconsistent with the traditional religiosity and derivations of political implications from religious doctrines.
In addition to cognitive consistency, the relationships between values are influenced by contemporaneous political-ideological divisions, in particular by the left-right dimension (Fuchs & Klingemann, 1990).Consequently, secularism and PM could be regarded as being 'naturally' associated with political left.Hypotheses about consistencies between values are usually well captured by 1 According to Deegan-Krause (2006) political cleavages are 'complete' when all three aspects are present: social division, political parties, and corresponding ideological dimensions.
linear and additive models.Many studies report on consistent associations between the examined orientations, which are also demonstrated by their interrelationships in present data (see Appendix A).However, the hypothesis about the interaction between values implies that political influence of one value orientation may depend on the levels of another value orientation.These aspects of relationships between values have been scarcely theorized, although there are research examples that provide tests of interaction between values (De Koster et al., 2013;De Koster & van der Waal, 2007).None the less, it is possible to illustrate the potential of explicit hypotheses about interactions.
Ideological orientations are often described as bi-polar dimensions.Thus, ethnocentrism is opposed by tolerant orientation.Materialist orientation is opposite to PM, religiosity to secularism, or atheism.However, the two extremes need not have complementary implications for relations with other values.One extreme may be more constraining regarding political expressions than the other.Two major constraining forces are hypothesized here.One is the aforementioned cognitive-affective consistency between specific orientations, which has been well elaborated in social psychology. 2 The other force is more political in nature.Associations between values are not always, or necessarily, logical.Political parties themselves are important.They do not just reflect social cleavages; they are active agents that can create divisions. 3Parties also 'aggregate' specific attitudes and orientations into more generalized ideological dimensions that may be idiosyncratic, i.e., time-and place-specific.In the Dutch case, for instance, we have a relatively unusual combination of anti-immigrant and socially libertarian attitudes, represented by the Lijst Pim Fortuyn (LPF).Green Left has been pronounced in its postmaterialist, ecological, left-wing and ethnically tolerant ideology.This factor, the actual combination of ideological orientation within a particular political space, is therefore another major influence affecting the politically-relevant interaction between values.
Returning to the example in question, it could be hypothesized that materialism is additionally restrictive with reference to the possible positions on the left-right dimension more than the postmaterialist orientation.In that instance, the effect of ethnocentrism onto left-right ideology ought to depend on the level of materialism.Among materialists, ethnocentrism should matter little for their left-right position.Postmaterialists tend to be more open in their political outlooks, and therefore their left-right identification can be stalwartly associated with other attitudes they adopt.This would imply that ethnocentrism affects right-wing identification much more strongly among postmaterialists than among materialists.
An additional illustration addresses religiosity.In democracies with clearly defined religious parties, high religiosity may limit one's range of acceptable ideological orientations and political parties.Non-religious individuals, on the other hand, may be more open towards different political perspectives.Therefore, PM may be associated with political preferences only among the non-religious individuals, while preferences of the religious individuals can be unrelated to their level of PM.Such preferences should be more closely associated with right-wing identification, and religious parties.
Additional hypotheses are not difficult to generate.Nationalist/ethnocentric orientation is strongly associated with right-wing preferences.However, nonnationalists could be left-wingers as well as right-wing libertarians.At this point, one's level of PM or religiosity might be decisive -an interaction between values appears likely.
Moreover, many studies report of the link between religiosity and ethnocentrism (or prejudice; Altemeyer, 2003;Glock & Stark, 1966;Gorsuch & Aleshire, 1974;Rokeach, 1960;Ziebertz, Kay, & Riegel, 2009).The underlying argument is that they both are representatives of a kind of a closed belief system.Therefore, an interaction could be expected -strong religiosity may limit variability in ethnocentric attitudes, and could lead to a very specific political outlook -preference for religious parties, or radical-right parties (Arzheimer & Carter, 2009) and right-wing identification.However, secular respondents may spread across the ideological spectrum depending on other values they have come to endorse.
To recapitulate, the explanation for such interaction could be sought in general ideological narratives.Such narratives often imply specific associations between values, attitudes, and political preferences (Maio et al., 2003).Yet, the relations that bind ideological systems are not strictly logical (e.g., Achterberg & Houtman, 2009), so different sides of one value dimension may seem to imply different political views depending on the other elements in the general ideological representation.Search for the further explanation could set out in two directions.One is towards more basic psychological processes, such as the need for cognitive-emotional consistency (Rosenberg, 1960).The other would lead towards the actual political discourse, including context-dependent history of political oppositions and ideological heritage.

