The influence of societal values on attitudes towards immigration

Abstract

This paper examines the influence of societal values on individual attitudes towards immigration and immigrants. We argue that conflict between individual and societal values leads individuals to be exposed to frames and opinions that are contrary to their values, evokes competing considerations and creates attitudinal ambivalence and volatility. To evade ambivalence, individuals whose values are in conflict with those of their society rely less on their core values to construct their attitudes. Using data from the first wave of European Social Surveys and relying on Heteroskedastic Maximum Likelihood Regression, we test our argument simultaneously for 18 European countries and show that deviations from society’s conservation and self-transcendence values lead to greater ambivalence in attitudes towards immigration and immigrants. Our results provide evidence of the importance of the social context and society’s shared values in influencing personal political attitudes and judgments.

Keywords

ambivalence attitudes towards immigration Heteroskedastic Maximum Likelihood Regression societal values values

Introduction

Scholars of the Columbia School have consistently stressed the crucial role of social context and ‘cross pressures’ on political attitudes and behavior (e.g. Berelson et al., 1954; Lazarsfeld et al., 1944). Diverse social environments provide various countervailing positions, expose people to both sides of the argument, and thus make them less confident and more ambivalent in their attitudes (e.g. Ben-Nun Bloom and Levitan, 2011; Berelson et al., 1954; Huckfeldt, Johnson et al., 2004; Huckfeldt, Mendez et al., 2004; Lazarsfeld et al., 1944; Levitan and Visser, 2009; Mutz, 2006; Visser and Mirabile, 2004). Regardless of persuasive empirical demonstrations of the effects of social context on public opinion, most attempts to explain political attitudes focus on personal motivations without considering their interaction with societal pressures and cues.

This is no different for the study of values and attitudes towards immigration and immigrants. Plenty of examples demonstrate the influence of values on the European immigration debate and the role of public opinion in influencing immigration policies (Lahav, 2004a, 2004b). These include Switzerland’s 2009 minaret ban referendum, in which campaigns connected the construction of mosques with the oppression of Muslim females; European immigration authorities actively enforcing migrants’ rights; and France’s cabinet examining a draft bill that imposes prison sentences on men who force their wives to wear a burka. Indeed, research has established the crucial role played by the core values of conservation and self-transcendence on attitudes towards immigration (Davidov et al., 2008; Sagiv and Schwartz, 1995; Schwartz, 2007). On the other hand, most works largely ignore the effect of societal pressures on individual preferences in immigration policies.

In this paper we intend to fill this gap. We argue that societal values influence issue frames, political discourses, and public debate, which are generally known to individuals through social interactions and the consumption of mass media. When personal values and social influences that reflect a society’s values are compatible, core values function as important guides for the construction of political attitudes. However, when there is tension between individual values and the values of a broader social environment, individuals are exposed to frames, ideas, and opinions that offer competing considerations. This leads them to experience greater ambivalence, which, in turn, results in more volatile attitudes. Since people are also inclined to seek consistency and stability in their belief structure (Eagly and Chaiken, 1993), they prefer to avoid being guided by values that induce ambivalence and uncertainty. Accordingly, increasing conflict between private and societal values leads to less reliance on values during the process of attitude formation when compared with a state of congruency between personal and societal values.

To test our hypotheses, we rely on Heteroskedastic Maximum Likelihood (ML) estimation, which models the error variance of three regression models that predict attitudes towards immigration. Using data from European Social Surveys, we test our argument simultaneously for eighteen European countries and demonstrate that as individuals increasingly deviate from society’s conservation and self-transcendence values, they experience more ambivalence in their attitudes towards immigrants and support for immigration, as indicated by the increased error variance in the Heteroskedastic ML model. Second, we show that the well-established effect of conservation and self-transcendence values on immigration attitudes is contingent on the congruence of personal and societal values; as tension between privately held and societal values increases, the effect of core values on immigration attitudes decreases.

This paper’s findings stress the dynamic interactions that exist between individual level attitude formation and the broader societal environment. Our attitudes are shaped not only by our own values and our societal climate, but also by the congruency between the two. By studying the role of social context in political attitude formation, this paper offers a fuller and more multilayered picture of comparative public opinion.

Value-driven ambivalence

In contrast to past conceptualizations of political attitudes, which regarded attitudes as objects saved in memory, current literature in political psychology claims that attitudes are constructed on the spot at the moment of appraisal. This makes attitudes vulnerable to the environment in which opinion is created (e.g. Barsalou, 1987; Zaller, 1992; Zaller and Feldman, 1992). According to this view, when people are asked for their opinions, they refer to a distribution of relevant considerations attached in their memory to the political concept under consideration (e.g. values, feelings, experiences and beliefs, which may be in tension with each other). They then build on those considerations accessible at the moment of questioning to compute their responses. Such probabilistic memory search can be conceptualized as a ‘sampling’ of the distribution of one’s considerations, in which attitude actually becomes one’s estimation of the central tendency measure of their distribution of associations on the subject at the time of appraisal (Feldman, 1995).

Accordingly, individual differences in attitudes may manifest themselves in two ways that correspond to the distribution’s central tendency measures. First, people may come up with different estimations of their distribution’s mean, which translates into a different response on an issue (e.g. strongly support vs. mildly oppose). Second, people may differ in the consistency of their attitude, which will be governed by the variance of the distributions of considerations on an issue (Alvarez and Brehm, 2002; Zaller and Feldman, 1992). In other words, two people may have the same mean for the distribution of considerations (e.g. strongly oppose) but differ on the variance, meaning that their answers could be similar on average, but that greater variance in distribution will lead them to give less predictable and more volatile responses in repeated sampling.

When one’s distribution for some issue consists of contrasting considerations, they are said to be ambivalent (see Lavine, 2001; Steenbergen and Brewer, 2004). Conflict between core values is cited as an important example of competing considerations that result in attitudinal ambivalence (e.g. Alvarez and Brehm, 1995, 1997, 2002; Feldman and Zaller, 1992; Rudolph, 2005). In this line of work, when two values are relevant to a political issue yet ‘push’ opinion in different directions, the person internalizing both values is expected to show ambivalence in her attitude on the issue. Such value-driven ambivalence has been detected in a range of attitudes towards policies, such as abortion, social welfare, affirmative action, and campaign finance reform (e.g. Alvarez and Brehm, 1995, 1997, 2002; Feldman and Zaller, 1992; Rudolph, 2005), as well as candidate evaluations and electoral choice (Basinger and Lavine, 2005; Lavine, 2001; McGraw et al., 2003).

