Everyone knows that education makes people more tolerant, right? Well, yes, if you focus on the traditional targets of intolerance that are generally studied in the social sciences, such as members of ethnic, racial, or religious outgroups. A college education seems to make people more cosmopolitan and less prejudiced. But what if you look at tolerance of ideological outgroups? New research by P.J. Henry and Jaime Napier finds that the education effect reverses. College graduates on both sides of the political spectrum are less tolerant of their political opponents than are people who have less education. In this post I summarize that article and draw out its implications for the current debate about free speech and the intellectual climate on college campuses.

Full Reference: Henry, P.J. & Napier, J.L. (2017). Education is related to greater ideological prejudice. Public Opinion Quarterly, 81, 930-942.

Abstract: Decades of research have shown that education reduces individuals’ prejudices toward people who belong to different groups, but this research has focused predominantly on prejudice toward ethnic/racial groups, immigrant groups, and general nonconformists. However, it is not clear whether education reduces other prejudices against groups along different dimensions, including ideological identification. An analysis of American National Election Studies data from 1964 to 2012 shows that education is related to decreases in interethnic/interracial prejudice, but also to increases in ideological (liberal vs. conservative) prejudice. This finding could not be explained simply by the greater polarization of the American electorate in the past twenty years. The results require rethinking how and why education is associated with reduced prejudice for certain groups but not others.

Introduction

Although increased education has long been associated with more tolerant attitudes towards historically underrepresented and disadvantaged groups (e.g., racial and ethnic minorities, immigrants, religious minorities, sexual minorities) there is good reason to suspect that increased education will produce greater ideological prejudice, and thus less tolerant attitudes towards one’s political opponents. Henry and Napier (2017) briefly review literature on the positive relationship between increased education and political sophistication (see e.g., Highton, 2009; Jacoby, 1988; Sidanius, Levin, van Laar, & Sears, 2008), before turning to the literature linking political sophistication with ideological bias (seen e.g., Bartels, 2008; Taber & Lodge, 2006; Zaller, 2004).

Henry and Napier (2017, p. 932) then explain how increased education may increase ideological prejudice. First, they note that political orientation is closely linked to one’s basic morals and values (see e.g., Graham, Haidt, & Nosek, 2009):

… one consequence of education may be to better understand how morals and values connect to political ideologies. Left- versus right-wing political philosophies, by their very description, stand in contrast to each other and aim for different societal ideals; moreover, liberals and conservatives are not marginalized minority groups but rather large groups that pose plausible symbolic threats to their political counterparts. Because the value threat that the outgroup represents may be more obvious with increased education, and because such symbolic threats are intertwined with evaluations and emotions (Sears, 1993), the more educated may therefore express stronger negative prejudices towards their ideological opponents than the less educated.

Education level was, therefore, hypothesized to have a positive association with ideological prejudice. Henry and Napier (2017) tested this hypothesis by using data from the American National Election Studies (ANES). Since the ANES has collected data in every American election cycle since 1948, Henry and Napier were able to test their hypothesis in multiple decades and additionally predicted that the link between education level and ideological prejudice was not a new phenomenon.

Method

Data was obtained from the ANES. The questions relevant to Henry and Napier’s (2017) data analysis were included in all samples collected from 1972 to 2012. An individual’s political ideology was determined from the ANES 7-item self-assessment of ideology (“extremely liberal,” “liberal,” “slightly liberal,” “moderate, middle-of-the-road,” “slightly conservative,” conservative,” and “extremely conservative”). Education was categorized into one of four levels (“less than high school,” “high school”, “some college,” and “college graduate”). The “college graduate” level was a combination of the 4-year college graduate and graduate/advanced degree categories in the ANES.

Group feeling thermometers that ranged from 0 (very cold and unfavorable) to 100 (very favorable and warm) provided the variables to assess ideological prejudice. A difference score between feelings towards liberals and feelings towards conservatives was created (i.e., feeling thermometer rating for liberals minus feeling thermometer rating for conservatives). Finally, a number of covariates were also included in the analysis: respondent sex (female or male), age, household income [1], and census region.

Results

A preliminary analysis revealed that Henry and Napier’s (2017) findings replicated previous work documenting that more education is associated with a more liberal self-identification, however, and again consistent with prior research (see e.g., Sullivan, Pierson, & Marcus, 1982), this effect was not particularly strong. When demographic variables were controlled for, college graduates were more liberal than others, but there was little association between education and ideology among non-college graduates.