Notes regarding Dutch political context
The Netherlands represents a particularly suitable setting for the study of the interplay between values, ideological orientations, and political preferences.The key motivation is the existence of an elaborate party system with parties having clear ideological orientations, particularly regarding the variables included in this study.
Despite the initial sharp secularization, the religious cleavage remains politically important.It is expressed is the existence of both a mainstream Christiandemocratic party (Christian Democratic Appeal, CDA), and several smaller denominational parties (Staatkundig Gereformeerde Partij, ChristianUnion).The Dutch have also been described as one of the most postmaterialist nations (Inglehart & Abramson, 1994).Parties such as Green Left (GroenLinks, GL), and even the Socialist Party (SP) provide political expression for postmaterialist preferences.The Europe-wide trend of anti-immigrant politics has also been early on and intensely expressed in the Netherlands, indicating the relevance of ethnocentric attitudes.Parties of this ideological orientation include Lijst Pim Fortuyn (LPF), Livable Netherlands (both non-existent since mid-2000s), and more recently Partij voor de Vrijheid (PVV).Labor Party (PvdA) represents the traditional European social-democratic left, while the Socialist Party is a blend of populist and new-left appeals.An additional benefit of focusing on the Dutch case is the availability of high-quality survey data.

Method Sample and data collection method
The analysis is based on the Dutch Parliamentary Election Study (DPES)4 data.DPES studies are post-election public opinion surveys, based on face to face interviews using probability samples of Dutch population.The newly created cumulative data set, with studies from 1971 to 2006, is freely accessible through the DANS archive.5 DPES surveys from 1994, 1998, 2002, and 2003 are used here since they contain the required variables.Basic characteristics of the samples are shown in Table 1.The reason to examine multiple surveys is to establish which findings seem to be more robust and consistent, while controlling for the political-cultural context.The incorporated DPES surveys employed the widely used standard 4-item operationalization of materialist vs. postmaterialist value orientation (for a recent summary see Abramson, 2014).Respondents are asked to rank four political goals according to their order of preference.Those who chose reducing inflation and maintaining order as their first two preferences are classified as materialists.Those selecting 'More say in politics', and 'Freedom of speech' as the most important goals are classified as postmaterialists.Respondents with value priorities that combine the two categories are coded as 'mixed' type, and placed in-between the two former categories on the continuum from materialism to PM. Distribution of respondents across the three categories is given in Table 2. Frequency of religious service is coded on a scale from 0 meaning 'not religious', to 5, meaning attending religious service at least once a week.The confessional attitude score is a composite measure that combines questions about the support for the existence of confessional political parties, trade unions, schools, and broadcasting organizations, in addition to the belief that 'Religion is a good guide in politics.' 6n order to construct an index of religiosity that captures both the behavioral aspect of one's relationship to institutional religiosity (religious service attendance), and views on the political role of religion (the confessional attitude score), a summary measure of religious orientation is used throughout the paper.This Religious Orientation Scale (ROS) is defined here as the average score on the religious service attendance item and the confessional attitude score.Both input measures were first rescaled to the range from 0 to 1, hence the ROS scale has the same range.The goal was to obtain a measure that is both more reliable and more general, i.e., which includes both initial aspects.

Ethnocentric orientation.
There are two indicators of nationalist/ethnocentric orientation that are used over several Dutch election studies.One addresses the attitude towards asylum seekers (agreement with this item means 'Send back as many asylum seekers as possible'), and the other focuses on the Ethnic minorities i.e., "foreigners and ethnic minorities", where high score means that they "Should completely adjust to Dutch culture". 7or the respondents, this typically means immigrants from Suriname, Morocco, Turkey, and recently Eastern Europe.Throughout the remainder of the paper we concentrate on the attitude towards ethnic minorities, interpreting it is as a measure of ethnocentric orientation.8Distribution of this variable over the four studies is given in Table 3.  9 Responses are given on a scale from 0, meaning "very unsympathetic, to 10, meaning "very sympathetic".The variable is closely related to voting behavior, since most respondents actually vote for parties they like the best.
Left-Right self placement is perhaps the most central measure of political-ideological orientation (e.g., Fuchs & Klingemann, 1990).The scale uses the customary 11-point format, from 0 (left) to 10 (right).

