Saturday, March 3, 2018

Authoritarianism, Affective Polarization, and Economic Ideology (with commentary) 1 of 2

Authoritarianism, Affective Polarization, and Economic Ideology

Christopher D. Johnston
Advances in Political Psychology
Volume 39, Issue Supplement S1
February 2018 

DOI: 10.1111/pops.12483


I consider two theories of affective polarization between Democrats and Republicans in the United States: (1) ideological divergence on size-of-government issues (Webster & Abramowitz, 2017) and (2) authoritarianism-based partisan sorting (Hetherington & Weiler, 2009). I argue that these alternatives cannot be easily disentangled, because politically engaged citizens seek out and assimilate information about economic policy from elites who are perceived to share their core traits and cultural values. In this way, the economic preferences emphasized by the first view are partly endogenous to the worldview divide emphasized by the second. Elite position taking on economic issues may elicit strong emotions among citizens because it reliably signals a commitment to one worldview or the other. I review new and existing evidence for this claim in both observational survey data and two experimental studies. I also consider the broader implications of these results for the distribution of economic opinion across indicators of human capital.


There is little doubt that Democrats and Republicans feel more negatively toward each other now than in the past (Abramowitz & Webster, 2016; Iyengar, Sood, & Lelkes, 2012; Iyengar & Westwood, 2014; Webster & Abramowitz, 2017), but debate continues over the causes and implications of “affective polarization.” A central point of disagreement concerns the relationship of affect to policy preferences and ideology. Some scholars argue that ideological distance between Democrats and Republicans has increased substantially over the last few decades (Abramowitz, 2010; Abramowitz & Saunders, 1998, 2005; Bafumi & Shapiro, 2009), and affective polarization is the result of sharp, principled disagreements over policy (Rogowski & Sutherland, 2016), especially on size-of-government issues related to redistribution [of wealth or resources] and social welfare (Webster & Abramowitz, 2017).

Others contend that affective polarization is rooted in perceptions of social rather than ideological distance, and partisans remain relatively moderate in their policy preferences—that is, they “agree disagreeably” (Mason, 2015a). There are at least two strains of this argument. The first focuses on the increasing alignment of various political and social identities. In this view, partisan identities have strengthened over time as cross-cutting social cleavages have become less common [in no small part owing to the downturn in union membership and political involvement; union members so often being socially conservative albeit economically liberal] and various group memberships have fused into a single left-right divide (Mason, 2015a, 2016). Affective reactions to the opposing party are stronger because the stakes of intergroup competition have been raised, not because partisans are more extreme on the issues (Huddy, Mason, & Aaröe, 2015). The second perspective argues that affective polarization is driven by the association of partisanship with the personality dimension of “authoritarianism” (Hetherington & Weiler, 2009). Intense emotions, in this view, arise from the perception that one's opponents are fundamentally different kinds of people. As Hetherington and Weiler (2009) suggest, “People come to perceive that their views of right and wrong and good and bad are diametrically opposed to those of their opponents, making it difficult to understand (or perhaps even respect) the worldview that makes those preferences possible” (p. 17).

In the present article, I wish to focus on the conflict between two of these views on affective polarization: (1) that it is rooted in sharp ideological disagreements on size-of-government issues (Webster & Abramowitz, 2017), and (2) that it is rooted in differences of personality and cultural worldview (Hetherington & Weiler, 2009). My argument is that these two perspectives can, to some degree, be reconciled because preferences on economic issues are infused with the same personality divide that structures preferences on “gut-level” issues such as gay marriage and terrorism. Economic issues have the capacity to evoke intense emotions because competing positions represent, or stand for, this deeper conflict. I put aside for the time being the work of Mason and colleagues on social sorting and identity strength, though I believe it is quite important. Also, while I focus on authoritarianism, my arguments are intended to apply to other related traits and dispositions (Jost, Glaser, Kruglanski, & Sulloway, 2003). Similar results to those presented here, using a variety of alternative measures of personality, along with a discussion of their conceptual overlap, can be found in Johnston, Lavine, and Federico (2017; see also Federico & Malka, 2018).

