INTRODUCTION
Income inequality in the United States and many other countries has been increasing (Mishel and Finio, Reference Mishel and Finio2013; Saez, Reference Saez2013), with little corresponding movement in public demands for redistribution. Despite theoretical expectations that increasing inequality should lead to redistributive pressure (Meltzer and Richard, Reference Meltzer and Richard1981), the literature provides many empirical examples to the contrary.Footnote 1 In response to evidence that popular reactions to growing inequality have been mixed, recent contributions to the literature have explored the conditions under which demand for redistribution is more likely to occur (Franko, Reference Franko2016; McCall, Reference McCall2016; McCall et al., Reference McCall, Burk, Laperrire and Richeson2017), and the psychological (Johnson and Fujita, Reference Johnson and Fujita2012; Day and Fiske, Reference Day and Fiske2017) and socio-economic (Franko, Reference Franko2017) variables that may be responsible for the patterns we observe. Scholarly debate continues about the specific mechanisms by which demand for redistribution does (not) emerge in response to increasing inequality.
In this paper, we focus on one specific mechanism that could explain why we observe political inaction in the face of growing inequality. In particular, we test a sub-prediction of system justification theory (SJT) (Jost et al., Reference Jost, Banaji and Nosek2004; Jost and Banaji, Reference Jost and Banaji1994). System justification theory posits that individuals are motivated to legitimize existing social arrangements, “even at the expense of personal and group interest” (Jost and Banaji, Reference Jost and Banaji1994, p. 2). System justification theory has been used to explain the prevalence of numerous attitudes that do not seem to serve the self-interest of the individual or their group (Laurin et al., Reference Laurin, Shepherd and Kay2010; Wakslak and Bauer, Reference Wakslak and Bauer2011; Brescoll et al., Reference Brescoll, Uhlmann and Newman2013), including attitudes toward economic inequality (Jost et al., Reference Jost, Banaji and Nosek2004).
The specific sub-prediction of system justification theory that we focus on is the hypothesis that inequality itself can produce or exacerbate the system justification motivation (SJM) (Jost et al., Reference Jost, Banaji and Nosek2004, Reference Jost, Blount, Pfeffer and Hunyady2003; Jost and Banaji, Reference Jost and Banaji1994). For conciseness, we call this prediction the “inequality-induced motivation hypothesis” throughout the paper. This hypothesis predicts that just experiencing or being aware of inequality will make people more motivated to justify their social, political, or economic systems.
This prediction builds on cognitive dissonance theory: the more people are disadvantaged by a system, the more starkly their experience of being disadvantaged contrasts with their motivation to believe that the system is just, and the more they therefore engage in system justification. On the individual level, this hypothesis implies that disadvantaged individuals exhibit higher levels of system justification than advantaged individuals (Jost et al. Reference Jost, Banaji and Nosek2004, p. 909, Hypothesis 17). Aggregated to the society level, this hypothesis implies that we should observe more system justification in more unequal societies (Jost et al. Reference Jost, Banaji and Nosek2004, p. 910, Hypothesis 18). This inequality-induced motivation hypothesis is thus an application of system justification theory to the specific question of what happens when there is more or less social and economic inequality in a society. In this paper, we test the inequality-induced motivation hypothesis in the context of economic inequality. In particular, we ask whether presenting realistic information about inequality to respondents results in increased system justification.
The prediction we test is different from the expectation that system justification influences how people react to inequality (which is a key prediction of SJT, but not one we test in this paper). The inequality-induced motivation hypothesis goes further: it posits that exposure to inequality can increase the SJM, i.e., as inequality increases, people become even more likely to react with motivated reasoning, and to see their system as fair.
If the inequality-induced motivation hypothesis is true, it could explain why increasing income inequality does not usually result in public dissatisfaction with inequality. As such, finding evidence in favor of this hypothesis would make a contribution to the broader literature on public responses to inequality, and the ongoing exploration of mechanisms that explain these responses. It would also yield further political predictions for the future, as it would suggest that inequality would tend to bring about even more inequality (as citizens became less and less likely to object to inequality due to system justification).
Thus far, the evidence for the inequality-induced motivation hypothesis has been observational, and has reached contradictory conclusions. Jost et al. (Reference Jost, Blount, Pfeffer and Hunyady2003) find that “members of disadvantaged groups sometimes support and justify the social order to an even greater degree than members of advantaged groups do” (p.14). However, Brandt (Reference Brandt2013) finds no support for this hypothesis using cross-national survey data. To the best of our knowledge, an experimental test of this hypothesis has never been published. In this paper, we experimentally test it by directly manipulating respondents’ perceptions of inequality.
