Abstract
People differ vastly in perceptions of inequality, some seeing a small elite at the top of their society with a vast impoverished mass at the bottom, others a prosperous society with most people in the middle. This was found first for two nations, Australia and Communist-era Hungary. We extend these results to 43 nations and to the post-Communist era. The results support our original middle-range “reference group and reality blend” hypothesis. We extend the original findings to show that, for analogous reasons, socioeconomic development makes societies seem more egalitarian, that societies’ actual income inequality shapes perceptions, and that the collapse of Communism dramatically increased perceptions of inequality. In sharp contrast to these diverse perceptions, ideals are shared, almost everyone preferring prosperous egalitarian societies. Data are from 92 large representative national samples in 43 nations with more than 100,000 respondents, analyzed by multilevel generalized least squares (GLS) methods.
Keywords
Introduction
Inequality and class are central aspects of human society, second only to religion/ethnicity as a source of solidarity (within) and conflict (between). The political history of the West since the 1800s is a story of class and inequality, a conflict between the equality advocated by the left and inequality favored by the right; modern Western politics have been shaped by that conflict, partly from the emergence of the welfare state, more from Communism’s long struggle in the West and post-war domination of Eastern Europe. But in the political struggle, it is not just (or even) the objective reality of inequality and class that matter, but how they are perceived. If inequality is not recognized, it may have no political consequences; but if it is recognized (or imagined), it will have consequences, perhaps grave ones. Reality is (at least partly) in the eye of the beholder.
Thus, it is essential to understand how people perceive their society. This article focuses on the diversity and origins of perceptions of inequality—examining the influence of different aspects of reality and reference group processes across a wide variety of societies at different levels of socio-economic development and with different degrees of income inequality. Thus, we will be examining the joint and several effects of individuals’ characteristics and of their national context on their perceptions of inequality. Future papers in this research program will examine the role that these perceptions play in the broader culture of inequality, in class conflict, and in politics.
In such an endeavor, a key issue is how to elicit people’s perceptions of inequality. Early studies asked purely verbal questions but strong underlying visual imagery was soon discovered. The (large and interesting) older literature is reviewed in detail in Evans, Kelley, and Kolosi (1992). Kolosi and colleagues began to develop a key visualization question in Hungary toward the end of the Communist era (Kolosi, Papp, Gombar, Pal, & Bara, 1980). This then evolved into the visual questions that define the modern approach fielded in a bilateral survey of Hungary and Australia (Evans et al., 1992). These have been widely used in the decades since, for example, in the International Social Survey Programme’s (ISSP) “Ideology of Inequality”/“Social Inequality” surveys (concepts and design: J. Kelley et al., 1998; data: Zentralarchiv fuer Empirische Sozialforschung, 2012) and other years. In all, data have now been collected from 92 surveys in 43 countries with more than 110,000 respondents.
This article is the first systematic comparative analysis of the new data.
Plan of the Article
The next section sets the stage by introducing the range of inequality perceptions as embodied in a set of images, spanning income distributions that range from the extreme inequality of a medieval agrarian society to a very prosperous post-industrial society. The Section “Theory” discusses theories and prior evidence on the sources of images of inequality: Evans, Kelley, and Kolosi’s reference group and reality blend theory (which the evidence here supports) and competing theories—materialist theory, dominant ideology theory, and postmaterialist theory (all of which fail to align with the evidence). All the theories lead to specific testable hypotheses. In the Section “Data, Measurement, and Method,” we introduce the ISSP data and detail our measurement procedures (including coding of the images) and methods of analysis (multilevel variance components models with random intercepts, estimated by generalized least squares [GLS]). Exploratory analysis indicated substantial differences in the effects of many predictor variables between European Communist-legacy societies and all others (which, for convenience, we will call “Western” societies), so we open the analysis by describing the effects of individual socioeconomic status (SES) and demographic characteristics on images of inequality, while allowing for all possible interactions between societal type and these variables. This is the evidence needed to address Hypothesis 1. The analytic results in this section are net effects from our multilevel models that include all the predictors, but, for clarity, we have grouped the reporting of the results into substantive blocks that correspond to the hypotheses developed earlier. After that, a subsection on the relationship between socioeconomic development and individuals’ images of society addresses Hypothesis 2. It, in turn, is followed by a subsection on national income inequality and images of inequality, which addresses Hypothesis 3. The subsequent subsection assesses the evidence about the impact of Communism and its aftermath on images of inequality (Hypotheses 4 and 5) and the final results subsection contrasts the diversity of perceptions across societies and across individuals’ social locations with strong consensus on ideal inequality (Hypothesis 6). Next, the Section “Discussion” opens by summarizing and synthesizing the results and their implications for the theories, and then locates these findings into the broader research tradition of the culture and social psychology of inequality.
Description: Images of Inequality
Images of inequality are very diverse. Some people hold an elitist image—a large mass at the bottom of society and a small, privileged elite at the top. By contrast, others perceive society as egalitarian with the largest classes in the middle or even upper middle—a society in which most people have a good position and access to the good life. Yet others have intermediate views. The five figures below represent the social class distributions in a range of plausible societies, each accompanied by a brief description (Figure 1). The diagrams have approximately equal area, but differ in class sizes, ranging from an extremely elitist society with tiny middle classes to a broad-shouldered egalitarian society in which the lower classes have dwindled away to almost nothing. Respondent’s selection of the inequality diagram that most closely resembles their society forms the focus of this article. Each diagram represents an image of societal inequality derived from prior research, as discussed below, along with the percentage of respondents choosing that diagram.
Different types of society.
Type A is a strongly elitist image of society, with a small elite at the top, negligible middle classes, and the great mass at the bottom. Inequality is great, with a coefficient of variation of fully 1.9. 1 It resembles medieval agrarian societies more than modern industrial or post-industrial societies, but earlier studies suggest that many low status groups see society that way (Goldthorpe, 1966; Kolosi et al., 1980; Mayer, 1975) It also represents the traditional Marxist image of capitalist societies.
Societies vary enormously in the degree of popular acceptance of Type A as an accurate depiction. Less than 20% think it accurately represents their society’s class structure in the established market-oriented societies, and (interestingly) in Hungary, before the fall of Communism. But, in stark contrast, a near majority of 48% in post-Communist Eastern Europe (including Hungary) think they now live in the extremely elitist Type A—a startling change in just a few years.
Type B represents a much more egalitarian society: a pyramid, with a small elite at the top, successively larger intermediate classes, and the largest class at the bottom. The average person clearly holds a better position in this society than in Type A; the modal class is still at the bottom but the average has risen to 26, and inequality has fallen to 1.2. This image, or perhaps Type C, is closest to traditional functionalist imagery and to Goldthorpe’s (1966) view of an early capitalist society. In every country, around 30% choose it.
Type C is similar to Type B, except that the lowest class is smaller and the middle classes correspondingly larger. The modal class position is no longer the lowest class; the mean has risen to 37 and inequality again fallen, to 0.7. This moderately egalitarian society is similar to the image Germans had of their society in the 1960s (Bolte & Neidhardt, 1967), to Goldthorpe’s (1966) view of a modern capitalist society, and to one variety of the American “middle mass” images (Vanneman & Cannon, 1987). In established market-oriented societies, as well as Hungary in Communist times, 20% to 25% chose this image.
Type D portrays a very egalitarian society with small top and bottom classes and most people in the middle. The modal class position is now the middle class, average class position is 50, and inequality has fallen yet again, to 0.5. This corresponds to a version of egalitarian society and is a second variety of “middle mass” society (Vanneman & Cannon, 1987; see also Stephenson, 2000). There is huge international variation in how many people see Type D as an accurate representation of their society: about 25% in established market-oriented societies, 33% under Communism, and less than 10% in post-Communist societies.
Type E portrays a society with an inclusive elite at the top, very large upper middle and middle classes, and only a vestigial bottom class; inequality is just 0.4. This resembles the traditional socialist ideal and the implicit model in theories of post-industrial society, and is a third variety of “middle mass” society (Vanneman & Cannon, 1987). Few people see their social structure as resembling Type E.
Although these are unusual and apparently complex questions, respondents answered them easily. The item non-response rate is only fractionally higher than for simple attitude questions, and test–retest reliability is a respectable r = .39 over a 2- to 3-year period in Australia (where the items were included in a panel directed by the authors with N = 3,627).
Theory
Reference Group and Reality Blended (Preferred Theory)
Part 1: The reference group contribution
Reference group theory holds that people’s social perceptions include as main ingredients their own experiences and the experiences of their families, friends, and co-workers, incorporating only limited doses of information about society as a whole or other abstract imagery (Lockwood, 1966; Merton & Kitt, 1950; Runciman, 1966; Stouffer, Suchman, DeVinney, Star, & Williams, 1949). Building on the foundations of H. H. Kelley (1967), cognitive psychologists see the reference group phenomenon as a special case of the “availability heuristic” (Kahneman, Slovic, & Tversky, 1982; Tversky & Kahneman, 1981), a systematic perceptual bias, whereby individuals derive their impressions from their small social circle—in effect, doing an informal survey of themselves and their social networks and summarizing their observations by a subjective regression analysis—and then generalize to the broader society.
These influences are likely to lead people to perceive their class as larger than it actually is. This tendency may be exacerbated by social–psychological identity defenses, which tend to exaggerate group size (Lindemann & Saar, 2014; Schmitt, Branscombe, & Kappen, 2003). For example, people of high objective status will overestimate the number of people like themselves, and hence will believe that the higher classes are large; similarly, people of low objective status will see the lower classes as large (Mayer, 1975). Hence, high status people will see society as more egalitarian than will people of lower objective status, because the fundamental difference between elitist and egalitarian images of society is the size of the higher classes (Figure 2). Prior research yielded ambiguous results on this hypothesis—support in Australia but not in Communist Hungary (Evans et al., 1992). These new samples from a larger range of countries should settle the issue.
Theory: SES.
Moreover, this logic implies that as education, occupation, and income rise with socioeconomic development, not only will one’s own SES rise but the pools of available network partners will include more and more high SES people. On average, that will raise the typical SES of others in one’s network, leading one’s societal image to include larger middle and upper middle classes.
Part 2: But reality bites
Of course reference group forces are not the only influences on class imagery. People’s interactions and information exposure not only are limited to their social networks but also include a myriad of small events and impressions ranging from interactions with strangers to encounters with institutions to media stimuli. These make the actual social structure an insistent presence, sometimes subtle, sometimes brash. A useful starting point here is the “reference group and reality blend” theory proposing that both reference group forces and the actual material structure of society both matter, and that they combine additively to generate images of society’s social class composition (Evans & Kelley, 2004b; Evans et al., 1992; J. Kelley & Evans, 1995). The perceptual recipe proposed by this theory to date has equal parts of reference group and reality, although the exact proportion has yet to be demonstrated empirically.
The hypothesis predicts different societal images for people in different social locations, as illustrated in Figure 2. These demonstrate how, according to this theory, the same actual society combines with the reference groups of people in different social strata to produce different societal perceptions: High status people tend to perceive a relatively egalitarian society, whereas those low in status perceive a relatively unequal, elitist society.
More formally, we propose that images of society are a weighted average of the actual distribution, the image derived from reference groups, other factors X, and a random error term e:
In the theoretical argument, we have assumed that actual inequality (e.g., Gini) and reference groups are equally important, so that the standardized regression coefficients would be equal, but one could, of course, imagine a different ratio, an issue to which we will return in the “Discussion” section. The other factors Xi include (as we will see) socioeconomic development of nation (GDP per capita at PPP), government history (Communist or not), and demographic factors.
Part 3: Hypotheses derived from the reference group and reality blend theory
First, the blend theory proposes that the individual’s own objective class position and reference group shape perceptions:
Next, reference groups are shaped not only by the individual’s position in the socioeconomic hierarchy but also by the social locations of potential network members in the society. Consider a society where many people are well educated, with high status jobs and ample incomes. The supply of middle and high SES network candidates is large; so, on average, any person’s collection of relatives, friends, and co-workers will have more middle and high SES members than in a society where middle and high SES members are rare. Conversely, in a society where most people are poorly educated and have low status jobs and modest incomes, their relatives, friends, and co-workers will usually be poorly off themselves. Economic development is largely a matter of high levels of education, skilled jobs, and high earnings—indeed the correlations between a nation’s level of development (GDP per capita at parity purchasing power) and its mean education, occupational status, and income are around .70 (Breznau, Lykes, Kelley, & Evans, 2011; Przeworski, 1997). Hence, in developed nations, more potential reference group members will have high SES. That availability will increase the chances that one’s reference group members will have high SES. Therefore, reference groups will typically be higher in the socioeconomic hierarchy (Curtis, 2013; Horvat & Evans, 2011) and hence, from Hypothesis 1, produce more egalitarian images of society:
Of course, reality enters into the blend as well, although diluted by reference group forces:
Finally, we consider a telling natural experiment: The years of Communism and the transition beyond it when the actual social structure changed sharply as the old system collapsed and the unevenly emerging market economies erratically opened new opportunities which, in turn, created new inequalities. We are only beginning to understand the perceptual and attitudinal ramifications of this huge exogenous shock (Curtis, 2013; J. Kelley & Zagorski, 2005). Figure 3 illustrates what this implies.
Theory: Communism.
By the last decades of Communism, government-imposed compression of incomes had created a short, fat social structure in reality (Milanovic, 1998). There was less income inequality than in the past or in contemporary Western nations, but standards of living were also low, much lower than in the West (Figure 3, first row, far left).
Combined with the ordinary worker’s modest reference group (first row, middle column), that gives a very low status and egalitarian image (far right). Thus, we hypothesize as follows:
With the fall of Communism (Figure 3, second row), new opportunities opened up, rapidly and vastly enriching a few, sometimes by talent and determination, sometimes by corrupt means, sometimes both (Figure 3, second row, left). These “tall poppies” were few but very visible in societies that had only recently vigorously advocated equality, touting it as one of Communism’s great virtues compared with capitalism’s vulgar excesses—think of old cartoons featuring the sleek stout boss debouching from a large, gleaming car with grimy workmen trudging home from a smoke-belching factory in the background. Many ordinary people found the new inequality shocking.