Authoritarianism, Affective Polarization, and Economic Ideology

Within political psychology, authoritarianism is typically conceptualized as a prioritization of social order and security that manifests in conformity, traditionalism, submission to authority, and aggression toward “deviants” and outgroups (Adorno, Frenkel-Brunswik, Levinson, & Sanford, 1950; Altemeyer, 1981, 1988; Feldman, 2003; Feldman & Stenner, 1997; Stenner, 2005). Individual differences in authoritarianism are thought to arise from differences in cognitive style and epistemic and existential needs (e.g., Duckitt, 2001; Hetherington & Weiler, 2009; Jost et al., 2003; Kruglanski, Pierro, Mannetti, & De Grada, 2006; Lavine, Lodge, Polichak, & Taber, 2002; Oxley et al., 2008). Relative to low scorers, individuals who score high on measures of authoritarianism are more attuned to potential threats and negative outcomes, find uncertainty and ambiguity aversive, and tend to see the world in sharply categorical (“black and white”) terms. In turn, they seek to limit the scope of individual freedom and various forms of diversity to maintain an ordered, predictable, and secure environment. In contrast, those who score low tend to focus on achieving gains and positive outcomes and are more comfortable with uncertainty, ambiguity, and fuzzy boundaries. In turn, they are attracted to diversity, draw weaker distinctions between in- and outgroups and tend to be skeptical of traditional values and sources of authority. [Just don't get the notion that you -- or anyone else --  will ever be able to explain any concept as complex and formal operational as this to concrete operationals who were stunted by parental and other denigration of their burgeoning intellects when they were children in Erikson's "initiative" and "competence" formation stages. These people are simplistic and come from peer-pressured social proof that anything as ostensibly complex as what they were abused for not being able to grasp as children is untrustworthy and affectively intolerable. It's very easy to see now how their "identities" coalesced around concrete operational, all-or-nothing, all-good-or-all-evil, black-&-white cognition. AND... that the future of mankind looks pretty dismal as a direct result. As Wright's Darwinian propagation of the species thesis suggests, our genetic heritage works as much against progress as for it. The "what is" that one who can see all this is forced to accept is that most humans are still animals dominated by their lower limbic systems, a reality that is not going to change in this generation, nor, perhaps in any generation of humans Before the Deluge.]

Consistent with this conceptualization, Hetherington and colleagues (Hetherington & Suhay, 2011; Hetherington & Weiler, 2009) demonstrate that authoritarianism strongly shapes citizens' preferences on a variety of issues concerned with these trade-offs, such as gay rights, immigration, and the war on terrorism. Further, as these issues have become more important to partisan branding, authoritarianism has become tightly linked to partisanship in the United States (see also Cizmar, Layman, McTague, Pearson-Merkowitz, & Spivey, 2014). This helps explain why politics in the twenty-first century seems more vitriolic than in the past: Partisans are divided by things that are easy to conceptualize, emotionally evocative, and which underpin one's broader lifestyle choices, all of which increases the perceived social distance of one's opponents and promotes bias against them (Iyengar et al., 2012; Iyengar & Westwood, 2014).

Empirically, authoritarianism is indeed closely tied to affective polarization in contemporary politics. Consider the top panel of Figure 1.1 Here I plot the kernel densities of relative party affect (in 2016) for non-Black, politically engaged citizens in the top and bottom terciles of authoritarianism. I exclude Black and African-American identifiers, as they overwhelmingly identify with the Democratic Party, leaving little room for other factors to operate.2 I focus on politically engaged citizens because engagement is central to the ideological divergence perspective on affective polarization (Abramowitz, 2010). Moreover, as I will demonstrate below, engagement has a large moderating effect on the influence of authoritarianism on political attitudes (Federico & Tagar, 2014; Johnston et al., 2017). The divide is stark: Those in the top tercile feel much “warmer” toward the Republican Party than the Democratic Party, while those in the bottom tercile show the opposite pattern. Among these respondents, the correlation between authoritarianism and affect is a substantial .42.

Figure 1.
Party affect and economic policy conservatism across authoritarianism among politically engaged citizens. Data is from the 2016 ANES. Relative party affect is measured on a −100 to 100 scale. Economic conservatism is measured in standard deviation units. “Low” authoritarianism represents respondents in the bottom tercile of authoritarianism, and “high” represents those in the top tercile. The figures include non-Black respondents in the top quartile of political engagement.

My claim is that this authoritarian divide between engaged Democrats and Republicans has important implications for values and policy preferences on size-of-government issues and thus for the structure of mass ideology comparing the engaged to the unengaged. As my colleagues and I have argued elsewhere, engaged citizens seek out and assimilate information about economic issues from elite actors who are perceived to share their traits and cultural worldview (Johnston et al., 2017; Johnston & Wronski, 2015). There are multiple reasons for this.