While this literature looks at ambivalence that emerges from a conflict between an individual’s core values, we argue that by providing competing considerations on an issue, societal values may also be a source of ambivalence for those who do not share the dominant orientations of their society. The crucial role that social and cultural environment plays in molding attitudes by providing cross-pressures and countervailing positions is well-established in the literature (e.g. Lazarsfeld et al., 1944; Berelson et al., 1954). Since people are motivated to refer to others in order to assess the accuracy and appropriateness of their own views (Festinger, 1954; Kelley, 1952), exposure to dissimilar values and beliefs exerts lower confidence in the correctness and appropriateness of their attitude, attitude volatility, motivation for attitude change, and ambivalence (Ben-Nun Bloom and Arikan, forthcoming; Huckfeldt, Johnson et al., 2004; Huckfeldt, Mendez et al., 2004; Keele and Wolak, 2006; Meffert et al., 2004; Levitan and Visser, 2009; Visser and Mirabile, 2004).

Socially shared values influence the social and political context within which political issues are discussed, including the context in which immigration debates take place. When an individual is exposed to societal values that are in tension with her core values, this exposure provides conflicting considerations that will manifest themselves in attitude instability. People are also inclined towards consistency in their attitudes (Eagly and Chaiken, 1993) and prefer to avoid a state of ambivalence, which involves unpredictability and uncertainty. Thus, individuals will rely less on a consideration that exerts ambivalence in attitude formation. Indeed, ambivalence in partisanship was found to decrease reliance on party identification as a political cue (Basinger and Lavine, 2005), while value conflict among religious and democratic values was seen to lead to decreased consideration of democratic attitudes when forming relevant political attitudes among the religious (Ben-Nun Bloom et al., 2011). Accordingly, we also expect those who experience personal–societal conflict to be more reluctant to rely on values as a guide when constructing their attitudes towards immigration. Next, we discuss societal and personal values, and their influence on individual attitudes towards immigration.

Personal and societal values and attitudes towards immigration

Values are desirable goals that are central to evaluations, attitudes and choices, and are also ‘the criteria people use to select and justify actions and to evaluate people (including the self) and events’ (Schwartz, 1992: 1). Personal values likewise motivate political behavior and influence political attitudes, political action, partisanship, and vote choice (see Feldman, 2003, for a review). Values also structure political knowledge and provide constraint across issues and response stability across time (Feldman, 2003).

Shalom Schwartz’s theory of universal human values established the foundations of current research in cross-cultural psychology (Feldman, 2003), and is also relevant for the study of attitudes towards immigration. Schwartz (1992) charts human values on a two-dimensional plane, with one axis contrasting openness to change with conservatism, and the second contrasting self-transcendence with self-enhancement. Both types of higher-order values – self-transcendence and conservatism – are strong indicators of attitudes towards immigration (Davidov et al., 2008; Green, 2009; Sagiv and Schwartz, 1995). Since immigration brings about changing traditions and norms, it is seen as an obstacle to the conservation values of tradition, conformity, and security. Individuals who place value on honoring social institutions, tradition, and norms, and who are concerned with preserving the social order, are less willing to accept immigrants to their countries (Davidov et al., 2008; Sagiv and Schwartz, 1995). Conservation values are also tied to an authoritarian belief system, which is strongly associated with ethnocentrism and thus generates stronger opposition to immigration (Davidov et al., 2008). On the other hand, universalism and benevolence, which represent the higher-order value dimension of self-transcendence, are related to tolerance and concern with the well-being of others, and lead to higher support for immigration and positive attitudes towards immigrants (Sagiv and Schwartz, 1995).

Accordingly, our first hypothesis expects to find that higher-order conservation values are associated with negative evaluations of immigrants, less support for allowing immigration, and support for stricter conditions for immigration (Hypothesis 1a), while self-transcendence values are related to willingness to allow immigrants into the country, a preference for fewer restrictions on immigration, as well as positive attitudes towards immigrants (Hypothesis 1b).

While the effect of personal values on individual behavior is already well established, less is known about the relationship between personally held and socially shared values – especially in cases where an individual’s personal values emphasize goals that conflict with those of the broader social environment. We argue that societal values influence the context within which political phenomena take place, and that when in conflict with personally held values, they affect the stability of individual attitudes and reliance on personal values as a guide.

Societal values¹ have evolved over long periods of time and function as the broadest and most fundamental context for social interaction (Johnston and Klandermans, 1995). Even when individuals do not fully adopt their society’s dominant values, their mental schemas and ways of thinking, acting, and behaving are affected by social values (Smith et al., 2006). In fact, individuals tend to use terminology and principles central to their society’s shared values when discussing policy preferences, even if their personal value orientations are in conflict with those of society (Feldman and Zaller, 1992). Even beneath sharp societal disagreements, citizens of the same country may show more than a little consensus on certain issues since they share a common moral vocabulary (see Bellah et al., 1985).

One of the ways in which societal values influence attitude formation is by media and elite framing of key issues. Political frames are affected by shared societal values (Hertog and McLeod, 2001) and they are a powerful source of influence since they work to structure the social and political world in a symbolic and meaningful manner (Reese, 2001). Frames also define and diagnose the social and political problems and prescribe solutions to them (Gamson, 1992). If, for example, immigration were framed in terms of the problems caused by the influx of immigrants into a country, then responses to immigration would focus on how to curb it. Yet, if immigration is framed in terms of immigrant rights and integration, then the debate will center on the definition of integration or how best to achieve it.

Despite some diversity in issue frames within any nation, there is also evidence of considerable hegemonic framing, as well as substantial cross-national variation on issue frames, campaigning styles, and the political information available to viewers (Semetko and Mandelli, 1997). For example, in the Netherlands, newspapers representing different ideological standpoints use similar frames for discussing immigration (Roggeband and Vliegenhart, 2007). Poverty in American discourse is often framed in a way that places responsibility on the individual (Iyengar, 1991), whereas an article about poverty is likely to be framed differently in most European social welfare states (Semetko and Mandelli, 1997). There is also evidence of cross-national variation in issue frames that can be attributed to different societal values. Public discourse and issue frames in universalist nations, such as Sweden and Norway, generally reflect concern for the successful integration of immigrants into their society, immigrants’ rights, and battles against discrimination (Benito, 2007; Hagelund, 2002), while a large sector of the media in nations, such as Greece or Italy, which place greater emphasis on conservation values, assumes nationalistic and xenophobic standpoints, and even the moderate media shy away from criticizing extremist frames (Cotesta, 1999; Pavlou, 2001; Triandafyllidou and Veikou, 2002).