Moreover, emerging inequality was not the only shock. Bettering the lot of the working class was the central legitimation theme of European Communism, and the stringent censorship imposed by Communist regimes enabled them, for many years, to persuade their citizens that the working class was more prosperous under Communism than in the West. In 1987, shortly before the Russians went home and the Eastern European Communist regimes collapsed, Hungarians perceived their country as a prosperous Type D, and expected to move toward an even more prosperous and egalitarian Type E 30 years in the future (Evans et al., 1992). The shock of realizing how affluent the working class actually was in European market societies was felt as a betrayal: People had endured much because they had trusted—and understood their fellow citizens to trust—the Communist regimes, at least in this (Kuran, 1995; Verwiebe & Wegener, 2000; Wegener, 2000). Faith in the Communist regimes collapsed, although there was a lingering preference for more governmental intervention in the economy than in the West (Breznau, 2010; Sikora, 2005; Sikora & Kelley, 1999).
Characteristically, responses to betrayal involve blackening the character of the betrayer—and sometimes blackening an entire set of institutions of the same kind as the betrayer—even beyond what is warranted by the action (Jasso, 2008; Joskowicz–Jabloner & Leiser, 2013; Koehler & Gershoff, 2003). This may be especially true in matters of justice (Jasso, 2006). Determined not to be fooled again, the betrayed often tilt into the opposite perceptual error (Nannestad, 2008; Pidgeon et al., 2005; Poortinga & Pidgeon, 2004). In this case, that would be imagining the elite to be larger and richer than they are, and the lot of the working class even worse than it is (Kreidl, 2000). The last row of Figure 3 shows how the actual inequality in these societies (third row, first image) and the betrayal-induced distorted perception (which leads people to imagine their own position and that of their reference group peers to be worse than it objectively is; third row, second image) combine to generate a mental image of a dramatically inegalitarian society (third row, last image), one close to elitist Type A. Thus, we hypothesize as follows:
Communist elites in China seem to have evaded being labeled with betrayal by adopting market reforms in the 1990s (Nee, 1991), reforms which produced sustained economic growth, lifting many out of poverty. It may also have helped that the Asia capitalist exemplars illustrating Communism’s economic inefficiency (Taiwan, Singapore) were small and arguably atypical, unlike the overwhelming, inarguably prosperous example of the European Union (EU) that destroyed Eastern European Communist illusions.
Part 4. Extension: Hypothesis about ideals, about what society ought to be
Our arguments so far are strictly about perceptions of what societies are in fact, and have no implications for ideals about what societies ought to be. But ideals are also important, in part because the discrepancy between ideal and perceived levels of inequality is a key source of political conflict in the modern world.
The post–World War II world has embraced a broadly egalitarian zeitgeist (Bell, 1972; Jasso, 1980; Robinson & Bell, 1978). Complete equality is not the ideal—that is not endorsed in any nation yet studied (Evans & Kelley, 2007; J. Kelley & Evans, 2009). Instead, the ideal is what might be thought a “modest and civilized” degree of inequality, one providing clear rewards for hard work and talent but moderating excessive windfalls for the fortunate, and alleviating harsh deprivation for the unfortunate (Evans, Kelley, & Peoples, 2010; Organisation for Economic Co-Operation and Development [OECD], 2011). This zeitgeist suggests egalitarian societies such as Type D or Type E as ideal for people under all types of institutional regimes and at all levels of socioeconomic development:
Figure 4 summarizes our arguments.
Predictions from alternative theories.
Rejected Alternative Theories
Materialism (Figure 4, row 4)
In the classic materialist view, people perceive the actual structure of society directly: Reality matters, reference groups do not, directly contrary to our Hypothesis 1. On materialist arguments, high socioeconomic development (GDP per capita) will readily be perceived but will shape perceptions of societal inequality only insofar as development is correlated with inequality. In fact, they are correlated at about r = −.40 (see also Zagorski, Evans, Kelley, & Piotrowska, 2014), so there should be a modest effect. This agrees with our reference group and reality blend arguments in direction but not in magnitude: Our blend argument implies that the reference groups processes amplify the actual development effects (Figure 4, column 2). Similarly, materialist arguments imply that the sudden appearance of successful entrepreneurs in the early post-Communist period should generate inegalitarian imagery insofar as inequality actually increases (column 5). But beyond that, the collapse of Communism should have no effect. The blend argument agrees in direction but again implies a stronger effect, one reinforced by reference group processes.
As for ideals, materialist arguments suggest low SES people would favor egalitarian societies (out of self-interest) whereas high SES people should favor inegalitarian ones (also out of self-interest). One could also argue that reality elicits ideals as in Homans’s misinterpretation of Pope, “Whatever is, is right.” 2 In this view, a society’s actual inequality will shape its citizens’ inequality ideals (Homans, 1974; Trump, 2013).
Postmaterialism (Figure 4, row 5)
The postmaterialist hypothesis (Inglehart, 1990; Inglehart & Baker, 2000) suggests that the abundant affluence that high socioeconomic development generates will detach societal imagery from its material base, weakening links between actual social structure and perceptions of society’s class composition. This implies that actual SES should have no effect, actual inequality should have no effect, and that socioeconomic development should have an impact only insofar as it is correlated with non-economic improvements—prior research shows strong connections among economic growth, democracy, rule of law, and low levels of corruption (Breznau et al., 2011; Przeworski, 1997; Rothstein & Holmberg, 2011). Postmaterialism implies that societies increasingly embrace egalitarian ideals as they develop. Note, however, that time-series analyses for the United States do not tend to support this Maslovian hierarchy-of-needs interpretation (Ferguson, Kellstedt, & Linn, 2013).
Dominant ideology and system justification (Figure 4, row 6)
The dominant ideology thesis, with many adherents in political science, psychology, and sociology, proposes that elites successfully promulgate social constructions of reality that serve their interests (Enns, Kelly, Morgan, Volscho, & Witko, 2014; Gramsci, 2012; Rytina, Form, & Pease, 1970). The related system justification hypothesis, widely accepted by social and political psychologists, provides a plausible mechanism explaining how this comes about—why people low in the pecking order comply with social constructions of understandings that are contrary to their interests. It posits an intrinsic human need to believe in a just world; this makes us susceptible to imagery of our society as just, whatever it is actually like (Furnham, 2003; Jost, Banaji, & Nosek, 2004; Napier & Jost, 2008; Rutjens & Loseman, 2010). Thus, dominant elites can prey upon the citizenry’s system justification needs (Rubin & Hewstone, 2004).
This theory suggests that there should be few or no effects of SES, socioeconomic development, or actual inequality (contrary to our Hypotheses 1, 2, and 3). But there should be strong effects of regime orientation, with Communist societies perceiving themselves as egalitarian in accordance with their Communist elite’s ideology. The overthrow of Communist elites and de-legitimation of their ideology in the post-Communist era should then lead to a decline in egalitarian perceptions, but not to a strong decline because the new governing elites in Eastern Europe have been a transitory mix of (sometimes extreme) left, middle, and right. This prediction agrees with our reference group arguments in direction but not in magnitude (Figure 4, Hypothesis 5).
As for ideals, dominant ideology and system justification arguments imply that, in conformity with their dominant elites, market-oriented societies favor inegalitarian ideals and Communist societies favor egalitarian ones. Post-Communist societies would be mixed, egalitarian from history and childhood socialization, but mixed in conformity with their current elites’ mixed orientations (which cover the ideological spectrum).