First, economic issues are often “hard”: means-oriented, technical, and lacking the emotion-laden symbols of cultural issues (Carmines & Stimson, 1980). In turn, citizens wish to delegate opinion formation to elite actors, but “choosing whom to believe” is itself a difficult problem—it is not always obvious who shares one's interests and who is seeking to manipulate opinion for their own gain (Lupia & McCubbins, 1998). One strategy is to listen to elites who share readily observable traits and values—what Caprara and Zimbardo (2004) call a “‘dispositional heuristic’ that anchors impressions, as well as inferences and voters' beliefs, to those traits used to describe oneself and other people… . The more apparent perceived similarity with their candidates, the more positively they are perceived” (p. 587).

[But the real, get-up-out-of-the-box-of-the "consensus trance" issue here is actually less a matter of which beliefs than it is belief, itself. Belief is -- after all -- not the result of empirical observation. Belief is no more than conditioned, in-doctrine-ated, in-struct-ed, socialized, habituated, normalized and institionalized repetition of verbal concepts that (in spectral or continuum organization) may or may not have much relationship to what actually is so at any given moment. And though (according to this and other research, including that of Hetherington, et al) they tend to be more empirical than belief bound, liberal followers are only marginally more empirical than their conservative counterparts.

Ever since James Buchanan's arrival at the University of Virginia in the 1950s (see J. Mayer, and N. MacLean, both 2017), the supposedly "conservative cynicism" of the 1830s (around slavery as the bedrock of wealth acquisition in the American south), 1880s (around "manifest destiny" for wealth accumulation during the "gilded age") and 1920s (around stock peculation and manipulation for wealth accumulation) has found increasing re-empowerment via the use of egregiously corrupted scientific empiricism and pseudo-empiricism.

I can only speculate, but I seriously doubt that Ronald Reagan -- thought not his spouse -- had any idea at all that the "neo-Libertarians" were taking over the Republican Party in the 1980s. But by the Newt Gingrich era in the 1990s, cynical wealth accumulation was using "empirical science" to equate the liberal social welfare initiatives of the 1930s, 1960s and early 1970s with socialism and communism. And with generous funding from such as the Koch brothers and Sciafe family, able to begin to sell their "science" to millions of dispossessed rust-belt whites in racist and xenophobic packaging.

The Bush family benefited, of course. But the Trump phenomenon is the obvious  demonstration of the "Law of Unintended Consequences" in 2018. Because they are easily manipulated believers -- rather than empirical observers -- the concrete operational, belief-bound, and simple minded tens of thousands of formerly "liberal" union Democrats in the rust belt gave the cynically manipulative neo-Libertarians a shot at major power.

But the elected agent of the neo-Libertarian movement is only a vice president. And the movement will have to find a way to get rid of the seemingly contradictory or uncertain fellow at the desk in the Oval Office before they can get on with the sort of sea change that is well beyond what only occurs when the radicals of one party gain control over all four branches of federal government. Because in the present circumstance, the Republican Party is itself no longer the party of the Bushs, Ronald Reagan, Gerald Ford, Dwight Eisenhower or even Herbert Hoover. It is the party of the extreme, demi-feudal and possibly even demi-facist elitism of the "manifest destiny" presidents of the "fat cat gilded age" in the late 19th century.]  

Authoritarianism may structure economic preferences indirectly by shaping [the believers'] evaluations of elite actors who signal their worldview through position taking on cultural or lifestyle issues.3Thus, citizens may genuinely try to form opinions that match their economic interests, but the dispositional heuristic they use to delegate may be poorly suited to the task.

Second, citizens may avoid taking policy positions associated with people and groups with opposing personality traits and cultural values. As Kahan and colleagues have argued (see Kahan, 2015 for a review), there are both psychological and social reasons for conforming policy positions to those who share one's broader cultural worldview, and these may trump instrumental considerations of what effects a policy will have for the self or society. First, endorsing policies associated with disliked groups causes internal psychological conflict. Citizens will be motivated to avoid and resolve this imbalance, and peripheral policy attitudes are likely to give way to more central predispositions (Heider, 1958). Indeed, as Tesler (2015) argues, basic values and attitudes toward social groups tend to be primed by political conflict, while policy positions tend to change to conform to these more stable predispositions. For example, opinions on health insurance reform changed to match one's racial attitudes during the Affordable Care Act debates. As authoritarianism is generally considered a [belief-and-value-driven] predisposition in the mold of other core traits and values (e.g., Feldman, 2003), consistency pressures should lead economic policy opinions to change to match the positions taken by those who share one's authoritarian or non-authoritarian values.

