Abstract
This study tests specific competing hypotheses from social dominance theory/realistic conflict theory (RCT) versus system justification theory about the role of social status. In particular, it examines whether system justification belief and effects are stronger among people with low socioeconomic status, and in less socially developed and unequal nations than among better-off people and countries. A cross-national survey was carried out in 19 nations from the Americas, Western and Eastern Europe, Asia, and Oceania using representative online samples (N = 14,936, 50.15% women, Mage = 41.61 years). At the individual level, system justification beliefs, right-wing authoritarianism, social dominance orientation, national identification, sociopolitical conservatism, sex, age, and social status were measured. At the national level, the human development index and the Gini index were used. Multilevel analyses performed indicated that results fit better with the social dominance/RCT approach, as system justification was higher in high-status and developed nations; further, associations between legitimizing ideologies and system justification were stronger among high-status people.
Keywords
In this article, we examine how group status is associated to beliefs justifying status quo in the social system. We analyze whether social status is positively or inversely related to system justification at the individual level, and also whether the effect of legitimizing system ideologies is moderated by group status and by country-level contextual variables. Different approaches have proposed competing hypotheses for the relationship between group status and legitimizing ideologies. These may broadly be grouped into two camps.
Social Dominance Theory/Realistic Conflict Theory (RCT)
According to social dominance theory, most stable, functioning societies are organized as group-based hierarchies (Sidanius & Pratto, 1999, 2004). Almost by definition, this form of social organization is detrimental to lower power, lower status groups. The question prompted by these two facts, then, is, why is group dominance stable and almost universal across cultures?
Social dominance theory argues that one reason that group-based dominance hierarchies are stable is that legitimizing ideologies help coordinate beliefs, actions, and institutional practices that maintain hierarchy. For example, in the United States, belief in equal opportunity leads people to assume that people who are not doing well economically are there because of their own lack of effort. Therefore, it is widely believed in the United States that inequality is not a matter of injustice but “failure to take advantage of equal opportunity” (Lerner, 1980; Pratto, Sidanius, Stallworth, & Malle, 1994). In other words, the ideology of equal opportunity legitimizes inequalities in the social system, and maintains system stability despite inequality in the United States.
But social dominance theory does not assume that the maintenance of inequality implies that subordinated groups acquiesce or approve of their inferior power position (Sidanius & Pratto, 1999). Rather, social dominance theory predicts that the extent to which people tolerate group-based inequality is in general lower among people with inferior, rather than superior, group position (Sidanius, Liu, Shaw, & Pratto, 1994). This prediction has been borne out in scores of nations and studies (e.g., Pratto et al., 2000). A meta-analysis of group difference effect sizes on social dominance orientation (SDO), or general orientation toward approving or disapproving group inequality showed robust differences between dominant or high-status and subordinated or low-status groups (Lee, Pratto, & Johnson, 2011). In addition to predicting that men and women and salient superior/inferior groups within societies, such as those defined by religious, racial, immigration status, ethnic, or other cleavages, will differ on SDO (with the higher powered group being higher in SDO), social dominance theory makes two further predictions regarding SDO and endorsement of hierarchy-legitimizing ideologies or “myths,” which are understood as those ideologies that support unequal distribution of wealth, power, and other resources that define group-based hierarchies (Sidanius et al., 1994; Sidanius & Pratto, 1999).
The first prediction is that SDO will correspond to endorsement of legitimizing myths. For example, people higher on SDO are more in favor of political–economic conservatism, racism, nationalism, and sexism, and these ideologies in turn predict positive attitudes toward social policies and political parties and candidates that sustain hierarchy and negative attitudes toward those that would attenuate the level of inequality in a society. The rationale behind this association is that these hierarchy-enhancing ideologies provide support for group-based hierarchies (Sidanius, Levin, Federico, & Pratto, 2001). For instance, Pratto, Stallworth, and Conway-Lanz (1998) found that SDO significantly predicted nationalism and conservatism in the United States, and these in turn predicted support for beginning the 2003 Gulf War against Iraq. According to the ideological asymmetry hypothesis of social dominance theory, this association between SDO and nationalism and conservatism should emerge primarily for members of high-status groups (see, for example, Levin, Sidanius, Rabinowitz, & Federico, 1998; Rabinowitz, 1999; Sidanius, Feshbach, Levin, & Pratto, 1997). In other words, the prediction of nationalism and conservatism of SDO should be true primarily (e.g., stronger, more consistent across cultures) among hegemonic, high-status group members. This is because the societal myths that disadvantage them are not as compelling in subordinate groups; they have psychological forces such as social identity that weaken this relationship. For example, identifying more with one’s ethnic, gender, or sexual orientation group than with their complements correlates more strongly with SDO among the higher power groups (e.g., Whites, men, and straight people) than among lower power groups (e.g., Blacks and Hispanics, women, and lesbian, gay, bisexual, and transgender [LGBT] people; Pratto & Stewart, 2012; men and women; Wilson & Liu, 2003).
These predictions of social dominance theory comport with realistic group conflict theory, which was a major influence in the development of the integrative theory of social dominance. The classic position of RCT (Huddy, 2003; Sherif, Harvey, White, Hood, & Sherif, 1988) assumes that people try to maximize their rewards and minimize their costs associated to their social positions and group membership (Moghaddam, 2008). Given this antecedent, the theory proposes that intergroup behavior will reflect group interests (Brown, 2010), and people will prefer social arrangements that maximize in-group interests (King, Knight, & Hebl, 2010), such as redistributive policies among low-status individuals and greater concentrations of power or money among high-status people or groups. This theoretical framework has received long-standing empirical support in social psychology (e.g., Brief et al., 2005; Echebarria-Echabe & Fernández Guede, 2003; Insko et al., 1992).
