Special Thematic Section on " The Social Psychology of Citizenship , Participation and Social Exclusion " Schooling , Citizen-Making , and Anti-Immigrant Prejudice in France

Are schools an effective institution to build citizenship and to transmit values associated with a given citizenship regime? A survey of 300 middle and high school pupils showed that for pupils, the representation of the French citizenship model is structured in two dimensions, ‘republican citizenship’ (or colorblind equality) and ‘new laïcité’ (or secularism), replicating previous research among adults. Moreover, the results support the schools’ effectiveness in the transmission of republican values by showing that in the mainstream track, older high school pupils endorse more strongly than younger school pupils both the principle of republican citizenship and new laïcité. The fact that this is not the case for pupils in a professional track suggests that these results are not simply a question of age but of schooling. Finally, support is found for a theoretical model suggesting that these two principles of the French citizenship model mediate the effect of schooling on prejudice. The implications of these results for current theories of intergroup relations are discussed.

Several studies suggested that the republican model was highly salient within the school system in France (Legendre, 2002;Massignon, 2011;Perroton, 2000;Raveaud, 2003).According to Raveaud (2003), the French pupil "is introduced to a model of citizenship where citizens are engaged in the public sphere only as a public person -a person who has neither age nor sex, nor religion, nor skin color, a universal person" (p.26, our translation).
These features appear strongly related to the dimension of republican citizenship identified by Kamiejski et al. (2012).Concerning the dimension of laïcité, Massignon (2011) reported the results of a survey of French pupils aged from 14 to 16 years old and concluded that "A majority of the pupils had internalised the rules of laïcité at school, notably the distinction introduced by the 2004 law between visible religious symbols in the school setting, which were forbidden, and discreet religious symbols, which were permitted" (p.171).
These studies support the idea that schools play an important role in the transmission of republican citizenship (Raveaud, 2003) and new laïcité (Massignon, 2011).However, they do not provide any direct evidence that this is actually the case.For example, Massignon (2011) tested a single age group and did not examine the evolution of attitudes across the curriculum.Furthermore, the current capacity of French schools to transmit values has been regularly questioned in these last years (see Charuau, 2011) given their resource constraints.So far, few empirical studies examined the role of the school in the transmission of republican principles.The main objective of the present study was to scrutinize this issue.If republican principles are transmitted by the school, then one can predict that pupils will be more likely to support these principles as they progress through the school curriculum.
Yet this would not necessarily imply that the school is responsible for the change.A control group of young boys and girls who live in France but do not go to school would be necessary to reach such a conclusion.Because all children in France are required by law to go to school such a control group is not a realistic possibility.However, the effect of schools on the endorsement of republican principles may be more pronounced for some pupils than others, depending on the type of program that they attend.To rule out the effect of age or maturation, the present study was designed to consider the role of two potential moderators of the effect of school on support for republican principles: the school type or program and identification with the school.Indeed, all of the scientific literature on the phenomena of socialization led us to believe that these two variables may play a significant role (Baudelot, Leclercq, Chatard, Gobille, & Satchkova, 2005).

Type of Schools
Several studies showed that depending on the academic discipline, schools can have very different effects on prejudice and socio-political orientations (Baudelot et al., 2005;Collard-Bovy & Galand, 2003;Dambrun, Kamiejski, Haddadi, & Duarte, 2009;Guimond, 2010;Guimond & Dambrun, 2003;Guimond & Palmer, 1990;Guimond, Palmer, & Begin, 1989;Ma-Kellams, Ruiz, Lee, & Madu, 2014;Sidanius, Pratto, Martin, & Stallworth, 1991).These studies were conducted at the university level where there is indeed a strong differentiation between academic disciplines.But this differentiation may appear much earlier in the school curriculum.In France, pupils are educated in a common program during four years (from 11 to about 14 years old), and then, they can choose between either a mainstream or a professional track of study for three subsequent years.The mainstream track corresponds to regular high school -'lycée d'enseignement général et technologique ' (general and technical high school).This track allows pupils to pursue higher education upon completion.The professional track -'lycée professionnel'corresponds to vocational schools and is intended to train pupils for specific occupations.In France, vocational schools are devalued (Jellab, 2009;Mauger, 1998;Moreau, 2008) in comparison to mainstream schools.A study conducted in 17 countries (van de Werfhorst, 2007) showed that educational track (vocational or general) in particular had an impact on the practices of active citizenship (i.e., political interest, participation in voluntary organ-izations, etc.): "Vocational tracks are designed to prepare pupils for work through the provision of work-relevant skills, leaving less room for citizenship classes" (p.12).Citizenship issues may be less of a concern for people educated in vocational programs compared to those educated in general programs.Thus, we hypothesize that in France, the two principles of the republican model will be transmitted less effectively in professional track schools than in mainstream track schools.