Second, there can be substantial social costs—such as loss in reputation—for taking political positions that deviate sharply from those within one's social network. Citizens will seek to conform their opinions on salient issues to the positions they expect to be held by others in their social milieu, especially when they can (or must) do so in a visible way (Kahan, 2015; Sinclair, 2012). To the extent they are likely to spend time with people who share their core traits and values, personality-related cues at the elite level can help citizens predict where members of their social network are likely to fall on a given issue and thus reduce the probability of making costly errors. In this sense, dispositional heuristics can help to coordinate opinions and reduce tension within social groups when issues are complex or obscure.

The upshot is that authoritarianism is likely to structure information seeking about economic policy. Due to the current structure of party conflict—in which Republicans and conservatives tend to send cues congenial to those whose who score high in authoritarianism, while Democrats and liberals send cues congenial to those who score low—high and low authoritarians are likely to end up with right- and left-wing views on economic issues, respectively. However, this should be highly conditional on exposure to elite cues. That is, there should be a strong, positive interaction between political engagement and authoritarianism in predicting economic opinion in contemporary American politics (Johnston et al., 2017; see Malka, Soto, Inzlicht, & Lelkes, 2014 for evidence supporting this hypothesis in other contexts that have a similar structure to the United States).

Consider the bottom panel of Figure 1. Here I plot the kernel densities of economic policy conservatism in the 2016 ANES for the same groups as the top panel.4 The divide in party feelings is mirrored in a less sharp, but still substantial, division in economic preferences. Engaged citizens who score low in authoritarianism are quite liberal, while citizens who score high in authoritarianism have a more widely dispersed set of preferences overall and a much higher probability of holding conservative views.5 Authoritarianism appears to pull the distribution of economic ideology to the right. Indeed, the correlation between authoritarianism and economic preferences for these citizens is .36. For comparison, the correlation with income is only .02. This simple demonstration illustrates that the two broad frameworks for understanding affective polarization discussed above—personality and polarization on size-of-government issues—cannot be cleanly separated. The economic beliefs and preferences of politically active citizens are deeply intertwined with basic traits and cultural values related to authoritarianism. In part, polarization on size-of-government issues appears driven by authoritarianism.

Empirical Tests in Cross-Sectional Survey Data

We can test these claims in a more rigorous fashion. I consider three dependent variables in the 2016 American National Election Study: relative party affect, the economic policy conservatism scale examined above, and an additional “economic values” scale constructed from three forced-choice items measuring support for limited government (see the appendix for items used to construct all variables). I regressed each of these three variables on authoritarianism, political engagement, the interaction of authoritarianism and engagement, and a set of controls including age, gender, Hispanic or Latino identification, educational attainment, household income, verbal ability,6 marital status, number of children in the household, employment status, household union membership, and southern residency. Regression tables are included in the appendix, and I focus on key estimates in the figures below. As above, I consider only non-Black respondents for all models.
Figure 2 shows the relationship between authoritarianism and relative party affect across levels of political engagement. Consistent with Johnston et al. (2017) and Federico and Tagar (2014), there is a strong and significant interaction. At low levels of engagement (5th percentile), there is little change in party affect across levels of authoritarianism, and respondents are, on average, relatively neutral in their feelings about the two parties. At the 50th percentile of engagement, however, there is a strong and significant relationship7, such that high authoritarians show relative warmth toward the Republican Party, and low authoritarians show relative warmth toward the Democrats. The marginal effect for authoritarianism, for a change from minimum to maximum, is 42 points on a scale that ranges from −100 to 100. The marginal effect for highly engaged citizens is striking. A change from low to high authoritarianism is associated with a change in relative party affect of 73 points.

Overall, this simple analysis provides strong evidence that affective polarization is tightly linked to authoritarianism among politically active citizens. Importantly, however, among inattentive citizens, there is no relationship of personality to affect. As we will see, this has important implications for how authoritarianism structures preference formation on economic issues.

Figure 2.
Party affect and authoritarianism across political engagement. Data is from the 2016 ANES. Party affect ranges from −100 to 100.

