1In France, as in many countries, research has established the existence of discrimination in connection with the ethnic or racial origin of individuals (Lippens et al. 2023; Quillian and Midtboen 2021). In its 2020 report, the French antidiscrimination platform Défenseur des droits pointed out that “origin-based discrimination remains widespread in France” (p. 3). According to the most recent meta-analyses, when comparing the rate of racial discrimination in hiring in some twenty countries (in Europe, North America, and Australia), it appears that discrimination is a problem in all countries and that with respect to certain criteria, the rate of discrimination in France is higher than in all other countries (Quillian et al. 2019, Thijssen et al. 2022). How can we explain the origin of these discriminatory behaviors? This article deals precisely with this question. The answer requires an interdisciplinary approach wide enough to address the many sociological, psychological, economic, historical, and political factors involved.
2This article presents the theoretical, methodological, and empirical contribution of experimental social psychology, a discipline whose contribution to the study of these issues remains unrecognized in France. More precisely, it will first examine the theories behind the intergroup approach in social psychology, and in a second part, the methods for measuring discriminatory behaviors in the laboratory. It is imperative to develop these methods to be able to test hypotheses about the origin of discriminatory behavior. The third and last part will discuss the most recent research on institutional discrimination and implicit measurement methods, which, according to the innovative theoretical framework developed by Payne and Hannay (2021), offer new ways of studying contemporary manifestations of institutional discrimination.
3Social psychology makes a distinction between the cognitive (stereotyping), affective (prejudice), and behavioral (discrimination) aspects of intergroup relations (Yzerbyt and Demoulin 2019). Stereotypes belong to the category of beliefs. They consist in associating certain psychological characteristics (positive or negative) with members of a group: for example, women are “sensitive and affectionate”; men are “rude” (Guimond 2010). In social psychology, the notion of prejudice, on the contrary, refers to inevitably negative assessments and affects. Prejudice is an attitude (not a belief) that reflects an aversion or a negative opinion of another person, because of the person’s membership in a group. We speak, for example, of prejudice against Blacks (racism), against women (sexism), against Jews (anti-Semitism), or against foreigners (xenophobia, Bourhis 2021). Finally, the concept of discrimination is exclusively used to refer to the actions and behaviors of individuals (and not what they think or feel).
4Although there are connections between these three concepts (stereotyping, prejudice, and discrimination) in the sense that, for example, racist prejudice or attitudes are often the motivation behind acts of racial discrimination, it is important to keep these distinctions for several reasons. First, it has been confirmed that members of one group may widely share the same stereotypes about another group while only a very small minority will also show prejudice against that group (Leyens 2015). Second, it has also been documented that discriminatory behavior can occur without necessarily any prejudice, as will be explained further. But these distinctions are not what mainly characterizes the intergroup approach (Guimond 2004). Rather, the main tenet of the intergroup approach is to focus on group membership when analyzing these phenomena for the purpose of developing a collective psychology that can extend beyond the limits of individual psychology.
5From this point of view, the crucial aspect of stereotyping, prejudice and discrimination is that they involve individuals who are generally perceived by others and who perceive themselves as belonging to different groups (White/Black, male/female, national/immigrant, etc.). We use the word ingroup to qualify the group to which the individual belongs and outgroup to qualify other groups. Thus, discrimination will be defined as the practice of treating an individual differently and unfavorably because the individual is perceived as belonging to an outgroup. When an apartment rental is refused to a person who is Black, Arab or Muslim (i.e., the person is perceived as belonging to an outgroup), this is an instance of discrimination. For European proponents of the intergroup approach, “to discriminate is to treat members of the outgroup differently than those who are members of our group” (Yzerbyt and Demoulin 2019 : 71). This unjustified difference in treatment between the ingroup and the outgroup defines discriminatory behavior. As Deaux et al. (1993) point out in a major social psychology textbook, “discrimination refers to specific behaviors toward members of a group that are unfair in comparison with behavior toward members of other groups” (p. 355).
6As research has progressed, it has become important to distinguish three forms of discrimination: ingroup favoritism, outgroup discrimination, and outgroup favoritism (Bourhis 2021, Brewer 1999, Guimond 2010). This distinction was made because some forms of discrimination imply a relative preference for the ingroup, i.e., a more favorable treatment of the ingroup, without necessarily demonstrating a rejection of the outgroup. This is known as ingroup favoritism. It is perhaps the most common form of discrimination, and it aims to favor the ingroup without seeking to inflict ill-treatment on the outgroup.
7Other cases of discrimination obviously involve persecuting members of an outgroup or harming them psychologically or physically. In these cases, we speak of anti-outgroup discrimination or active denigration of the outgroup and thus, a more extreme form of discrimination found in hate crimes and in episodes of open conflict involving attempts at ethnic cleansing or genocide (Badaan and Jost 2020, Bourhis 2021).
8Finally, when there is a relative preference for the outgroup, i.e., a discrimination in favor of the outgroup and not in favor of the ingroup, we speak of outgroup favoritism (Bourhis 2021). In this respect, we know that White children who have the choice of playing with a White doll or a Black doll inevitably choose the doll of their group, the White doll. This is ingroup favoritism. In contrast, Black children show a strong tendency to choose the White doll over a Black doll (see Guimond 2010). This is outgroup favoritism. This well-known phenomenon observed among many groups of low social status is often interpreted as reflecting a rejection of one’s own group, a “self-hatred” due to internalized negative stereotypes (Jost and Banaji 1994). However, this interpretation is not as well established as its proponents claim. Other interpretations have been better documented empirically (see Faniko et al. 2022, Guimond 2010). But before we address this subject in greater detail, we need to gain a better understanding of the contribution of the intergroup approach by examining the research behind it, starting with the work of Sherif (1966).
9The intergroup approach was first developed in social psychology by Musafer Sherif (1966) whose research was based on a general theory of social conflict, known throughout the humanities and social sciences as realistic conflict theory (Guimond 2010, Klein, Wollast et al. 2022). Sherif’s central theoretical proposition is that the competitive or cooperative nature of intergroup relations largely determines the attitudes and behaviors of group members toward each other. Through research conducted in a naturalistic setting at a teenage summer camp, Sherif sought to demonstrate that when the young boys were divided into two groups and made to compete for scarce resources, negative intergroup attitudes and discriminatory behaviors could be observed. Within a short period of time, well-behaved boys became thugs capable of insults, threats, and aggressive behavior toward outgroup members (Yzerbyt and Demoulin 2019). This would happen regardless of the psychological characteristics of the boys involved, and even if the best friend from one of the boys before the study was on the opposing side. This “best friend” then became an “enemy”. This led to the distinction between interpersonal and intergroup relations.
10Drawing on Sherif, a book by Taylor and Moghaddam (1994) defines the concept of intergroup relations as “any aspect of human interaction that involves individuals perceiving themselves as members of a social category or being perceived by others as belonging to a social category” (p. 6). For Sherif, and contrary to the dominant theories of social psychology at the time (e.g., Adorno et al. 1950), hostility in interpersonal relations cannot be considered as equivalent to hostility in intergroup relations. It is therefore misleading to believe that scientific knowledge about interpersonal relations can be generalized to understand intergroup relations. Interpersonal relations involve intimate and relatively sophisticated knowledge of another person. While hate crimes, which primarily target individuals who belong to a racial, ethnic, or religious minority (Badaan and Jost 2020) involve perpetrators who may travel thousands of miles to harm their victims without having the least knowledge of their personal characteristics (Dunbar 2018).
11Sherif’s work served as a foundation for the development in Europe of the social identity theory (Tajfel 1972, Tajfel and Turner 1979). This theory has significantly contributed to a renewed interest in the intergroup approach. Today, it represents one of the most important theoretical approaches in social psychology (Brown 2020).
12In 1970, Tajfel published an article in Scientific American that presented a new experimental paradigm known as the “minimal group paradigm”. The author conducted a laboratory study of the minimal conditions under which members of one group exhibit discriminatory behaviors toward members of another group. All the economic, political, or historical factors that could lead to discrimination were carefully removed from the experimental situation by creating two artificial groups. Under a trivial pretext (preference for an abstract painting by Klee or Kandinsky), participants were assigned to one group (“you are in the Klee group”), and they learned of the existence of another group (the Kandinsky group). No one knew the identity of the other members of either the Klee or Kandinsky group and there was no interaction between the members of the two groups. Also, since the two groups had been created completely artificially and randomly, there was no history of rivalry or mutual hatred between the two groups, which were therefore considered “minimal groups”. After this categorization into two groups, the experimenter explained to the participants that the purpose of the experiment was to study decision making. Each participant had to decide how to divide monetary resources between two anonymous individuals, one of whom was presented as a “Klee group member” and the other as a “Kandinsky group member”. The purpose was to measure discriminatory behavior and determine whether more resources were granted to one group and less to the other group, i.e., a differential treatment, with no justification other than membership in the Klee or Kandinsky group.
- 1 1 Subsequent research (see Bourhis 2021) has shown that this unusual egalitarian tendency occurred (...)
13Under this paradigm, the goal of the study was to add, one by one, variables supposed to be causes of discrimination (e.g., intergroup competition) and to measure their effects. However, the initial findings, which were subsequently confirmed in many other laboratories, showed that no variables other than social categorization were necessary. Some participants distributed resources in a strictly egalitarian manner,1 but the greatest tendency in the responses was to give more to members of one’s group, and even to try to maximize the difference between the two groups in favor of the ingroup. These discriminatory behaviors occurred in the absence of prejudice or any history of rivalry between the Klee and Kandinsky groups. Throughout further research, “matrices” were developed to measure different response strategies used by participants in these experiments (parity/equality; ingroup favoritism; maximization of intergroup difference, etc.). Claude Flament (Aix-Marseille University) has played an important role in the development of these “matrices” (Bourhis et al. 1997). Social identity theory was developed to explain the results obtained in these experiments.
14Since participants were in no case allowed to allocate a monetary reward to themselves, Tajfel and Turner ruled out from the outset the role of self-interest as a reason for discrimination. Rather, they considered it an issue of identity and self-esteem. Categorization in a group confers an identity on the individual, a place in the social system. This identity is assessed through intergroup comparisons. Since individuals usually prefer to have a positive image of themselves, social identity theory posits that our desire is to have a positive social identity. We want to belong to a group that will allow us to obtain a positive social identity, that is, a group that distinguishes itself positively from other groups. Discrimination is a behavior that allows us to obtain this identity.
15In the minimal group paradigm, the number of points or resources assigned to the two groups was the only dimension of intergroup comparison. Individuals therefore sought to ensure that this comparison was favorable to their group. The tendency to maximize the difference in gains between the ingroup and the outgroup, even if it means sacrificing a larger gain for one’s group, is evidence of this motivation for a positive social identity. In short, while Sherif describes an objective competition for limited resources, Tajfel and Tuner refer instead to “social” competition for a positive social identity, a symbolic stake that has nothing to do with a material reward, but which is essentially linked to the process of intergroup comparison.
16From these basic principles, the social identity theory, presented as complementary to Sherif’s, proposes a new conceptual framework for understanding the behavior of individuals in various contexts. Without going into detail, it is sufficient for our purpose to remember a simple but fundamental idea: discrimination is an intergroup behavior based on an awareness of ingroup/outgroup categorization. In the experiments on minimal groups, the participants could have demonstrated an individual behavior, regardless of the Klee/Kandinsky categorization, which was an artificial and essentially irrelevant categorization. Instead, they engaged in the behavior of an individual acting as a member of a group. This theory thus emphasizes that to understand discrimination, we must understand how and why individuals act as members of a group and not as individuals. In other words, from the perspective of social psychology, this theory invites us to consider that when individuals categorize themselves as members of a group, it leads to changes in their psychology (at the level of cognitions, emotions, and motivation, etc.) that must be taken into account.
17These changes have been more clearly demonstrated with the complementary self-categorization theory subsequently developed by Turner in 1987 (see Turner et al. 1994, Turner and Reynolds 2013). While social identity theory describes ingroup-outgroup categorization, Turner’s theory (1987) deals with self-categorization and argues that the self-concept, far from being a rigid psychological structure, is extremely flexible and malleable. Depending on the context, an individual can adopt an almost infinite number of self-categorizations: as a friend, brother, father, boss, environmental activist, Parisian, French, European, etc. In practice, Turner (1987) points out that there are three main levels of self-categorization: the level of personal identity (I am a singular individual different from others); the level of social identity (I am an interchangeable member of a group), and the supra-ordinate level (I am a human being). The psychology of this individual (the individual’s values, opinions, personality, behavior) can change significantly depending on which level of self-categorization is activated at a specific moment (Guimond 2010, Turner et al. 1994, Turner and Reynolds 2013).
18The relevance of these processes can be illustrated by Verkuyten and Hagerdoorn’s (1998) experiment that aimed to study prejudice toward Turks in the Netherlands. In the first experimental condition, Dutch students answered a series of questions that made their national identity salient. In the second condition, they first completed a self-esteem questionnaire that made their personal identity salient. All participants were then tested for their level of right-wing authoritarianism, a personality trait that according to Altemeyer (1998) is the main factor behind ethnic prejudice, and for their level of prejudice toward Turks in the Netherlands.
19Verkuten and Hagerdoorn’s (1998) findings, later confirmed by other researchers (see Guimond 2022), show that in the “personal identity” condition, the most authoritarian Dutch students effectively demonstrated the most prejudice toward Turks. In agreement with Altemeyer (1998), it thus seems that prejudice is essentially a matter of individual personality. However, in the “national identity” condition, Verkuyten and Hagerdoorn (1998) did not find any correlation between right-wing authoritarianism and the measure of prejudice. In their experiment, intergroup perceptions, such as stereotypical perceptions of national group characteristics, predicted the manifestation of anti-Turkish prejudice.
20In other words, what motivates and determines the behavior of individuals closely depends on the level of self-categorization. The variables that shape people’s behavior in the context of interpersonal relations, such as personality traits for example, can become completely irrelevant in an intergroup context where people define themselves with respect to their social identity (or national identity). However, as soon as individuals interacting together talk about “the Dutch” and “the Turks” for example, social interactions will cause individuals to perceive themselves as members of groups, and not as simple individuals, and self-categorization at the level of their social identity will inevitably become the most relevant.
21Social identity and self-categorization theory thus offer important insights into the way stereotyping, prejudice and discrimination operate. The process of social categorization, i.e., the tendency for all of us to categorize individuals, is generally considered to be the foundation for the development of stereotyping, prejudice, and discrimination. The resource allocation task used in the minimal group paradigm was subsequently used to measure discriminatory behavior in many studies (see Anier et al. 2018). A serious limitation of the initial studies, however, is that these minimal groups lack social status. This limitation was later corrected (see Sachdev and Bourhis 1987), but the fact remains that social identity theory originally addressed the issue of discrimination in a completely egalitarian social system where the Klee and Kandinsky groups had the exact same social status. Social dominance theory has sought to remedy this problem.
22In 1999, Jim Sidanius, professor of psychology and African American studies at Harvard University, and Felicia Pratto, professor of experimental social psychology at the University of Connecticut in the United States, presented a new theory of intergroup relations in a book that incorporated previous theories. Their starting point is the observation that in all complex societies, there is a hierarchical social organization: some groups have more power, wealth, and prestige than others. Why is this so? To understand phenomena of oppression, conflict, and discrimination, this theory suggests that the central question must be: what produces and perpetuates this hierarchical social organization composed of at least one dominant group at the top of the social pyramid and one or more dominated groups at its base? The theory offers a holistic and multi-level explanation that includes individual and institutional factors and draws as much from Vilfredo Pareto’s sociology of elites as from Adorno et al.’s (1950) theory of authoritarian personality, and even from evolutionary psychology.
23According to this theory, social hierarchy is the result of the combined operation of three factors: individual discrimination, institutional discrimination, and asymmetries in the behavior of group members with different social status. If we add up all the individual acts of discrimination in a society with respect to employment or access to education, for example, the members of one group end up being in advantaged social positions (e.g., Whites in the United States), while members of other groups, who are discriminated against (e.g., Blacks in the United States), are at the bottom of the social scale.
24In addition to this individual aspect, Sidanius and Pratto (1999) extensively documented, in several chapters, how institutions within Western democracies operate in a way that also leads to unequal treatment and reinforces the hierarchical structure of society. We will come back to this aspect of institutional discrimination in the last part of this article.
25Behavioral asymmetry is the third factor that strengthens social hierarchy. It is based on a large body of research, including studies on social identity theory, that shows that members of low-status groups frequently demonstrate egalitarian behaviors and even self-deprecating tendencies, while members of high-status groups virtually never behave in this way. The outgroup favoritism discussed above is most prevalent among low-status group members. Thus, these asymmetries necessarily perpetuate the social hierarchy.
26A striking example of asymmetry was observed among scientists themselves in their assessment of the contribution of different disciplines to the solution of complex problems. Urbanska et al. (2019) surveyed 280 researchers from 26 different countries. Among them, 160 were working in a discipline traditionally recognized as a science (physics, mathematics, engineering, etc.) and 120 in the social sciences (economics, sociology, psychology, etc.). Participants were asked to say how much colleagues from the natural and social sciences could contribute to advancing research on complex problems. The “scientists” showed a clear pro-ingroup bias by indicating that the social sciences could not contribute as much as other scientific disciplines. In contrast, social scientists on average had an egalitarian judgment, recognizing that the contribution of the traditional sciences could be as important as that of the social sciences. This asymmetry in judgments thus most likely contributes to perpetuating the hierarchy between scientific disciplines.
27Social dominance theory suggests that this discrimination and asymmetry in behavior can be explained by the dissemination in the culture of a set of legitimizing myths that justify social practices. Racism, which justifies inequalities between Blacks and Whites, sexism, which justifies inequalities between men and women, or conservatism, which justifies inequalities between rich and poor, are examples of such myths.
28In contrast to traditional approaches to social stratification in sociology, social dominance theory considers that these myths do not serve to simply justify unequal practices. Rather, it suggests that there are two broad categories of myths: myths that enhance social hierarchy and inequality (HE myths), such as those previously mentioned, but also myths that conversely attenuate social hierarchy (HA myths) and justify equality. Almost all existing and imaginable ideologies can be classified as HE or HA using this framework. Multiculturalism is thus considered an HA myth that aims at achieving equality between cultures. Feminism is also considered an HA myth that aims at achieving equality between men and women.
29The notion of “myth” is used to mean that what matters in these belief systems is not whether they are legitimate or correspond to reality, but the fact that people believe them to be true. Human social systems are thus understood to be subject to the influence of HE forces that increase inequalities between groups and at the same time to HA forces that counterbalance them by reducing inequalities. The HE (Hierarchy-Enhancing) and HA (Hierarchy-Attenuating) distinction is also applied to social institutions and social roles. Sidanius and Pratto’s theory therefore recognizes the existence of institutionalized HE forces (the police, the secret service, banks) but also of HA institutions (social services, food banks, etc.).
30How can we explain why individuals embrace these legitimizing myths? Why do some individuals choose a career in an HE institution rather than an HA one? To answer these questions, the theory postulates that each individual has a general psychological orientation toward intergroup relations known as SDO (for Social Dominance Orientation) and that this new variable will explain why individuals endorse legitimation myths and choose certain social roles. This is a key original theoretical contribution of social dominance theory, one that has attracted the most attention in social psychology. SDO reflects a general attitude or preference for unequal and hierarchical intergroup relations, as opposed to a preference for egalitarian intergroup relations.
31A 16-item questionnaire measuring this orientation was presented and validated by Pratto et al. (1994). The scale was translated and validated in the french language by Duarte et al. (2004). By validated, we mean mainly that the scale, in English and in French, measures a psychological orientation that is distinct from other existing measures. Thus, SDO scores are strongly associated with left-right political orientation. In France, far-right sympathizers have the highest scores of all political parties on this SDO scale while the far left and the left obtain the lowest scores (Guimond et al. 2015). Despite the significant link between SDO and political orientation, SDO scores accurately predict prejudice toward stigmatized groups or individuals’ level of sexism regardless of their position on the left-right scale (Pratto et al. 2006, Roebroeck and Guimond 2018).
32A large body of research now confirms that across many cultures and contexts, SDO is one of the most powerful variables known to predict individuals’ racist or sexist attitudes (Altemeyer 1998, Pratto et al. 2006, Osborne et al. 2021). This includes experimental demonstrations supporting the causal role of SDO (Guimond et al. 2003). In the words of Van Bavel and Packer (2021), SDO represents “the psychological foundation for much of the racism that exists around the globe” (p. 154). There is also growing evidence suggesting that SDO does not correspond to a personality trait as Altemeyer (1998) believed but rather to ideological beliefs resulting from a socialization process (see Guimond 2022). However, few studies have examined the relationship between SDO and discriminatory behaviors (Sidanius et al. 2007).
33In social psychology, there are many techniques for measuring stereotyping and prejudice, including the Implicit Association Test (IAT), which will be discussed in depth in the last part of this article. However, apart from the resource distribution task discussed above, the discipline has few methods for measuring discriminatory behavior. This probably explains why there is much more research on stereotyping and prejudice than on discrimination.
34Michinov et al. (2005) proposed a new computerized method to at least partially deal with the lack of methods for measuring discrimination in social psychology. This method is based on social dominance theory, according to which discriminatory behaviors are intended to maintain and perpetuate a hierarchical system. Discrimination is not only a treatment that is unfair, but it is above all a treatment designed to keep members of stigmatized groups in inferior social positions and to protect the advantageous social positions of members of groups at the top of the hierarchy.
35Michinov et al.’s (2005) method was to ask people to participate in an experiment about “decision making and leadership in organizations”. Participants had to carry out different “tests” or tasks designed to study their “organizational behavior”. The discrimination measure was the “organizational chart” task. Participants were presented on a computer screen with six people, both women and men, who work in the organization, identified by their first names. The participants’ task was to place each of these people (and themselves) in an organizational chart with four hierarchical levels (see Figure 1). They were asked to drag the image of each person and place it on the chart. The computer recorded their choices and recorded their response times and the order in which targets were placed in the hierarchy (see Michinov et al. 2005). The highest level of the organizational chart showed only one box marked “group leader”, which indicated that it was the highest position in the hierarchy.
36Among the six individuals, four were named Alexandre, Pascal, Laetitia and Sandrine, and two had a distinctive Arab-Muslim first name: Abdel and Rachid. To measure origin-based discrimination, the position assigned to these last two individuals in the hierarchical structure was of particular interest.
Figure 1. Computer screen displayed to participants in Anier et al.’s (2018, 2019) experiments to present the discrimination measure task. The six employees work for a (fictional) retail company called “Pelletier et fils”.
37Since the participants had no information about these individuals other than their first name and thus their group affiliation, any systematic tendency to place Abdel and Rachid at the bottom of the organizational chart was a discriminatory behavior, i.e., the unfair treatment of an individual based on the assumption that this individual belongs to an outgroup. We therefore coded the position assigned to these two individuals in the organizational chart, from 1 (high position) to 4 (lowest position). The probability that a target was placed totally randomly in a position was calculated, and this theoretical probability was compared with the choice made by the participants (e.g., using a chi-square test). If there was no statistically significant difference between the theoretical probability and the actual choice, there was no discrimination. Abdel and Rachid were placed in different positions randomly. However, if there was a significant difference indicating that a choice was not random, and Abdel and Rachid were placed at the bottom of the pyramid, this revealed discrimination.
38In a first study using this task, Michinov et al. (2005) wanted to compare the participants’ behavior with their score on the SDO scale. The hypothesis was that people with high SDO scores, who are motivated to maintain and perpetuate the social hierarchy, would show more discrimination than people with low SDO scores. The results obtained with 81 French students of European origin confirmed this hypothesis. For participants with a low SDO (N = 42), the chi-square test was non-significant, suggesting that they randomly positioned Abdel and Rachid in the organizational chart. But for participants with high SDO scores (N = 39), the test was significant. For these participants, a lower-than-expected frequency was observed in positions 1 and 2, at the top of the pyramid, and conversely, a higher-than-expected frequency in positions 3 and 4, at the bottom of the pyramid structure. These results suggest that high-SDO participants did not respond randomly. Instead, they acted in a discriminatory manner toward targets perceived as North African.
39This study shows a relationship between social dominance orientation and behaviors that maintain and reproduce the social hierarchy. However, according to the theory presented earlier, individual discrimination can be explained by the belief in legitimizing myths, which in turn would be due to the influence of SDO. We therefore measured anti-North African prejudice, a legitimizing myth, and hypothesized that belief in this myth would mediate the influence of SDO on discrimination. This is exactly what was confirmed. Not only is SDO related to behaviors that perpetuate group social hierarchy, but prejudice, a variable on which research has focused for the past 30 years, is effectively involved in this relationship and explains why SDO is linked with discriminating behavior.
40These results and this postulated causal chain from SDO to prejudice and from prejudice to discrimination confirm an important part of Sidanius and Pratto’s (1999) theory. However, these results are tentative given their correlational nature and are based on a small sample of women only. Several limitations with respect to this method could be corrected in the future (artificial nature of the procedure, the lack of stakes, the different number of people belonging to a minority or majority group, etc.). This method lacks the external validity of data obtained using correspondence testing (Petit et al. 2013) or from observations of behavior in a field experiment (Aranguren et al. 2021). It can, however, be seen as a complementary way of testing hypotheses about the mechanism of discrimination, which is much more difficult to achieve in correspondence testing conditions or in field studies.
41For example, this organizational chart task has more recently been used to test hypotheses about contextual determinants of discrimination (see Anier et al. 2018, 2019). The results of these experiments will be discussed in the final section on institutional discrimination. These new studies also sought to improve the realism of the procedure by using photos of real people from a bank of stimuli developed for psychological experiments. Thus, apart from the first names that reinforce group membership, all the photographs were equivalent in terms of facial expression (see Figure 1). This improvement was intended to address the criticism of artificiality leveled at the original Michinov et al. (2005) experiment that used stylized drawings of six faces to represent the targets to be placed in the organizational chart (see Appendix of Michinov et al. 2005). It also explains the variation across studies in the gender of the minority targets due to whether suitable photographs were available.
42The results obtained in these new studies using photographs of real people’s faces support the idea that the organizational chart task can be a useful instrument for measuring origin-based discrimination. Unpublished work further shows that this task could also be used to measure gender discrimination (Guimond, Faure and Anier 2016), which can be measured by simply examining the position of women (Sandrine and Laetitia) in relation to men in the organizational chart. We observed that this position was significantly correlated with a measure of gender bias. Those who placed women at the bottom of the hierarchy were those who expressed the strongest sexist attitudes. Further research is needed to confirm the validity of this method by testing the consistency of the results with other types of discrimination measures. Moreover, this procedure concerns individual discriminatory behavior. Yet, it has been argued that discrimination and racism are not limited to individual attitudes or behaviors.
43The idea of institutional discrimination and controversies about the concept of “systemic racism” have recently gained traction in Europe and North America (Guimond 2022). In France, in 2020, the government commissioned an “Information Mission” to examine “the emergence and growth of various forms of racism and the responses to them”. The extensive report it produced with the help of recognized experts in sociology, history, and economics was submitted on March 9, 2021 (https://www.vie-publique.fr/rapport/279117-lemergence-et-levolution-des-differentes-formes-de-racisme). The report introduces the concept of institutional racism and thus recognizes to a certain extent the existence of this phenomenon, but it fails to mention research in social psychology, which is why it is worthwhile to briefly highlight its contributions here. At least two series of studies should be discussed: the evidence of institutional discrimination from the perspective of social dominance theory and recent research on the Implicit Association Test (IAT, Greenwald et al. 1998), which will lead us to observations concerning the contextual determinants of discrimination in France.
44Part III of the book by Sidanius and Pratto (1999) is the lengthiest. It has four chapters and is titled “The circle of oppression: the myriad expressions of institutional discrimination” (Sidanius and Pratto 1999 : vii). It begins with a definition: “By institutional discrimination we are talking about the way that social institutions such as schools, businesses, and government bureaucracies disproportionately allocate positive social value (e.g., high social status, good health care, good housing) to dominants and disproportionately allocate negative social value (low social status, poor housing…) to subordinates” (p. 127). This is followed in four chapters by quantitative evidence based on available research that confirms the existence of discrimination in housing (chapter 5), in the labor market (chapter 6), in education and health care (chapter 7) and in the dynamics of the criminal justice system (chapter 8) not only in the United States but in many Western countries including France and Canada. In each chapter, the studies are described in detail, starting with those that suggest the existence of discrimination without being able to demonstrate it formally (soft evidence) and progressing to those which, using more robust methods, leave little doubt (i.e., correspondence tests, see L’Horty and Petit, this issue). In each field, we find that the most rigorous studies confirm the existence of discrimination already mentioned in the initial studies. By their scope and scale, these findings confirm the existence of a hierarchical system of domination that affects practically all institutions, hence the term institutional discrimination.
45More recent work using modern statistical techniques seems to be in full agreement with Sidanius and Pratto (1999). For example, 20 years later, at the end of a meta-analysis of 97 studies using the correspondence testing method in 9 countries, Quillian et al. (2019) concluded, “In every country we consider, nonwhite applicants suffer significant disadvantage ... our results are consistent with perspectives such as social dominance theory (Sidanius and Pratto, 1999) that emphasize the pervasiveness of discrimination against nonwhites in Europe and North America” (p. 486).
46These empirical findings point to an aspect that distinguishes individual from institutional discrimination, namely the way in which discriminatory behavior is distributed across the population (see Galvan, Alvarez, Cipolli, Cooley, Muscatell and Payne 2022). If the actions of a small number of individuals explain the vast majority of discrimination cases, and if we therefore observe a distribution according to the Pareto principle, it would be logical, according to Galvan et al. (2022) to speak about individual discrimination. Discrimination would be due to a few individuals and most people would not display such behavior. If, on the contrary, we observe, as seems to be the case for Sidanius and Pratto (1999) and Quillian et al. (2019), that everyone can potentially discriminate and that discriminatory behavior is not limited to a few isolated individuals but seems to be widespread in the population and in all societies, this tends to support the idea of institutional discrimination. In this case, discrimination remains an “individual” behavior as defined at the beginning of this article, but it is observed with such frequency that it comes to have a “systemic” character that extends beyond the actions of a few individuals.
47Social dominance theory is thus the first of the major theories of social psychology to offer an analysis of ethnic or racial discrimination suggesting that the phenomenon is observed with such magnitude across Western democracies and across different domains of social life that it is justified to speak of “systemic discrimination” (Bourhis 2021), “structural racism” (Rucker and Richeson 2021), “institutional racism” (Dhume 2016, Jones 2002), “systemic racism” (Banaji et al. 2021), or “state racism” (Dhume et al. 2020). The precise limits of what defines these different concepts, from institutional discrimination to state racism, remain a matter of debate and no conclusion will be provided here. But as Primon and Simon (2018) suggest, we can assume that while not identical, “the notion of institutional racism is very similar to that of institutional discrimination” (p. 192).
48In both cases, the idea is that ethnic or racial discrimination is not simply due to a few particularly malicious or “racist” individuals but is rooted in institutions and social structures. Talking about institutional discrimination, Sidanius and Pratto suggest that the origin of certain discriminations can be found in the way institutions, and not simply individuals, operate. This is consistent with the definition of “institutional” or “systemic” racism found in the literature, which refers to institutional policies, laws, or practices that contribute, sometimes unintentionally, to favoring certain groups and penalizing others (Guimond 2022).
49However, beyond the wealth of information gathered by Sidanius and Pratto (1999), one may wonder to what extent they succeeded in demonstrating in their book the distinct origins of individual and institutional discrimination. The SDO variable plays an important role, as we have seen, in explaining individual prejudice and discrimination, but it can also play a role in institutional discrimination. Social dominance theory distinguishes HE social roles that maintain and perpetuate social hierarchy and inequality (e.g., lawyers for large corporations, police officers) from HA social roles (e.g., social workers, members of the human rights league). It posits that SDO scores predict career and social role choices and thus, that individuals with high SDO scores will turn to HE social roles involving more power and status and vice versa for individuals with low SDO scores. Many studies have confirmed this hypothesis (Pratto et al. 2006). The result, and this can serve to explain institutional discrimination, is that some institutions are staffed with people with particularly high SDO scores, while others gather and recruit people with low SDO scores.
50In one study, Sidanius et al. (1998) demonstrated that, consistent with theory, Los Angeles police officers differed from other groups (college students, citizens drawn for jury duty, public defenders) by having particularly high SDO scores, whereas Los Angeles area public defenders differed from other groups by having particularly low SDO scores. More recently, Swencionis et al. (2021) showed that there are systematic and statistically significant relationships between the SDO scores of White police officers in the United States and the use of physical force by those same officers (as measured from police authority administrative data). The more White police officers support hierarchy and inequality (high SDO scores), the more physical force they use in their interactions with citizens, which supports the assumptions of social dominance theory.
51These findings are important. They suggest that the SDO variable is not only associated with instances of ingroup favoritism but also with anti-outgroup discriminatory behavior (see definitions of the forms of discrimination at the beginning of this article). Sidanius and Pratto were sensitive to criticism leveled against social identity theory according to which the type of discrimination identified in experiments using the minimal group paradigm was ingroup favoritism, i.e., a preference for the ingroup, but no real hostility for the outgroup (Brewer, 1999). For this reason, social dominance theory was also used to attempt to explain real cases of oppression and persecution of minority groups and not only ingroup favoritism. The study by Swencionis et al. (2021) and other published research by Sidanius and Pratto clearly sought to achieve this.
52However, the emphasis on the role of SDO is not the most compelling way to explain institutional discrimination. To explain discrimination by a psychological orientation such as SDO does not question the way institutions operate, at least not directly. It may even support the position of those who deny the existence of institutional discrimination. The report of the French government’s information mission cited above provides examples of this denial that can be summarized by the testimony of Mario Stasi, President of the International League against Racism and Anti-Semitism (LICRA) on September 15, 2020: “Is there racism in the police? Yes! Is the police racist? No (…) Is there systemic racism, state racism? No, we do not live in apartheid. There are no racist laws or institutions in our Republic, but there are of course individual deviations and, in the police as elsewhere, racist individuals” (p. 46). This is a firm denial of the possibility of systemic racism in France.
53What social psychology teaches us does not seem to support this view (for a recent review of research, see Banaji, Fiske and Massey 2021). Perhaps anticipating such reactions, Sidanius and Pratto (1999), setting aside the role of SDO, assembled a body of evidence that leaves little doubt about how criminal justice systems in democratic countries treat outgroup members at all stages of the legal process, and more particularly, about the systemic nature of police violence against ethnic and racial minorities.
54These police actions are now often documented by photos and videos published on social media. Should we consider such events “individual deviances”, “system failures”, or isolated cases? A quick and superficial analysis of the situation might lead us to think so. A closer look at the work of researchers in the field offers a different perspective. For Sidanius and Pratto (1999), who studied this issue long before the era of social media and the long list of cases everyone has witnessed recently, these events illustrate how a system of domination routinely operates. The authors of social dominance theory have identified five general principles (called the laws of law) that characterize how institutional discrimination occurs in the criminal justice system, and they show that research supports these principles.
55One of these principles, the fourth law of law, states for example: “The degree of negative sanctions against security forces for abuses of power will tend to be exceedingly small, especially in cases of abuse against subordinates” (p. 218). This principle is particularly enlightening because, if there is no institutional discrimination but only isolated instances of individual discrimination by a few police officers who are particularly “racist,” then it should be expected that these individuals would be sanctioned. The fact that these individuals are not punished may therefore be important evidence against the idea that these are simply cases of individual discrimination.
56The analyses and the evidence gathered on this subject by Sidanius and Pratto in 1999 point toward institutional discrimination. Police officers guilty of misconduct are not only rarely sanctioned, but they are even sometimes promoted. For example, Sidanius et al. (1998) examined the report of the Warren Christopher Commission in the United States that was set up in 1991 following the violence committed against Rodney King by several Los Angeles police officers. The report analyzes all allegations (3,419 cases) of excessive use of physical force by members of the Los Angeles Police Department (LAPD) between January 1986 and December 1990. The vast majority of these allegations (96.8%) were not sustained. According to Sidanius et al. (1998): “Of those complaints found to have merit, only 7.6% of them resulted in the removal of the police officer involved. Putting all of this information together, we see that the probability of having the police officer involved removed from the LAPD for complaints of excessive force or improper tactics is 0.00228, or essentially zero” (p. 153).
57Sidanius and Pratto (1999) cite the work of other researchers in the United States and in various countries showing that racial prejudice was significantly and positively related with the officers’ performance evaluation by their supervisors, and thus, that brutality against minorities “often results in institutional rewards” (p. 219). As the researchers note, there are few studies on these topics, but those that have been conducted are consistent with the fourth principle of social dominance theory. A few surveys in France also support social dominance theory by implicating not only individual police officers but also the social context in which these officers work (see Gatto and Dambrun 2012).
58In addition to studies related to social dominance theory, other developments in social psychology confirm the importance of its contribution to the study of the institutional aspects of discrimination. Along with the decline of the concept of biological racism (also known as old-fashioned racism) based on the belief in the existence of inferior races, and with the emergence of new, more subtle forms of racism, social psychology has contributed to the development of a whole series of new methods inspired by semantic priming paradigms in cognitive psychology that allow for “implicit” measures of stereotyping and prejudice (for a review of early developments in this regard, see Dambrun and Guimond 2003). In reference to these new methods of measurement, the concept of “implicit bias” has progressively gained acceptance and has spread very widely outside the circle of social psychology and scientific research, with certain advantages but also several disadvantages.
59As the general population is becoming aware of this concept and of the idea that we might all be “racist” without being aware of it, more and more researchers working on the subject are alerting us to certain assumptions that have proven to be wrong. The debate is not over, but it is now clear that the concept of “implicit bias” does not cover all the meanings ascribed to it. Thus, in an undoubtedly imperfect effort to clarify the issue, we will refer to measures of “unintended bias” rather than “implicit bias.”
60In this regard, Corneille and Hütter (2020) have shown that the meaning of the concept “implicit” in psychology generates confusion because the same word is used with very different meanings without anyone noticing. These new methods therefore raise conceptual as well as empirical problems. We now know, after 20 years of research, that the most influential of these new measures, the Implicit Association Test (IAT, Greenwald et al. 1998), has neither the acceptable temporal stability nor the expected validity to predict behavior (Kurdi and Banaji 2017, Payne and Hannay 2021). The IAT can be adapted to measure attitudes toward various groups. The test is open to anyone (including in French) on the “Project Implicit” website hosted by Harvard University. In what follows, we will primarily discuss the IAT that purports to measure attitudes toward Whites versus attitudes toward Blacks, and our comments do not necessarily apply to other types of implicit measures that have been developed (but which, to date, have not been used as frequently as the IAT).
61From the beginning, the main argument for using the IAT was that because responses to this test are by construction practically uncontrollable, they solve the social desirability problems of questionnaire-based measures of racist prejudice that allow racist individuals to report themselves as non-racist. A person may effectively decide, when answering a questionnaire, to give a socially acceptable answer even if this answer does not correspond to that person’s true attitude. IAT-type measures are supposed to allow access to “real” individual attitudes without control from the person taking the test and should therefore be better at predicting behavior. At least that is the way the IAT has been presented.
62Yet, according to Oswald et al.’s (2013) meta-analysis of studies examining the relationship between the IAT and measures of racial discriminatory behavior, this is not the case. Overall, the IAT is not a better predictor of discriminatory behavior than measures obtained from questionnaires. Thus, this new measurement method does not appear to be the ideal instrument to study individual discrimination (see Meissner et al. 2019). On the contrary, as Gawronski (2019) points out, given that a person’s “implicit biases” are highly unstable over time (even though they were initially assumed to be extremely stable and difficult to change), while explicit biases are quite stable, the implications are that explicit survey measures provide a better basis for predicting future behavior. However, the situation was reversed thanks to the work of Payne and colleagues (see Payne and Hannay 2021) who found that a novel way of studying institutional discrimination, not individual discrimination, was to measure unintended bias at a collective level, rather than at the individual level.
63First, in terms of the measure’s reliability and validity, Payne and Hannay (2021) show that by using IAT scores at the collective level, e.g., the IAT scores of the population of a city, a region, or a state, etc., rather than individual scores, they could obtain good temporal stability of the measure and improve its power to predict behavior (see, for example, Zerhouni, Rougier and Muller 2016). This led Payne and Hannay (2021) to propose a new interpretation of the very nature of what is called “implicit bias.” While most theories assume that these new measures such as the IAT provide access to an individual characteristic (an attitude, belief, or trait), Payne and Hannay (2021) suggest instead that they measure a characteristic of the environment or social context in which individuals find themselves. It appears that collective racist biases are stable as long as the environment is stable, and that a change in the social context could change them.
64How might this new perspective be of interest for the study of institutional discrimination or what Payne and Hannay (2021) call “systemic racism”? First, the collective IAT should be viewed not as a direct measure of “systemic racism” or “institutional discrimination” but as a tool for studying its manifestations. For some, like Rucker and Richeson (2021), it can be seen as a consequence of systemic racism, but this is only a hypothesis at this point.
65One of the main differences between individual and institutional discrimination noted above is the conceptualization of their origin. Individual discrimination is explained by the psychological characteristics of individuals, whereas institutional discrimination is explained by the characteristics of institutions or the social context in which individuals find themselves. To the extent that the social or institutional context leads people to be biased and to behave in ways that adversely affect members of a group, even without intending to do so, this is called institutional discrimination (rather than individual discrimination).
66Therefore, two main predictions can be made to confirm the existence of institutional discrimination using a measure such as the collective IAT: 1) collective IAT scores should be higher and demonstrate stronger unintended biases in certain institutional contexts or types of environments; and 2) the existence of this higher collective bias in certain contexts should predict the occurrence of discriminatory behaviors, regardless of individuals’ personal predispositions to exhibit such behaviors. Current research clearly confirms these two predictions.