The framing of issues by the media and elite is obviously not the only factor that exposes individuals to societal influences. People interact with others in various environments. Although it may be argued that people select their social networks and prefer to interact with those who hold similar values, studies find that social networks often provide competing political attitudes and points of view that have an impact on individual attitudes (e.g. Huckfeldt, Mendez et al., 2004).

In one way or another, individuals are aware of information and opinions that reflect the influence of socially shared values even if these are in conflict with their own personal values (DiMaggio, 1997). In addition, most people are not sufficiently well informed and cognitively active to develop counterarguments to the influences to which they are exposed (Iyengar, 1991; Zaller, 1992) and thus end up retaining nearly all the information that reaches them, usually without questioning its source or credibility (DiMaggio, 1997). When the social and political context emphasizes values that are contrary to the individual’s desired goals, this information still provides competing considerations that the individual takes into account when constructing her responses to policy preferences. Because the considerations that derive from an individual’s core values compete with socially derived influences, they push her opinion in a different direction, leading to greater ambivalence. For instance, in a society in which conservation values are widely shared, individuals who hold less conservative values are still exposed to arguments about threats to social order due to the rising number of immigrants in the society; whereas in a society where self-transcendence is valued, individuals who do not hold these values strongly are exposed to frames that emphasize tolerance and concern with the well-being of migrants. In both cases, individuals are exposed to influences that reinforce the effect of competing considerations. As a result, as individuals deviate from their country’s mean level of conservation or of self-transcendence values, their attitudes towards immigration and immigrants should be less stable, and manifest greater ambivalence (Hypothesis 2).

While Hypotheses 1a and 1b expect values to be used as a key guide in the formation of immigration attitudes, exposure to competing considerations should decrease the extent to which the value is relied upon in constructing the attitude. Classic theories in psychology claim that people seek consistency and are motivated to avoid change in existing attitudes, and thus prefer to derive their attitudes from stable constructs (see Eagly and Chaiken, 1993). However, the inherent instability of ambivalent constructs decreases reliance on them in constructing attitudes so as to avoid potential change and inconsistency (Basinger and Lavine, 2005; Ben-Nun Bloom et al., 2011). Accordingly, as an individual is exposed to competing influences that are contrary to his or her core values, the probability that they will rely on this value when forming their political attitudes should decrease. On the other hand, those with greater congruence between their positions and those of society as a whole tend to rely on the values that represent both their and their country’s outlook. Therefore, we expect an interaction between individual values and deviation from society’s values to diminish the effect of values on immigration attitudes (Hypothesis 3).

The model, data and variables

Our hypotheses are tested via Heteroskedastic ML Regression, which enables the researcher to infer ambivalence from the estimation of error variance (e.g. Alvarez and Brehm, 1995, 2002; Franklin, 1991). Unlike OLS regression, Heteroskedastic ML Regression allows one to test hypotheses on the basis both of variables that affect the predicted value of a dependent variable (mean equation), and of variables that increase or decrease instability, i.e. error variance, in the dependent variable (variance equation). A positive coefficient for an independent variable in the variance equation indicates that, holding all else constant, it has a positive effect on the error variance, which means less predictability in the dependent variable. We define a normal function for the model, and, since the variance in the dependent variables is non-negative, we define an exponential equation to constrain it accordingly.

The data come from the 2002–2003 wave of the European Social Survey (ESS), which contains a battery of questions on attitudes towards immigration in twenty-one European countries. Since some of the countries lack the Schwartz values or some independent variables, the final analysis includes data from eighteen countries.² All variables, with the exception of age, are coded to vary 0–1. For full measures and summary statistics see the online appendix at the authors’ website.

In our analysis, we use three different dependent variables as measures of attitudes towards immigration and immigrants: positive attitudes towards immigrants, constructed from six questions about whether respondents see the presence of immigrants in their country as a positive thing; support for allowing immigrants to one’s country, composed of six questions that tap the readiness of individuals to accept immigrants into their country; and support for less strict conditions for immigration, which taps the extent to which respondents support less or more strict conditions for allowing immigrants into their country. Where necessary, items are recoded in such a way that higher values represent higher support for immigration and more positive attitudes towards immigrants.

For each of the three measures, we use a confirmatory factor analysis (CFA) to test for the internal consistency of the items and differential item functioning by using the multi-group CFA model (Reise et al., 1993; Van de Vijver and Leung, 1997). We first run a baseline model in which all the factor loadings for each scale are constrained to be equal in all countries, while factor means, variances, covariances, and residuals are freely estimated for each country. Next, we evaluate modification indices and introduce several modifications to the models by relaxing the measurement invariance constraints for certain items. The final partial invariance models all have fit indices that are above the acceptable thresholds. We then save the factor scores for each latent variable, in which the items that perform differently across countries are given different weights to ensure a common measurement scale (Reise et al., 1993).³

The key independent variables in the analysis are the Schwartz values of conservation and self-transcendence. We use them to compute both personal values and the individual’s deviation from societal values. The items are recoded so that higher scores represent greater acquiescence with the statements. Adhering to the instructions provided in the ESS Codebook, we first compute the scores of the values by taking the means of the items that index them, and then correct them for scale use by computing the centered scores of these values by subtracting the mean of the items that index the values from the individual’s mean score on all value items.⁴ In order to measure an individual’s deviation from his or her country’s dominant value orientation, we take the absolute value of the z-score. That is, we measure how many standard deviations a respondent is from his or her country’s mean level on that higher-order value in absolute values.

We also include a number of control variables in the mean and variance equations. One expects interpersonal trust and higher income and education levels to be associated with positive attitudes towards immigrants and support for immigration. Those who identify with the political right and those who feel they have a low level of personal safety are expected to hold negative attitudes towards immigrants, oppose immigration, and support stricter conditions for allowing immigrants into the country, while those who have more contact with immigrants and those with an immigrant background are expected to be more supportive of immigration and have more positive attitudes towards immigrants. In addition, we also control the effects of gender, age, and religiosity.