Although both social dominance theory and RCT have proposed that high-status groups will legitimize the social arrangements to a higher degree than low-status groups, Kunst, Fischer, Sidanius, and Thomsen (2017), based on the former, suggested a complementary approach. In their argument (consistent with that of social identity theory, see Tajfel & Turner, 1986), societal stability is a moderator, because when social systems are stable, challenging the status quo is highly costly and risky for low-status individuals. According to this view, group-based hierarchies would not be as detrimental for these groups when societal stability is high.
System Justification Theory
An alternative approach to social dominance theory and realistic group conflict theory is system justification theory (Jost & Banaji, 1994, 2004). Such as social dominance theory, system justification theory addresses why unequal social systems are maintained, and it posits that people’s tendency to justify the system is an essential psychological process toward that end (Jost, Banaji, & Nosek, 2004). According to system justification theory, people are actively motivated to justify the social, political, and economic arrangements to which they belong. In that sense, they perceive that these systems are legitimate and fair, even if they contradict their own material interests (Jost et al., 2004; e.g., Jost, 1997; Proestakis & Brañas-Garza, 2016). This theoretical approach proposes that the main psychological mechanisms of the system justification are related to the fulfilling of epistemic, existential, and relational needs (Hennes, Nam, Stern, & Jost, 2012). So, system justification allows people to achieve certainty, to reduce external threats, and to share a common reality with other individuals, even in the face of social inequality. In that way, Kay and colleagues (Friesen, Kay, Eibach, & Galisnky, 2014; Kay & Friesen, 2011; Kay, Laurin, Fitzsimons, & Landau, 2014; Knight, Tobin, & Hornsey, 2014; Tullet, Kay, & Inzlicht, 2015) have proposed that people prefer to have control over their lives and environment (even if it is not objectively true and it is a sort of delusion), and because of that preference, they perceive these constructed social arrangements as fair and legitimate.
One of the most controversial claims of system justification theory posits that those who are most disadvantaged by unequal social systems are those with the highest motivation to perceive the system as fair and good: “this hypothesis is that members of disadvantaged groups are even more likely than members of more advantaged groups to provide ideological support for the very social system that is responsible for their disadvantages” (Jost, Pelham, Sheldon, & Sullivan, 2003, p. 30). Thus, system justification theory predicts that the hierarchy-enhancing belief in a “just” social system should be higher among lower power groups than among higher power groups. Specifically, the so-called strong SJT hypothesis (Jost, Pelham, Sheldon, & Sullivan, 2003) posits that system justification beliefs (i.e., those ideologies that justify the status quo; Jost & Hunyady, 2005) will be stronger among low-status than high-status groups and individuals, under certain circumstances. Within system justification theory, several ideologies have been identified, such as the protestant work ethic, meritocratic ideology, fair market ideology, among others (Jost & Hunyady, 2005). Most of them have been treated as cross-cultural ideologies, but recently Osborne, Yogeeswaran, and Sibley (2017) showed that in addition to these, there may be also culture-specific ideologies, such as postcolonial belief systems in New Zealand.
According to system justification theory, among high socioeconomic status people, the motivation for justifying the self, group, and system is compatible. However, among low socioeconomic status people, the motivation to justify the system contradicts the need to have a positive view of the self and in-group (Jost, Pelham, Sheldon, & Sullivan, 2003). This argument leads to the reasoning that low-status individuals, under certain circumstances, are more prone to justify the system, because they experience cognitive dissonance between justification motives. These individuals solve this dissonance by emphasizing system justification (Jost et al., 2003). In other words, it is suggested that low-status group members will be more inclined than high-status group members to provide more intense justifications to rationalize and uphold the prevailing systems and status quo, in particular in contexts involving high power distance and unequal societies (Jost, Gaucher, & Stern, 2015; Jost et al., 2003).
The particular circumstances in which the low-status individuals are more motivated to justify the system are related to contextual factors (Jost et al., 2003). The first factor is the societal income inequality, because in those highly unequal nations, cognitive dissonance should be higher. Thus, system justification tendencies are supposed to increase in societies with more substantial group inequalities. This idea is cogent to results showing that high power distance or authoritarian values are more prevalent in nations with high income inequality, measured by the Gini index (Hofstede, 2001). The second factor is the democratic context, because in democratic nations, people will perceive more control over their own outcomes, and, thus, dissonance should be higher. The last factor is group identity, so low social status identity should be associated with increases in system justification (for critical reviews, see Owuamalam, Rubin, & Spears, 2016; Owuamalam, Rubin, Spears, & Weerabangsa, 2017).
It is important to emphasize that system justification theory does not explicitly propose that national-level income or development would be related to system-justifying beliefs. However, previous research on the strong form of SJT has paralleled individual-level arguments with those at the country or national level (e.g., Caricati, 2017; Caricati & Lorenzi-Cioldi, 2012). In addition, from a theoretical point of view, and following statements of system justification theory (Jost et al., 2003), we expect that a country’s development should be a contextual factor, as low-status people from less developed countries will be objectively more deprived than those in more developed contexts.