China and the East
Communism in China seems to be following a different path than in Eastern Europe, although the processes of status attainment in China have been closely similar to those in other poor countries (Evans, Kelley, & Yang, 2016). The Chinese Communist Party (CCP) maintained its grip on power when its Russian progenitor collapsed. This enabled the CCP to liberalize the Chinese economy and foster essentially capitalist economic growth while maintaining regime continuity: There is no drama of collapse and, perhaps, less trust initially. That should, we argue, leave perceptions of inequality in China somewhere between the egalitarian perceptions of classic Communism (Hypothesis 4) and whatever is to come in China’s future. If vast illegitimate wealth were to explode onto the scene (e.g., among the elite’s family members—already a sensitive topic in China), then we would predict the same result as in Eastern Europe: hugely exaggerated perceptions of inequality (Hypothesis 5). But if China manages to develop to Western European levels without that, we predict that perceptions of inequality will in time converge to Western patterns (Hypotheses 1 and 2).
Perceptions of inequality in Japan and Taiwan have, we argue, already converged to images typical of long-established market economies. Japan has for generations been a successful market-oriented society. Taiwan was never Communist, has long had a market economy, and is steadily approaching Western European levels of affluence. Like China, it has a Confucian culture but that, we argue, does not imply Confucian exceptionalism on matters of inequality. Rather, we suggest that prosperous, market-oriented people have similar perceptions of inequality in any society, Eastern or Western, where family, friends, and co-workers are socially homogeneous.
Data, Measurement, and Method
Data: International Social Survey Programme (ISSP) Inequality, Rounds I to IV
The data used here are from the ISSP’s “Ideology of Inequality” surveys conducted in various years between 1987 and 2009 by a consortium of mostly academic survey organizations. Completion rates averaged around 60%, counting losses both at the interview and the drop-off stages, with most of these reported rates being AAPOR/WAPOR Response Rates 3 or 5, or Cooperation Rate 1. These rates compare favorably with other surveys in many industrial nations. Comparisons with the national census show the surveys to be largely representative of the population. The surveys are widely used in cross-national research. Details are in the online supplement and on the ISSP website, www.issp.org. The authors originally proposed the “Ideology of Inequality” survey (later sometimes called the “Social Inequality” survey) and chaired the Questionnaire Design Committee (also called the “Design Committee”) for the first three rounds (1987, 1992, and 1998).
Descriptive statistics for each country are in Figure 5, with the most egalitarian countries listed first. The striking changes in Hungary between the Communist era and afterwards are noteworthy. The United States is not exceptional.
Perceptions of actual and ideal type of society and other characteristics.
Measurement
Perceived structure of society
For the multivariate analyses, perceived structures of society were scored according to their implicit level of equality as defined in the original source (Evans et al., 1992, Table 1; see also Figure 1). Specifically, we score 0 for the most elitist type of society to 100 for the most egalitarian, with intermediate types scored in proportion to their coefficient of variation. Thus, elitist Type A is scored as 0, B as 47, C as 80, D as 93, and egalitarian Type E as 100.
Sensitivity analysis reveals that the results are highly robust to differences in scoring. In particular, a conventional Likert-type equal-interval scoring (0, 25, 50, 75, and 100) leads to the same substantive conclusions, as does an ordinal probit specification making no assumptions about the intervals.
National income inequality (Gini coefficient)
The Gini coefficient is the usual measure of inequality, so we will use it here, although a strong argument can be made for the standard deviation (Allison, 1978). Gini coefficients are the average for the 3 years preceding the date of each survey (averaging 3 years gives greater reliability than using a single one; e.g., Evans & Kelley, 2004b). Data are from the World Bank with a few interpolations by the authors.
SES
For individual-level objective socioeconomic position, we include education, occupation, and family income, scoring them conventionally. Because the component of the theory that focuses on an individual’s social location concerns a summary location, with the combined effect of multiple aspects of social location being of interest (rather than the decomposition), we use a multiple-item measure combining the different aspects. All surveys included detailed questions on these.
Complete measurement details are in the online supplement.
Missing data
Missing data are treated using complete case analysis (listwise deletion). With our very large number of cases and reasonable rates of item-missing data, there would be little or no gain to using missing data imputation methods. Provided that the data are MAR, both complete case analysis and FIML imputation methods are unbiased, with the FIML imputations being slightly more efficient (Enders, 2001). The gain in efficiency is negligible with large sample sizes, so, following Occam’s razor, we deploy the simpler technique.
Method
We use a variance components multilevel model estimated by GLS with random intercepts by society (Hox, 1995) estimated in Stata 14 using the xtreg procedure. A corresponding mixed-effects multilevel regression (Stata’s xtmixed) produces identical results. Multilevel modeling is the optimal choice for this analysis because it provides the correct standard errors for the national-level variables and does not assume independence of cases within nations, unlike ordinary least squares (OLS). Aggregate analysis would be, of course, subject to the ecological fallacy (e.g., Breznau, 2015, 2016). Details are in the online supplement.
Replication and Extension
A well-documented data set in a form convenient for reanalysis, together with complete Stata code for all the analyses in this article, is freely available online on our website, www.international-survey.org.
Results
SES and Reference Groups: Hypothesis 1
Western nations: Description
In long-established market societies, people in the top half of the socioeconomic hierarchy perceive a more egalitarian society than do those in the bottom half (Figure 6, columns 1 and 2). Overall, only 11% of high status people see their society as the most elitist Type A (Egalitarianism Score of 0), most seeing it instead as a middling Type B (29%, Egalitarianism Score of 47), or Type C (27%, Egalitarianism Score of 80) or even D (31%, Egalitarianism Score of 93). In contrast, low status people in these societies are much more likely to think their society is elitist Type A (21%) or middling Type B (36%) rather than more egalitarian Types C (19%) or D (20%). In sum, in these societies, high status people on average see their societies as well toward the egalitarian end of the spectrum, 66 points out of 100, whereas low status people see their societies as somewhat less egalitarian, on average near the middle, 55 points.
Perceived type of society, percentage distributions separately for individuals low and high in socioeconomic status (Hypothesis 1).
We have argued that these divergent perceptions arise because high SES people generally have high SES reference groups (family, friends, acquaintances, and co-workers), which biases their societal perception upward, whereas low SES people usually have low SES reference groups, which biases their perceptions downward (Hypothesis 1; Figure 2).
To discover whether these perceptual differences still hold even net of personal and national characteristics, we turn to multilevel analysis. The regression coefficients, first differences, and significance levels for our multilevel models are in Figure 7.
Perceived and ideal structure of society: Multilevel models separately for the Western European culture area and for Communist and formerly Communist nations, 43 nations, 1987-2009.
Western nations: Analysis
These SES differences remain clear even when taking into account the actual level of inequality in the society, its level of development, and demographic differences of age and gender in the multilevel analysis (Figure 8, left panel; details in Figure 7, panel 1). Specifically, people in the highest decile of the Western socioeconomic hierarchy will, on average and other things being equal, perceive their society as a relatively egalitarian 65 points out of 100. In contrast, those in the lowest decile socioeconomically will see it as only a middling 53 points. This is a difference of 65 − 53 = 12 egalitarian points (shown in Figure 8, left hand panel, top row). That is a substantial difference, much larger than age or sex differences (which run just 1 or 2 points), as large as GDP differences, and two thirds as large as the most important single influence, differences between the top and bottom decile of Western nations in actual income inequality as measured by the Gini coefficient. These are first differences in the sense of King, Honaker, Joseph, and Scheve (2001), comparable with each other—like standardized partial regression coefficients—but in concrete units.