Results
Interactive effects of values are examined for four election years -1994, 1998, 2002, and 2003, due to the limitations of the available data.Independent variables include the three measures of value orientation, and three interaction terms, for each of the pairs of values.Dependent variables are left-right selfplacement and sympathy for specific Dutch political parties.In the study, we will present results for the left-right self-placement scale, and for sympathy towards Green Left party.Results for several other Dutch parties are presented in Appendix B. The focus at this point is on the most robust findings, where the effects are relatively strong and consistent over multiple election studies.
Interactive effects are particularly apparent in the case of the left-right self-placement variable, as shown in Table 4. 10 Consistently over the 4 election studies, all the included value orientations are significant predictors of the leftright ideology, and the associations are in the expected direction.Right-wingers tend to be less postmaterialistic, more religious and more ethnocentric.Crucially, the interaction between PM and ethnocentrism has a significant effect as well.
9 Question text for 'sympathy score' items: And now I would like to know from you how sympathetic you find the political parties.You can give each party a score between 0 and 10.With this 0 means that you find this party not sympathetic and 10 means that you find this party very sympathetic.What score would you give the [PvdA]? 10 Note that we did not include any of the so-called control variables (age, gender, education).
The reason is that the present goal is not to construct a comprehensive explanatory model of left-right identification or party preferences, but rather to demonstrate the argument that values exhibit their political influence interactively.
Hence, the effects of the two variables are not independent of one another.The remaining two interactions proved insignificant.In order to enhance a more intuitive interpretation of the interactive effects, graphical presentation may be helpful.We adopt a method recommended by Franzese and Kam, termed 'Interpreting effects through differences in predicted values ' (2009, p. 25). Figure 1 shows how the influence of ethnocentrism on left-right ideology changes across different levels of PM.The graph is based on estimates obtained for 1994 DPES study, and includes realistic values (full range) for the concerned variables.It is apparent that among the materialists, variations across the spectrum of the ethnocentrism scores make little difference -they remain stably on the right wing.Postmaterialist orientation, on the other side, increases the influence of ethnocentrism.Among these respondents, even small changes in ethnocentrism are consequential for their ideological self-understanding.
Figu re 1. Interaction between postmaterialism and ethnocentrism in predicting left-right self identification (DPES 1994): impact of ethnocentrism As Bambor, Clark, and Golder point out, "It is a feature of multiplicative interaction models that they are symmetric." (2006, p. 10). Figure 2 demonstrates the way in which levels of ethnocentrism interact with the influence of PM onto left-right ideology.For respondents with average ethnocentrism scores, PM is moderately associated with left-wing preference.Among the strongly ethnocentric, PM is entirely ineffective as an ideology predictor.However, among the tolerant respondents, PM makes large difference.
Figu re 2. Interaction between postmaterialism and ethnocentrism in predicting left-right self-identification (DPES 1994): impact of postmaterialism Comparable tendencies can be observed regarding sympathy towards several of Dutch political parties.Given the relatively fragmented and volatile party system over the last two decades, we analyzed attitudes only towards the major parties.Here, we present the results for the Green Left party, while the remaining findings are given in Appendix B.
According to the results presented in Table 5, all three orientations are significant predictors of the attitude towards GL, correspondingly in the four surveys.Clearly, the strongest is the negative effect of ethnocentrism.Regarding the interactive effect, the results are akin to those reported for the left-right scale -significant effects are obtained for the interaction between PM and ethnocentrism in three out of four studies.In the 2002 survey, there is also a significant interaction between PM and religious orientation.Figure 3 provides a more intuitive insight into the obtained results.Materialist orientation constrains the effects of ethnocentrism -the materialists dislike GL regardless of their level of ethnocentrism.For postmaterialist respondents, ethnocentrism seems to be decisive for their attitude towards this party.Likewise, the level of PM does not make much difference if ethnocentrism is high -those individuals dislike GL anyway.Low ethnocentrism clears the path for PM to influence one's attitude towards the GL.Described interactions are among the most outstanding and consistent in the analyzed set.Comparable interactions between PM and ethnocentrism are also observed for the Socialist Party (SP) and Christian Democrats (CDA) (in the latter case, the interaction coefficient is in the reverse direction; see Appendix B).Weaker and inconsistent interaction between PM and religiosity is observed for liking the GL (2002 survey), and liking of the extreme right LPF (in 2003).

