Running head: THE SOCIAL VALUE OF BJW IN CONTEXT 1 The Social Value of Expressing Personal and General Belief in a Just World in Different Contexts

Research conducted in France and Portugal has consistently found that expressing high versus low Belief in Personal a Just World (BJW-P) is more socially valued. Results concerning the Belief in a General Just World (BJW-G) have been mixed. We propose this reflects a higher resistance of BJW-P social value to contextual changes. Testing this idea was the main goal of three experimental studies conducted in France, Germany and Portugal. In Study 1 (N = 283) participants expressed higher BJW-G when asked to convey a positive versus a negative image in a job application at a bank. The opposite pattern showed up when they applied for a job at a Human Rights NGO, an employment assistance institution and a trade union. Participants expressed higher BJW-P in all contexts, except at the trade union (no significant differences). In Study 2 (N = 489) participants judged bogus candidates who expressed high or low BJW-P/G while applying for a job at the same contexts. The patterns of judgments replicated those of self-presentations in Study 1. In Study 3 (N = 158), participants were asked to judge targets who expressed high vs. moderate vs. low BJW-P at a trade union. The former target was more socially valued than the other two. High versus low BJW-P expression was associated with higher stamina and less unadjusted self-enhancement. We conclude that in Western societies the expression of BJW-P is more central to the legitimation of the status quo, and that of BJW-G is more context sensitive.


5
are rewarded" (our translation). For Vermunt and Steensma (2008), the BJW is a strongly internalized norm that guides individuals' judgments regarding what they and others do. As we will develop in the next section, most research on the very expression of BJW has put into evidence both spheres of justice, especially BJW-P, as normative discourses that serve important legitimizing functions (Alves & Correia, 2008;Testé, Maisonneuve, Assilaméhou, & Perrin, 2012; for a review, see Gangloff, Soudan, & Auzoult, 2014).
The Sociocognitive Approach and the Expression of BJW-P and BJW-G as

Judgment Norms
Research conducted in France and Portugal has studied the social value associated with the expression of both spheres of BJW. This line of research has followed the socio-cognitive approach, which focuses on the study of judgment norms.
Judgment norms are prescriptive statements which may or may not reflect reality, and in which people may or may not believe (Dubois & Beauvois, 2003). The prescriptive normativity of these statements or discourses lies not in their objective or perceived truth but in the social value (or social credit) they carry for legitimizing and perpetuating social arrangements (for instance, the "norm of internality", Dubois, 2000;Jellison & Green, 1981). The socio-cognitive approach identified and theorized on various experimental paradigms to study judgment norms. In the specific case of research on BJW expression, two main experimental paradigms have been used: the self-presentation paradigm and the judge paradigm (Gilibert & Cambon, 2003).
Using the self-presentation paradigm, Alves and Correia (2008, Study 1) asked Portuguese participants to fill out either the BJW-P scale (Dalbert, 1999) or the BJW-G scale (Dalbert, Montada, & Schmitt, 1987) in such a way that they would self-present positively or negatively (i.e., to convey a positive or a negative image of themselves, Running head: THE SOCIAL VALUE OF BJW IN CONTEXT 6 respectively). Those participants used higher scores of both BJW-P and BJW-G when they were asked to self-present positively. (Alves & Correia, 2008, Study 1). This pattern was stronger in the case of BJW-P than BJW-G. Indeed, when participants aimed to self-present positively, they used even higher BJW-P than BJW-G. When they aimed to self-present negatively, they used even lower BJW-P than BJW-G. Alves and Correia (2010b, Study 2) replicated this pattern using more specific self-presentation goals (e.g., to be seen as competent, successful or pitiful), and concluded that the expression of BJW, especially that of BJW-P, could be used strategically to convey specific images to an (unspecified) audience.
In France, Duchon and Gangloff (2008) asked 60 unemployed men and women to convey a positive or a negative image of themselves by filling out Rubin and Peplau's (1975) BJW-G scale. They found that women, but not men, used significantly higher scores to self-present positively versus negatively. Gangloff and Duchon (2010) asked 504 employed and unemployed individuals of both sexes to convey either a positive or a negative image of themselves by responding to a BJW-G scale specifically devised for the labor context. Scores were higher in the positive than in the negative image condition, regardless of participant sex and employment status.
Using the judge paradigm, Alves and Correia (2008, Study 2) found that the targets were evaluated (i.e., judged) differently on a general impression measure depending on the way they had allegedly responded to the BJW-P or BJW-G scales.
Specifically, participants judged the target who had expressed high versus low BJW more favorably (but no difference between the BJW spheres). Alves and Correia (2010a, Study 1) also found that targets who expressed high versus low BJW in bogus interviews were judged more favourably (for an extension of this research, see Alves & Correia, 2013). This effect was even stronger when participants judged targets Running head: THE SOCIAL VALUE OF BJW IN CONTEXT 7 expressing BJW-P versus BJW-G. Specifically, the target who expressed high BJW-P versus high BJW-G was even more socially valued, whilst the target who expressed low BJW-P versus low BJW-G was even less socially valued. Finally, Alves et al. (2015) found that non-victims, non-innocent victims and innocent victims were judged more favorably when they expressed high versus low BJW-P. This is consistent with the notion of BJW-P as a judgment norm -individuals approve of its expression even though they do not necessarily judge it as a faithful account of one's reality (Alves & Correia, 2010a, Study 2).
Research conducted in France with the judge paradigm arrived at similar conclusions, especially regarding the expression of BJW-P. In four experimental studies, the targets who expressed high versus low BJW-P were judged more positively (Testé and Perrin, 2013; see also Testé et al. (2012). In contrast, the targets who expressed high versus low BJW-G were judged less positively in three studies and marginally more favorably in one. This led Testé and Perrin (2013) to propose the normativity of BJW-G expression as ambivalent. