Equality Hypocrisy, Inconsistency, and Prejudice: The Unequal Application of the Universal Human Right to Equality

In Western culture, there appears to be widespread endorsement of Article 1 of the Universal Declaration of Human Rights (which stresses equality and freedom). But do people really apply their equality values equally, or are their principles and application systematically discrepant, resulting in equality hypocrisy? The present study, conducted with a representative national sample of adults in the United Kingdom (N = 2,895), provides the first societal test of whether people apply their value of “equality for all” similarly across multiple types of status minority (women, disabled people, people aged over 70, Blacks, Muslims, and gay people). Drawing on theories of intergroup relations and stereotyping we examined, relation to each of these groups, respondents’ judgments of how important it is to satisfy their particular wishes, whether there should be greater or reduced equality of employment opportunities, and feelings of social distance. The data revealed a clear gap between general equality values and responses to these specific measures. Respondents prioritized equality more for “paternalized” groups (targets of benevolent prejudice: women, disabled, over 70) than others (Black people, Muslims, and homosexual people), demonstrating significant inconsistency. Respondents who valued equality more, or who expressed higher internal or external motivation to control prejudice, showed greater consistency in applying equality. However, even respondents who valued equality highly showed significant divergence in their responses to paternalized versus nonpaternalized groups, revealing a degree of hypocrisy. Implications for strategies to promote equality and challenge prejudice are discussed.

I have a dream that one day this nation will rise up, live out the true meaning of its creed: "We hold these truths to be self-evident, that all men are created equal." -Martin Luther King, Jr. (1963) The very first article in the Universal Declaration of Human Rights (UDHR) adopted by the General Assembly of the United Nations Organization in 1948 states that "All human beings are born free and equal in dignity and rights. They are endowed with reason and conscience and should act toward one another in a spirit of brotherhood." Yet, Martin Luther King Jr.'s dream of equality is far from fulfilled. In this article, we consider how and why people may espouse equality as an important principle, yet not apply that principle equally to all groups. We report an empirical test of this phenomenon using a nationally representative survey of nearly 3,000 adults in the United Kingdom.

Universal Conceptualization of Human Rights
Fifty years after the adoption of the UDHR, Willem Doise formally theorized and tested the existence of social representations of human rights. For example, across four cultures (Costa Rica, France, Italy, and Switzerland), Clémence, Doise, de Rosa, and Gonzalez (1995) demonstrated remarkable convergence in the understanding of what constitutes violations of human rights (see also Doise, 2002). In a more comprehensive test, Doise and colleagues asked participants from 35 countries to evaluate the 30 articles of the UDHR (Doise, Spini, & Clémence, 1999). The structure of the different rights was remarkably convergent across cultures, and closely resembled the original structure presented by the French jurist, Renée Cassin at the United Nations General Assembly in 1948.
In several studies, Doise and colleagues showed that people's commitment to these principles is related to their wider values. For example, people who value universalism and selftranscendence are more likely to be involved with human rights issues and to believe that governments are not doing enough to protect human rights. In contrast, those who value conservatism and endorse right-wing political ideologies favor restricting individual rights to guarantee the functioning of society (Doise et al., 1999;Spini & Doise, 1998). Moreover, they tend to endorse the power of governments and other institutions to decide upon the distribution of human rights (Moghaddam & Vuksanovic, 1990).

Human Rights as a Function of Intergroup Relations
In addition to these individual differences in conceptualizations of human rights, intergroup relations research suggests that support for human rights may depend on power and status relations between groups. For example, research has shown that intergroup ideologies such as social dominance orientation (SDO) and right-wing authoritarianism (RWA) negatively affect human rights support (e.g., Cohrs, Maes, Moschner, & Kielmann, 2007;McFarland & Mathews, 2005;Stellmacher, Sommer, & Brähler, 2005). People high in SDO prefer hierarchical (rather than egalitarian) relations between social groups, while the opposite is true for people low in SDO (Pratto, Sidanius, Stallworth, & Malle, 1994). Similarly, people high in RWA tend to be less favorable toward according the same rights to all groups. This is because people high in RWA believe this would allow unwarranted means of social control to socially subordinate groups (e.g., religious minorities).
There are also differences between minority and majority groups' emphasis on people's rights versus people's duties. Specifically, members of minority or low power groups give higher priority to their personal rights, whereas members of majority or high power groups give higher priority to the duties that low power groups need to enact (Moghaddam & Riley, 2005). Moghaddam and Riley argue that such divergence was evident during the U.S. civil rights and women's rights movements, whereby these minority groups highlighted their human rights, whereas majority groups focused on the duties of these minorities (e.g., to obey the law, at that time restricting the minorities' rights).
Similarly, Azzi (1992) demonstrated that participants who belonged to, or were primed to identify with, a minority ethnic group were more likely to advocate equal distribution of procedural resources (i.e., political power) between a simulated ethnic minority and majority group. Conversely, participants who belonged to, or were primed to identify with, a majority ethnic group were more likely to advocate a proportional distribution of procedural resources.
In line with these findings, Louis and Taylor (2005) advocated a relativist advocated of human rights, highlighting that affordance of rights varies across contexts, time, the social groups people belong to, and the social identities they espouse. People interpret human rights relative to their ingroup, and so the interpretation is affected by the group's status position within the societal hierarchy (see also Worchel, 2005).
The picture is rendered more complex when we consider that people typically have multiple group-based identities, hence more than one ingroup (Crisp & Hewstone, 2007). By implication, people also have multiple outgroups toward whom their endorsement of human rights may also vary. This suggests that inconsistency in rights endorsements could arise because different ingroup-outgroup relationships involve different frames of comparison.

Compatibility of Universalist and Relativist Approaches
It seems to us that the universalist (Doise et al., 1999) and relativist (Louis & Taylor, 2005;Worchel, 2005;cf. Kymlicka, 2001) positions can be reconciled. There could be a universal conceptualization of human rights, but these principles can be applied differently due to the hierarchical nature of human societies, and the intergroup relations they embody. Therefore, we consider that people's endorsement of the value of equality may not translate into application to specific groups, because social identities, power hierarchies, and ingroup norms come into play, all of which might place greater value on some groups than others.

Defining Equality Hypocrisy
Empirically, individuals in Western societies generally support the abstract goal of human rights. For example, in 2002, poll results showed that 90% of Americans rated human rights as a goal that is very important or somewhat important (Chicago Council on Foreign Relations, 2002, cited in McFarland & Mathews, 2005. McFarland and Mathews argue that this may reflect social desirability concerns because endorsement of rights is an essential part of North American, and more generally Western, ideology. The researchers found that when comparing people's preference for human rights versus national selfinterest goals, "promoting and defending human rights in other countries" was ranked only as 12th out of 15 goals. This reveals that individuals may preach human rights more than they are prepared to practice them, at least when choosing between the importance of global rights versus national priorities. Staerklé and Clémence (2004) explored inconsistency between values and application in two schools in Switzerland. Adolescents who valued human rights highly judged sanctions that violated human rights to be less acceptable when applied to a murderer than to a pedophile rapist, when applied to a thief than to a drug dealer, and when applied to "handicapped" children rather than to immigrant children. In studies using minimal groups, Maio, Hahn, Frost, and Cheung (2009) showed that varying the situational salience of equality values could also affect whether they were applied to resource distribution between groups. Support for the human right to equality logically implies support for equality for everyone regardless of their race, gender, religious affiliation, sexual orientation, age, and physical ability. Despite evidence that many people agree with the notion that all human beings should be treated equally, research on intergroup prejudice leads us to expect that, when asked more concretely, people will differentiate which groups most "deserve" these rights, thereby revealing equality hypocrisy. Specifically, equality hypocrisy occurs when people express strong support for equal rights for all, but then differentially favor equal rights for some groups above those of others. We believe that equality hypocrisy is inherent in many, possibly all societies. The present study explores its forms and possible influences in the United Kingdom-a country that is usually regarded as relatively modern, progressive and liberal.

Intergroup Prejudice
Intergroup relations research has long established that people are prone to express ingroup bias, and that this might result from ingroup commitment (Brewer, 1999), intergroup competition (Sherif, 1966) or the motivation to selfenhance and establish positive ingroup distinctiveness by evaluating ingroups more favorably than outgroups (Tajfel & Turner, 1979). People's ingroup commitment might simply mean that they view all outgroups as less deserving than the ingroup. Potential intergroup competition may motivate people to deny equality to groups that are viewed as competing with the ingroup (either ideologically or materially). Moreover, people may garner positive ingroup distinctiveness, self-esteem and competitive superiority by ensuring that lower status groups are not afforded the same "rights" as a majority ingroup. Although these ideas have been tested with regard to single specific outgroups (see Abrams, 2015;Dovidio & Gaertner, 2010;Hewstone, Rubin, & Willis, 2002), there does not appear to be any existing research that shows whether people apply ingroup preference when they apply their values in the context of multiple outgroups, or whether the type of outgroup would necessarily affect how they apply the value of equality. This is surprising given that most people live in societies that do present multiple outgroup categories.

Motivations to Control Prejudice
Research has shown that the personal and social motivations to control prejudice strongly predict its expression toward specific outgroups (e.g., Butz & Plant, 2009;Crandall & Eshleman, 2003;Devine & Monteith, 1993;Gonsalkorale, Sherman, Allen, Klauer, & Amodio, 2011;Plant & Devine, 2009). People who are high in internal motivation to control prejudice show lower prejudice in public as well as private contexts. This is because they want to be free of prejudice (Plant & Devine, 2009). People low in internal motivation but high in external motivation to control prejudice only show lower prejudice in public, but not in private, contexts. This is because they want to be seen as unprejudiced, but not necessarily to be free of it (Plant & Devine, 2009). For example, Legault, Gutsell, and Inzlicht (2011) showed that, compared to a control condition, when people were primed with autonomous motivation to regulate prejudice (i.e., internal motivation) they showed less explicit and implicit prejudice whereas when primed with the societal requirement to control prejudice (i.e., external motivation) they expressed more explicit and implicit prejudice.
Although motivation to control prejudice is compatible with advocacy of equality, and although a liberal interpretation of such motivation is that it is consistent with a free and fair society, these concepts are not necessarily synonymous. For example, it is possible to envisage that someone could be unconcerned about their own prejudice but still advocate the principle of equality for all, perhaps for religious, moral, or material reasons. Moreover, it is plausible that someone who is highly motivated not to be prejudiced could still be perfectly willing to accept that society should tolerate inequality. Finally, someone whose primary concern is not to appear prejudiced may be motivated either because they value equality or because they prefer inequality but do not wish to be seen to do so. Whether these motivations to control prejudice similarly affect the ascription of rights to different types of group, and whether they do so independently of equality values, are interesting and unexplored questions in both intergroup relations and human rights research.

Minority Groups
Why might we expect uneven affordance of equality to different minority groups? Different societal groups are perceived and stereotyped differently. Seminal work by Fiske and colleagues (Fiske, Cuddy, Glick, & Xu, 2002) in the United States demonstrated that groups are evaluated along two primary characteristics: warmth and competence. The combination of these two primary characteristics gives rise to the perceived stereotypicality of groups and to differential qualities of prejudice. Thus, groups that are seen as high in warmth and low in competence are more likely viewed with pity and to suffer so called benevolent, or paternalistic prejudice (e.g., the "elderly," "housewives," "disabled," and "blind" people). Groups that are considered low in warmth but high in competence are on the receiving end of envy and envious prejudice (e.g., the "rich," "Asian," "Americans," "Jews," men). Groups that are considered low in both competence and in warmth elicit contempt (e.g., the "poor," "welfare recipients," "Hispanics"), and groups that are considered high in competence and in warmth elicit admiration; these are often majority status ingroups (e.g., "Christians," "middle class," "White people," "students"). Of course, there are various blends of moderate levels of these qualities, forming a middle cluster. In the United States these include "migrant workers," "gay men," "Arabs," "Muslims," "blue-collar workers," and "southerners" (Cuddy, Fiske, & Glick, 2007;Fiske et al., 2002). Numerous studies have demonstrated that different groups are evaluated in terms of the warmth-competence stereotype dimensions, which in turn informs people's emotional and behavioral reactions toward these groups (e.g., Cuddy et al., 2007;Glick & Fiske, 2001). Cross-cultural examinations have also supported the four clusters of stereotype content related to specific groups (see Fiske & Cuddy, 2006;Cuddy et al., 2009).

The Present Research
For the purposes of this research we examine whether participants assign the human right of equality differentially to different status minorities as a function of the evaluative implications of stereotypes associated with these groups. We expect status minority groups that are known to be stereotyped as warmer but less competent (hence paternalized) will be judged differently from status minorities that are stereotyped as colder, or potentially more threatening (hence not paternalized).

Policy Context
In 2005, the U.K. Labour government prepared to merge the roles of distinctive commissions (e.g., Commission for Racial Equality, Disability Rights Commission, Equal Opportunities Commission) with those of other NGOs representing age, sexuality and religion under the umbrella of a new Equality and Human Rights Commission. To better understand the implications of this merger, the Women and Equality Unit within the Department for Trade and Industry for the "Equalities Review" com-missioned and conducted this research, which provided part of the foundation for establishing the Equality and Human Rights Commission (The Equalities Review, 2007). It was the first single piece of integrated U.K. research to attempt to understand prejudice and values about human rights in relation to all six "equality strands," corresponding to gender, age, disability, ethnicity, religion, and sexuality. This provided a unique opportunity to discover how, across a whole population, views about the rights of these distinct groups would relate to overall values about key human rights.
Paternalistic stereotypes depict social groups as pitied and instigate feelings of compassion and sympathy and a desire to help these needy groups. Paternalized groups are those that are targets of "benevolent" prejudice, which accords those groups low status and competence but relatively high levels of warmth. As a result they are treated as dependent and needy, deserving of sympathy, but are effectively pinned to low status and power positions. The dilemma for these groups is that they lose the "benefits" of patronage and charity if they challenge for higher status positions. Such prejudice is by no means benign. For example, female victims of acquaintance rape are more likely to be blamed by perceivers who are higher in benevolent sexism (Abrams, Viki, Masser, & Bohner, 2003). Based on the stereotype content model Cuddy, 2004, personal communication), among the six equality strands in the Equalities Review, we expected people to apply these stereotypes to women, older people, and disabled people. In contrast, Black, Muslim, and gay people were expected to pose various types of threat (culturally or materially) and as liable to be viewed as competitors vis-à-vis majority White British society. Thus, we classified these as nonpaternalized groups. We hypothesized that the representative sample would assign equal rights more readily to paternalized than to nonpaternalized groups. The present research examines how equality values and motivation to control prejudice relate to equality hypocrisy, equality inconsistency and prejudice. We examine the following issues in relation to judgments involving women, people over 70, disabled people, gay and lesbian people, Muslims, and Black people.

Societal Equality Hypocrisy
If, on average, people in society claim to value equality as a universal right more than they are willing to attach importance to the wishes and equality of opportunity for specific social groups this suggests that the society manifests what we term equality hypocrisy. The hypocrisy arises because valuing equality more highly for some groups than others is logically incompatible with valuing universal equality. Our first question is whether there is societal evidence that the level of endorsement of equality values is not matched by support for equality for specific groups in society (equality hypocrisy).

Individuals' Equality Inconsistency
Societal hypocrisy could exist because all individuals favor certain groups more than others. However, these average societal differences do not reveal a further aspect of equality hypocrisy-some individuals may differentiate levels of importance they attach to the equality rights of different groups more than other individuals do. That is, individuals may differ in the extent to which they show equality inconsistency. Such inconsistency is potentially hypocritical because it seems perverse to advocate greater equality for some groups at the expense of others. Therefore we consider the extent to which individuals attach different importance to satisfying the wishes, and ensuring equal employment opportunities for each group (equality inconsistency). We propose that, matching the societal level differences, individuals' equality inconsistency will expose a contrast between paternalized and nonpaternalized groups, whereby the latter are liable to be regarded as less deserving of equality.

Prejudice
We examine a measure of prejudice in the context of employment: expressions of comfort in having a boss who is from each minority group (a specific form of social distance; Bogardus, 1933). Because of their common link in terms of intergroup relations, we expect equality inconsistency to be mirrored by a similar pattern of preferences in social distance. We also investigate the extent to which equality inconsistency and prejudice are predictable from an individual's support for the value of equality and their internal and external motivation to control prejudice.
In summary, we expect that while people may agree with the general value of equality they may not support equality equally for all minority groups (equality hypocrisy). Furthermore, on the basis of intergroup relations theory we expect that people may place higher value on equality for paternalized than nonpaternalized groups (equality inconsistency). We expect that the gap in importance attached to equality for paternalized versus nonpaternalized groups should be lower among individuals who value equality for all, and who are internally or externally motivated to control prejudice.

Participants and Design
Data were collected as part of a specially commissioned representative national survey in Britain in 2005 (Abrams & Houston, 2006), a time when Britain had a Labour (left-wing) government led by Tony Blair that had widespread popular support and was strongly promoting universal human rights. The sample comprised 1,289 men (44.5%) and 1,606 women (55.5%); total N of 2,895. Age ranged from 16 to 93 years (M ϭ 46.07, SD ϭ 19.14). The majority of participants (87.5%) were White British, 4.8% were Black, 6.4% were Asian, and 1.3% was coded as missing. Furthermore, the majority of participants (92.5%) were non-Muslim, nondisabled (78.3%), and heterosexual (88.7%). Of the participants, 35.2% were in full-time employment, 11.3% were in parttime employment, 21.9% were unemployed, 25% were retired, and 6.7% were in full-time education. Of the participants, 60.3% had left full-time education before 18 years of age, 13.2% held qualifications up to 18 years ("Alevel"), 13.5% had completed a university degree, and 13% had completed another type of college qualification (e.g., Business and Technology Innovation Council, BTEC). Politically, the sample was slightly left of center (on a 6-point scale that ranged from 1 ϭ definitely left to 6 ϭ definitely right, the mean was 3.35, SD ϭ 1.30). Data reported in this article were from a larger survey that assessed a range of societal perceptions and attitudes (see Abrams & Houston, 2006, for details and descriptive statistics).
To keep the survey to a manageable length, and because we had high statistical power due to sample size, three versions of the survey were administered to separate nationally representative samples each comprising approximately 1,000 respondents. All versions included measures of equality values, measures of the importance of equality, and motivation to control prejudice, but the different versions included specific questions about opportunities and social distance for two target groups only. Specifically, each version asked about one group that we considered subject to paternalistic prejudice and one that was more likely to be subject to traditional nonpaternalistic prejudice. Version A asked these questions in relation to women and homosexuals, Version B asked these questions in relation to people over 70 and Muslims, and Version C asked these questions in relation to disabled people and Black people.

Procedure
Data were collected between May 20 and June 1, 2005, among Ն16-year-olds from England, Scotland, and Wales. The survey was administered by a polling company, TNS/ Omnimas, in their omnibus face-to-face computer-assisted personal interviews surveys to a nationally representative sample. Left-and right-scale anchor points were counterbalanced between participants (e.g., agree and disagree were anchored alternately on the left or right end of a scale for Likert scale items), and item orders were rotated within sections of the survey.

Measures
Equality value. Participants were asked to rate their agreement with the statement, "There should be equality for all groups in Britain." Responses ranged from 1 (strongly disagree) to 5 (strongly agree).
Motivations to control prejudice. We used two items to measure internal and external motivation to control prejudice. These were selected on the basis of factor analyses of pilot study data involving over 400 British participants. The items were the highest loading items on the internal and external factors from Plant and Devine's (1998) scales. Participants were told, "People sometimes make an effort not to be prejudiced. To what extent would you disagree or agree that each of the following reasons describes your view?" The item for internal motivation to control prejudice was, "I attempt to act in nonprejudiced ways toward other groups because it is personally important to me." The item for external motivation to control prejudice was, "I try to appear nonprejudiced toward other groups in order to avoid disapproval from others." Participants responded from 1 (strongly disagree) to 5 (strongly agree).
Group rights. Participants were informed, "Not all groups in society want the same thing as the majority. How important do you feel it is that the particular wishes of each of the following groups is satisfied?" Participants responded for each of the six groups (i.e., women, people over 70, disabled people, Muslims, Black people, lesbian women, and gay men). The response options were 1 (not at all important), 2 (not very important), 3 (neither important nor unimportant), 4 (quite important), or 5 (very important).
Group equality. Participants were asked to tick a box to indicate whether they believed "attempts to give equal employment opportunities to [relevant minority group, depending on version] in this country have gone too far or not far enough?" The response options were 1 (gone much too far), 2 (gone too far), 3 (about right), 4 (not gone far enough), or 5 (not gone nearly far enough).
Social distance. The measure of social distance gauges respondents' anticipated emotional responses to varying levels of closeness toward members of different target groups. Depending on version, participants were asked, "How comfortable or uncomfortable do you think you would feel if a suitably qualified [target group person] was appointed as your boss?" They responded using a scale from 1 (very uncomfortable) through 3 (neither comfortable nor uncomfortable) to 5 (very comfortable). To some extent this measure may also tap respondents' willingness to work for members of the relevant social group, and therefore has implications for potential prejudice or discrimination in the workplace.

Preliminary Analyses
Correlation analyses revealed some significant but small relationships between participants' equality value or motivations to control prejudice on the one hand and gender, ethnicity, age, religion (whether Muslim), sexual orientation (whether heterosexual), but not disability, on the other (see Table 1). Analysis of covariance (ANCOVA; controlling for demographics) tested for differences between versions (A, B, C). These revealed no significant effect of version on equality value, F(2, 2,892) ϭ 2.67, p ϭ .069, 2 ϭ .002, nor on internal, F(2, 2,892) ϭ .45, p ϭ .638, 2 Ͻ .001, or external, F(2, 2,892) ϭ .05, p ϭ .956, 2 Ͻ .001, motivations to control prejudice. To adjust for the relationships in subsequent analyses all demographic variables were included as covariates.

Equality Hypocrisy: Equality Value Versus Group Rights
Our first goal was to establish whether there was evidence of equality hypocrisy. We examined the percentage of respondents who selected each response option for the equality values item and the group rights items. Figure 1 shows that, whereas 84% of respondents claimed they value or strongly value equality for all groups, fewer than 65% considered it quite important or very important to satisfy the needs of Black people, fewer than 60% considered it quite or very important for Muslims, and fewer than 50% considered it quite or very important for homosexual people. Descriptively, this amounts to an equality hypocrisy gap of between 15% and 30%.
Equality hypocrisy can be evaluated statistically by comparing the mean responses of equality value levels with mean levels of group rights and group equality for specific groups. Because the response scales for equality value and the other measures differ, we are cautious about making direct comparisons, but they seem meaningful to the extent that the highest score for all measures (5) reflects a high priority for equality, whereas a midscale score reflects a neutral preference. With these caveats in mind, pairwise comparisons between equality value and each of these other measures were all highly significant (df Ͼ 801, ts Ͼ 4.15, ps Ͻ .0001). Compared with equality value, respondents judged the group rights of paternalized groups to be closer to the maximum, whereas they judged the group rights of nonpaternalized groups to be further from the maximum. Thus, some respondents clearly do not attach equal importance to the rights of different groups. Overall, these descriptive differences show clearly that people's willingness to espouse equality as a value is greater than their willingness to ascribe the same rights and equality to different groups.