Influences on perceived structure of society.
Formerly Communist nations: Description
The extraordinary history of European Communism’s rise, long ideologically charged rule, and sudden fall has left a dramatic impact on citizens’ perceptions of society. Most strikingly almost half now see post-Communist society as highly elitist, Type A (Figure 6), a striking perception we will take up in detail later. (Figure 6, columns 3 and 4 labeled, “Previously Communist” gives the percent distributions of the perceptions.) Nonetheless, reference group processes like those found in the West have a modest impact, as they do also on values about meritocracy (Smith & Mateju, 2012). Those in the top half of the SES hierarchy in post-Communist nations are a little less likely than those in the bottom half to see their society as highly elitist, 41% to 49%, and in all their mean perception is a little more egalitarian, 39 points to 33 points (Figure 6, columns 3 and 4 labeled, “Previously Communist”).
Formerly Communist nations: Analysis
The lesser importance of SES in Communist-legacy societies than in Western societies, which we saw in the descriptive analysis above, persists even after taking into account actual inequality, GDP, and demographic factors in multilevel models (Figure 8, panel B on the right; details in Figure 7, panel 2). Detailed analyses (not shown) indicate that these differences have to do only with income, whereas in the West they have to do more or less equally with income, education, and occupational status. Socioeconomic development of the nation, as indexed by GDP, is about equally potent in both Western and Communist-legacy societies. By contrast, national income inequality is the dominant influence on these perceptions in the West, but fails to have a statistically significant effect in the post-Communist societies, perhaps because the emergence of market-based inequalities was such a shock that differences in the magnitude of those inequalities paled by comparison. This interpretation is buttressed by the extremely large difference (a huge 37 net of other influences) between perceptions before and after the end of Communism in previously Communist nations (Figure 8, panel B on the right, bottom bar).
Summary
High status people, especially in the long-established market societies, tend to perceive their societies as more egalitarian than low status people do, as shown by the differences in percent distributions for the two groups (Figure 6) and by the multilevel regression coefficients for SES (Figure 7). This pattern is as predicted by our preferred reference group and reality blend theory.
By contrast, these results are clearly inconsistent with the alternative theories: materialism, postmaterialism, dominant ideology, and system justification (Figure 4, Hypothesis 1). Prior research found ambiguous results about the impact of SES in a two-country comparison (Evans et al., 1992), but the results of the current analysis involving many more countries make it clear that high SES groups do have more egalitarian perceptions, especially in the West.
Development and GDP: Hypothesis 2
In advanced nations, both individuals and their reference groups will typically have more education, higher occupational status, and higher incomes than would be common in less developed nations. That higher SES should, as we have just seen from Hypothesis 1, produce more egalitarian images of society (Hypothesis 2).
National differences: Description
GDP has a strong positive relationship with perceived structure of society: The richer the society, the more egalitarian it is perceived to be (Figure 9). For example, Brazil (“Bzl,” toward the lower left) is a relatively poor country whose people perceive it to be inegalitarian with most of the population concentrated on the bottom rungs of society: The estimated mean “egalitarianness” is exactly what we would expect given Brazil’s level of socioeconomic development. At the other end of the development spectrum, Norway (“Nor,” at the upper right) is a very rich society with perceptions of “egalitarianness” exactly as expected from the general pattern for other countries in the study.
Perceived structure of society by GDP.
But a few nations depart from the general pattern. Among advanced nations, Danes (“Den” toward the upper right) see themselves as rather more egalitarian that would be typical for nations as rich as they are. Among poorer nations, Ukrainians and Bulgarians see themselves as rather less egalitarian than would be typical whereas Filipinos and Chinese see themselves as a little more egalitarian.
But the really striking difference is Communism: Hungarians in Communist times see their society as dramatically more egalitarian than is typical for nations at their modest level of development (Figure 9, “Hng_red” in the upper middle). These 1987 data are the only ones in existence for a Communist society. Just a few years later in post-Communist times, Hungary is in no way exceptional (“Hng_mkt” at the lower left near “Rus” Russia).
Multilevel analysis shows that the Hungarian difference under Communism is not due to actual inequality although that was low (0.23, just a shade lower than typical for capitalist Scandinavian nations, Figure 5); nor was it due to anything else we know about (Figure 8, panel B). Net of all the other measured influences, Hungarians in Communist times saw their society as fully 37 points more egalitarian than would be typical for Eastern Europe in the post-Communist era.
Western nations
Americans perceive their nation as rather less egalitarian than would be typical for so rich a nation (Figure 9, far right). But the difference is not dramatic: 17% see the United States as an elitist Type A, compared with 13% among other advanced nations (Figure 10, columns 1 and 2). Descriptively, GDP seems to matter greatly in the West. For example, in developing nations, 38% see themselves as elitist Type A (column 3). Less than half as many see themselves that way in the advanced nations (columns 1 and 2).
Perceived type of society: Percent distributions separately for low and high GDP nations within each culture area (Hypothesis 2), 43 nations, 1987-2009.
GDP differences remain large in the West in the multivariate analysis even after adjusting for societal inequality, individual SES, and other factors: There is a 12-point difference in perceptions between the richest and poorest decile (Figure 8, panel A). This is just as important as individual SES and about two thirds as important as the largest single influence, societal income inequality.
Communism and its aftermath
Even among the post-Communist nations of Europe with all their dramatic changes and complex history (Balaev & Southworth, 2007; Oberschall, 1996; Sikora, 2005; Smith & Mateju, 2012; Walder, 2003), GDP has a strong link to perceptions (percent distributions with no controls are shown in Figure 10, columns 4 and 5). Among more prosperous Eastern European nations (basically, East-Central Europe), 37% see themselves as elitist Type A—about the same as developing nations in the West. But among poorer Eastern European nations (basically, Eastern Europe), fully 54% see themselves as elitist Type A.
In all, in post-Communist Europe, people in the more prosperous countries see their country on average as 42 points out of 100 on egalitarianism, compared with only 28 in poorer nations, consistent with prior findings emphasizing the importance of self-justification in the citizenry’s social construction of inequality in these countries (Slomczynski & Wilk, 2002). The multilevel analysis confirms that development influences inequality perceptions in these countries, approximately equally as strong as in the West, even controlling for many other influences (Figure 8, panel B).
Theoretical implications
These results are consistent with our reference groups and reality blend argument that the more developed the society, the more will its people perceive it as egalitarian (Hypothesis 2). Moreover, the results are clearly contrary to the dominant ideology and system justification hypotheses, which anticipate no impact of development (Figure 4, row 6, column 2).
The implications for materialism theories are more ambiguous (Figure 4, row 4, column 2). Insofar as development is correlated with lower inequality (which it is) then that should produce more benign images of society. So differences would be in the same direction as the findings. But they would not be of the great magnitude we find, producing at most a 5-point increase in egalitarian perceptions, perhaps a third of the GDP differences we actually find. 3
The implications for postmaterialism are similar: Insofar as development is correlated with non-economic gains (e.g., more democracy, less corruption), then these might produce more egalitarian images of society, but not by anything like the magnitude in these results.