The arguments proposed by the system justification theory have received empirical support in system justification among ethnic groups (Henry & Saul, 2006; Sengupta, Osborne, & Sibley, 2015), and among social status groups (e.g., in Turkey; Dirilen-Gumus, 2011). Nevertheless, prior studies either lack empirical evidence regarding the broader hypotheses we propose or neglected to observe these relationships in cross-cultural contexts (Brandt, 2013; Caricati, 2017; Caricati & Lorenzi-Cioldi, 2012; Trump & White, 2014). Specifically, Caricati (2017) directly tested the strong form of SJT in a sample composed by 36 countries using data from the International Social Survey Programme (ISSP). The study showed that high-status individuals perceived that income differences in their societies were fairer than low-status individuals. In addition, he did not find that the contextual variables, theoretically related to collective social status and inequality, moderated the relationship between status and perceived fairness (in accord with previous studies; see Brandt, 2013; Caricati & Lorenzi-Cioldi, 2012).
The present study compares the predictions made by social dominance theory and RCT with those proposed by system justification theory regarding the role of social status. We extend previous studies in the area (e.g., Brandt, 2013; Caricati, 2017; Caricati & Lorenzi-Cioldi, 2012), testing whether both subjective and objective social–economic status are inversely related to system justification, and the role of psychosocial dispositions and ideologies, such as SDO, right-wing authoritarianism (RWA), conservatism, and national identification in this relationship. This is the main contribution of the article, because to date, studies testing the strong form of SJT have focused only on the social–economic status variable, and have excluded psychosocial dispositions and ideologies of theoretical interest. In addition, we test the influence of national-level variables, such as social development and income inequality. Finally, we focus on subjective social–economic status, because it is a measure that can be used across cultures. Other indices of status, such as ethnicity, are highly context dependent and cannot be directly compared across countries.
Predictors of System-Justifying Beliefs
SDO has proved to be reliably related to measures that directly or indirectly suggest support for system-justifying beliefs. Although these beliefs are theoretically related to support for the status quo (Jost & Hunyady, 2005), given that most social arrangements are already unequal (Jost et al., 2015), the preference for unequal group relationships is, thus, also associated with system-justifying beliefs (for a similar argument, see Brandt & Reyna, 2012; Sengupta et al., 2015). For example, SDO has been associated with opposition to equality (Sibley & Duckitt, 2010); decreasing support to the underprivileged, as well as less support for women in leadership positions (Pratto et al., 2013); prejudice and conservatism (Ho et al., 2012); support for the death penalty, opposition to homosexual civil rights (Pratto et al., 2000); denying climate change (Jylhä & Akrami, 2015); and preference for advantaged countries (Does & Mentovich, 2016); among other variables. In addition, SDO tends to be lower in more democratic and equalitarian countries (Fischer, Hanke, & Sibley, 2012). Based on these findings, we argue that SDO will be empirically related to system justification.
A conceptually distinct, although related individual difference variable orienting a person toward system justification beliefs is RWA. RWA expresses the desire for coercive social control, obedience, and respect for the tradition; whereas, SDO expresses the desire for intergroup domination (Duckitt & Sibley, 2009). RWA emerges as a response to the perception of a dangerous world; whereas, SDO emerges as a response to a threatening and fear-inducing environment. Onraet, van Hiel, and Cornelis (2013) found that countries where there was a high level of threat, as measured by country-level variables, had higher levels of RWA.
System justification has been linked to RWA (Osborne & Sibley, 2014), following a similar rationale, parts of which have already been described for SDO (for details, see Duckitt & Sibley, 2009, 2010). In another study (Wilson & Sibley, 2012), SDO and RWA predicted system justification, and these effects were qualified by interactions with political conservatism, so that the lowest levels of conservatism (or highest levels of political liberalism) were found in those reporting lower levels of both SDO and RWA. Thus, RWA and SDO appear to function as predictors of system justification. In addition, we propose that SDO and RWA predict system-justifying beliefs, and not the reverse causal path, because both SDO and RWA are thought of as dispositional variables (Altemeyer, 1998; Pratto et al., 1994). Although there are contextual factors that influence both SDO and RWA (e.g., social roles within society), there are individual differences beyond them (for a critical review, see Duckitt & Sibley, 2010).
A related construct is political conservatism, though it has in recent years received less theoretical attention from social psychologists. Conservatism has been equated to the left- and right-political preferences, and in the United States has been operationalized as the preference for the Democratic or Republican Party. A theoretical approach underpinning this is motivated social cognition (Carney, Jost, Gosling, & Potter, 2008; Jost & Amodio, 2012; Jost, Stern, Rule, & Sterling, 2017; Nam, Jost, & Van Bavel, 2013), which using a cross-cultural meta-analysis (Jost, Glaser, Kruglanski, & Sulloway, 2003) defines conservatism as resistance to social change and opposition to equality (for a critical review, see Feldman & Johnston, 2014). According to this approach, conservative ideology is more appealing to individuals with high needs for structure and order, and low tolerance for uncertainty, and because of this, they endorse conservative beliefs. Given this definition, system justification theory proposes that conservatism is related to system justification, with conservative people showing stronger system-justifying beliefs (Kay, Czaplinski, & Jost, 2009; Napier & Jost, 2008).
Relatedly, another factor that legitimizes status quo is national identification, as high identifiers tend to reject class struggle and social conflict within a nationality. National identification is conceived as the categorization of people within a country as members of a national in-group, regardless their class, ethnicity, and so forth. In other words, national identity emphasizes that all people are members of a superordinate in-group that is supposed to share a common fate and common interests. National identification has proved to be a cognitive tool that allows people to justify the system (Carter, Ferguson, & Hassin, 2011). Although research on system justification theory has not been clear about the causal order and mechanisms involved in the relationship between national identity and system justification, some studies suggest that the attempt to create harmony by emphasizing a common in-group identity results in a greater acceptance of inequality in society. Jasko and Kossowska (2013) found that when the participants’ in-group is disadvantaged, making the superordinate identity salient (i.e., Polish national identification) led to a stronger justification of inequalities in the financial system, which is coherent with the statement that national identification can distract attention from differences between groups within a nation, focusing attention on what is common between them, and this in turn predicts system-justifying beliefs (Sengupta & Sibley, 2013).
National identity provides a superordinate identification among people from the same country and this identification makes the common elements more cognitively accessible than the differences between social groups within the nation. So, people endorsing national identity should be more motivated to see commonalities, and less the objective differences among social groups in their country. Following these arguments, we expected national identification to act as a predictor of system-justifying beliefs. Moreover, and based in the same rationale, we propose that national identification moderates the relationships between the proposed legitimizing ideologies and system justification. Among those who are less identified with their national context, the role of other individual difference variables, such as RWA, SDO, and political conservatism should be stronger, as national identity can smother the effect of other individual differences.
Hypotheses
Given the presented theoretical and empirical antecedents, in this article, we will test the following hypotheses in a sample of 19 countries:
Method
Sample
The sample was collected during September 2015, as part of the survey conducted by the Digital Influence Project, which includes a research partnership between lead scholars in Europe and New Zealand and others around the world. A curated online panel of more than 10 million people by international polling firm Nielsen was employed with stratified quota samples to represent national populations based on census information on gender, age, and region. In this survey, 22,034 individuals participated from 22 countries (52.15% women, Mage = 40.63 years, SD = 14.574). Overall, cooperation rate was relatively high, averaging 77% across the panel (American Association of Public Opinion Research, 2016; for details, see Gil de Zúñiga & Liu, 2017). We selected cases with valid data in all variables for analysis, and we excluded Taiwan, because of the lack of available national-level variables, and India and South Africa, given that we obtained only city samples there. This procedure retained 14,936 participants (50.15% women, Mage = 41.61 years, DE = 14.634) from Argentina, Brazil, Chile, China, Estonia, Germany, Indonesia, Italy, Japan, South Korea, New Zealand, Philippines, Poland, Russia, Spain, Turkey, United Kingdom, Ukraine, and United States. The sample sizes ranged from 647 (Chile) to 962 (Russia).
Instruments
System Justification Scale (SJS)
The main dependent variable of the study is a brief version of the SJS (Kay & Jost, 2003). We selected and adapted four items from the original version: “In general, I find society to be fair”; “In general, my country’s political system operates as it should”; “Everyone in my country has a fair shot at wealth and happiness”; and “My country’s society is set up so that people usually get what they deserve.” The responses ranged from 1 (disagree completely) to 7 (agree completely). This scale was highly reliable in the global sample (α = .86).
SDO Scale
We used a brief version of the SDO Scale (Pratto et al., 2013): “In setting priorities, we must consider all groups” (reversed and recoded item); “We should not push for group equality”; “Group equality should be our ideal” (reversed and recoded item); and “Superior groups should dominate inferior groups.” The answers ranged from 1 (strongly oppose) to 7 (strongly support). The scale was acceptably reliable in our sample (α = .64).
RWA
We selected three items from the Right-Wing Authoritarianism Scale (Altemeyer, 1998): “Obedience and respect for authority are the most important virtues children should learn”; “Our country needs a powerful leader, to destroy the radical and immoral elements in society today”; and “In these troubled times, laws have to be enforced without mercy, especially when dealing with the agitators and revolutionaries who are ‘stirring things up.’” The answers ranged from 1 (disagree completely) to 7 (agree completely). This scale was acceptably reliable (α = .67).
National identification
We used a modified version of the National Identification Scale (Huddy & Khatib, 2007), with agreement (1-7) with the following statements: “Being [Nationality] is very important to me,” “I feel that I am a typical [Nationality],” “The term [Nationality] describes me well,” and “I identify with my nationality.” The scale was highly reliable (α = .92).
Conservatism
We used a three-item measure of liberal–conservatism, with the self-placing in a 0 (strong liberal) to 10 (strong conservative) continuum (Pratto, Stallworth, & Sidanius, 1997). The items were related to economic, political, and social issues. The scale was highly reliable (α = .89).
Subjective social status
It was included as an item to assess the perceived social status: “On a scale of 1 to 10, with 10 being people who are the most well off in society, and 1 being the people who are least well off, where would you describe your position?”
Income
We included the house income recoded in five groups: 0 to 10, 11 to 30, 31 to 70, 71 to 90, and 91 to 100 percentiles.
National-level variables
We included two national-level variables. The first variable is the Gini coefficient (Yitzhaki, 1979), where a high score expresses high income inequality. We obtained data from the Organisation for Economic Co-operation and Development (2017) and United Nations Development Programme (UNDP; 2017b) databases. The second variable is the human development index (HDI; UNDP, 2017a), which is a multidimensional measure of country development, and an index of social development, including income, life expectancy, and educational levels. High scores mean higher human development.
Individual-level control variables
We controlled for the effect of age and gender (1 = female, 0 = male).
Procedure
Participants were recruited from Nielsen (a multinational survey company based in the United States that commissions survey panels in other countries through partnership agreements) online panels. This company contracts cross-cultural panels with local survey organization that have more than 10,000,000 individuals enrolled. Nielsen included individuals in the sample using stratified quota techniques, to obtain a demographic distribution similar to the national level on gender, age, and region of residence. This procedure led to a response rate of 77%. The survey was conducted online from September 14 to 24, 2016, using Qualtrics (for details, see Gil de Zúñiga & Liu, 2017).
Analyses
We conducted a series of multilevel linear regression analyses, given the hierarchical structure of data (Gelman & Hill, 2007). One level of analysis pertained to participant characteristics (individual-level variables: system-justifying beliefs, RWA, SDO, national identification, conservatism, sex, age, and subjective social status). These variables were nested within the higher order countries’ characteristics (societal variables: HDI and Gini). This approach examines the pattern of individual variables inside each country, and the way in which country-level features moderate this pattern (Tabachnick & Fidell, 2007). First, we assess an intercept-only model, in which we did not introduce any independent variables, but we included a random intercept. Second, we included the individual-level predictors as fixed effects. Third, we included the national-level variables. Finally, we added the individual- and cross-level interactions, such as the SDO × status interaction that concerns the ideological asymmetry hypothesis. It is important to remark that we analyzed the objective and subjective social status interactions in different models, to avoid biases due to the high number of predictors.
All regressions were estimated using the restricted maximum likelihood estimator, because of the few level 2 units in our sample (Hox, 2010). In addition, we group-mean centered the individual predictors, and grand-mean centered the country-level variables (Gelman & Hill, 2007).
Results
Preliminary Analyses
The descriptive statistics by country are presented in Table 1. The results show significant differences in system justification by country, F(18, 14,917) = 141.18, p < .001,
Descriptive Statistics.
Note. SJS = System Justification Scale; SDO = social dominance orientation; RWA = right-wing authoritarianism; ID = identification.
On SDO, there were also significant differences by country, F(18, 14,917) = 84.89, p <.001,
On RWA, there were also significant differences by country, F(18, 14,917) = 146.24, p < .001,
For the National Identification Scale, there were significant differences by country, F(18, 14,917) = 78.13, p < .001,
Finally, in the Conservatism Scale, which ranged from 0 (strong liberal) to 10 (strong conservative), there were also differences by country, F(18, 14,917) = 31.94, p < .001,
The individual-level correlation matrix is presented in Table 2 and collective level in Table 3. The results showed that SJS was positively and significantly related to SDO (r = .199), RWA (r = .169), national identification (r = .290), and conservatism (r = .197), as expected. Being male and being older were also associated with SJS. The four system predictors of SJS were significantly and positively related to each other, except for the association between national identification and SDO, which was close to zero (r = −.028). In this way, bivariate individual results were globally congruent with Hypothesis 1. SJS was positively associated to subjective social status (r = .279) and income (r = .059). This result disconfirmed the strong SJT hypothesis and supported the SDO/RCT Hypothesis 2a. Collective-level correlations showed that SJS was marginally related to national identity (r = .448). Finally, national identity and RWA characterized less developed nations (r = −.576 and –.729, respectively).
Correlations at the Individual Level (N = 14,936).
Note. SJS = System Justification Scale; SDO = social dominance orientation; RWA = right-wing authoritarianism; ID = identification.
p < .05. **p < .01. ***p < .001.
Correlations at the Collective Level of Nation/Country (N = 19).
Note. SJS = System Justification Scale; SDO = social dominance orientation; RWA = right-wing authoritarianism; ID = identification; HDI = human development index.
p < .05. **p < .01. ***p < .001.
Main Analyses
Multilevel linear regressions are shown in Tables 4 and 5. Model 1 presents the intercept-only model, in which the constant is modeled as a random effect. This model allows us to estimate the intraclass correlation (ICC), which is a measure of variance due to the hierarchical structure of data. According to this statistic, 14.29% of the variance in system justification was due to cross-national differences, ICC = .1429, SE = 0.040, 95% confidence interval (CI) = [0.081, 0.240]. This means that if we perform a linear regression, the results will be biased, because the residuals are not independent of country, justifying the use of multilevel models (Hox, 2010).
Multilevel Linear Regressions: System Justification.
Note. SDO = social dominance orientation; RWA = right-wing authoritarianism; HDI = human development index; AIC = Akaike information criterion; BIC = Bayesian information criterion.
Multilevel Linear Regressions: System Justification.
Note. SDO = social dominance orientation; RWA = right-wing authoritarianism; HDI = human development index; AIC = Akaike information criterion; BIC = Bayesian information criterion.
Individual-level variables were included in Model 2, explaining 18.4% of the individual-level variance and the 0.7% of intergroup variance, according to the intragroup and intergroup R2, respectively. This model led to a decrease in the –2 log likelihood (Δ = 1,518.56), Akaike information criterion (AIC; Δ = 3,021.11), and Bayesian information criterion (BIC; Δ = 2,960.23), with a significant likelihood ratio test,
In Model 3, we included Gini (income inequality) as a national-level predictor. This model did not improve the individual variance explained, but did obtain a better cross-national R2. This change was reflected in a change in –2 log likelihood (Δ = 0.14), although according to AIC (Δ = −1.70) and BIC (Δ = −9.32), the fit was worse than in Model 2, which is also supported by a nonsignificant likelihood ratio test,
In Model 4, we included the individual- and cross-level interactions. We obtained a better fit than in Model 3, according to both intragroup and intergroup R2, and the decreasing of –2 log likelihood, Δ = 26.43,
The social status by RWA interaction term was significant. The simple slope analysis showed that the effect of RWA was significant among high social status people (+1 SD, b = 0.08, SE = 0.011, z = 7.03, p < .001), but not among low social status people (–1 SD, b = 0.02, SE = 0.011, z = 1.68, p = .092). The social status by conservatism interaction was also significant, so that the effect of conservatism was stronger among high-status people (+1 SD, b = 0.06, SE = 0.006, z = 9.43, p < .001) than among low-status people (–1 SD, b = 0.03, SE = 0.006, z = 5.17, p < .001). The last significant interaction term was national identification by SDO, so that the effect of SDO was stronger among less nationally identified people (–1 SD, b = 0.23, SE = 0.014, z = 16.46, p < .001) than among highly nationally identified people (+1 SD, b = 0.18, SE = 0.014, z = 12.39, p < .001).
The results observed in Model 4 supported SDO/RCT Hypothesis 4a, because legitimizing ideologies showed a stronger effect in high-status persons. In addition, Hypothesis 5 was supported, because the effect of SDO was stronger among less nationally identified people. Finally, Hypothesis 6 (i.e., effect of social status will be stronger among unequal countries) was not supported, given that the interaction term was nonsignificant.
In Model 5, we replaced the subjective social status measure with the self-reported income in the interaction terms. We obtained a better fit than in Model 3, according to the decreasing of –2 log likelihood, Δ = 11.63,
The income by Gini interaction was significant, so that the effect of income on system justification was significant and negative among unequal countries (+1 SD, b = −0.05, SE = 0.014, z = −3.60, p < .001), and nonsignificant among egalitarian countries (–1 SD, b = −0.00, SE = 0.013, z = −0.38, p = .707), providing support for Hypothesis 6b, which is at odds with the strong form of SJT (i.e., the effect of social status is negative and stronger among unequal countries). Also, the national identification by SDO was significant, so that the effect of SDO on system justification was stronger among less identified individuals (–1 SD, b = 0.23, SE = 0.014, z = 16.45, p < .001), than among high identified individuals (+1 SD, b = 0.18, SE = 0.014, z = 12.55, p < .001).
In Model 6, we included the individual-level predictors and the HDI as a national-level predictor. Respecting Model 2, the intergroup and intragroup R2s were similar, and we did not obtained a better fit according to AIC (Δ = −1.93), BIC (Δ = −9.55), which is also supported by a nonsignificant likelihood ratio test,
In Model 7, we included individual- and cross-level interactions. We obtained a better fit than in Model 6 according to intragroup R2, and the decreasing of –2 log likelihood, Δ = 44.89,
The main effect of subjective social status was qualified by interactions with RWA, conservatism, and HDI. In the first case, the simple slope analysis revealed that the effect of RWA was significant among high social status people (+1 SD, b = 0.08, SE = 0.011, z = 7.02, p < .001), but not among low-status people (–1 SD, b = 0.02, SE = 0.011, z = 1.78, p = .074). The social status by conservatism interaction was significant, so that the effect of conservatism was stronger among high-status people (+1 SD, b = 0.06, SE = 0.006, z = 9.65, p < .001) than among low-status people (–1 SD, b = 0.03, SE = 0.006, z = 5.13, p < .001). The social status by HDI interaction was significant, so that the effect of social status was stronger among highly developed countries (+1 SD, b = 0.21, SE = 0.008, z = 25.78, p < .001) than among less developed countries (–1 SD, b = 0.16, SE = 0.009, z = 18.15, p < .001). This last finding provides support for the SDO/RCT conflict Hypothesis 7a (i.e., the effect of social status will be stronger in more developed nations).
The main effect of national identification was qualified by interaction terms with SDO and HDI. In the first case, the effect of SDO was higher among less identified people (–1 SD, b = 0.23, SE = 0.014, z = 16.82, p < .001) than among those highly identified (+1 SD, b = 0.18, SE = 0.014, z = 12.51, p < .001). For the national identification by HDI interaction, the effect of national identification was stronger in less developed countries (–1 SD, b = 0.23, SE = 0.011, z = 22.53, p < .001) than among more developed countries (+1 SD, b = 0.16, SE = 0.010, z = 16.18, p < .001). These results suggest that national identification is more effective as a legitimizing ideology in less developed societies.
Finally, in Model 8, we substituted a more “objective” measure of status, the percentile income variable for the subjective social status measure used previously in the interaction terms. We obtained a better fit than in Model 6, according to –2 log likelihood, Δ = 30.77,
The main effect of national identification was qualified by interaction terms with SDO and HDI. In the first case, the effect of SDO was higher among less identified people (–1 SD, b = 0.23, SE = 0.014, z = 16.79, p < .001) than among those highly identified (+1 SD, b = 0.18, SE = 0.014, z = 12.69, p < .001). For the national identification by HDI interaction, the effect of national identification was stronger among less developed countries (–1 SD, b = 0.23, SE = 0.010, z = 22.38, p < .001) than more developed countries (+1 SD, b = 0.17, SE = 0.010, z = 16.31, p < .001).
Supplementary Analyses
Although our hypotheses were tested using fixed effects in the multilevel models, an anonymous reviewer suggested including random effects. We found significant standard deviations for all the individual-level variables: SDO (SD = 0.104, SE = 0.202, 95% CI = [0.071, 0.152]), RWA (SD = 0.115, SE = 0.020, 95% CI = [0.081, 0.163]), conservatism (SD = 0.054, SE = 0.010, 95% CI = [0.038, 0.077]), national identification (SD = 0.110, SE = 0.019, 95% CI = [0.078, 0.155]), social status (SD = 0.048, SE = 0.010, 95% CI = [0.032, 0.072]), and income (SD = 0.057, SE = 0.013, 95% CI = [0.036, 0.091]), when considering the random effects in separate models. In other words, the effects of these individual-level variables on the dependent measure varied from country to country.