Equality Inconsistency
The group rights data indicate equality hypocrisy vis-à-vis equality values, but they also reveal differences in the application of rights to different groups (equality inconsistency). The next analyses examined group rights, group equality, and social distance judgments to establish whether there were systematic statistical differences between different target groups (i.e., equality inconsistency). We hypothesized that participants would place greater importance on equality for paternalized groups (women, people over 70, and disabled people) than for nonpaternalized groups (Muslims, Black people, and homosexuals). Group rights. A six-level (target group: women, people over 70, disabled people, Muslim people, Black people, and homosexuals) repeated measures ANOVA was conducted. The effect of target group was significant, F(5, 13,830) ϭ 20.32, p Ͻ .001, 2 ϭ .01. All pairwise differences were significant at p Ͻ .0001 aside from a nonsignificant difference between women and people over 70. Group rights were rated highest for disabled people (M ϭ 4.22, SE ϭ .02), then for women (M ϭ 4.15 , SE ϭ .02), people over 70 (M ϭ 4.14 , SE ϭ .02), Black people (M ϭ 3.78 , SE ϭ .02), Muslims (M ϭ 3.62 , SE ϭ .02), and finally, homosexuals (M ϭ 3.38 , SE ϭ .02). Importantly, consistent with our hypothesis a planned comparison between the three paternalized and three nonpaternalized groups showed a highly significant difference. Group rights were rated higher for paternalized (M ϭ 4.16, SD ϭ .81) than for nonpaternalized (M ϭ 3.59, SD ϭ .96) groups, t(2,894) ϭ 38.38, p Ͻ .0001, d ϭ .64.
Group equality. Because advocacy of equal employment opportunity for different Figure 1. Means for strength of endorsement of the value of universal equality ("equality for all groups") and of importance of the rights and advocacy of greater equality of opportunity for specific groups. Higher means represent stronger endorsement. The equality value response scale is from strongly disagree to strongly agree; the group rights scale is from not at all important to extremely important; the group equality scale is from gone much too far to not gone nearly far enough. Error bars depict standard errors.
pairs of groups was measured in different versions of the survey, we analyzed these judgments using a 2 (Type of Group: Paternalized, Nonpaternalized) ϫ 3 (Version: A [women, homosexuals], B [people over 70, Muslims], C [disabled, Black people]) mixed analysis of variance (ANOVA) with survey version as a between participants factor.
Results revealed a significant main effect of type of group, F(1, 2,454) ϭ 12.72, p Ͻ .0001, 2 ϭ .01. As predicted, paternalized groups (M ϭ 3.73, SE ϭ .02) were rated higher than nonpaternalized groups (M ϭ 3.02, SE ϭ .02). There was also a significant main effect of survey version, F(2, 2,454) ϭ 5.41, p ϭ .005, 2 ϭ .004, whereby advocacy of group equality was rated higher in Version C (Black people and disabled people) than in Version A (women and homosexuals; p ϭ .008), and compared to Version B (people over 70 and Muslim people; p ϭ .003).
There was also a significant type of Group ϫ Version interaction, F(2, 2,454) ϭ 16.37, p Ͻ .001, 2 ϭ .01. Simple effects of type of group within version showed that, regardless of survey version, group equality scores were significantly higher (all ps Ͻ .0001) for the paternalized groups (women, people over 70, and disabled people) than for the nonpaternalized groups (homosexuals, Muslim people, and Black people, respectively).
Within the paternalized groups, group equality scores were higher for people over 70 (M ϭ 3.30, SE ϭ .03) and for disabled people (M ϭ 3.34, SE ϭ .03) than for women (M ϭ 3.18, SE ϭ .03; p ϭ .003 and p Ͻ .0001, respec-tively), but there was no significant difference in group equality ratings for people over 70 and disabled people (p ϭ .314).
Within nonpaternalized groups, advocacy of group equality was rated significantly lower for Muslim people (M ϭ 2.70, SE ϭ .03) than for homosexuals (M ϭ 3.07, SE ϭ .03) and Black people (M ϭ 3.08, SE ϭ .03; ps Ͻ .0001). There was no significant difference between advocacy of equality for homosexuals and Black people (p ϭ .820).

Is Equality Inconsistency Dependent on Equality Value?
A plausible reason for equality hypocrisy across the population as a whole might be that those who more strongly value equality for all will indeed espouse greater equality for any particular group. Those who value equality less may express more divergent views about the importance of equality for different groups. To test this idea we divided the sample according to whether their general equality value scores were at the midpoint or below (not valuing equality) or above the midpoint (valuing equality). We then examined the scores on dependent variables for the paternalized versus nonpaternalized groups. These analyses employed mixed ANOVAs (Equality Value: High vs. Neutral and Low) ϫ (Type of Group: Paternalized, Nonpaternalized). We examined responses to three dependent variables, group rights, group equality, and social distance. Results are depicted in Table 2. Note. N ϭ 2,895. SE ϭ standard error; df ϭ degrees of freedom. All main and interaction effects were significant at p Ͻ .05, excluding the target group main effect on social distance (boss), which was nonsignificant (p ϭ .113).
All three ANOVAs revealed a significant interaction between equality value and type of group. For all three dependent measures, pairwise comparisons showed that all four means differed from one another (ps Ͻ .05). The pattern is consistent across dependent variables. Respondents who valued equality more highly did indeed advocate higher group rights, group equality, and desire less social distance for each specific group. However, even though these respondents valued equality highly, they significantly favored paternalized groups over nonpaternalized groups, meaning that equality hypocrisy persists. Indeed, when we inspected the mean scores on group rights and group equality among respondents who had selected the strongly agree option for equality values, even these respondents significantly favored paternalized groups over nonpaternalized groups on both measures (ps Ͻ .001). For the social distance measure, the difference was highly significant among those who agreed (p Ͻ .001), and nonsignificant (though in the same direction) among those who strongly agreed (p ϭ .131).

Motivation to Control Prejudice and Equality Inconsistency
To examine the predictive effects of individual differences in motivation to control prejudice and equality value on equality inconsistency we computed within-person variance scores from ratings of paternalized and nonpaternalized groups. For the group rights variable we were able to compute variance using ratings of all six target groups. For the group equality and the social distance variables the variances were computed using the target pair in the relevant survey version (i.e., women and homosexuals; disabled and Black people; people over 70 and Muslims). Whether or not version was controlled for (by creating two dummy variables) made no difference to the findings. Because these scores tap within-respondent variance in judgments about the different groups, higher scores reflect greater inconsistency.
We hypothesized that internal motivation to control prejudice should be associated with lower equality variance. Second, given that survey responses were observable (by the interviewer) we also expected external motivation to control prejudice to be associated with lower equality variance. Therefore, equality value and both types of motivation to control prejudice should be associated with lower equality variance. In principle, if all three are high, there should be no equality variance because someone who values equality for all, and who does not wish to be or be seen to be prejudiced should view the rights and equality of all groups as equally important. We also propose, therefore, that equality variance should be maximized if equality value and both types of motivation to control prejudice are all low.
To test whether internal and external motivation to control prejudice moderated the relationship between general equality values and equality variances for each measure, we used Hayes' (2013) PROCESS macro (Model 3 for multiple moderation). In separate analyses of the withinperson variance of each dependent variable (group rights, group equality, social distance), equality value was the independent variable and internal and external control were separate moderators.
Simple slopes analyses (Model 1) were conducted to probe the Equality Value ϫ Internal Motivation to control prejudice interaction. External motivation to control prejudice was also retained in the model and entered as a covariate. This revealed that equality value only predicted variance in group rights at low levels of internal motivation (B ϭ Ϫ.17, SE ϭ .02, p Ͻ .0001) but not at high levels of internal motivation (B ϭ Ϫ.04, SE ϭ .03, p ϭ .118; Figure 2a). Importantly, the effect of internal motivation was smaller when equality value was high (B ϭ Ϫ.05, SE ϭ .02, p ϭ .022) than when equality value was low (B ϭ Ϫ.17, SE ϭ .02, p Ͻ .0001).
Similarly, simple slope results for the Equality Value ϫ External Motivation interaction (with internal motivation as a covariate) revealed that equality value only pre-dicted variance in group rights at low levels of external motivation (B ϭ Ϫ.17, SE ϭ .03, p Ͻ .0001) but not at high levels of external motivation (B ϭ Ϫ.05, SE ϭ .03, p ϭ .073; Figure 2b). Moreover, when equality value was low external motivation had no effect on variance (B ϭ .005, SE ϭ .02, p ϭ .816). However, when equality value was high, respondents with higher external motivation also showed greater variance in their responses (B ϭ .10, SE ϭ .02, p Ͻ .0001).
To summarize the overall pattern, we note two points. First, the variance was greatest when equality value, internal motivation, and external motivation were all low. Variance was smallest when equality and internal motivation was high but external motivation was low. Second, the relationship between levels of equality and variance was strongest when both internal and external motivations were low and smallest when both were high. Post hoc inspection of the simple slope for equality value within levels of internal and external motivation showed they were significant (ps Ͻ .05) except when both internal and external were high, B ϭ Ϫ.03, SE ϭ .04, p ϭ .347.
Simple slopes for the three-way interaction showed that the effect of equality value was greatest when both external and internal motivations were low (B ϭ Ϫ.25, SE ϭ .04, p Ͻ .0001). However, equality value no longer predicted group equality when internal motivation was high (B ϭ Ϫ.07, SE ϭ .04, p ϭ .093) or external motivation was high (B ϭ .04, SE ϭ .06, p ϭ .466), or when both external and internal motivations were high (B ϭ .02, SE ϭ .05, p ϭ .616). Similarly, only when the equality value was low did internal motivation (B ϭ Ϫ.15, SE ϭ .03, p Ͻ .0001) and external motivation (B ϭ Ϫ.20, SE ϭ .04, p Ͻ .0001) significantly relate to variance in group equality. To summarize the overall pattern, the variance was large when equality value, internal motivation, and external motivation were all low. Variance was smallest if any one of these variables was high. The relationship between levels of equality and variance was stronger when both internal and external motivation were low than when either were high (see Figure 3).
Social distance. Higher internal motivation to control prejudice (B ϭ Ϫ.21, SE ϭ .05, p Ͻ .001), higher external motivation to control prejudice (B ϭ Ϫ.13, SE ϭ .05, p ϭ .010), and higher support for the equality value (B ϭ Ϫ.15, SE ϭ .04, p ϭ .002) significantly predicted lower variance in social distance. There was Figure 2. Plots for the Equality Value ϫ Internal Motivation to Control Prejudice interaction and the Equality Value ϫ External Motivation to Control Prejudice on variance in group rights. Low and high refer to values 1 standard deviation below and above the variable's mean, respectively. also a significant Equality Value ϫ Internal Motivation to Control Prejudice interaction (B ϭ .11, SE ϭ .03, p ϭ .001). Simple slopes analysis (Model 1 with external motivation entered as a covariate) revealed that equality value only predicted variance in social distance at low levels of internal motivation (B ϭ Ϫ.18, SE ϭ .04, p Ͻ .0001) but not at high levels of internal motivation (B ϭ .06, SE ϭ .04, p ϭ .181; see Figure 4). Similarly, internal motivation only predicted variance at low levels of equality value (B ϭ Ϫ.21, SE ϭ .04, p Ͻ .0001). Stated differently, variance was higher when both internal control and equality value were low than  when either variable was high. There was also a marginal Equality Value ϫ External Motivation to Control Prejudice interaction on variance in social distance (B ϭ .07, SE ϭ .04, p ϭ .068), which followed a similar pattern. Although the three-way interaction of Equality Value ϫ Internal Motivation to Control Prejudice ϫ External Motivation to Control Prejudice on group rights was nonsignificant, B ϭ Ϫ.06, SE ϭ .04, p ϭ .124, we conducted post hoc tests of simple slopes for comparison with the effects on group equality, the relationship between equality value and group rights was significant only when both internal and external were low, B ϭ Ϫ.27, SE ϭ .06, p Ͻ .001. In summary, as with group equality, the variance in social distance was large when equality value, internal motivation, and external motivation were all low. Variance was smaller when any one of these variables was high. The relationship between levels of equality and variance was stronger when both internal and external motivation were low than when either were high.

Discussion
Can a society in which a large majority claims to value the human right of equality for all regard itself as meeting the requirements of Article 1 in the UDHR? From this study of the United Kingdom during one of its more liberal eras, the answer appears to be that espousing the general value of equality is not sufficient. The present research exposes clear evidence of equality hypocrisy because people were less willing to endorse equal rights for specific groups than they were for all groups. Moreover, this hypocrisy was manifested both at the aggregate level characterizing society as a whole (see Figure 1), and within individuals who chose to prioritize the equal rights of particular groups more than other groups (showing equality inconsistency).

Evidence for Equality Hypocrisy
Whereas previous research has highlighted the potential mismatch between overall human rights support and application to specific groups (e.g., Staerklé & Clémence, 2004), the present research examined whether people apply their equality values to the same extent across different intergroup contexts and different types of minority groups. Arguably, this is a stronger test of equality hypocrisy as it determines whether people do apply the principle of equality equally across different types of minority. Our findings showed clear support for the existence of equality hypocrisy. Specifically, respondents advocated equality as a value more strongly than they advocated equality for nonpaternalized minority groups. They also judged the rights of some groups to be more important than the rights of others. Strikingly, 22% were prepared to assert that equality had gone "too far" for Muslims.

Evidence for Equality Inconsistency
We proposed that differences in the application of equality to different groups would reflect differences in paternalistic stereotypes associated with each group . In particular, we expected that because paternalized groups pose little threat to the status or power of other groups, respondents would be more willing to grant equality to those groups than to nonpaternalized groups. Specifically, we proposed and found that respondents advocated equality more strongly for women, older people and disabled people, than for Blacks, Muslims and homosexual people. Importantly, differential equality in favor of paternalized groups occurred regardless of whether respondents were asked to consider all six of these groups or whether they were asked to consider one of three different pairings of the groups. This evidence suggests strongly that equality inconsistency in favor of paternalized groups is not an artifact of demand characteristics or measurement procedures, but is a robust effect.

Predicting Individual-Level Equality Inconsistency
We then pursued the question of why equality inconsistency between paternalized and nonpaternalized groups exists and whether it shares a common basis with intergroup prejudice. We reasoned that people who value universal equality more highly should be more consistent in their application of equality across different groups. In addition, prior research has established that people may moderate their expressions of prejudice depending on both their personal (internal) motivation to be unprejudiced, and social (external) motivation to be unpreju-diced. If application of equality values is related to intergroup prejudice then these two motivations should also result in greater consistency in the application of equality across specific groups. However, we could not be sure whether equality values would subsume prejudice motivations, whether these different motives and values would have independent additive effects or whether they would interact. As far as we are aware this issue has not been explored in previous research.
Across different measures, the results showed that the motivations to control prejudice and equality values had interactive effects. Either high equality value or high internal motivations to control prejudice were sufficient to reduce inconsistency in judgments of the rights of different groups. Similarly, consistency in social distance (prejudice) responses was greater if either equality value or internal motivation to control prejudice were high, than if both were low. We note that the main effect of external motivation to control prejudice differed across measures. Future research may need to consider why this might be.
Taken together, these findings are both encouraging and concerning. It is encouraging that we have identified three possible ways to promote greater application of Article 1 of the UHDR. One is to simply reinforce the basic value of equality. Another is to promote motivation to be unprejudiced, and the third may be to reinforce the idea that being seen to be prejudiced is highly undesirable. The latter strategy implies that people may in fact remain prejudiced, but simply not show this publicly. However, reducing public prejudice may have beneficial indirect effects through changing social norms (cf. Aronson, 1992;Berkowitz, 2005).
Less encouraging is the persistence of significant equality inconsistency even among people who we might expect to show none. Specifically, even those who most highly valued equality showed equality inconsistency. We believe that this reflects the pervasiveness and power of societal intergroup relations and stereotypes, and indicates a need for future research to explore ways to break the social and psychological barriers in the treatment of these different kinds of groups. Our findings suggest that it may be helpful if equality and diversity training can promote equality consistency through multiple routes, including appealing to people's equality value and also their motivations to be unprejudiced. The findings also highlight the importance of incorporating an intergroup relations perspective within equality and diversity training. For example, one promising strategy encourages people to think of multiple counterstereotypic social categories, thereby leading to greater egalitarianism and lowered generalized prejudice toward a multitude of both paternalized and nonpaternalized groups (Vasiljevic & Crisp, 2013).
Another promising intervention is the value self-confrontation technique, which aims to either change or stabilize people's beliefs, attitudes, values, and behavior (Grube, Mayton, & Ball-Rokeach, 1994;Rokeach, 1973Rokeach, , 1975). Rokeach's classic studies asked participants to rank a number of values, among which were equality and freedom. Freedom was usually ranked high, and equality rather low, which served as the main point given in the feedback, whereby Rokeach drew people's attention to the wide discrepancy in valuation of freedom and equality. Rokeach surmised that participants would be dissatisfied with this discrepancy, which would lead them to change their values, attitudes, beliefs, and behaviors. The value selfconfrontation technique has been extensively tested and results have been promising, especially considering the longitudinal effects of this technique (Altemeyer, 1994;Ball-Rokeach, Rokeach, & Grube, 1984;Rokeach, 1973). It would be interesting and promising to apply this self-confrontation technique to equality inconsistency.
Based on intergroup relations theories, we proposed that equality hypocrisy and equality inconsistency could arise for several reasons. Equality hypocrisy (the general failure to apply espoused equality values) may reflect ingroup biases due to ingroup commitment, intergroup competition, or social identity distinctiveness and esteem motivations (Abrams, 2015;Abrams & Hogg, 1988;Ellemers, Spears, & Doosje, 2002).