Figure 3 shows the relationship between authoritarianism and economic preferences for each of the two measures of the latter (policy and values). As shown in Panel A, and consistent with the descriptive findings above, authoritarianism is strongly related to economic policy conservatism among politically engaged citizens. At the median value of engagement, a change in authoritarianism from low to high is associated with an increase in conservatism of .60 standard deviations. At the 95th percentile of engagement, the relationship is very strong: A change in authoritarianism from minimum to maximum is associated with a 1.28 standard deviation increase in economic conservatism. No other predictor in the model even approaches the magnitude of this latter relationship. The next best predictor of policy conservatism is household income, for which a change from minimum to maximum is associated with a .40 standard deviation increase. The effect size for income is thus smaller than the effect of authoritarianism at even median values of engagement.8

Figure 3.
Economic preferences and authoritarianism. Data is from the 2016 ANES.

At the lowest levels of engagement, by contrast, the relationship of authoritarianism to economic conservatism is statistically significant, but in the exact opposite direction. That is, among unengaged citizens, authoritarianism is associated with support for redistribution, social insurance, and market regulation. Johnston et al. (2017) argue that politically unengaged citizens are neither concerned with, nor attentive to, party politics and thus largely ignore the cultural conflicts that drive engaged citizens to dislike the opposing party with such intensity. This argument is consistent with the minimal relationship between authoritarianism and party affect among the unengaged in Figure 2. Johnston et al. further argue that economic liberalism is a “functional match” for high authoritarians, because government intervention in the economy promotes order, certainty, and security through social insurance programs and market regulation (e.g., Social Security, FDA testing requirements). This suggests that citizens who “tune out” party politics are more likely to convert authoritarianism into liberal than conservative economic preferences. The results in Panel A of Figure 3 support this hypothesis using previously unexamined data.9

Panel B of Figure 3 shows the results for limited government values. For this straightforward measure of preferences over the size of government, the pattern is even stronger. Recall that values are measured on a 4-point scale, ranging from 0 to 3, where the score indicates the number of free-market-supportive statements with which the respondent agreed. At high levels of engagement, the effect of authoritarianism is 1.29 points. This means that, for the politically attentive, high authoritarians agree with over one additional statement relative to low authoritarians, on average. At low levels of engagement, by contrast, a change from low to high authoritarianism is associated with a statistically significant, .94-point decrease in economic conservatism. In other words, for unengaged citizens, low authoritarians agree with about one additional free-market-supportive statement relative to high authoritarians. In the cases of both policy and values, the relationship of authoritarianism to economic conservatism is strong and significant, but exactly opposite in direction at low and high levels of political engagement. This pattern reveals why past work has found a minimal relationship between authoritarianism and economic ideology (Feldman & Johnston, 2014; Hetherington & Weiler, 2009; Stenner, 2005): Unless the interaction with engagement is explicitly included in the model, the opposite signs will cancel, producing a small or null “main effect.”

In sum, using the most recent wave of the ANES, I find a tight connection between authoritarianism and economic preferences. Among the politically engaged, authoritarianism is among the strongest predictors of economic preferences, and citizens scoring high in authoritarianism are far more conservative than their low-scoring counterparts. This suggests that we cannot treat personality and economic ideology as competing explanations of affective polarization—they are deeply intertwined. Finally, I replicate the “reversal effect” presented in Johnston et al. (2017): Authoritarianism promotes economic liberalism among the politically inattentive.

Experimental Evidence for Authoritarianism-Based Cue-Taking

What accounts for this reversal of the role of authoritarianism and its powerful effect on economic conservatism among the politically engaged? In previous work (Johnston et al., 2017; Johnston & Wronski, 2015), my colleagues and I argue that engaged citizens seek out information about economic issues from elite actors who share their traits and cultural values. In this section, I provide direct evidence for these claims and review two experimental studies which demonstrate the role of authoritarianism in shaping elite cue-taking on economic policy among the engaged. Both experiments are also reported in Johnston et al. (2017, Chap. 6).10

The first experiment focuses on partisan and ideological cue-taking.11 These data include three measures of personality: authoritarianism (measured as above), the need for nonspecific cognitive closure (Kruglanski & Webster, 1996), and conservation versus openness to change values (Schwartz, 1992). Conceptually and empirically, these three constructs are closely related (Johnston et al., 2017; Jost et al., 2003; Kruglanski et al., 2006). Following Johnston et al. (2017), I use factor scores from the superordinate factor indicated by each of the three constructs, with higher scores indicating a “closed” rather than “open” personality. For ease of exposition and consistency with the previous and subsequent discussions, I will simply refer to this measure as “authoritarianism” in what follows.12 The primary hypothesis is that citizens who score high on authoritarianism will disproportionately seek out and assimilate information about economic issues from Republican and conservative elites, while those who score low will seek out and assimilate from Democrats and liberals. As suggested by Figure 2, however, this should be primarily the case for the politically engaged. That is, the effect of elite cues on economic opinions across authoritarianism should increase as a function of political engagement.