A number of factors may also influence the error variance since they may affect the cognitive capacity or the existence of relevant contextual knowledge. Young adults may have less crystallized attitudes. Therefore we expect error variance to decrease as age increases. Education is expected to reduce stochastic variance since it implies more contextual knowledge and may also be a proxy for cognitive capacity. Respondents with higher income may experience less ambivalence due to less competition with immigrants for material resources. Interpersonal trust is also expected to decrease ambivalence since those who are more trusting should be more willing to embrace immigration. Those who feel personally unsafe in their environment as well as those with immigrant backgrounds may have more crystallized attitudes towards immigrants, which is expected to decrease the error variance. On the other hand, personal contact with immigrants may lead individuals to have different types of experiences that lead to less stable attitudes. Finally, religiosity may be a source of cognitive constraint for individuals and decrease attitude ambivalence. In addition to these variables, we also provide controls for the effects of gender and political right. Although we do not expect personal values to have a systematic effect on ambivalence, we still include the values of conservatism and self-transcendence in the variance equation since they pose an alternative intuitive explanation for any results of value conflict, and are thus essential controls.

The context within which immigration debates take place may also influence one’s attitudes towards immigration. Although evidence on the effect of contextual factors is mixed (see Green, 2009; Sides and Citrin, 2007), a large immigrant population as well as poor economic conditions may lead to greater opposition to immigration and immigrants. We include two measures for the presence of immigrants – immigration flow to the country in the last five years (in percentages), the percentage of foreign-born population in the country (as of 2001), and two measures for the well-being of the economy: logged GDP per capita averaged over the 1997-2002 period, and unemployment rates (in percentages, as of the year 2001).

Results

The dependent variables are each regressed on individual values of conservation and self-transcendence, measures of deviation from societal values, and a set of control variables as well as country-level fixed effects both in the mean equation (Table 1) and in the variance equation (Table 2).

Table 1.

Personal and societal values and attitudes towards immigration: mean equation

Mean equation	Positive attitudes towards immigrants	Support for allowing immigrants to one’s country	Support for less strict conditions for immigration
Values
Difference from country level conservation	−.033 (.008)^**	.015 (.010)	.029 (.010)^**
Difference from country level self-transcendence	−.015 (.010)	−.003 (.012)	.080 (.011)^**
Conservation	−.146 (.008)^**	−.159 (.010)^**	−.117 (.010)^**
Self-transcendence	.276 (.010)^**	.301 (.012)^**	.225 (.012)^**
Individual level controls
Feels unsafe	−−.054 (.004)^**	−.048 (.004)^**	−.038 (.004)^**
Contact with immigrants	.070 (.003)^**	.064 (.004)^**	.033 (.004)^**
Parent immigrated to country	.022 (.003)^**	.016 (.004)^**	−.008 (.003)^**
Trusts people	.136 (.004)^**	.117 (.005)^**	.058 (.004)^**
Political right	−.075 (.004)^**	−.074 (.005)^**	−.075 (.005)^**
Religiosity	.015 (.003)^**	.002 (.004)	−.024 (.004)^**
Age	.000 (.000)	−.001 (.000)^**	−−.001 (.000)^**
Male	.003 (.002)*	.004 (.002)^**	−.006 (.002)^**
Education	.106 (.004)^**	.137 (.005)^**	.033 (.004)^**
Income	.040 (.007)^**	.035 (.009)^**	−.006 (.008)
Countries fixed effects
GDP per capita, last 5 years (logged)	.029 (.004)^**	.012 (.005)^**	.054 (.005)^**
Migration, last 5 years (percentage)	−.005 (.001)^**	−.003 (.001)^**	.003 (.001)^**
Foreign born 2001 (percentage)	−.002 (.000)^**	.003 (.000)^**	−.002 (.000)^**
Unemployment 2001 (percentage)	.000 (.000)	.003 (.000)^**	.000 (.000)
Constant	.022 (.041)	.153 (.054)^**	−.253 (.047)^**
Log likelihood	15369.35/	9851.11/	11917.11/
Hetero/Homo	14774.85	9533.05	11381.13
N	28012	27617	27972

Table entries are maximum likelihood coefficients, std. errors in brackets; ** significance in two-tail 95% confidence level, * one-tail 95% confidence level.

Table 2.

Personal and societal values and attitudes towards immigration: variance equation

Variance equation	Positive attitudes toward immigrants	Support for allowing immigrants to one’s country	Support for less strict conditions for immigration
Values
Difference from country level conservation	.189 (.042)^**	.119 (.043)^**	.093 (.042)^**
Difference from country level self-transcendence	.421 (.048)^**	.177 (.048)^**	.566 (.049)^**
Conservation	.059 (.042)	−.185 (.043)^**	−.200 (.042)^**
Self-transcendence	−.203 (.050)^**	−.016 (.050)	.647 (.050)^**
Individual level controls
Feels unsafe	.071 (.016)^**	−.002 (.017)	−.089 (.017)^**
Contact with immigrants	.122 (.017)^**	.040 (.018)^**	.131 (.018)^**
Parent immigrated to country	.074 (.015)^**	.007 (.015)	−.016 (.015)
Trusts people	−.161 (.017)^**	−.077 (.018)^**	.061 (.018)^**
Political right	.079 (.020)^**	−.103 (.021)^**	−.133 (.021)^**
Religiosity	−.015 (.015)	−.069 (.016)^**	−.115 (.015)^**
Age	.001 (.000)^**	.000 (.000)	−.003 (.000)^**
Male	.046 (.009)^**	−.004 (.009)	−.020 (.009)^**
Education	−.215 (.020)^**	−.150 (.020)^**	.011 (.019)
Income	−.034 (.038)	−.097 (.038)^**	−.080 (.038)^**
Countries fixed effects
GDP per capita, last 5 years (logged)	.001 (.022)	−−.110 (.023)^**	−.035 (.024)
Migration, last 5 years (percentage)	.035 (.003)^**	.038 (.003)^**	.002 (.003)
Foreign born 2001 (percentage)	−.006 (.001)^**	−.005 (.001)^**	.005 (.001)^**
Unemployment 2001 (percentage)	.015 (.002)^**	.005 (.002)^**	.000 (.002)
Constant	−2.089 (.217)^**	−.442 (.225)^**	−1.606 (.231)^**
Log likelihood Hetero/Homo	15369.35/ 14774.85	9851.11/ 9533.05	11917.11/ 11381.13
N	28012	27617	27972

Table entries are maximum likelihood coefficients, std. errors in brackets; ** significance in two-tail 95% confidence level, * one-tail 95% confidence level.

Mean equation

Table 1 presents the results for the mean equation, in other words, estimates for the effects of privately held values, deviation from societal values, and control variables on immigration attitudes in all three models.

Values

The higher-order values of conservation are associated with more negative attitudes towards immigrants, less support for allowing immigrants into the country, and support for harsher conditions for immigration. On the other hand, holding all else constant, as individuals increasingly value self-transcendence they show more support for immigration and evaluate immigrants more positively. These findings are consistent with both our hypotheses and previous findings (e.g. Davidov et al., 2008b).

One must also note that a comparison of the magnitude of the effect of independent variables in the mean equation, made possible by coding all individual level variables except age between 0 and 1, reveals that these higher-order values have the strongest effect on attitudes towards immigration in all three models. Self-transcendence is the single strongest explanatory variable, explaining on average 27% of the range in immigration attitudes (28% of attitudes towards immigrants, 30% of support for allowing immigrants into the country, 23% of support for easier conditions for immigration). The second strongest is conservation, which explains between one eighth and one sixth of the variance in the three dependent variables (15%, 16%, and 12%, respectively).

Our hypothesis on the effect of deviation from societal values is interactive, and we do not expect deviation in itself to have any systematic effect on attitudes towards immigration in the mean equation. In fact, results show that the effect of deviation from a country’s values is either insignificant or inconsistent and small if individual-level values and other controls remain constant. Deviation from a society’s self-transcendence values is not a significant predictor of attitudes towards either immigrants or support for allowing immigration. It weakly increases support of less strict conditions for immigration, explaining 8% of the range of the dependent variable. Similarly, deviation from a society’s conservation values is not statistically significant in explaining support for allowing immigrants into a country. It is weakly associated with attitudes toward immigrants, weakly increases support for less strict conditions for immigration, and explains only about 3% of the range of the two latter dependent variables. Therefore, it seems that when the effect of personal values is controlled, deviation from societal values does not lead to a systematic and significant change in attitudes towards immigration and immigrants.

To test the third hypothesis, which states that the effect of personally held values on attitudes towards immigration will decrease as an individual deviates more from her society’s values, we rerun the three models and include interactions between personal values and deviation from society’s values. All other specification choices remain intact. Five of these six interactions fall in the expected direction, and three of them are statistically significant. Figure 1 presents the effects of the significant interactions.⁵

Figure 1.

Interactions between deviation from societal values and personal values: mean equation

The interactive effect of conservation values and deviation from society’s conservation values is statistically significant when predicting attitudes towards immigrants (b = .16, p = .00), but the interaction is not statistically significant for self-transcendence values even though the coefficient is in the expected direction (b = −.12, p = .14). The upper left panel of Figure 1 shows the significant interaction for conservation values. Holding the mean of all independent variables constant, the value of conservation vs. openness leads to more negative attitudes towards immigrants when an individual’s value orientation is congruent with that of the society (minimum deviation). However, as respondents increasingly deviate from their society’s conservation values, the effect of personally held conservation values on dependent variables decreases, and nearly flattens out when deviation is at its maximum.

The lower panel of Figure 1 replicates this effect on support for less strict conditions for immigration. Again, the interactive effect of conservation values as well as the deviation from society’s values is statistically significant (b = .13, p = .07),⁶ while the interaction for self-transcendence falls in the expected direction but is statistically insignificant (b = –.10, p = .26). Individuals rely more on their personal values to construct their attitudes towards conditions for immigration if their conservation values are compatible with those of their society, but reliance on conservation value decreases as individuals deviate from their society’s values. The effect of personal values on the attitude is almost nonexistent in those whose deviation is at maximum.

Finally, the effect of interactive terms for self-transcendence values in predicting support for immigration is statistically significant (b = –.26, p = .01). These effects are presented in the upper right panel on Figure 1. As expected, the effect of self-transcendence values on attitudes towards allowing immigrants to a country is stronger when individuals’ values are congruent with those of society, while this effect completely disappears in those whose values are in total conflict with those of society. Interactions for the conservation values are insignificant and do not assume the expected direction (b = –.11, p = .12).

Individual level controls

Control variables usually have the expected effects on dependent variables. Holding all other variables constant, having contact with immigrants, interpersonal trust, and higher education increase the endorsement of immigrants, support for allowing immigration to one’s country, and support for less strict conditions for immigration. While higher income, being male, and having a parent who was not born in one’s country have a positive effect on the first two dependent variables, they do not have a significant effect on the third dependent variable. As expected, a sense of personal insecurity in one’s environment and identification with the political right are associated with less positive attitudes toward immigrants and generate more opposition to immigration as well as support for stricter conditions for immigration. Age has a negative effect on the latter two dependent variables. Finally, religiosity does not have a systematic effect on any of the three attitudes to immigration.

Country-level fixed effects

As hypothesized, holding all else equal, higher national income is associated with positive attitudes towards immigration in all three models. We also find that both a stronger migration flow to a country and a stronger presence of foreign-borns lead to more negative attitudes towards immigration in two of the three models.⁷ Finally, unemployment has no significant effect on immigration attitudes in two of the three models, but surprisingly increases support for allowing immigrants to one’s country.

Variance equation

The key hypothesis of this paper concerns the effect of deviation from society’s values on the variance of the error residual. To test this hypothesis, we examine the heteroskedastic component of the model.⁸ Table 2 presents the results for the variance equation, in other words, the estimates for the effects of privately held values, value deviation, individual-level determinants, and country fixed effects on the error variance in immigration attitudes for all three models.

Values

Hypothesis 2 expects higher deviation from society’s values to induce ambivalence in attitudes towards immigration, thus increasing the error around the dependent variable. Indeed, the significant positive coefficient for the two deviation variables in all three models implies that noise around the attitude significantly increases with increasing deviation from societal values, when personal values and all other control variables are held constant.