The Role of School Identification
Several studies highlighted the key role of school identification in the processes of socialization and learning (Bliuc, Ellis, Goodyear, & Muntele Hendres, 2011;Guimond, 2000Guimond, , 2010)).The research focusing on the role of school identification is based on social identity theory (Tajfel & Turner, 1979) and self-categorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987).According to self-categorization theory (Turner et al., 1987), individuals who identify more with their group will adopt the standards, beliefs and behaviors of the group to a greater extent than those who identify less.Many studies showed that individuals who are well-integrated within a new community and identify strongly with it are more likely to adopt its norms (see Bizumic, Reynolds, Turner, Bromhead, & Subasic, 2009;Guimond, 2000;Newcomb, 1943).Thus, we predict that the transmission of republican principles will be greater among the pupils who identify more with their school.

Implications for Prejudice and Intergroup Relations
Do French schools that adopt the republican model of citizenship -a model that does not recognize any group membership whatsoever -have a positive or a negative effect on anti-immigrant prejudice?Based upon the discussion above, we propose to test the model represented in Figure 1, whereby the effects of French schools on intergroup attitudes are mediated by the transmission of republican principles.No existing study has, to our knowledge, ever tested such a model.However, evidence exists to underpin it.
In the present study, we considered whether support for republican principles is related to anti-immigrant prejudice.
Based on the work of Kamiejski et al. (2012), we can expect, as suggested in Figure 1, that the endorsement of republican principles will predict anti-immigrant prejudice.More precisely, support for the egalitarian principle of republican citizenship should reduce anti-immigrant prejudice whereas support for the new laïcité should increase it.
A wide range of studies also showed that intergroup prejudice is strongly related with obtained education level among adults (Baudelot et al., 2005;Carvacho et al., 2013;Mayer, Michelat, & Tiberj, 2008;Vogt, 1997).The effect of education on prejudice is often stronger than the effects of age or gender and remains even when occupational status (Hernes & Knudsen, 1992;Wagner & Zick, 1995) and parental attitudes (Hello, Scheepers, Vermulst, & Gerris, 2004) are taken into account.Note that most of these studies investigated the effect of education level in itself (the amount of education), without necessarily looking at the education type (academic discipline, type of school) that may also play a role in the evolution of prejudice (Baudelot et al., 2005).
According to the model of cultural socialization proposed by Selznick and Steinberg (1969), the effect of formal education on prejudice can be explained by the transmission of official culture, constituted through ideal norms and values of a given society.Consistent with this explanation, education was found to be more effective in reducing prejudice in democratic countries than in countries in which democracy was less fully established (Hello, Scheepers, & Gijsberts, 2002;Weil, 1985).That is to say, because the societal norms and values in democratic systems differ from those in less democratic systems, schools are likely to transmit different principles across these and therefore would impact upon intergroup attitudes differentially.Our model (see Figure 1) is thus quite compatible with the thesis of Selznick and Steinberg (1969).We propose to test this model, examining the mediating role of support for the republican principles in the explanation of the effect of schooling on prejudice.

Hypotheses
In sum, this study considered four main hypotheses: (1) Support for the principles of republican citizenship and new laïcité will become stronger as pupils progress through the mainstream career track (in general and technical high school) but not necessarily for those in the professional career track (in vocational school).
(2) This effect of schooling on support for republican principles will be moderated by school identification.
Thus, the effect of school (middle school and vocational school versus general and technical high school) on support for republican principles will be greater among pupils who identify strongly with their school than among weak identifiers.
(3) Support for new laïcité will be positively related while support for republican citizenship will be negatively related to anti-immigrant prejudice.
(4) The effect of schooling on prejudice will be mediated by the transmission of republican principles of citizenship.

Method Participants
In total, 358 pupils participated in this survey.Data from 58 participants were excluded from the analysis either because they were tertiary pupils (n = 5); because they were undertaking a specific curriculum not considered by this study (e.g., specialized school, n = 22); or because they did not indicate in a correct or logical manner their type of school and education level (n = 31).The final sample consisted of 300 high and middle school pupils from more than 20 different schools in the region of Clermont-Ferrand.They were divided into three main groups for the analysis as a function of their grade: 96 middle school pupils (6th-9th grade; 43 girls, 51 boys, 2 unspecified), aged from 12 to 16 years (M = 14.04,SD = 0.87); 160 high school pupils in a general and technical high school (10th-12th grade; 103 girls, 57 boys), aged from 13 to 19 years (M = 16.46,SD = 1.08); and 44 high school pupils in a vocational school (10th-12th grade; 32 girls, 12 boys), aged from 15 to 18 years (M = 16.95,SD = 0.94).

Procedure
The survey was conducted between November 2012 and January 2013.Participants were recruited in Clermont-Ferrand, a mid-size city in the center of France, during orientation and information forums which were designed to introduce pupils to different training/study programs and fields of activity.These forums were open to all.Pupils attended these forums individually, accompanied by their parents, or in class.A total of four experimenters, all students in psychology, invited them to complete a brief questionnaire on site for about five minutes.Tables and pens were made available to participants.The survey was presented as a study of social issues.It was stated that the questionnaire was anonymous and that there were no right or wrong answers.The items were always presented in the same order.At the end of the study, participants were debriefed and thanked.

Measures
Participants were asked to indicate their level of agreement on a five-point scale (1 = strongly disagree, 5 = strongly agree) for the following measures: School identification.Identification with the school was measured with a five-item scale adapted from Galand's (2001) scale of the sense of belonging at school (see Appendix A for item wording).Two items were recoded to control response styles.A principal component factor analysis showed that one of the items did not load on the same factor as the other ("I have many friends in my school.").This item was thus removed, increasing the Cronbach's alpha from α = .66(five items) to α = .71(four items).
Support for new laïcité.Endorsement of the principle of new laïcité was measured with a four-item scale taken from Kamiejski et al. (2012).The Cronbach's alpha coefficient of this scale denoted a moderate to weak internal reliability (α = .63).All items are displayed in French (as used here) and translated in English in Table 1.
Support for republican citizenship.Endorsement of the principle of republican citizenship was measured with a five-item scale (α = .84)adapted from Kamiejski et al. (2012; see Table 1 for item wording).
Prejudice against immigrants.Prejudice was measured with a four-item scale adapted from Dambrun and Guimond's (2001) scale of generalized prejudice (see Appendix B for item wording).Two items were positively and two were negatively worded to control response styles, α = .72.
Education level and school type.Participants were asked to check the box corresponding to their current level of education among several answers as well as their type of school (middle school; general and technical high school; vocational school).Since the distribution by education level was quite unbalanced, only school type was considered in the analyses.

Support for the Republican Model: Two-Dimensional Structure
We performed a principal components analysis with Varimax rotation on the nine items purporting to measure the level of support for the republican principles.The extraction was based on the eigenvalue greater than 1 rule.The analysis confirmed the existence of two dimensions corresponding to two distinct principles: new laïcité and republican citizenship (see Table 1).
Indeed, the five items concerning republican citizenship loaded substantively on the first factor.All these items were related one way or another to a colorblind ideology, the idea that in order to treat everybody equally, one should be blind to cultural, ethnic or religious differences.The four other items loaded substantively on the second factor.Their content was related to the principles of new laïcité (i.e., separation of church and state, ban on religious symbols in public schools, as related to the law of 2004).These two dimensions were only slightly correlated, r(300) = .16,p = .004.The first and the second dimensions accounted for 36.56% and 20.07% of the total variance, respectively.

Schooling and Support for Republican Principles
We conducted two regression analyses to examine the impact of school type and school identification on support for the principle of republican citizenship, on the one hand, and support for new laïcité on the other hand.For these regression analyses, all statistical assumptions were met (independence, normality and constant variance of residuals, no multicollinearity of the independent variables) except in the case of the measure of republican citizenship which showed a problem of heteroskedasticity.Our statistical procedure was designed to handle this problem.
Based on the idea of a transmission of the republican model within French schools, our central prediction was that support for the republican principles would vary according to the type of school (middle school, general and technical high school, vocational school) and become stronger among older pupils in the mainstream track (general and technical high school).Specifically, we expected that pupils in general and technical high schools would express more support for new laïcité and republican citizenship compared to both younger pupils in middle schools and pupils of approximately the same age in vocational schools.The latter group was not expected to express more support for these principles than middle school pupils.To test this hypothesis, we created two orthogonal contrasts: the first contrast (C1) compared pupils in general and technical high school with pupils in middle school and vocational school: 'general and technical high school' (+1) versus 'middle school' (-0.5) and 'vocational school' (-0.5).The second contrast (C2) compared the latter two groups together: 'vocational school' (+0.5) versus 'middle school' (-0.5), while the group in 'general and technical high school' was not considered (0).
In order to test a possible interaction effect of school identification with the type of school on endorsement of republican principles, we created two interaction variables that correspond to the product of school identification transformed in centered form, with each of the two contrasts: IDINTERC1=IDFC*C1 and IDINTERC2=IDFC*C2.
We first conducted a regression analysis in which support for republican citizenship was regressed on age and gender as control variables, C1 and C2 contrasts, school identification, the interaction between C1 and school identification (IDINTERC1) and the interaction between C2 and school identification (IDINTERC2).To correct for heteroskedasticity, an analysis with robust standard errors was conducted.The variables in this model explained 12.3% of the variance in support for republican citizenship, F(7, 289) = 5.67, p < .001.The results of this regression analysis revealed first, a significant effect of gender on support for republican citizenship, β = -.13,t(289) = -2.33,p = .02.Girls endorsed republican citizenship more strongly than boys.There was no significant effect of age on endorsement of republican citizenship, β = -.01,t(289) = -0.13,p = .894.However, as predicted, it appeared that type of school (general and technical high school versus middle school and vocational school) had a significant effect on support for republican citizenship, β = .17,t(289) = 2.58, p = .01.Consistent with our hypothesis, the pupils in general and technical high schools endorsed the principle of republican citizenship (M = 4.54, SD = 0.69) more strongly than pupils in vocational schools (M = 4.18, SD = 1.08) and in middle schools (M = 4.17, SD = 0.84).
On the other hand, pupils in vocational schools did not support republican citizenship significantly more (M = 4.18, SD = 1.08) than younger pupils in middle schools (M = 4.17, SD = 0.84), β = .02,t(289) = 0.25, p = .801.A significant effect of school identification on support for republican citizenship was also observed in the same analysis, β = .23,t(289) = 2.69, p = .008.The more pupils identified with their school, the more they endorsed republican citizenship.The interaction effect between the C1 contrast and school identification on endorsement of republican citizenship was non-significant, β = .001,t(289) = 0.02, p = .986.Contrary to our hypothesis, the effect of type of school (general and technical high school versus middle school and vocational school) on endorsement of repub-lican citizenship did not seem to be dependent on identification with the school.There also was no significant interaction effect between the C2 contrast (middle school versus vocational school) and school identification on support for republican citizenship, β = .11,t(289) = 1.44, p = .151.
We then conducted the same analysis using endorsement of the principle of new laïcité as the dependent variable.
The variables in this model explained 4.6% of the variance in endorsement of new laïcité, F(7, 289) = 3.03, p = .004.The results of the regression analysis showed no significant effect of gender, β = .08,t(289) = 1.40, p = .162,nor age, β = .10,t(289) = 1.05, p = .294,on support for new laïcité.Consistent with our hypothesis, we found a significant effect of the school type (middle school and vocational school versus general and technical high school) on support for new laïcité, β = .21,t(289) = 3.14, p = .002.Pupils in general and technical high schools endorsed the new laïcité (M = 3.42, SD = 0.86) significantly more than pupils in middle schools (M = 3.13, SD = 0.91) and than pupils in vocational schools (M = 2.79, SD = 1.11).On the other hand, pupils in vocational schools did not endorse this principle more strongly (M = 2.79, SD = 1.11) than pupils in middle schools (M = 3.13, SD = 0.91), β = -.16,t(289) = -1.88,p = .061.In fact, the results indicated a non-significant reverse trend with pupils in vocational school being less likely to support the new laïcité than pupils in middle school.Different from the results for republican citizenship, the effect of school identification on endorsement of new laïcité was not significant, β = .01,t(289) = 0.22, p = .828.The interaction effect between the C1 contrast and school identification on support for new laïcité was non-significant as well, β = .000,t(289) = 0.005, p = .996.Contrary to our hypothesis, the effect of type of school (middle school and vocational school versus general and technical high school) on support for new laïcité did not seem to depend on identification with the school.Finally, there also was no significant interaction effect between the C2 contrast (middle school versus vocational school) and school identification on endorsement of new laïcité, β = .03,t(289) = 0.45, p = .651.