An Important Applied Issue: Relevance to Policy
Our research shows how attitudes to human rights are expressed in ways that appear inconsistent with people's core values. We tested these questions in a social and political policy context that was actively promoting equality, and that was engaged with the goal of protecting and advocating human rights. After the 2007-2008 world banking crisis, the Labour Government was succeeded by a Conservative/Liberal-Democrat coalition. One of the coalition's earliest acts was to cut the budget and size of the Equality and Human Rights Commission dramatically. The coalition government launched sustained criticism of the judgments of the European Court of Human Rights, and bemoaned the imposition of undue "political correctness" from outside the United Kingdom. In this rhetoric a sustained theme has been that of undeserving groups (those espousing different values, foreigners stealing British jobs, welfare scroungers, feckless youth, and so on). Politicians have argued that equal rights should only be granted to these groups if they assume equal "responsibilities" (an economic and structural impossibility). We consider that the success of these rhetorical strategies lies in their capacity to activate intergroup motives and to drive a wedge between the rights of minority status groups that are paternalized versus nonpaternalized. Narratives that contrast the deserving and undeserving groups or subgroups (among the poor, immigrants, etc.) are particularly insidious as they are likely to combine paternalistic prejudices (e.g., benevolent sexism) with nonpaternalistic prejudices to sustain the status quo. Paternalistic prejudice can ostensibly demonstrate tolerance and consideration of human rights, while nonpaternalistic prejudices demonstrate defense of ingroup values and freedoms. Yet, in this type of rhetoric, support for minorities is conditional on their posing no threat and remaining dependent, while denial of rights to nonpaternalized minorities is justified with more overtly "hostile" forms of prejudice that focus on the threats to ingroup culture, economy or security posed by such groups.

Limitations and Future Directions
The present research has several limitations. One is that we did not use identical response scales to measure equality value and equality judgments relating to specific groups. Although the response anchors were necessarily different, and may have introduced differences in item difficulty, these differences might also be construed as a virtue in the sense that they reduced the risk of common measurement effects and reduced likely social desirability effects in terms of trying to appear consistent.
We are aware that it is preferable to use multiple items to measure constructs in psychological research. Single items are likely to yield smaller effects and this may account for some of the small effect sizes in the present research. However, the advantage of a very large representative sample and the use of pretested items that are representative of particular constructs is that what is lost in measurement error is partially compensated for in statistical power. In addition, small effect sizes can sometimes underpin important substantive effects (Prentice & Miller, 1992). The social relevance and generalizability of our findings are greatly enhanced by use of a large and nationally representative sample, but we recognize that additional experimental research could help to explore the relevant processes and mechanisms in greater detail.
An empirical limitation is that the research was conducted only in one cultural setting. Kymlicka (2001) argues that whereas Western cultures can ideologically accommodate both individual freedom and group rights under the umbrella of "equality," the same is not true in all cultures. Notwithstanding that caveat, we have several reasons for believing that the findings and general processes at work will generalize, at least to most Western cultures. First, there was some cultural heterogeneity within our national sample, and the findings emerged when multiple demographic variables were accounted for as covariates. Second, the general phenomenon of equality hypocrisy, which we observed across different types of group, echoes the findings from other cultural contexts that inconsistency exists between general equality values and application to a single minority. Third, the general principles underpinning the stereotype content model have been shown to have good cross-cultural replicability (Fiske & Cuddy, 2006;Cuddy et al., 2009). Therefore, even if the particular groups that are more paternalized differ between cultures, we would still expect that people would more willingly endorse equality for paternalized groups. It would be very useful for future research to explore cross cultural differences in equality hypocrisy to illuminate the generalizability of the role of paternalization. Related to this question is whether there are important nuances and differences in equality hypocrisy as applied to nonpaternalized groups, and particularly whether there are situations in which they can attract perceptions of being highly competent without also posing a threat to majority groups.
There are several other interesting avenues for future research. One would be to investigate how other values articulate with group motives, and how moral principles may be strategically incorporated into group-based judgments (cf. Abrams, Rutland, Ferrell, & Pelletier, 2008;Abrams, Randsley de Moura, & Travaglino, 2013;Haidt, 2007;Rutland, Killen, & Abrams, 2010). Another is to test whether equality inconsistency can be reduced by priming specific motives and values (cf. Legault et al., 2011;Maio et al., 2009). Finally, although our measure of social distance implies certain behavioral orientations and emotional responses to minority group members (Kraus, 1995), it would be useful if future research could examine behavioral outcomes more directly. To conclude, we hope that the present research provides a unique source of evidence and new insight into equality hypocrisy and that, extending earlier work (Doise et al., 1999;Mogghadam & Riley, 2005), this will invigorate further social psychological research in this area.