We provide respondents with information about income inequality in the United States, and then measure their levels of SJM. Across several types of informational treatments and multiple measures of system justification, in one nationally representative sample and several convenience samples, we find no evidence for the hypothesis that inequality increases the SJM. Our null results are precisely estimated and not due to a lack of statistical power. Political non-responsiveness to inequality may be explained by many mechanisms (including system justification). But in this paper, we find no evidence that inequality-induced system justification is one of them.
HYPOTHESIS
We empirically evaluate two predictions derived from system justification theory (Jost et al. Reference Jost, Banaji and Nosek2004, pp. 909–910). We first evaluate the hypothesis (H1) that higher inequality causes higher system justification.
Second, we look for evidence that would be consistent with the individual-level, cognitive dissonance based mechanism by which H1 is expected to occur. H1 is predicated on cognitive dissonance resulting from a combination of observing inequality and wanting to believe the world is just. Therefore, this effect is expected to be the strongest among disadvantaged individuals, for whom the contrast between their own disadvantage and the supposed fairness of the system is strongest. Our second hypothesis (H2) is therefore that we should observe a larger effect of inequality on system justification among disadvantaged respondents.
METHODS AND RESULTS
Study Details
We conductedFootnote 2 a nationally representative online survey experiment through the GfK (formerly Knowledge Networks) panel in December 2014–January 2015.Footnote 3 A total of 1,020 participants were allocated to one of the six groups (one of the two treatment groups, then one of the three outcome variables).
Design
The experimental design went through rigorous pre-fielding peer review as part of the funding process. The final research proposal, which includes power analyses and hypotheses, is available online at http://www.tessexperiments.org/data/trump605.html and thus serves as a form of pre-registration.Footnote 4
Our main experimental treatment exposed participants to information about economic inequality. We showed participants one popular measure of income inequality: the Gini coefficient, which has been increasing relatively steadily in the United States since the 1970s. All participants in our study were exposed to accurate information about the Gini coefficient over the period 1968–2010, graphed in a similar way (data from Solt, Reference Solt2009). However, they were randomly assigned to one of two different conditions. In the “low-inequality” or “control” treatment condition, the y-axis was set from 0 to 100 such that the trend looked relatively flat, while in the “high-inequality” treatment condition the y-axis was truncated to make the upward trend appear much steeper (see Figure 1). Y-axis truncation is a common and effective means of creating deceptive visualizations (Pandey et al., Reference Pandey, Rall, Satterthwaite, Nov and Bertini2015), and has previously been used to generate treatments that successfully increase SJM (Kay et al., Reference Kay, Gaucher, Peach, Laurin, Friesen, Zanna and Spencer2009).
Figure 1 The Treatment and Control Conditions.
We randomly assigned each respondent to answer one of three system justification measures, all drawn from prior literature on the subject. The first outcome measure was a system justification scale (Kay and Jost, Reference Kay and Jost2003; Jost and Kay, Reference Jost and Kay2005) where participants indicated their agreement or disagreement (on a nine-point scale) with eight statementsFootnote 5 regarding the fairness of the overall social system of the United States. The items formed a reliable scale (Cronbach’s α = 0.88), and we refer to it as the SJM measure.
Our second measure uses institutional trust questions, as in Brandt (Reference Brandt2013). These questions ask the participants to indicate how much confidence (“A great deal,” “Only some,” or “Hardly any”) they have in the people running the following institutions: the military, major companies, banks and financial institutions, the executive branch of the federal government, the United States Supreme Court, and Congress. Averaging across respondents’ trust in the six institutions resulted in a scale with Cronbach’s α = 0.74, and we refer to this as the “institutional trust” measure.
Finally, we include a slightly compressed version of a system justification scale that directly measures legitimation of the system in the economic realm: the economic system justification scale (α = 0.78) (Jost and Thompson, Reference Jost and Thompson2000).Footnote 6 We will refer to it as the economic SJM measure. For ease of interpretation, we rescale all three measures to a 0–1 scale.
Results
We find no evidence that information about increasing economic inequality in the United States increases the respondents’ SJM.
Before presenting the results in more detail, we pause to give a brief overview of the data, focusing in particular on the power we obtain in this study. The study is powered to detect an experimental treatment effect of 0.3 standard deviations. This means that after rescaling all outcome variables to a 0–1 scale, we are powered to detect experimental treatment effects of 0.04 (SJM), 0.04 (institutional trust), and 0.03 (economic SJM), respectively. In our estimate, therefore, this study is sufficiently powered to detect any substantively significant changes in SJM.
To evaluate H1, that inequality increases system justification overall, we present simple differences-of-means in the levels of system justification across treatment and control conditions. Results from these t-tests are shown in Figure 2; the point estimates are presented with 95% confidence intervals.