Since, with the exception of age, all variables in this equation are coded to vary 0–1, their relative explanatory power on dependent variables is comparable. In both models, deviation from society’s self-transcendence values has either the strongest or the second strongest effect on the variance around immigration attitudes. In addition, all else being equal, deviation from self-transcendence values is a stronger explanatory variable of the error in immigration attitudes when compared with deviation from conservation values, although both effects are strong and significant.

We did not have any particular expectations for the effect of conservation and self-transcendence values on ambivalence towards immigration issues. Indeed, the effects of these variables are not consistent and systematic. First, holding conservation values to a greater extent has a positive but insignificant effect on attitudes towards immigrants, and generally decreases ambivalence in support of immigration and imposing less strict conditions. Second, valuing self-transcendence values to a greater extent decreases ambivalence in attitudes towards immigrants, has no effect on ambivalence in support for allowing immigration, and increases the noise in support of less strict conditions for immigration. Still, the key point is that regardless of a person’s level on these values, the absolute difference between her value level and societal values increases the error in her response on all three dependent variables, and thus indicates greater ambivalence in attitudes towards immigration and immigrants.

Individual-level controls

Several factors are controlled for in the variance equation. As expected, religiosity, which acts as a value system that supplies cognitive constraint, decreases ambivalence in two of the three models. Education decreases ambivalence in support for immigration and attitudes towards immigrants, and thus supports the hypothesis that, ceteris paribus, error variance decreases as cognitive skills increase. Similarly, higher income levels reduce ambivalence in two of the dependent variables, suggesting that less competition for material resources leads to more crystallized attitudes towards immigration. As expected, higher interpersonal trust decreases ambivalence in support for allowing immigrants and attitudes towards immigrants, but the variable has an unexpected positive coefficient in support for less strict conditions for immigration. Finally, personal relationships and contact with immigrants contribute to rich idiosyncratic experiences and increase the noise around immigration attitudes. Contrary to our expectations, the age of respondents did not have any systematic effect on ambivalence. Although we did not have any particular expectation for the effects of feeling safe, having a parent with an immigrant background, gender, and an ideology regarding ambivalence, we included them in our variance equations as controls. These variables also have inconsistent and unsystematic effects on the error variance.

Country-level fixed effects

The variance equations also provide controls for certain contextual variables. Results show that holding all else constant, both a stronger migration inflow and higher levels of unemployment in the country lead to more ambivalence in attitudes towards immigration in two of the three models. However, the stock of foreign-born population and the economic standing of a country show non-systematic or no effect on the error variance.

Robust analysis

We conducted a series of robust analyses. First, results were robust to model specification changes, such as the inclusion or omission of contextual and individual-level variables in the mean and variance equations. Second, results were replicated while running non-imputed models for the initial Heteroskedastic ML estimation. Finally, although Hypothesis 3 was fully supported, we also tested whether the variance in the model is in fact constant. Since the homoskedastic model is a reduced form of the heteroskedastic model, we used the likelihood ratio to compute this test. In all three models χ² test statistics from this likelihood ratio permit a rejection of the null hypothesis of homoskedasticity at the 5% significance level. This further confirms our claim about the influence of social–individual value conflict on the variance of immigration attitudes.

Conclusion

Political attitudes are formed in collective contexts and are to a great extent affected by the social environment. Yet, research on public opinion has largely neglected the crucial role played by social context in the construction of attitudes. As Mutz et al. (1996: 5) state, ‘more often than not, our topics of study and the methods we employ fail to take into systematic account the power of situations to influence political attitudes.’ However, social scientists have recently revived their interest in the effect of inter-personal interactions on attitude processes with the understanding that political attitudes cannot be explained by individual-level variables alone (see Levitan and Visser, 2009; Mutz, 2006; Visser and Mirabile, 2004). In addition, research has started to pay attention to the effect of institutional and cultural context on individual attitudes (Blekesaune and Quadagno, 2003; Green, 2009; Sides and Citrin, 2007; Weldon, 2006). This paper adds to these efforts by stressing the influence of societal values on the relationship between individual values and attitudes towards immigration.

At a time when the EU has overtaken the US as the world’s premier destination for people seeking a better life abroad (Theil, 2010), opposition to immigration is reflected in immigration laws (such as Italy’s 2009 immigration bill and laws undermining immigrants’ rights, and the constitutional amendment banning minaret construction in Switzerland), election outcomes (e.g. the recent success of PVV in the Dutch general elections), and European public opinion. To date, some studies have confirmed that values explain immigration-related attitudes (Davidov et al., 2008b; Sagiv and Schwartz, 1995). This paper’s contribution lies first and foremost in investigating the role of societal values in the ambivalence that surrounds attitudes towards immigration. We verified the hypothesis that when personal values are in conflict with collectively held values, personal values actually lead to less crystallized and more ambivalent attitudes towards immigration. When an individual’s values are at odds with the societal values, the latter provide countervailing positions, which push the attitude in a different direction, and thus lead to more unstable attitudes. The findings for three different dependent variables in a comparative dataset that includes observations from eighteen countries are robust. This strengthens the generalizability of our conclusions. Although our hypothesis is verified for both value dimensions, the effects are stronger for conservation values.

Another contribution of this paper lies in the discovery of a boundary condition for the effect of personal values on immigration attitudes. We find that the effect of personal values on attitudes towards immigration is at times contingent on the congruency of individual values with social values. As individuals deviate from societal values, and accordingly show increased attitudinal instability, they rely less on their values to construct their attitudes.

Our results provide empirical evidence for the importance of social pressure on the formation and nature of political attitudes. All individuals are exposed to various social influences through socialization, enculturation, and the media. Such influences may provide competing or congruent considerations depending on the initial predispositions of the individuals. Even if people may choose the media sources from which they get information, or the people with whom they interact, they are exposed to broader social influences and competing political values. All the same, this paper does not speak to the specific mechanisms by which people are exposed to societal values or to individual differences that make some people more susceptible to value conflict then others; we leave this path open to future research.