67With respect to the former, Payne et al. (2019) examined the collective racist bias of the U.S. population based on the geographic location of their residence. In this type of research, the open access data came from the Project Implicit website, which provided the implicit and explicit responses of millions of participants. Payne et al. (2019) observed that White residents of U.S. states who in 1860 were highly dependent on slavery as a labor force exhibited a greater collective pro-White bias more than 150 years later than residents of U.S. states less involved in slavery. Therefore, depending on the social environment (living in a certain neighborhood, city or region), variations could be observed in the collective IAT scores. According to Banaji et al. (2021), these results suggest that “Americans who live in regions with greater historical legacies of slavery must be acquiring the particles of race bias embedded in the social atmosphere”. Moreover, this illustrates the capacity of social psychological methods to empirically test hypotheses that connect historical events or contexts with contemporary unintended bias measured on millions of individuals, which traditional methods used in history were unable to produce. Specifically, Payne and Hannay (2021) provide new empirical support for hypotheses derived from work in history suggesting that slavery in the United States has left traces to the present day and that this period of racist laws and racial inequality must be considered when explaining racism in the United States today. The suggestion that there are discriminatory behaviors in contemporary societies that find their origin in bygone historical eras has often been associated with the idea of institutional discrimination (Van Bavel and Packer 2021 : 166).
68As for the second prediction, the one that is arguably most important to establish the existence of institutional discrimination, recent studies conducted in separate laboratories have also clearly confirmed it (see Banaji et al. 2021, Bourhis 2021, Payne and Hannay 2021). Hehman et al. (2018), for example, show that the disproportionate use of lethal force by police officers against Blacks occurs in areas of the United States where the collective IAT indicates a stronger pro-White racial bias. In other words, this research establishes a significant link between the discriminatory behavior of police officers in the U.S. (instances of anti-outgroup discrimination) and the collective IAT.
69While most theories originally have assumed that individuals who showed strong individual biases (as evidenced by the individual IAT) would discriminate more than individuals with weak biases, which would be consistent with individual discrimination, recent research shows that the collective IAT, rather than the individual IAT explains in part the variation in discriminatory behavior. As Hehman et al. (2018) note, these findings suggest that contexts or environments characterized by the presence of unintended racist bias may be the source of police behaviors that unduly penalize the Black population, regardless of whether individual police officers’ attitudes are more or less “racist.” This is one of the clearest demonstrations of institutional discrimination: discriminatory police behavior is predicted by a basic characteristic of the institutional context, and not by a psychological characteristic of the officers. An independent study by Stelter et al. (2022) focused on a different police behavior considered by research as discriminatory (see Van Bavel and Packer 2021), namely, the fact that Black drivers in the United States are disproportionately stopped compared to Whites. The results confirm the phenomenon and highlight the role of the collective IAT. Stelter et al. (2022) conclude, “researchers analyzing racial disparities in policing should not only focus on individual officers as individuals but also consider the contexts in which police operate” (p. 493). These data, which concern American society and were obtained by psychologists, empirically support the existence of institutional discrimination and counter explanations according to which the phenomenon only concerns a few “racist” police officers.
70Once this is established, we must acknowledge that each of these studies has certain limitations and that several interpretations of the results of a correlational nature are possible. For example, when interpreting their own results, Hehman et al. (2018) also point out that Blacks in certain areas may be more violent when interacting with the police, which can justify the more frequent use of force by police, and which in turn may influence prejudice toward Blacks of people living in the area (Hehman et al. 2018).
71A suitable interpretation of the results, however, must be consistent with all the relevant studies that contribute various elements of validity (see Banaji et al. 2021, Bourhis 2021, Guimond 2022, Payne and Hannay 2021). From this point of view, the interpretation that seems to be in line with current knowledge is summarized by Bourhis (2021): “All the members of these mostly White communities in each district share... this implicit prejudice against Blacks, which creates an intergroup social climate that legitimizes and makes it acceptable for police officers to use extreme force against Black minorities” (p. 405). Such an interpretation of intergroup behavior has been examined in detail in France, including in experimental laboratory studies, as part of a research program designed to understand how state’s national policies can influence citizens’ prejudice and discriminatory behavior.
72There have not been any studies in France using the collective IAT equivalent to those in the U.S. However, research conducted in social psychology by different means has shown results that are theoretically consistent with those observed in the United States and confirm an interpretation of the results obtained with the collective IAT in line with the one proposed by Bourhis (2021). The results that link the collective IAT of an aggregate with individual behavior can be related to the role of social and cultural norms and to the process of normative influence long studied in social psychology and sociology (Guimond 2010). It can be hypothesized that the collective IAT measured for a city or region gives an indication of the system of norms in that city or region, i.e., the “social atmosphere” mentioned by Banaji et al. (2021) or the “intergroup social climate” described by Bourhis (2021). Individual behavior, including discriminatory behavior, is regulated by social norms that define what is appropriate or legitimate conduct (see Anier et al. 2019). This is the conclusion of a decade of research conducted in various countries, including France, aimed at explaining the influence of national immigration and integration policies on intergroup attitudes and behaviors (see Guimond 2019 for a synthesis of these studies including descriptions of longitudinal and experimental studies). These studies are directly relevant to the topic here, as the very definitions of the concepts of “institutional discrimination” or “systemic racism” refer to the influence of institutional policies, laws, or practices. Thus, Deaux et al. (1993) note that: “Racism is not just a matter of individual prejudices or attitudes. At a national level, the laws and practices of a society can enforce and perpetuate discrimination against one group by another” (p. 359). Can the laws and practices of a society “perpetuate discrimination”?
73This question has been asked before (see Berry, Kalin and Taylor 1977). But it is only in recent years that evidence has confirmed it: inclusive and egalitarian national policies are associated with a decrease in hostility toward ethnic or religious minorities (Guimond 2019, for a recent example, see Kende et al. 2022). Moreover, the impact of these policies on the national population can be explained by the fact that the laws defining these policies have created and strengthened specific social norms, independently of individuals’ personal predispositions (cf. Guimond 2019, 2022, Schlueter et al. 2013). For example, with regards to the multiculturalism policy that was introduced in Canada more than 50 years ago, we have shown that this policy is linked to a norm that is favorable to cultural diversity, regardless of personal attitudes toward multiculturalism, a norm that does not exist in countries such as Germany or France that have never had a national multiculturalism policy (Koopmans et al. 2005, Guimond et al. 2013; 2014). Furthermore, results have confirmed that this diversity norm helps explain why anti-Muslim prejudice is lower in Canada compared to other countries. Thus, the normative context favorable to cultural diversity (theoretically stemming from a national multiculturalism policy) is linked to a decrease in negative attitudes toward outgroups.
Figure 2. Perceptions de la norme relative au multiculturalisme, selon le pays (notes moyennes) (1 = pas du tout d’accord; 7 = tout à fait d’accord)
74Figure 2 shows the mean scores for the perception of the multiculturalism norm in five countries (see Anier et al. 2019, Study 1). Participants were university students who answered a series of questions regarding what most people in their respective countries think (e.g., “Most people think that ethnic and racial minorities living in [France/Germany/Canada/…] should be helped and their cultural heritage preserved,” for more details regarding the reliability and validity of such measures, see de la Sablonnière et al. 2020, Guimond et al. 2015). Responses were expressed on Likert-type scales ranging from 1 (strongly disagree) to 7 (strongly agree). The higher the mean for these responses, the more participants believe this norm is important in their country, with a maximum possible score of 7. In the case of multiculturalism, it was confirmed that the perception of a norm favorable to cultural diversity could be found in Canada, i.e., the country that has implemented a national multiculturalism policy for several decades. In contrast, in France, the perceived multiculturalism norm was low, with a mean score of 3.14 below the midpoint of the response scale.