Effects on Anti-Immigrant Prejudice
We wanted to examine the implications of the transmission of republican principles from middle school to general and technical high school for intergroup prejudice.For this, we first examined the evolution of prejudice from middle school to general and technical high school.We conducted a regression analysis in which prejudice were regressed on age and gender as control variables, school type (middle school versus general and technical high school), school identification and the interaction between school type and school identification.For this analysis, all statistical assumptions were met (independence, normality and constant variance of residuals, no multicollinearity of the independent variables).
The variables in this model explained 6.3% of the variance in prejudice, F(5, 247) = 4.37, p = .001.The results of the regression analysis showed no significant effect of age, β =.03, t(247) = 0.35, p = .728,nor gender, β =.08, t(247) = 1.24, p = .218,on prejudice.There was a significant effect of identification with the school, β = -.14, t(247) = -2.08,p = .038,and a significant interaction effect between identification with the school and school type, β = -.18,t(247) = -2.81,p = .005,on prejudice.A 'simple slopes' analysis showed that among less identified pupils, the level of prejudice was significantly higher in general and technical high schools than in middle schools, β = .23,t(252) = 2.71 p = .007.However, among the more identified pupils, the level of prejudice was marginally lower in general and technical high schools than in middle schools, β = -.15,t(252) = -1.68p = .094.There was no significant effect of school type on intergroup prejudice, β =.03, t(247) = 0.34, p = .731.Prejudice did not decrease nor increase significantly from middle school (M = 2.63, SD = 1.00) to general and technical high school (M = 2.70, SD = 1.02).However, this did not necessarily mean that the school did not exert indirect effects on prejudice through transmission of republican principles.Indeed, school type (middle school versus general and technical high school) had a significant effect on support for new laïcité and republican citizenship, two variables that can be linked to prejudice (Kamiejski et al., 2012).Thus, the school could have indirect effects on prejudice through support for these two principles of the republican model.

Indirect Effects of School on Prejudice Through Support for New Laïcité and Republican Citizenship
As explained by Preacher and Hayes (2008), a mediation can exist without total effect of an independent variable (X) on a dependent variable (Y).It is indeed quite possible that indirect effects of X on Y cancel each other out, making thus the total effect of X on Y non-significant.Therefore, in our case, school may have both positive and negative indirect effects on intergroup prejudice, through support for the two republican principles that are linked differently to prejudice.These indirect effects may to some extent cancel each other out.We thus wished to examine the specific indirect effects of school on prejudice via support for republican citizenship and new laïcité.
Specifically, we wanted to test a model predicting that the school has an effect on prejudice through the level of support for the republican principles.For this, we used the 'INDIRECT' macro for SPSS, developed by Preacher and Hayes (2008) and available on the website www.quantpsy.org.As recommended by Preacher and Hayes (2008), we conducted a bootstrap analysis on a model with prejudice as the dependent variable, support for new laïcité and for republican citizenship as mediators, and school type (middle school versus general and technical high school) as the independent variable (see Figure 2).Twenty-thousand bootstrap samples were programmed.
The variables in this model explained 21.3% of the variance in prejudice, F(3, 252) = 24.02,p < .001.The total effect of school on prejudice was non-significant, b = .07,t(252) = 0.55, p = .579.However, when we controlled the effects of supporting republican principles, there was a direct and significant effect of school on prejudice, b = .24,t(252) = 2.04, p = .043.As expected, support for republican citizenship predicted intergroup prejudice negatively, b = -.62,t(252) = -8.19,p < .001,while support for new laïcité predicted intergroup prejudice positively, b = .18,t(252) = 2.79, p = .006.We then examined the indirect effects of schooling on prejudice through endorsement of new laïcité and republican citizenship.An indirect effect is significant if the confidence interval (CI) does not contain 0. There was a significant total indirect effect of school on prejudice through endorsement of new laïcité and republican citizenship, b = -.17,95% CI = [-.31; -.04].School significantly reduced prejudice among pupils through endorsement of republican principles.We then investigated the indirect effects of school on prejudice through new laïcité and through republican citizenship separately.School had a significant indirect effect on prejudice through endorsement of republican citizenship, b = -.22,95% CI = [-.37; -.10].Schooling increased the support for the principle of republican citizenship, and supporting republican citizenship decreased prejudice.
There was also a significant indirect effect of school on prejudice through endorsement of new laïcité, b = .05,95% CI = [.01;.14].This indirect effect suggested that school increased support for new laïcité, and supporting new laïcité increased prejudice.So one indirect effect of the school on prejudice involved a decrease in prejudice (through republican citizenship) and the other an increase in prejudice (through new laïcité).These results confirmed, as shown in Figure 2, that the effect of schooling on prejudice was mediated by endorsement of the principles of republican citizenship and new laïcité.