In addition, we observed significant associations between the intercept and the standard deviation of the slopes only for national identification (r = .608, SE = 0.154, 95% CI = [0.224, 0.828]), which suggests a stronger effect of this variable among those countries with higher average system justification.
Discussion
The aim of this study was to analyze the impact of social status on system-justifying beliefs. In addition, we compared predictions based on social dominance theory with those proposed by system justification theory. Descriptive results showed that overall means were below the score of agreement with a positive view of the social system, suggesting that most of the participants did not agree with beliefs justifying the social system. False consciousness seems to be a minority phenomenon, as evaluated by explicit items—also belief in the legitimacy of group domination was weak. Survey and qualitative studies have found that false consciousness is neither absolute nor dominant and that people share a mixed set of critical and acceptance beliefs (van Dijk, 1998). At the opposite end, national identification and authoritarianism were more endorsed. As was expected by the statements proposed by both system justification theory (Jost & Banaji, 1994) and social dominance theory (Sidanius & Pratto, 1999), SDO, RWA, conservatism, and national identification were positively related to system justification.
The strong SJT hypothesis (Jost et al., 2003), which is a specific theoretical claim within this approach, posits that system justification belief and effects are stronger in low-status people, and in less socially developed and unequal societal contexts. It was argued that low-status group members suffer strong cognitive dissonance and conflict between ego, group, and system justification motives and these conflicts lead to more intense justifications to rationalize and uphold the prevailing systems and status quo (Jost et al., 2003). However, the social dominance/realistic conflict hypothesis (Sherif et al., 1988; Sidanius & Pratto, 1999) suggests that high status or possessing an advantaged social condition makes it more likely for people to support the existing social system. Results largely supported the social dominance/realistic conflict position on Hypotheses 2a (i.e., higher social status will positively predict system-justifying beliefs), 3a (i.e., system justification will be associated with equality and social development at the country level), and 4a (i.e., the effect of SDO, RWA, and conservatism will be stronger among high-status people). These results are compatible with previous studies that have tested the strong form of SJT, albeit using socioeconomic independent variables rather than psychological measures (e.g., Brandt, 2013; Caricati, 2017; Caricati & Lorenzi-Cioldi, 2012; Trump & White, 2014).
Nevertheless, it is important to remark that we obtained two results congruent with the strong SJT hypothesis: a negative association between income and system-justifying beliefs and a significant interaction term between income and Gini, so that the effect of income was negative among unequal countries and nonsignificant among egalitarian countries. It is important to recall that status correlated positively with system justification, so that this negative coefficient could be due to a suppressor effect (Cheung & Lau, 2007). Thus, after accounting for demographic variables (i.e., gender and age), SDO, RWA, national identification, conservativism, subjective social status, and objective social status all negatively predicted system-justifying beliefs. One possible explanation could be related with the pervasive tendency to identify with the middle class among people from the whole social continuum (Castillo, Miranda, & Madero Cabib, 2013; Evans & Kelley, 2004). If that phenomenon is occurring, then any measure of perceived social status would imperfectly reflect actual social statuses in our samples. Nevertheless, according to the strong form of SJT, subjective social status is more important than objective status, because if people do not perceive their disadvantaged position, cognitive dissonance cannot occur (for a review of cognitive dissonance theory, see Aronson, 1992; Cooper & Fazio, 1984; Festinger, 1962; Harmon-Jones & Judson, 1999; Joule & Beauvois, 1997). These complex results could be of great interest for future research, given that previous studies testing the strong form of SJT in cross-cultural samples, and including contextual factors, used either objective or subjective social status, but not both variables (e.g., Brandt, 2013; Caricati, 2017; Caricati & Lorenzi-Cioldi, 2012).
According to social dominance theory (Sidanius & Pratto, 1999, 2004), there are differences among social status groups in the relationship between legitimizing myths and SDO, which has been called ideological asymmetry (Sidanius et al., 2001; Sidanius & Pratto, 1999). Our results extended this hypothesis, because we observed the relationship between legitimizing myths (or ideologies) and psychosocial dispositions (i.e., SDO), and system justification or the perceived fairness of the social system. The pattern of results showed that the effect of the former variables was congruent with ideological asymmetry, as it was stronger among high-status groups in the case of RWA and conservatism. However, the interaction of status by SDO was not significant as expected. It is important to remark that the measure of social status was a subjective report of individual-level status. This limited operationalization might explain the lack of an SDO × status interaction, which has been found in studies using a more “objective” measure of status, such as ethnicity (Rabinowitz, 1999). Another explanation is that the relatively strong association between conservatism and SDO could have suppressed the interactive effect of the last variable.