Reality and Perceptions: Hypothesis 3
The reality of income inequality does influence perceptions, but only to a limited degree. Western societies that are actually very unequal as measured by the Gini coefficient are perceived as just moderately unequal, those in the highest decile on the Gini coefficient differing from those in the lowest decile by just 12 points in their perceptions of societal egalitarianism (multilevel analysis, Figure 8, left panel). By comparison, the actual range of perceptions is much larger: Perceptions often vary by 40 or 50 points between societies (see the range in the raw data, Figure 5). Hence, actual inequality in the sense measured by the Gini coefficient (the usual measure of income inequality) is only a small part of the story. And, for post-Communist societies, inequality in the sense of the Gini coefficient does not affect perceptions at all (Figure 8, right panel).
Theoretical implications
These modest but real effects of actual inequality are as predicted by our preferred reference group and reality blend theory and by materialist arguments as well (Figure 4, Hypothesis 3). But they conflict with the predictions from postmaterialist, dominant ideology, and system justification theories.
Communism and Its Aftermath: Hypotheses 4 and 5
Under Communism, government policy created a strongly egalitarian and widely accepted distribution of income, with the elite earning only two or three times as much as ordinary workers (Frentzel-Zagorska & Zagorski, 1993; J. Kelley & Zagorski, 2005; Lane, 1982). Society was, quite correctly, perceived as egalitarian. Judging from the Hungarian 1987 data—the only time these questions were asked in a European society still under Communist rule—only 8% thought society was elitist Type A (Figure 11, column 1). Communism collapsed a couple of years later.
Perceived type of society: Percent distributions for selected nations with different experiences of Communism, 1987-2009.
Only 3 or 4 years after that, in stunning contrast to Communist times, a majority of Hungarians, 56%, thought their nation had become elitist Type A (Figure 11, column 2). This view is echoed in all other post-Communist Eastern European nations at the time and in subsequent decades (column 3). This was a truly dramatic change.
These differences were not just a consequence of changing income inequality (although that did increase), or of changing perceptions of the earnings of high and low status occupations, or of changes in GDP (which at first declined and then steadily increased), or of demographic change. The multilevel analysis shows that net of all these things, the fall of Communism made society seem fully 37 points (out of 100) less egalitarian, by far the largest impact of anything we have been able to measure (Figure 8, panel B, and Figure 6). This seismic shift is especially startling, because political attitudes over this period, at least in East Central European democracies, showed strong convergence toward the small persistent cultural and social position–related influences typical in the West (Mateju & Vlachova, 1998).
Chinese exceptionalism? In short, no
By 2009, when we first have data from mainland China—some 20 years after their market reforms began—they perceived their society as more elitist than did Eastern Europeans in Communist times, 22% seeing it as elitist Type A (Figure 11, column 4). But that is still far less elitist than contemporary post-Communist societies in Eastern Europe, where a majority see themselves as elitist Type A. We speculate that Communist China has so far been able to avoid the sense of Communist betrayal that pervades Eastern Europe.
Evidence from Taiwan—Chinese in culture but with no history of Communist domination and a prosperous, long-established market economy—suggests that Chinese culture has no distinctive impact on perceptions of society type (Figure 11, column 5). Taiwan looks much like many other moderately prosperous market economies (compare it with long-established market Western nations in Figure 10, columns 1 and 2). In this respect, we suspect that perceptions in mainland China too will converge toward the long-established market pattern, as Japan already has, rather than following the distinctive Eastern European post-Communist path. Growing inequality they have, but marketization took place under the auspices of the Communist Party, which has then been able to take credit for growing GDP. At least so far, no sense of betrayal is evident.
Theoretical implications
These results do not entirely accord with our preferred reference groups and reality blend theory: Those arguments lead to the right prediction in direction, but suggest a change smaller in magnitude than is found (Hypothesis 5, in Figure 4, row 1, column 5).
It seems that something is distinctive about post-Communist Eastern Europe above and beyond its social composition, inequality, and socioeconomic development. Elite and citizenry reactions to massive political shocks can be highly variable (Rohlinger, 2009). In this case, we argue that the citizenry experienced a massive sense of betrayal: Eastern Europeans sacrificed much on the promise of out-performing capitalism by providing working class prosperity and social equality. The obvious, overwhelming affluence of the working class in the market economies of Western Europe showed that Eastern European Communists had betrayed their ideals and misled their citizens.
Materialist, dominant ideology, and system justification theories correctly predict that Communist societies will be perceived as egalitarian (Figure 4). They also predict that post-Communist societies will be perceived as inegalitarian but, like our preferred reference group and reality theory, they get the direction right but the magnitude much too small. (Postmaterialism made no prediction on this.)
Type of Society—Ideals and Reality: Hypothesis 6
We have seen that people in different countries have drastically different perceptions of what their societies are actually like, ranging from overwhelmingly egalitarian in Scandinavia and Hungary under Communism (Figure 9, near the top) to overwhelming inegalitarian in post-Communist Eastern Europe (Figure 9, toward the bottom), and everything in between.
But despite these huge differences in the perceived reality of inequality, there is virtual worldwide consensus that societies ideally ought to be egalitarian, with scores between 85 and 95, near Type C and Type D (Figure 5, column 8). Indeed, almost all the countries’ ideals fall in a narrow band between Type C (80) and Type D (93). This shows a broad international consensus in the degree of desired inequality.
There is consensus in market economies: Despite middling Types B and C being common in reality (Figure 12, column 1), preference is firmly for the highly egalitarian Type D (53%; column 2) or even the extreme Type E (25%). Hungary in Communist times was even more extreme: middling views of actual egalitarianism (not much different from market economies) but an overwhelming preference for egalitarianism, including 42% for the extreme Type E (columns 3 and 4). Formerly Communist nations see themselves as dramatically inegalitarian, as we have seen (48% in elitist Type A) but their ideals are at least as egalitarian as anyone else’s (the vast majority preferring egalitarian Types D and E; columns 5 and 6). These results are strongly contrary to the Homans thesis that ideology directly mirrors social structure.
Rejected alternative theory: Dominant ideology.
In all countries, the ideal degree of equality exceeds the actual degree, that is, the ideals are all closer to Type E than are the actual means. But the size of the gap varies greatly, at a minimum around 10 in Denmark and Norway and a maximum around 70 in Bulgaria, Latvia, and Ukraine (Figure 5). The dramatic variations in these gaps have important implications for politics.
This preference for egalitarianism is widely shared within societies. The multivariate analysis shows that there are no significant SES differences, no significant differences between rich societies and poor, no significant difference between Communist societies and market economies, or even between previously Communist nations and the rest of the world (Figure 7, panel 3).
Theoretical implications
The fact that the ideals are so widely shared in vastly different societies—long-established market economies, Communist dictatorships, and post-Communist transition economies alike; high GDP and developing alike, all these together explaining barely 1% of the variance in ideals—argues decisively against dominant ideology and system justification theories (Figure 4, row 6). It also decisively rejects the argument that “is” implies “ought” (Homans, 1974; Kelly & Enns, 2010). Moreover, the pervasiveness of the ideal is also evidence against the postmaterialism hypothesis that values and attitudes change decisively with development.
Instead, the egalitarian zeitgeist pervades this diverse array of countries, at least in this historical period, and regardless of institutional history or level of development.