Discussion
The central findings of this research were that public schools in France appeared to shape the level of anti-immigrant prejudice of their pupils through the transmission of principles related to the republican model of citizenship.
Consistent with previous work, a principal component analysis showed firstly that the endorsement of the republican model was structured in two factors: new laïcité and republican citizenship.We thus found among 12 to 19 years old pupils the same mental representation of the republican model than in adults (Kamiejski et al., 2012).
These results support the validity of our measuring instrument for this younger age group.In our study, girls adopted the republican citizenship principle to a greater extent than boys.However, there was no difference in support for new laïcité between girls and boys.This pattern of results is not necessarily surprising considering the literature on gender differences in values.Indeed, many studies showed that women attach more importance to equality than men in their attitudes and values (Dambrun, Duarte, & Guimond, 2004;Pratto et al., 1994;Schwartz & Rubel, 2005).Previous research in France showed that republican citizenship is an egalitarian principle, negatively related to social dominance orientation, whereas this was not the case for new laïcité (Kamiejski et al., 2012).

Evolution of Support for Republican Principles
We then examined the evolution of endorsement of republican principles throughout the school curriculum.As expected, we found that pupils in general and technical high schools endorsed the principles of republican citizenship and new laïcité significantly more strongly than pupils in middle schools and in vocational schools.Moreover, those in vocational schools did not support these principles more than middle school pupils.Evolution concerning the endorsement of republican principles from the middle school to general and technical high school could therefore not be explained by an effect of age.Indeed, we did not observe similar trends among vocational school pupils, who were of approximately the same age as general and technical high school pupils and were older than middle school pupils.Although these results support our hypothesis and allow us to rule out maturation as a possible explanation, several qualifications are in order.An important limitation of the present study is that pupils in vocational schools did not represent a perfectly adequate control group.As such, the difference in support for republican principles between pupils in vocational schools and those in general and technical high schools could be explained, not by education, but by other characteristics.For example, pupils from a lower social class or from immigrant's background tend to be overrepresented in vocational schools (Jellab, 2009).This could explain why we found that pupils in vocational schools were less likely to support republican principles.Unfortunately, it was not possible, given the nature of the present study, to request information on social class or immigration background or religion.Further research should seek to control these variables.
Similarly, it may be argued that, compared to other pupils, those in vocational schools often experience learning difficulties and end up with lower levels of academic achievement.Thus, the difference in support for republican principles between pupils in vocational schools and those in general and technical high school could be the result, not of the school context, but of a difference in cognitive sophistication among pupils (i.e., in cognitive ability).
Cognitive sophistication may play a role in understanding democratic principles (Meeusen, de Vroome, & Hooghe, 2013) and their implementation in everyday decision making (Lind, 2000).However, some recent studies on civic participation suggested that "The assertion that education is irrelevant compared to cognitive ability in determining good citizenship is untenable" (Hauser, 2000, p. 580).Hauser (2000) found that education had a positive effect on civic participation, independently of cognitive ability, whereas much of the association between cognitive ability and civic participation was due to the correlation between ability and education.Nevertheless, it would be desirable to control the effects of cognitive ability in the future.As noted by Emler and Frazer (1999), "education (…) is a complex variable containing a range of distinct yet correlated processes and outcomes" (p.256).There is a need to consider how these different factors could be involved in socialization effects.In the present study, one such factor that we considered was the role of school identification.

Influence of School Identification
Contrary to what was expected, school identification did not play a moderator role of the school effect on support for republican principles.We did not find an interaction between school identification and school type but only a main effect of identification.The more pupils identified with their school, the more they supported the principle of republican citizenship.Regardless of the type of school they attended, the sense of belonging to the school community seems to be promoting the development of civic attitudes among pupils.On the other hand, support for new laïcité was not related to school identification.This suggests that endorsement of new laïcité could be influenced not only by the school, but also by the social and political context.These results strengthen the importance of distinguishing the two principles of republican citizenship and new laïcité.
Different results were obtained when considering prejudice as the dependent variable.There was a main effect of school identification and an interaction effect between school identification and school type (middle school versus general and technical high school) on prejudice.Progression in education level was associated with greater prejudice among the weakly identified pupils but with less prejudice among the strongly identified.Given that there was an interaction effect between the type of school and school identification on prejudice, but not on republican principles, we conclude that school identification plays a role in the evolution of prejudice, independently of the transmission of republican principles.