The findings also have implications for political behavior. Literature on the subject demonstrates several effects of ambivalence on political attitudes, including reduction in the predictability and stability of attitudes (e.g. Lavine, 2001; Zaller and Feldman, 1992), decrease in attitudinal extremity, declining confidence in judgments, improvement in the level of balance, or accuracy in political judgments (Lavine, 2001; Meffert et al., 2004; Zaller and Feldman, 1992). In addition, it has been shown that individuals embedded in attitudinally heterogeneous networks tend to experience ambivalence that leaves them more susceptible to persuasion appeals (Bassili, 1996; Visser and Mirabile, 2004), and prone to increased consumption and scrutiny of external information, such as news (Levitan and Visser, 2009; Scheufele et al., 2004). In this sense, a certain amount of tension with the societal status quo may allow for truly critical thinking, openness, self-doubt and information seeking – characteristics of the ideal citizen in a democracy.

Ultimately, our findings yield not just a greater theoretical understanding of how those around us influence the formation and maintenance of our attitudes, but also a fuller picture of the different ways in which social contexts affect political cognition and public opinion, not only by strictly changing people’s opinions, but also by affecting their confidence in their own attitude, their attitude stability, and their reliance on their own values when forming a political opinion.

Footnotes

Acknowledgements

An earlier version of this paper was presented at the Annual Meeting of the Midwest Political Science Association, 22–25 April 2010. We would like to thank panel discussant Elizabeth A. Suhay for her helpful comments and criticisms. All remaining errors are our own.

Notes

Author Biographies

Gizem Arikan is Assistant Professor in the Department of International Relations at Yasar University, Izmir, Turkey. Her research focuses on the effect of political culture and values on public opinion and policy. She also conducts research on religiosity and attitudes towards democracy.

Pazit Ben-Nun Bloom is Assistant Professor in the Department of Political Science at The Hebrew University of Jerusalem, Israel. Her research examines the role of morality, values and religiousness in political behavior.

References

Alvarez

Brehm

(1995) American ambivalence toward abortion policy: development of a heteroskedastic probit model of competing values. American Journal of Political Science 39(4): 1055–1082.

Alvarez

Brehm

(1997) Are Americans ambivalent towards racial policies? American Journal of Political Science 92(2): 345–374.

Alvarez

Brehm

(2002) Hard Choices, Easy Answers. Princeton, NJ: Princeton University Press.

Barsalou

(1987) The instability of graded structure: implications for the nature of concepts. In Neisser

(ed.) Concepts and Conceptual Development: Ecological and Intellectual Factors in Categorization. Cambridge: Cambridge University Press, 101–140.

Basinger

Lavine

(2005) Ambivalence, information, and electoral choice. American Political Science Review 99(2), 169–184.

Bassili

(1996) Meta-judgmental versus operative indices of psychological properties: the case of measures of altitude strength. Journal of Personality and Social Psychology 71: 637–653.

Bellah

Madsen

Sullivan

Swidler

Tipton

(1995) Habits of the Heart; Individualism and Commitment in American Life. Berkeley CA: University of California Press.

Ben-Nun Bloom

Arikan

(2011) A two-edged sword: the differential effect of religious belief and religious social context on attitudes towards democracy. Political Psychology 32(4): 643–665.

Ben-Nun Bloom

Levitan

(forthcoming) We’re closer than I thought: social network heterogeneity, morality and political persuasion. Political Psychology.

10.

Ben-Nun Bloom

Zemach

Arian

(2011) The religious experience as affecting ambivalence: the case of democratic performance evaluation in Israel. Democratization 18(1): 25–51.

11.

Benito

(2007) Sweden. In: Triandafyllidou

Gropas

(eds) European Immigration: A Sourcebook. Aldershot: Ashgate, 335–346.

12.

Berelson

Lazarsfeld

McPhee

(1954) Voting: A Study of Opinion Formation in a Presidential Campaign. Chicago, IL: Universuity of Chicago Press.

13.

Blekesaune

Quadagno

(2003) Public attitudes towards welfare state policies: a comparative analysis of 24 nations. European Sociological Review 19: 415–427.

14.

Cacioppo

Berntson

(1994) Relationship between attitudes and evaluative space: a critical review, with emphasis on separability of positive and negative sub-strates. Psychological Bulletin 115(3): 401–423.

15.

Cotesta

(1999) Mass media, ethnic conflicts and migration: a research on Italian newspapers in the nineties. Studi Emigrazione 135: 387–499.

16.

Davidov

Meuleman

Billiet

Schmidt

(2008a) Values and support for immigration: a cross-country comparison. European Sociological Review 24(5): 583–599.

17.

Davidov

Schmidt

Schwartz

(2008b) Bringing values back in: the adequacy of the European Social Survey to measure values in 20 countries. Public Opinion Quarterly 72(3): 420–455.

18.

DiMaggio

(1997) Culture and cognition. Annual Review of Sociology 23: 263–287.

19.

Eagly

Chaiken

(1993) The Psychology of Attitudes. New York: Harcourt Brace Jovanovich.

20.

Feldman

(1995) Answering survey questions: the measurement and meaning of public opinion. In: Lodge

McGraw

(eds) Political Judgment: Structure and Process. Ann Arbor, MI: University of Michigan Press, 249–270.

21.

Feldman

(2003) Values, ideology, and the structure of political attitudes. In: Sears

Huddy

Jervis

(eds) Oxford Handbook of Political Psychology. New York: Oxford University Press, 477–509.

22.

Feldman

Zaller

(1992) The political culture of ambivalence: ideological responses to the welfare state. American Journal of Political Science 36(1): 268–307.

23.

Festinger

(1954) A theory of social comparison processes. Human Relations 7(2): 117–140.

24.

Franklin

(1991) Eschewing obfuscation? Campaigns and the perception of US Senate candidates. American Political Science Review 85(4): 1193–1214.

25.

Gamson

(1992) Talking Politics. New York: Cambridge University Press.

26.

Green

(2009) Who can enter? A multilevel analysis on public support for immigration criteria across 20 European countries. Group Processes and Intergroup Relations 12(1): 41–60.

27.

Hagelund

(2002) Problematizing culture: discourses on integration in Norway. Journal of International Migration and Integration 3(3/4): 401–415.

28.

Hertog

McLeod

(2001) A multiperspectival approach to framing analysis: a field guide. In: Reese

Gandy

Grant

(eds) Framing Public Life: Perspectives on Media and Our Understanding of the Social World. New York: Taylor & Francis, 139–162.

29.

Hofstede

(1980) Culture’s Consequences: International Differences in Work-related Values. London: Sage.

30.