The focal hypothesis was tested with a multilevel model with a random intercept predicting the favorability of liberals over conservatives, as measured by the difference score created from the differences in the feeling thermometer responses. Results revealed that, as predicted increases in education for both liberals and conservatives were associated with increased prejudice. Analyses of the simple effects showed that each increase in education corresponded to more prejudice for both liberals and conservatives. In other words, liberals who did not complete high school were less ideologically prejudiced than those who did; liberals with a high school diploma only were less prejudiced that those with some college experience; and liberals with some college experience were less ideologically prejudiced than those who completed college. This pattern was also evident among conservatives.

Henry and Napier (2017, p. 936-937) further investigated if the association between education and ideological prejudice may have emerged as a result of the sharp rise in polarization between liberals and conservatives in the United States in recent years (see e.g., Iyengar, Sood, & Lelkes, 2012; Napier & Luguri, 2016; Pew Research Center, 2017):

Results showed a small but significant main effect of year (b = .16, SE = .04, p < .001) that was qualified by an interaction with ideology (b = -.07, SE = .02, p < .001). Analyses of the simple slopes showed that ingroup favoritism increased among liberals over time (b = .26, SE = .05, p < .001), but not among conservatives (b = .05, SE = .05, p = .285). As can be seen in figure 1, despite the main effect of time for liberals, the relationship between education and ideological prejudice was very stable over time for both liberals and conservatives.

Thus, although ideological prejudice has increased over time among liberals, the pattern of association between education and ideological prejudice has remained fairly stable among liberals and conservatives.

Discussion

Henry and Napier (2017) demonstrated clear and consistent support for their hypothesis – that greater levels of education are associated with greater levels of ideological prejudice – in their analysis of ANES data obtained from 1972 to 2012. Perhaps most importantly (Henry & Napier, 2017, p. 937, emphasis added):

… in contrast to interethnic bias, which is negatively related to education, ideological prejudice is stronger among those with high (vs. low) levels of education. These results call into question the notion that education promotes tolerance towards those who are different.

To explain this contrary finding, Henry and Napier (2017) offer multiple possibilities for future research:

• Education promotes the creation of more crystalized ideological belief systems (e.g., Jacoby, 1988) that are more closely linked to moral values. Political opponents who endorse different, and, possibly, opposing values may, therefore, pose more of a symbolic threat to those with more education.
• The social status of the groups being judged (liberals and conservatives) differs from the vast majority of research on the association between education and prejudice, in that neither group is marginalized. There is now good reason to suspect similar relationships between education and prejudice may exist between groups that are elective, that are defined by values and ideas, and do not have obvious status differentials.
• The current study could not test whether greater ideological prejudice, as measured by a feeling thermometer, produces more discriminatory behavior towards political opponents.

Thus, the results suggest that the long-held conclusion “…that education endows a type of tolerance toward different others is bounded by dimensions of difference” (p. 940) may require a reassessment. There are clear links between more education and tolerant attitudes for some people and groups, but there is now evidence that more education can also exacerbate intolerant attitudes towards some people and groups.

Why These Findings are Important

These findings may be the first that document a link between higher levels of education and greater prejudice. Such findings are consistent with the suggestion that the ideological conflict in the United States is particularly bitter at the more elite levels of society (see Gelman, 2008). Additionally, some recent analyses on support for free speech among college students and college graduates have emphasized the positive association between higher levels of education and greater political tolerance. These analyses have also argued that liberals are more politically tolerant than conservatives (see here and here). Other analyses have argued that current college students are not as supportive of free speech as college graduates (see here for some evidence that this is not so for those who were 18-24 years old in 2016; see here for some evidence that is not so for current college students) and that liberals and conservatives and largely equally intolerant of each other (see e.g., Brandt, 2017; Gibson, 2006; Sullivan et al., 1982). The findings of Henry and Napier (2017) lend more support to the latter analyses, compared to the former.

Footnotes

1. Income group was not recorded in the 2002 ANES, so Henry and Napier (2017) set each 2002 respondent’s income to the grand mean. Patterns of results and significance levels did not differ based on the inclusion or exclusion of the 2002 respondents from the analysis.