Summary
Our results strongly suggest that the reference group and reality blend hypothesis provides a better description of people’s societal imagery than does any major competing hypothesis. People’s perceptions reflect a complex mix of objective experience with subjective projections from their own reference group. Because socioeconomic development induces educational and occupational upgrading, the pools of potential reference group members in the upper strata expand with development. Correspondingly, the pools of lower strata reference group candidates contract. Thus, on average, the reference group for someone in any given SES location shifts upwards during socioeconomic development, leading to a positive effect of socioeconomic development on perceived equality, all else equal. Together with reality, reference groups play an important role.
Thus, our results continue a growing tradition of findings that began with a two-nation comparison and have since blossomed into a database of 44 nations, some of them spanning two decades. Results on the original Evans, Kelley, and Kolosi Hypothesis 2 (Evans et al., 1992) were ambiguous in their two-nation comparison, but are clearly supported in this study of many more nations:
Hypothesis: The higher one’s objective status, the more egalitarian one perceives society to be.
The political implications of this are clear in societies where political conflict is largely about class and inequality, for example, the United States, Sweden, or Australia (Evans & Kelley, 2004a; J. Kelley & Evans, 1995).
This evidence is counter to materialist interpretations of people’s images of the stratification system, including Marx’s, which anticipate perceptions to directly reflect reality (except potentially in the case of Marx’s “false consciousness,” in which case the predictions collapse into those for the dominant ideology hypothesis). This does not mean that people are completely deluded. Reality is present, but it is only part of the perception recipe. Findings with respect to these hypotheses are in keeping with prior research, but provide confirmation with new data and on a grander scale. Such confirmation is especially valuable give the replication crisis across the sciences (John, Loewenstein, & Prelec, 2012; Pautasso, 2010). We propose that this should be the working hypothesis in this domain.
This article extends the logic of the reference groups and reality blend thesis to apply to the consequences of national development: Socioeconomic development increases the supply of high status people through educational and occupational upgrading processes, so, as socioeconomic development proceeds, the average person’s reference group will be increasingly likely to contain high SES members. As a result, the reference group implication is that even above and beyond one’s own social position, one’s nation’s level of socioeconomic development will influence perceptions of inequality:
Hypothesis: The more developed one’s society, the more egalitarian one perceives it to be.
Again, the political implications of this are clear in societies where politics has mainly to do with class and inequality, rather than, for example, religion (Evans & Kelley, 2004a; J. Kelley & Evans, 1995).
The results of our multilevel analysis clearly support this contextual hypothesis. Earlier research on this topic was hampered from systematically assessing societal effects because of the small database: It is only with years of labor in this vineyard by many survey research teams that the field has garnered data from enough countries to analyze the societal/contextual effects as well as the individual-level effects.
Although reference groups play a vital role in the reference groups and reality blend hypothesis, they appear only indirectly through their many effects, especially on perceptions of inequality. With two exceptions, none of our data sets has direct information on reference groups. But the data that do exist, for the United States and Australia, are informative (Figure 13). These two surveys include information on the perceived location in the social hierarchy of respondent’s occupation, most of respondent’s friends, and most of respondent’s co-workers and they also include a social-legacy measure of the perceived location of respondent’s parents. An OLS analysis including controls for respondent’s actual stratification position show that people with higher reference groups do indeed see their society as more egalitarian, as our theory posits. Thus, not only are the results for many nations consistent with the reference groups’ interpretation but also direct tests detect reference group influence in these two nations.
Reference groups shape perceptions of the type of society, those with higher reference groups seeing their society as more egalitarian.
The “reality” part of the blended reference groups and reality hypothesis is also partly supported in our analysis of many nations, although the picture is more complex. An important feature of our analysis is that it includes both GDP and the Gini as predictors. They are correlated and, moreover, theory and logic clearly imply that both should affect our cognitive maps of inequality, so omitting either would risk confounding-variables bias, whereby the predictive power of the omitted variable is incorrectly attributed by proxy to the variable that is included. This risk is very real. It has, for example, been shown that the field was fundamentally misguided about the apparently substantial effect of inequality on subjective well-being, because the early models included only the Gini and omitted GDP. Better models including both show that low levels of development reduce subjective well-being, but, net of that, income inequality does not: The problem is poverty, not inequality (J. Kelley & Evans, 2016). By contrast, in the case of perceptions of societal inequality, both prosperity and inequality seem to matter, at least in most contexts.
We find that in countries without a Communist past, actual inequality (as measured by the Gini) is reflected in people’s perceptions of the structure of their society. This is a robust link and is net of GDP and individual characteristics. However, it appears to be context dependent: In previously Communist countries, the impact of actual inequality is not statistically significant, so it is probably absent, or simply too small to matter. This is in contrast to the prosperity linkage (as measured by GDP per capita), which everywhere leads people to see their society as more egalitarian.
This analysis of 44 countries makes it clear that there is more going on than anticipated by the reference group and reality blend thesis proposed in the foundational studies (e.g., Evans et al., 1992) based on the sparse data then available, in a time when Communism still dominated Eastern Europe. Emerging from the broader range of nations and more recent post-Communist data included here, an extension is needed. These new analyses provide evidence also of what we may call the “institutional illusion” hypothesis: The more pervasive a society’s elite-controlled institutions and the stronger the elite’s control over information resources, the more will public perceptions incorporate the imagery perpetuated by the elite. More specifically, under Communism, Eastern Europeans’ perceptions of their social class system were substantially more egalitarian than predicted by their socioeconomic development and actual inequality, suggesting that you can fool a majority of the people at least some of the time. This is the only aspect of the findings that aligns with the dominant ideology hypothesis: In societies with impermeable boundaries where a dominant elite controls all information resources and institutions, it is possible to perpetuate at least some illusions (in this case, the prosperity of the working class vis-à-vis their peers in capitalist countries). But, even here, the dominant ideology hypothesis provides no prediction about what happens in the aftermath of the collapse of such a dominant elite.
After the collapse of Communism, the citizenry’s perceptions plunged toward an extremely inegalitarian image of society. We suspect that this is partly a reaction of disappointment and anger at realizing that the previous image was an illusion (Nannestad, 2008; Zagorski, 1994). As Shakespeare (Sonnet 94) put it, “Lilies that fester smell far worse than weeds.” When a pervasive (successful) deception is revealed, trust and confidence are shattered and the new social construction of reality rising from the rubble is probably unrealistically negative about the situation and relationships in which one was deceived (Cleary & Stokes, 2009; Weber & Carter, 2003).
Thus, societal inequality imagery has multiple causes—reference groups, reality, and regime. The public’s images of inequality are a complex mixture not only of reality (as expected) but also of their reference groups (determined by the socioeconomic structure of their society) and by their nation’s institutions and political history. Perceptions of inequality are complex phenomena mixing economics, sociology, and politics in ways we are only beginning to untangle.
Simpler than the perception is the ideal. In all societies and all social classes, there is a preference for “top heavy” egalitarian societies. This is consistent with Jasso’s posited “universal longing for equality” (Jasso, 1980; see also Robinson & Bell, 1978, on the egalitarian zeitgeist). Research on ideal occupational earnings (also called legitimate pay or legitimate rewards) has reached a different conclusion: universal rejection of equality or even near equality (J. Kelley & Evans, 1993, 2009; Kolczynska & Merry, 2016). We propose that both things are true, but that they mean slightly different things to people: People want status differences in occupational earnings preserved (although they disagree over the size of those differences), but they strongly prefer societies where most people are high SES.