Huckfeldt

Johnson

Sprague

(2004) Political Disagreement: The Survival of Diverse Opinions Within Communication Networks. New York: Cambridge University Press.

31.

Huckfeldt

Mendez

Osborn

(2004) Disagreement, ambivalence, and engagement: the political consequences of heterogeneous networks. Political Psychology 25(1): 65–95.

32.

Iyengar

(1991) Is Anyone Responsible? Chicago, IL: University of Chicago Press.

33.

Johnston

Klandermans

(eds) (1995) Social Movements and Culture. Minneapolis, MN: UCL Press.

34.

Keele

Wolak

(2006) Value conflict and volatility in party identification. British Journal of Political Science 36(4): 671–690.

35.

Kelley

(1952) Two functions of reference groups. In: Swanson

Neewcomb

Hartley

(eds) Readings in Social Psychology. New York: Holt, 410–414.

36.

Lahav

(2004a) Immigration and Politics in the New Europe: Reinventing Borders. Cambridge: Cambridge University Press.

37.

Lahav

(2004b) Public opinion toward immigration in the European Union: does it matter? Comparative Political Studies 37(4): 1151–1183.

38.

Lavine

(2001) The electoral consequences of ambivalence toward presidential candidates. American Journal of Political Science 45(4): 915–929.

39.

Lazarsfeld

Berelson

Gaudet

(1944) The People’s Choice: How the Voter Makes Up His Mind in a Presidential Campaign. New York: Columbia University Press.

40.

Levitan

Visser

(2009) Social network composition and attitude strength: exploring the dynamics within newly formed social networks. Journal of Experimental Social Psychology 44(3): 640–649.

41.

McGraw

Hasecke

Conger

(2003) Ambivalence, uncertainty, and processes of candidate evaluation. Political Psychology 24(3): 421–448.

42.

Meffert

Guge

Lodge

(2004) Good, bad, and ambivalent: the consequences of multidimensional political attitudes. In: Saris

Sniderman

(eds) Studies in Public Opinion: Attitudes, Non-attitudes, Measurement Error, and Change. Princeton, NJ: Princeton University Press, 63–92.

43.

Muthén

(2006) Mplus user’s guide. Los Angeles: Author.

44.

Mutz

(2006) Hearing the Other Side: Deliberative Versus Participatory Democracy. New York: Cambridge University Press.

45.

Mutz

Sniderman

Brody

(eds) (1996) Political Persuasion and Attitude Change. Ann Arbor, MI: University of Michigan Press.

46.

Pavlou

(2001) The ‘smugglers of fear’: racist discourse and immigrants in the press of a potential metropolis. In: Marvakis

Parsanoglou

Pavlou

(eds) Immigrants in Greece. Athens: Ellinika Grammata, 127–162.

47.

Reese

(2001) Framing public life: a bridging model for media research. In: Reese

Gandy

Grant

(eds) Framing Public Life: Perspectives on Media and Our Understanding of the Social World. New York: Taylor & Francis, 7–32.

48.

Reise

Widaman

Pugh

(1993) Confirmatory factor analysis and item response theory: two approaches for exploring measurement invariance. Psychological Bulletin 114(3): 552–566.

49.

Roggeband

Vliegenhart

(2007) Divergent framing: the evolution of the public debate on migration and integration in the Dutch parliament and media, 1995–2004. West European Politics 30(3) 524–548.

50.

Rudolph

(2005) Group attachment and the reduction of value-driven ambivalence. Political Psychology 26(6): 905–928.

51.

Sagiv

Schwartz

(1995) Value priorities and readiness for out-group social contact. Journal of Personality and Social Psychology 69: 437–448.

52.

Scheufele

Nisbet

Brossard

Nisbet

(2004) Social structure and citizenship: examining the impacts of social setting, network heterogeneity, and informational variables on political participation. Political Communication 21(3): 315–338.

53.

Schwartz

(1992) Universals in the content and structure of values: theoretical advances and empirical tests in 20 countries. In: Zanna

(ed.) Advances in Experimental Social Psychology. New York: Academic Press, 1–65.

54.

Schwartz

(2004) Mapping and interpreting cultural differences around the world. In: Vinken

Soeters

Ester

(eds) Comparing Cultures: Dimensions of Culture in a Comparative Perspective. Leiden: Brill, 43–73.

55.

Schwartz

(2007) Universalism values and the inclusiveness of our moral universe. Journal of Cross-Cultural Psychology 38(6): 711–728.

56.

Semetko

Mandelli

(1997) Setting the agenda for cross-national research: bringing values into the concept. In: McCombs

Shaw

Weaver

(eds) Communication and Democracy: Exploring the Intellectual Frontiers in Agenda-setting Theory. Mahwah, NJ: Lawrence Erlbaum Associates.

57.

Sides

Citrin

(2007) European opinion about immigration: the role of identities, interests and information. British Journal of Political Science 37(4): 477–504.

58.

Smith

Bond

Kagitcibasi

(2006) Understanding Social Psychology Across Cultures: Living and Working in a Changing World. Thousand Oaks, CA: Sage.

59.

Steenbergen

Brewer

(2004) The not-so-ambivalent public: policy attitudes in the political culture of ambivalence. In: Saris

Sniderman

(eds) Studies in Public Opinion: Attitudes, Non-attitudes, Measurement Error, and Change. Princeton, NJ: Princeton University Press, 93–129.

60.

Theil

(2010) Europe’s big choice. Newsweek, 19 February.

61.

Triandafyllidou

Veikou

(2002) The hierarchy of Greekness. Ethnicities 2(2): 189–208.

62.

Van de Vijver

FJR

Leung

(1997) Methods and Data-analysis for Cross-cultural Research. Thousand Oaks, CA: Sage.

63.

Visser

Mirabile

(2004) Attitudes in the social context: the impact of social network composition on individual-level attitude strength. Journal of Personality and Social Psychology 87: 779–795.

64.

Weldon

(2006) The institutional context of tolerance for ethnic minorities: a comparative, multilevel analysis of Western Europe. American Journal of Political Science 50(2): 331–349.

65.

Zaller

(1992) The Nature and Origins of Mass Opinion. Cambridge: Cambridge University Press.

66.

Zaller

Feldman

(1992) A simple theory of survey response; answering questions versus revealing preferences. American Journal of Political Science 36(3): 579–616.