Intergroup Prejudice
The second major objective of our study was to examine the implications of change in endorsement of republican principles for intergroup prejudice.As expected, the progression in the general school curriculum influenced endorsement of republican citizenship and new laïcité, which in turn had effects on intergroup tolerance.Contrary to what has often been reported in the literature, we did not observe a reduction of prejudice with education level.
Pupils in middle school were as accepting of immigrants (with a mean of 2.63 on a 5-point scale) as pupils in general and technical high school (with a mean of 2.70).Furthermore, when the analysis took into account the indirect effects of education through support for republican principles, we saw that education had a direct and positive effect on prejudice.These results seem inconsistent with an explanation of school effects on intergroup attitudes through cognitive sophistication (Bobo & Licari, 1989).Indeed, if this was the case, we should observe an overall decrease of prejudice from middle school to general and technical high school, which would be linked to an increase in reasoning ability.
Our results nonetheless revealed significant indirect effects of school on prejudice through endorsement of republican citizenship and new laïcité.The school thus reduced prejudice among pupils by the endorsement of republican citizenship, while it increased them slightly through endorsement of new laïcité.These results are consistent with the model of Selznick and Steinberg (1969): It is the transmission of the official culture embodying the ideal norms and values of a society that affects prejudice.However, our study showed that the impact of schools on prejudice is not always in the direction of greater tolerance.There are also cases when some principles that are part of the official culture, such as new laïcité, increase prejudice, rather than reduce it.
While the new laïcité principle seems to have deleterious effects on tolerance, one should not conclude that laïcité is in itself an intolerant ideology.Several authors have stressed that there are different conceptions of the principle of laïcité (Baubérot & Milot, 2011;Geerts, 2010;Marchand, 2011;Milot, 2013).This could explain the fact that in contemporary France, the principle of laïcité is now advocated not only by the left-wing liberals but also by the right-wing conservatives, and even by the extreme right (Barthélemy & Michelat, 2007a, 2007b).In some of our recent work, the model of malleable ideology of Knowles, Lowery, Hogan, and Chow (2009) was used in order to explain why laïcité could be endorsed by individuals with opposing motivations (Roebroeck & Guimond, 2014).
In his book entitled Falsified Laïcité (2012), Baubérot developed the idea that the interpretation of the principle of laïcité, which has grown in recent years in France, has distorted its original meaning.This new laïcité, insisting that religious affiliations be kept in private, and establishing laws banning the wearing of religious symbols, appears to be intolerant of cultural and religious differences.It is this conception of laïcité that our measure was tapping.
Given the moderate to weak internal reliability score of our measure of new laïcité, we have however to remain cautious in our conclusions and continue to conduct studies on the subject.
According to Baubérot (2012), this new laïcité is opposed to the 'historical laïcité', valuing equality and tolerance.
In this regard, it is interesting to consider the last item in our scale of republican citizenship that is displayed in Table 1: "It is important to respect the equality before the law of all citizens, without distinction of origin, race or religion".This item was not part of the original scale developed by Kamiejski et al. (2012) Indeed, republican citizenship transmitted by the school has beneficial effects on tolerance.As shown in our results, the school had an impact on endorsement of republican citizenship, which then had a powerful and negative impact on prejudice.Moreover, our analyses suggested that it is support for republican citizenship which influences prejudice, and not the reverse.This is an important contribution of the present study.The results of Kamiejski et al. (2012) did not rule out the possibility that prejudice leads to the endorsement of republican principles rather than the other way around.In contrast, because we found that upper-level pupils did not differ from younger ones in their level of prejudice, a model suggesting that schooling has a direct effect on prejudice, and prejudice has an effect on endorsement of republican principles is clearly not supported by the data.With this study, we go beyond the simple bivariate association between principles of citizenship and prejudice.By introducing the role of schools, we placed these variables in a theoretically consistent framework and our findings provided strong support for this overall framework.Nevertheless, experimental studies will be needed to confirm causation.
How can we explain the beneficial effect of republican citizenship on tolerance?The principle of republican citizenship appears as an egalitarian principle, defending the equality of all citizens regardless of their religious or cultural memberships.A number of experimental studies showed that individuals who aim to be egalitarian in all situations could prevent the activation of stereotypes, even at a preconscious level (Moskowitz, Gollwitzer, Wasel, & Schaal, 1999).A study by Moskowitz and Li (2011) even showed that it is possible to get people to adopt egalitarian goals without their awareness, and thus prevent stereotypes activation.Moskowitz and Li (2011) noted that the colorblind ideology and multiculturalism are two forms of egalitarian goals.One can imagine that the principle of republican citizenship, transmitted by the school, leads individuals to adopt egalitarian goals, which would then inhibit the formation of negative stereotypes and prejudice.Consistent with this explanation, we observed in our results that the school had a positive direct effect on prejudice, but this effect was offset by a strong indirect effect of school on prejudice through endorsement of republican citizenship.
An important question for the future will be to examine if similar results can be obtained in other national contexts having different citizenship regimes.Based on studies related to multicultural education, it might seem surprising that republican citizenship would have positive implications for intergroup relations.Verkuyten and Thijs (2013) showed that multicultural education reduces prejudice among early adolescents in the Netherlands.If open discussion about cultural and ethnic differences (as in multicultural education) can reduce prejudice, one would not expect that a colorblind approach that de-emphasises cultural and religious differences, such as the one dominating French schools, would also have a positive impact.Furthermore, many have argued, in the United States and in France, that the colorblind model may represent a double-edged sword.De-emphasising racial differences may also be a way to overlook the existence of racial discrimination (Apfelbaum, Norton, & Sommers, 2012).According to Bonilla-Silva (2003), color-blindness in the United States has evolved into a shared social representation that helps preserve and maintain racial inequality rather than attenuate it (see also Knowles et al., 2009).Similarly, many have suggested that an approach that is gender-neutral or gender-blind often turns out in practice as supporting male dominance (Lister, 2001;Scott, 1998;Young, 1989).In fact, the experimental studies of Richeson and Nussbaum (2004) suggested that stressing color-blindness had deleterious effects on racial prejudice in the United States.Yet, all this work leads one to expect that supporting the republican model in France will be related to having more prejudice, not less as we found, and as Kamiejski et al. (2012) found.So why did we find that supporting color-blindness(or republican citizenship) in France had a positive effect on intergroup attitudes?Social desirability?No, because this would apply also to the principle of new laïcité, and in this case, we found that it has a negative effect on intergroup attitudes.We believe that the answer lies in the fact that supporting republican citizenship and supporting multiculturalism have one thing in common: They both reflect a genuine commitment