Figure 2 Estimated Effects of the “High Income Inequality” Treatment, Compared to “Low Inequality”/Control, on System Justification Measured in Three Ways (System Justification Scale, Institutional Trust Measure, and Economic System Justification Scale).
On two of the three measures of system justification, the treatment produces entirely null results: the estimates for institutional trust (estimated effect −0.01, p = 0.24) and economic system justification (estimated effect −0.02, p = 0.07) do not reach conventional levels of statistical significance, and the estimated differences are small enough to be substantively meaningless.
The result for the third measure, the SJM scale, is a substantively small but statistically significant reduction in system justification after exposure to the high-inequality treatment, (estimated effect −0.03, p = 0.03). In other words, we find an effect in the opposite direction to the hypothesis. Exposing people to more inequality makes them slightly less likely to say the system is fair.
Turning to H2, we ask whether individual status moderates reactions to information about inequality, such that disadvantaged individuals respond with more system justification than advantaged individuals. We add an interaction between treatment condition and household income to the analysis. Table 1 presents the regression results. Across the three outcome variables, we see one negative and significant (but substantively small) coefficient on the interaction, one non-significant negative, and one positive significant coefficient (the latter is in the opposite direction to the hypothesized effect). We conclude that the findings across all three outcome variables are not consistent with the hypothesis that inequality is more effectively causing system justification among the poor than among the rich. Additional analysis of H2 can be found in Supplemental Information, where we also present two smaller, additional studies with substantively the same null results.
Table 1 Study 1 Moderation Analysis. Results of Linear Regression with an Interaction Between Income and the “High Inequality” Treatment
Note: *p < 0.1; **p < 0.05; ***p < 0.01
CONCLUSION
We find no evidence that information about high inequality increases system justification. Across three different measures, in two cases the null hypothesis of no effect could not be rejected, and in one case we found an effect in the opposite direction to the hypothesis. Including interactions with household income does not reveal consistent moderation effects in the expected direction. In sum, we find no evidence to support the inequality-induced motivation hypothesis, and we find some small effects in the opposite direction of the hypothesis.
We do not think these null results are driven by treatment failure. Responses to our treatment check questions indicate that the treatments had the desired impact: when exposed to the “high inequality” treatment, respondents were significantly more likely (66% compared to 40% in the low-inequality group) to agree that “Income inequality in the United States has increased dramatically over time,” and significantly less likely to agree that “The share of total income of the very rich has not changed much over time in the United States” (18% versus 38%). Given this evidence of treatment effectiveness, in addition to the statistical power of our experiment and the precisely-estimated null point estimates shown in Figure 2 (as well as the extremely similar point estimates found in the two studies reported in the SI), we think it is unlikely that our null findings are simply a result of treatment failure or a lack of statistical power.
Further, we think this treatment captures one key way that inequality would shape system justification in the real world. Most Americans learn about economic inequality through media representations rather than direct personal experience (McCall, Reference McCall2013), so a graphic reporting income inequality is a realistic representation of how members of the public are likely to learn about inequality. The particular measure of income inequality that we used, the Gini coefficient, is an abstract construct, but has previously been used in observational research to explore the relationship between system justification and inequality (Brandt, Reference Brandt2013; Napier and Jost, Reference Napier and Jost2008), and graphs similar to our treatment appear in media stories about inequality.Footnote 7 It is possible that inequality experienced in more personal ways could produce more system justification; whether it does is an open empirical question that our data do not address. However, given that media is an important source of information about inequality, it is important to know whether the inequality-induced motivation hypothesis holds for information acquired through graphs like our treatment.
In order to be sure that the null findings of one study were not simply a fluke or due to poor experimental design, we also ran two similar studies on convenience samples, the detailed design and results of which are provided in the SI. These studies confirmed the null results in other samples, even when introducing an additional measure of system justification and a different inequality treatment. That we find consistently null results across several experiments and different operationalizations bolsters our confidence in the findings presented here.
We do not claim that these results undermine the existence of the SJM in general. A large body of work on system justification has established that people often consider inequality justified, and that system justification can serve important psychological purposes. We specifically focus on the inequality-induced motivation hypothesis: that the SJM becomes stronger when inequality is higher. We find no support for this hypothesis.
Increasing inequality in advanced societies has not led to corresponding increases in popular opposition to or concern about inequality. Given this context, we chose to study the inequality-induced motivation hypothesis because it posits an important mechanism that could explain this empirical pattern. Our results suggest that inequality-induced system justification is not a strong explanation for the lack of popular opposition to inequality.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/XPS.2018.2.