The design of the experiment is straightforward. All respondents indicated their opinions on four economic policy issues modeled on the ANES 7-point policy items: government-provided versus private health insurance, unemployment insurance expansion, regulation of the financial industry, and Social Security privatization.13 Respondents were randomly assigned to either receive (or not receive) elite cues associated with these issues. In the no-cues condition, competing positions on the issues were attributed to “some people” or “others,” while in the cues condition, positions were attributed to either “Democrats” and “Republicans,” or “liberals” and “conservatives,” such that each group was associated with its stereotypical approach to that issue (i.e., Democrats and liberals support a stronger government role).14 I combine these two conditions into a single “elite cues” condition.15 Engagement was measured as the average of political knowledge (six items) and political interest (r = .54). I include a similar set of controls, and I again exclude African-American and Black identifiers. All variables were coded to range from zero to one.

The regression results are shown in the appendix, and I focus on the key estimates in Figure 4. The y-axis represents economic preferences, with higher values indicating conservative views. The x-axis represents authoritarianism, and the three panels represent the 5th, 50th, and 95th percentiles of political engagement, respectively. The solid lines are for the no-cues condition, and the dashed lines are for the elite cues condition. There are two patterns of note in these results. First, the relationship of authoritarianism to economic conservatism becomes statistically significant and increasingly positive as engagement increases.16 While there is no significant relationship at low levels of engagement, I find a strong, positive relationship of authoritarianism to conservative policy preferences among the politically engaged in the no-cues condition.17 Second, the size of this latter relationship increases substantially moving from the no-cues to the elite-cues condition. In the no-cues condition, the effect of authoritarianism is .14 for citizens at the median of engagement, and .17 at the upper levels of engagement. In the elite-cues condition, these effects are .30 and .44, respectively. In other words, the presence of cues increases the size of the relationship between authoritarianism and economic preferences by 100% and 150% among moderately and highly engaged citizens, respectively. This increase is due to the tendency of citizens low in authoritarianism to assimilate the positions of Democrats, and citizens high in authoritarianism to assimilate the positions of Republicans. There is an asymmetry, however, such that cue-taking appears larger for the upper values than the lower values of authoritarianism. Among citizens at the 95th percentile of authoritarianism, moving from the control to the elite-cues condition, economic conservatism increases from .54 to .66 among moderately engaged citizens and from .57 to .77 among the highly engaged. This first experiment thus provides evidence that engaged citizens assimilate information about economic policy from party elites who share (or are perceived to share) their broader traits and values related to authoritarianism. In contrast, the unengaged show no response to party or ideology cues.

Figure 4.
Party cues experiment. Data is from the 2011 YouGov experiment.

In the second experimental study, I examine the extent to which cultural cues shape economic preferences in a way analogous to party and ideology cues. The logic is the same: Citizens seek out information about economic issues from those who are perceived to share their traits and cultural values. My colleagues and I included an experimental manipulation in the pre-election wave of the University of Minnesota's module on the 2014 Cooperative Congressional Election Study. Each respondent was presented with two short “platforms” for hypothetical candidates for public office. Each platform contained policy positions on three issues—abortion, gun control, and either financial regulation or Social Security privatization—and the two candidates always took competing positions on each issue. For the economic issues, respondents were randomly assigned to receive either financial regulation or Social Security privatization.18 Respondents were randomly assigned to receive either “stereotypical” or “counter-stereotypical” cultural cues. In the stereotypical condition, the candidate supporting gun control and abortion rights also supported financial regulation and opposed Social Security privatization (and vice versa for the other candidate). These positions are roughly consistent with the current ideological alignment of the parties in contemporary American politics. In the counter-stereotypical condition, the candidates took the opposite positions on the economic issues. Respondents were then asked their own opinion on the economic issues. Specifically, they were asked which candidate's position they prefer on Social Security or financial regulation.19 The goal of the study is to see whether engaged high and low authoritarians use the cultural cues—the positions on gun control and abortion—as information relevant to making judgments about financial regulation and Social Security. Authoritarianism was measured using the standard four-item battery. Engagement was measured as the average of political knowledge and interest (r = .50).