Thus, people want equality by educational and occupational upgrading rather than equality by earnings compression.
All in all, the evidence supports our reference group and reality blend hypothesis about influences on images of inequality. The findings strongly support four of its five predictions, and the fifth prediction (for Hypothesis 5) was in the correct direction but it alone does not predict the great magnitude of the change in inequality imagery in post-Communist Europe, and hence, needs to be supplemented with the institutional illusion hypothesis (see also Figure 4).
By contrast, the other theoretical approaches made many fewer correct predictions. The materialist approach and the dominant ideology approach each correctly predicted only one result: that their citizenries would perceive Communist-era European Communist societies as egalitarian. As with the blend approach, the materialist approach was in the correct direction for Hypothesis 5, but seriously under-predicted the magnitude of the change. The dominant ideology hypothesis did not have a strong prediction for the post-Communist situation. The postmaterialist approach made no correct predictions. These approaches may be useful in other domains, but in examining the sources of the general public’s images of inequality, the reference group and reality blend hypothesis is much the most strongly supported.
Discussion
There are many important aspects of people’s perceptions, interpretations, and evaluation of societal inequality, and this article analyzes only one of them: visualization of the structure of society. Hence, it is worthwhile considering the bigger picture: Where do these images of society and our current understanding of them as generated by reference groups, socioeconomic development, income inequality, and institutional history fit in the larger tradition of research on the culture and social psychology of inequality?
The Communist Legacy: Convergence or Path Dependence?
In terms of the objective sources of these perceptions, an important contribution of our findings to the analysis of subjective inequality is the dramatic importance of the Communist legacy, which is large in itself and moreover completely eradicates the effect of objective inequality. We see here that the Communist legacy has a profound effect strongly directing citizens of these countries toward visualizing their societies as grossly unequal. Similarly, some early work on the ISJP data suggested that people in European Communist-legacy countries were, at least in the early 1990s, much more likely than their peers in the West to see poverty as externally caused (Kreidl, 2000). However, over the transition period, we see several signs of convergence. For example, (a) the determinants of political party preference have shifted in the European Communist-legacy countries toward the general advanced-society pattern of many, generally small effects of different demographic and class-related characteristics (Mateju & Vlachova, 1998); (b) at least in the Visegrad countries, perceived occupational earnings roughly track objective trends (Buchel, 2013; Kolczynska & Merry, 2016); and (c) the legitimacy of different bases of reward is now very close in post-Communist countries to other countries at the same level of development (Evans et al., 2010; Peoples, 2013). Thus, an important challenge for future research will be to map which perceptions, attitudes, and values are shaped by an ongoing Communist legacy and which of them show no lingering Communist influence.
The traditional “Convergence or path dependence” question is too simple for this institutional impact. Instead we must ask the following questions: In which kinds of topics is there convergence? In which kinds of topics is there path dependence or even further divergence? Furthermore, given the strong difference in Gini effects on inequality perceptions between the European Communist-legacy countries and all others that we find here, it will be important for future research to check for interactions of this institutional legacy and contemporary inequality in effects on other potentially related outcomes such as subjective social class, subjective social status, perceptions of occupational earnings, and the like.
Socioeconomic Development
The analysis presented here is, to our knowledge, the first to assess the effect of socioeconomic development (as indexed by GDP) on images of inequality. We find that citizens’ perceptions of their societies move in a strongly egalitarian direction as those societies develop. In kindred research, socioeconomic development as indexed by GDP also has a powerful effect on other aspects of subjective inequality, specifically identity and system evaluation. For example, the impact (net of individual characteristics) of socioeconomic development lifting subjective social location (subjective social class) is well documented (Curtis, 2016; Evans & Kelley, 2004b; Lindemann & Saar, 2014). Our interpretation of the positive effect of GDP on perceptions of equality, emphasizing occupational upgrading as a process increasing the availability of high status reference group members, and thence, changing perceptions is also consistent with a finding that the percent of professionals in the country’s occupational structure increases subjective social status (Lindemann & Saar, 2014). Prior research also shows that socioeconomic development changes evaluations of inequality as well as perceptions: With socioeconomic development, people become less tolerant of inequality, at least based on an implicit measure of tolerance (J. Kelley & Evans, 2009).
Objective Inequality
Objective inequality has complex effects. In European Communist-legacy societies, objective inequality has little or no effect on perceptions of how egalitarian people believe their society is: Development matters, but inequality does not. By contrast, in the rest of the world, objective inequality has powerful effects on the perceived egalitarianism of society and there are also indications of effects on subjective social class (Andersen & Curtis, 2012).
These effects, nonetheless, are smaller than one would expect in light of the materialist hypothesis (one of the hypotheses rejected by the evidence) that subjective inequality simply mirrors objective inequality. Rather, this evidence is consistent with the blend hypothesis that reference groups mute and potentially distort perceptions of social inequality.
Future Research
The conceptual model in Figure 14 shows where our analyses fit within a larger research tradition on the culture and social psychology of inequality (Figure 14, on the left, in blue). The model begins with “reality” predictors on the left (individual level and national level) followed by their implications for the SES of individuals and their reference groups. Next, in causal order come perceptions of inequality for the society as a whole (the focus of this article) and also perceptions and ideals about inequality in occupational earnings. The next block is subjective social class; finally comes subjective social mobility. For simplicity, we have not included political orientation in the conceptual model because it is very likely involved in reciprocal relationships with nearly all the variables to the right of the diagram. This aspect of the model, in particular, will demand careful attention in future research.
Conceptual model for the larger program of research of which this article is a part.
An important next step will be expanding current models of subjective social class and subjective social mobility to assess the consequences (direct and indirect) of socioeconomic development and national levels of inequality (Andersen & Curtis, 2015; Curtis, 2016; Evans & Kelley, 2004b; J. Kelley & Evans, 2009). The Procrustes hypothesis (Evans et al., 1992) proposes that, all else equal, holding a strongly inegalitarian image of one’s society weighs down one’s subjective social class (little room at the top), hence the more egalitarian one’s image of society, the higher one’s subjective social class, and the greater one’s subjective social mobility. The sharp differences we have seen between European Communist-legacy nations and the rest of the world suggest there will be complications there, although probably not in China which, like Taiwan and Japan, seems unexceptional.
The collapse of European Communism dramatically increased perceptions of inequality, shifting from images of an egalitarian society with most people in the middle to almost medieval images of a society with a tiny elite at the top and a vast impoverished mass at the bottom. We have seen some reasons why this change came about. But the future evolution of these images, and their consequences for politics and social life, remain to be explored.
Footnotes
Authors’ Note
This study continues a line of work the authors began with Tamas Kolosi in the late 1980s, and we stress the continuing importance of his foundational contribution to the measurement of images of class.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Sponsorship of the data collection was provided by a host of funding agencies who are fully acknowledged in the electronic codebook for this module on the ISSP website: 
. Harmonizing and merging the data into a pooled cross-national file was directed by the authors and supported by two research and infrastructure grants from the Australian Research Council to the Melbourne Institute of Applied Economic and Social Research, the University of Melbourne.
Notes
Author Biographies