Finally, with respect to the interaction of national identification with HDI, our results were congruent with other studies that found that national identification is higher in less developed nations and that in these nations, beliefs legitimizing fighting to defend the nation in wars are strongly supported and influenced by beliefs in history as recording the progress of the nation (Paez, Bobowik, Liu, Basabe, & Hanke, 2016). A progressive view of history is endorsed in less developed, collectivistic, and hierarchical contexts (Hofstede, 2001). This could be perceived as at odds with previous studies suggesting that individualistic cultures, as contrasted with collectivistic and traditional ones, tend to have a more positive view of history, and to be more future and goal focused, in the service of economic capitalism development (Boniwell & Zimbardo, 2004). Industrialization and modernization enhance a positive view of societal evolution in developing nations such as China and India. The positive effect of national identification in system justification is, thus, probably more strongly emphasized as a civic duty for citizens in developing nations; whereas in more developed societies, a more disenchanted view of nation is permitted. This appears to be a dynamic link, however, as conservative nationalism is growing in many Western societies.
Limitations
There are several limitations that should be considered as future research directions. First, although we used a cross-national sample, we obtained data from just 19 nations. This is a low number of level 2 units in multilevel linear regression, and it could have biased our hypothesis testing, given a low level of statistical power for testing the effects of level 2 variables (Tabachnick & Fidell, 2007). In addition, we did not have samples from Africa, which is a region that represents an interesting field test for the strong form of SJT.
A second limitation is using brief versions of the constructs measured in our survey. If we could have used the complete versions of each instrument, we might have obtained variables with less measurement error. In addition, we were unable to achieve optimal levels of measurement invariance in our data, given that our scales (SJS, SDO, RWA, conservatism, and national identification) were short, with only three or four items each. However, given that we worked with an international online survey with many other research questions, including full versions of the scales was not possible.
A third limitation is the effect sizes found in the significant interaction terms. In general, in all the models in which we found significant county-level moderators, the differences between the coefficients at high versus low levels of the moderators were small. Indeed, when the simple slope analyses revealed that both coefficients were significant, the difference ranged from .03 to .07, which are modest values, given that the dependent variable was measured on a scale ranging from 1 to 7. Although it does not invalidate our findings, it is important to explicit this fact, given that more research will be necessary to confirm these results using a larger number of cross-cultural samples.
Finally, we use correlational data, which does not allow us to establish causal relationships. However, our results are akin or congruent with experimental studies that reject the strong form of SJT (Rubin, Badea, & Jetten, 2014; Trump & White, 2014).
Future Research
Our results provide contrary evidence to the strong form of SJT. However, a question remains unanswered in the field of system justification theory. Why are social arrangements so stable? According to system justification theory (Jost & Banaji, 1994), people are actively motivated to perceive the system as fair and legitimate, because of the hedonic benefit and the existential, epistemic, and relational needs that are fulfilled. This argument has received empirical support (Hennes et al., 2012; Jost & Hunyady, 2003; Thorisdottir, Jost, & Kay, 2009; van der Toorn & Jost, 2014). However, this stability needs the active or passive support of low-status people. Our results, and those from other studies (Brandt, 2013; Caricati, 2017; Caricati & Lorenzi-Cioldi, 2012), suggest that low-status people do not endorse system-justifying beliefs. For this reason, we suggest that the social systems’ stability does not come from an active motivation related to the endorsement of ideologies, but from the possession of cognitive biases such as hope for a better future (see Owuamalam, Rubin, & Issmer, 2016). According to the argument proposed by Owuamalam et al. (2017; Owuamalm, Rubin, & Issmer, 2016, and consistent with social identity theory, see Tajfel, 1979, 1982; Turner, Brown, & Tajfel, 1979), highly identified low-status individuals endorse system-justifying beliefs when they perceive societal arrangements as stable: Because if a social system is both stable and fair, then there is hope for improvements to a low-status position in the future.
If the system is perceived as unstable, then low-status individuals would be more likely to engage in social change activities rather than in system justification (Tajfel & Turner, 1986). An “augmenter” proposition was developed by Kunst et al. (2017), who argued that under social stability, the costs involved in challenging the status quo are too high, so low-status individuals would rather not try to fight against it. In both approaches, the cornerstone is system stability, which should be addressed in future studies on system justification. Nevertheless, we suspect that this variable should not alter the pattern of results regarding both SDO and RWA, because these concepts are thought as dispositional variables, so they should be less dependent on contextual factors than attitudes (Altemeyer, 1998; Pratto et al., 1994).
Finally, testing the strong form of SJT has been focused on the influence of contextual factors, such as inequality and collective (i.e., national) social status (e.g., Brandt, 2013; Caricati, 2017; Caricati & Lorenzi-Cioldi, 2012; Sengupta et al., 2015). Nevertheless, previous studies have shown that the general statements proposed by this theoretical approach are moderated by other contextual factors (for a detailed review, see Kay & Friesen, 2011), such as system dependence (e.g., Zhu, Kay, & Eibach, 2013), system threat (e.g., Jost, Blount, Pfeffer, & Hunyady, 2003), and system stability (e.g., Chernyak-Hai, Halabi, & Nadler, 2014). Given these antecedents, we anticipate that additional societal-level moderators may be required to see support emerge for the strong form of SJT in future research.
Supplemental Material
vargas_online_appendix – Supplemental material for A Comparison of Social Dominance Theory and System Justification: The Role of Social Status in 19 Nations
Supplemental material, vargas_online_appendix for A Comparison of Social Dominance Theory and System Justification: The Role of Social Status in 19 Nations by Salvador Vargas-Salfate, Dario Paez, James H. Liu, Felicia Pratto, and Homero Gil de Zúñiga in Personality and Social Psychology Bulletin
Footnotes
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: This research was supported by Grant FA2386-15-1-0003 from the Asian Office of Aerospace Research and Development. This article was partially supported by the grant PPSI2014-51923P to Dario Paez.
Supplemental Material
Supplementary material is available online with this article.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