Discussion and conclusions
Value orientations, such as PM, religiosity, and ethnocentrism, continue to be important for elucidating a variety of political attitudes and preferences in the Netherlands (e.g., De Koster et al., 2013, Todosijević et al., 2014).The present findings do not cast doubt on the already well established findings about the linear and additive effects of the examined values onto political attitudes.The findings, however, demonstrate that the picture is bound to remain incomplete unless interaction between values is taken into account.Only an analysis of interactive effects can answer questions such as: What happens when postmaterialist and ethnocentric orientations are combined?Whether and when such combinations of values have independent effects on political behavior, above and beyond linear effects of the component values?
The hypothesis that value orientations can generate their political effects interactively is based on several theoretical considerations.General structural models of values suggest that values form clusters, so that activation of one value is affecting probability of activation of other related values (e.g., Schwartz, 1992).Similar implications follow from the perspective of cognitive consistency theories, as some combination of values and attitudes may be seen more or less consistent with broader ideological narratives.
Parties as agents that influence aggregation of particular attitudes and orientation within a given polity are also important for the emergence of particular conditional relationships between various attitudes and political preferences.De Koster et al. (2013), for instance, showed that the support for Dutch populistright parties is associated with 'welfare populism', that is an interaction between egalitarianism and a critical view of the welfare state.
Currently presented analysis of the interactive effects of values showed that indeed the effect of one variable is not independent of the levels in other variables.The most persuasive evidence was acquired for strongly ideological variables, such as the left-right self-placement, or preference for ideologically clearly outspoken parties, such as GreenLeft.
Ideologies could be seen as specific 'packages' or clusters of attitudes and value orientations, created, developed, and affected by both intellectual and political traditions, and by contemporary political entrepreneurship.Thus, for instance, the 'new right' parties developed an original combination of antiwelfarism and egalitarianism (see De Koster et al., 2013).In contemporary politics, materialist and ethnocentric attitudes both belong to the same cluster defining the right-wing ideological outlook.They both constrain, or condition, what other attitudes one might adopt, or what more general ideological identification one might express.It is more difficult to imagine (Inglehartian) materialists on the left wing than postmaterialists on the (liberal) right wing.
Likewise, ethnocentrism is more firmly associated with the right wing than its opposite with the left (liberal right is not inconsistent with low nationalism).
In accordance with this reasoning, we obtained that PM and ethnocentrism indeed interact in a manner that ethnocentrism affects right wing identification conditionally, i.e.only among the postmaterialists, since materialists remain firmly on the right at any rate.In much the similar way, De Koster et al. (2013) obtained that interaction between egalitarianism and anti-welfare state attitude differentiates the new-rightist populist parties.
The overall implication is evident: political effects of values cannot be fully understood by looking at specific values in isolation or only at their additive effects.Presence or absence of other values often makes significant difference (see also De Koster et al., 2013, Međedović, 2013).Thus, ethnocentrism would have remained a modest factor influencing ideological identification (and GL sympathy) had the Dutch population consisted entirely of materialists.Likewise, it could be argued that ethnocentrism is an important determinant of Green sympathy only because the Dutch have a fair degree of postmaterialists.Had they been all materialists, ethnocentrism would not make a difference.
Regarding the sources of value interaction, it seems that the political discourse and actual ideological constellations are relevant, in addition to the more abstract ideological narratives.The influence of the Dutch political context over the last two decades is evident in the most visible interaction effectbetween ethnocentrism and PM.Much of the recent political controversy has revolved around the left-wing postmaterialists and right-wing ethnocentrics.On the other hand, the ethnocentric right in the Netherlands has always been predominantly secular, hence the lack of the interaction between religiosity and ethnocentrism. 11hese are evidently tentative interpretations of the observed results and require further theoretical elaboration and testing.Cross-national examination of conditional relationships between value orientations affecting political views would be particularly welcome.What would make such studies particularly useful is development of explicit hypotheses about conditional relationships between value orientations, taking into account specific political and cultural context.Such hypotheses could be tailored for explaining support for specific party families (e.g., De Koster et al., 2013), but also for other relevant dependent variables.An interesting extension of the research of interactions between values is to look at interactions between variables conceptualized and measured on different levels.Arikan & Bloom (2015), for instance, found that the association between individual-level ideology (left-right) and welfare preferences is affected by the cultural context, i.e. they found a significant cross-level interaction.
The main message, however, is clear: values exert their political influence in interactions with other values; difficulty being that a better understanding of the political roles of values at the individual level necessitates more complex theories than the classical theories of value consistency.

Table 1
Basic characteristics of the samples

Table 2
Distribution of the postmaterialism scale (in %) 2. Religious Orientation (ROS) Scale.DPES surveys contain several indicators of religiosity.

Table 3
Attitude towards foreigners and ethnic minorities -distribution of answers

Measures of political attitudes and preferences.
Sympathy for different political parties represents respondent's general evaluation of different political parties.
Dependent variable: Sympathy for Green Left party.Entries are standardized multivariate regression coefficients.Independent variables: PM -Postmaterialism, ROS -Religious orientation scale.