75Does norm perception necessarily reflect the individual attitudes of the group members? Not at all. First, the results in Figure 2 cannot be obtained when measuring personal attitudes, i.e., what participants themselves think about multiculturalism, because French and German students are personally very supportive of multiculturalism and almost as supportive as Canadian students. Second, a survey of a representative sample of the adult population in France (N = 1001) showed a particularly clear disconnect between perceived norm and individual attitudes (see Guimond et al. 2015). It confirms that the multiculturalism norm is perceived as low in France, in line with the results of Figure 2, and shows that this perception is very widely shared within the population, regardless of gender, age or even political orientation of the participants, in agreement with the idea that a norm is a socially shared belief. However, the survey also shows that when we examine personal attitudes toward multiculturalism in France, we obtain completely different results: personal attitudes favorable to multiculturalism characterize most people in France. In other words, what is perceived as the majority norm in France, i.e., to be rather opposed to multiculturalism, is in fact, at the level of personal attitudes, a minority position (only the sympathizers of the extreme right claim to personally oppose multiculturalism). We are thus faced with a well-known phenomenon of pluralistic ignorance (Guimond et al. 2015). The point here is that the perceived norm indicates a collective property that does not necessarily reflect individual attitudes. These results, obtained with a direct measure of the norm, are consistent with the theoretical position of researchers who used the collective IAT: “The region-level effect may be quite different than the individual-level effect” point out Hehman et al. (2018: 399).
76The theoretical basis of the concept of norm is that it is an emergent property of the social group that is not equivalent to a simple addition of individual positions (Guimond 2019, Sherif 1936). Similarly, the collective IAT measure reveals regional intergroup biases that are not a random aggregate of individual intergroup biases but that reflect the influence of certain contextual characteristics (Hehman et al. 2019). This is the theoretical position developed by Payne and Hannay (2021), and it represents a major advance in research using the IAT. Just as Figure 2 shows a strong relationship between living in a country and perception of the multicultural norm (regardless of individual attitudes), Payne and colleagues show that the collective IAT (in contrast to the individual IAT) is strongly related to various relevant properties of context (living in an area with more or less racial inequality, studying on a campus with low or high faculty diversity, etc.).
77Research have also been conducted on the influence of republican universalism and the principle of laïcité that characterizes French policy (Guimond 2019, Koopmans et al. 2005). In France, no laws dealing with multiculturalism have ever been passed, however several laws directly or indirectly concern laïcité. The theoretical model described previously with respect to multiculturalism was therefore applied to understand the situation in France: national policies and laws passed in France create or reinforce social norms, and these norms appear to influence the attitudes and intergroup behaviors of French citizens. In the case of Canadian multiculturalism, correlations have been established between country affiliation, perceived norm of multiculturalism and prejudice. In the case of French policy, similar correlational studies have been supplemented by experimental laboratory studies assessing the effect of “republican” norms on prejudice and discriminatory behavior. Moreover, in France, the situation is somewhat more complex than in other countries because the principle of laïcité is divided in the minds of the French population into two distinct orientations, one stemming from the 1905 law (historical laïcité) and the other, often positively correlated but distinct, stems from the laws of 2004 and 2011 (the new prohibitive laïcité, Guimond 2019, Lorcerie and Guimond 2021, Roebroeck and Guimond 2016). It must be noted that France is the first country in the world, along with Belgium, to have passed laws banning religious symbols in the public space, first with the 2004 law banning students in public schools from wearing conspicuous religious symbols, and the 2011 law commonly known as the burqa law. In line with Baubérot’s (2012) historical sociology analyses, our research suggests that these laws introduced a divide and contributed to this distinction between two views of laïcité associated with two distinct norms (for philosophical and historical aspects, see also Samuels 2016, Sptiz 2022). More importantly, these two laïcités have diametrically opposed implications for intergroup relations: the historical laïcité focused on equality and freedom is strongly associated with a decrease in prejudice and discrimination, while the new prohibitive laïcité is linked to an increase in prejudice and discrimination against ethno-religious minorities (Anier et al. 2019, de la Sablonnière et al. 2020, Guimond 2019, Nugier et al. 2020, Roebroeck and Guimond 2016, 2018).
78Figure 3 presents some of the results with respect to the perception of the new prohibitive laïcité norm in France and four other countries (see Anier et al. 2019, Study 1). The new laïcité norm was measured by two items rated by participants on a scale of 1 to 7: “In [France/Germany/Canada/etc.], most people consider that religious practices should be private and not public,” and “people in [France/Germany…] think it is normal to ban visible religious symbols in public schools”. These items were selected based on research conducted since 2008 that confirms that they are good indicators of this dimension of laïcité (Kamiejski et al. 2012).
Figure 3. Perceptions de la norme relative à la nouvelle laïcité, selon le pays (notes moyennes) (1 = pas du tout d’accord; 7 = tout à fait d’accord)
79In Figure 3, we see that France stands out from all the other countries with a new laïcité norm perceived as more widespread (the mean score obtained in France is significantly higher than the mean obtained in any other country). This should not be too surprising because France is the only country with laws banning religious symbols in public settings. However, subsequent research has shown that this new laïcité norm represents a normative context, a “social atmosphere”, or “intergroup social climate” that can directly affect prejudice and discrimination against ethno-religious minorities. For example, when participants in an experiment conducted at the Clermont-Ferrand laboratory were randomly divided into either a control condition or an experimental condition, which differed only in making the new prohibitive laïcité norm salient, more discrimination was observed in the experimental condition as measured using the organizational chart task described earlier (Anier et al. 2019, Study 2a, see also Anier et al. 2018). These findings corroborate a long list of studies that measure prejudice rather than discriminatory behavior (de la Sablonnière et al. 2020, Guimond 2019). Note that there is no indication that ethnic, racial or religious intolerance are embedded in laïcité. On the contrary, when the historical laïcité norm focusing on equality and freedom of conscience was made salient, under certain conditions there was a significant decrease in discriminatory behavior (Anier et al. 2019, Study 3), which is consistent with a long line of studies measuring prejudice. To sum it up, the data presented in Figures 2 and 3 represent characteristics of the social context in different countries that, under certain conditions, orient intergroup attitudes and behaviors.
80Discrimination is a problem in all societies, but it varies significantly in magnitude from one society to another (Quillian and Midtboen 2021, Thijssen et al. 2022). This suggests that an explanation should identify the general causes that concern all societies as well as the factors specific to certain societies that explain variations from one country to another (cf. Guimond et al. 2013). This presentation of the intergroup approach in social psychology shows that social dominance orientation (SDO) is a general factor that plays a role in prejudice and discrimination in most if not all societies, and that social norms (possibly derived from certain laws) that are specific to each society are factors that can account for variations from country to country. Recent research on regional intergroup bias in the United States suggests that beyond individual intergroup biases, it is possible to provide evidence of institutional or systemic aspects of discrimination and to predict, for example, disproportionate police violence against ethnic or racial minorities. Whether research was conducted in France or in the United States, there is growing evidence that the problem of prejudice and discrimination is not limited to the deviant behavior of a few individuals. These findings have fundamental implications for the development of strategies to reduce discrimination. They suggest that the context (and its norms) must be changed through powerful laws and policies that change individual behavior, and not the other way around.