Figure 1 .
Figure 1.Model of the effect of formal schooling (number of years in schools) on prejudice (intergroup attitudes) as a function of the transmission of republican principles and values.
suis opposé à ce que le gouvernement finance la construction d'édifices religieux.(NL1 / The government should not pay for the construction of religious buildings.que possible, les pratiques religieuses devraient être à caractère privé et non public.(NL2) / As far as possible, religious practices should be private and not public.me semble tout à fait normal que dans les écoles publiques en France les signes religieux visibles soient interdits.(NL3) / It seems to me perfectly normal to ban all visible religious symbols from public schools in France..68 .11 4. Je suis favorable à la séparation du religieux et de l'Etat en France comme dans les autres pays.(NL4) / I support the separation of religion and state in France as in other countries..75 .27 5. Je ne veux pas qu'on définisse les gens en France en fonction de leurs origines, de leur religion.(RC1) / I do not want that people in France be identified by their origin, their religion.société française est composée de citoyens avant toute chose, et non pas de communautés.(RC2) / French society is made up first and foremost of citizens, not of groups..14 .73 7. Un Français doit être pris en considération en tant que tel et pas en tant que membre d'une communauté particulière (culturelle, religieuse, sexuelle,…).(RC3) / French people should be considered as such and not as a member of a particular community (cultural, religious, gender ...).l'unité du pays, les individus doivent être considérés comme citoyen avant de l'être comme « black, blanc, beur », femme, homme, homosexuel ou hétérosexuel.(RC4) / For the unity of the country, individuals must be regarded as citizens before being seen as black, white, arab, woman, man, gay or heterosexual.est important de respecter l'égalité devant la loi de tous les citoyens, sans distinction d'origine, de race ou de religion.(RC5) / It is important to respect the equality before the law of all citizens, without distinction of origin, race or religion..06 .82Note.RC = Republican citizenship (Factor 1, 36.56% of total variance); NL = New laïcité (Factor 2, 20.07% of total variance).
. It was included in the present study to get a better sense of what is measured by the two scales.And we found that this item, arguably close to what historical laïcité means, did not relate with other items dealing with new laïcité.It loaded very strongly on the republican citizenship factor suggesting that this dimension probably reflects in part the alternative and more egalitarian view of laïcité.

Table 1
Items Measuring Support for the Republican Model of Citizenship: Factor Loadings From Principal Components Analysis With Varimax Rotation