The logistic regression results are shown in the appendix, and I plot the key estimates in Figure 5. The y-axis represents the probability of choosing the conservative policy option, and the x-axis represents authoritarianism. The panels correspond to increasing levels of engagement, and the solid and dashed lines correspond with the stereotypical and counterstereotypical cultural cue conditions, respectively. Looking first at the stereotypical condition, I find a pattern across engagement quite like that in the 2016 ANES. At the lowest levels of engagement, authoritarianism is negatively associated with conservatism, though this effect is inefficiently estimated, and not statistically significant (95% confidence interval = [−.45, .22]).20 At moderate and high levels of engagement, however, the relationship of authoritarianism to preferences is positive and statistically significant. A change in authoritarianism from low to high is associated with a change in the probability of a conservative choice of 28 and 39 percentage points, respectively.

Figure 5.
Cultural cues experiment. Data is from the 2014 CCES experiment.

The relationship changes sharply in the counter-stereotypical cues condition. While there is no statistically significant change in the relationship for unengaged respondents, the effect of authoritarianism changes significantly as a function of counter-stereotypical cues for both moderately and highly engaged respondents. Indeed, in both cases, authoritarianism has no effect at all in the counter-stereotypical cues condition. As seen in Figure 5, this is largely due to the behavior of individuals toward the upper end of authoritarianism moving away from the conservative policy because of its association, in the counter-stereotypical condition, with abortion rights and gun control. That is, many high authoritarians assimilate the economic policy position of the candidate who takes agreeable cultural policy positions, driving down the overall relationship.

In sum, two experiments demonstrate that engaged citizens take cues on economic policy as a function of authoritarianism and related traits. Citizens scoring high in authoritarianism change their policy positions to match those of Republicans and (cultural) conservatives, while those scoring low change their positions to match Democrats and (cultural) liberals. In both experiments, however, there is evidence for an asymmetry in cue-taking, with high authoritarians showing greater movement across conditions. The possibility that this is a general pattern in the economic domain would be interesting to pursue in future work.

Broader Implications of Authoritarianism-Based Preference Formation
In this final empirical section, I consider the possible broader implications of the patterns discussed above. Authoritarianism—as well as constructs conceptually and empirically associated with authoritarianism, such as ethnocentrism and moral traditionalism—are correlated with indicators of human capital, especially cognitive ability and education (e.g., Hodson & Busseri, 2012; Onraet et al., 2015; Stenner, 2005).21 I first replicate this finding in the 2016 ANES. I focus on two indicators of human capital: educational attainment and cognitive ability measured in terms of verbal ability. The former indicates the highest level of education attained by the respondent: less than a high school degree, high school degree or equivalent, some post-secondary training or a two-year degree, Bachelor's degree, or graduate degree. The latter is measured with the commonly used “wordsum” task (Thorndike, 1942), which has been included on the General Social Survey for decades, and is part of the Wechsler Adult Intelligence Scale. The wordsum task has also been used in past research to examine the relationship of cognitive ability to authoritarianism and related constructs (Brandt & Crawford, 2016; Stenner, 2005).22 Scores on the wordsum task are correlated with general cognitive ability and intelligence (Kan, Wicherts, Dolan, & van der Maas, 2013; Wechsler, 1958)23; however, I will refer to this measure as “verbal ability” in what follows.

I regressed the childrearing measure of authoritarianism on education, verbal ability, and a set of controls. The results are reported in Table 1. Consistent with past work (e.g., Stenner, 2005), education and verbal ability are the two strongest predictors of authoritarianism in the model. A change in education from less than a high school degree to a graduate degree is associated with a decline in authoritarianism of twenty-three percentage-points.24 A change in verbal ability from minimum to maximum is associated with a decline in authoritarianism of thirty-three points. Overall, these results confirm previous findings that measures of human capital are strongly related to the childrearing values. With respect to other predictors, both age and African-American identification have significant and substantial relationships to authoritarianism.25

Continued at Authoritarianism 2 of 2.

No comments:

Post a Comment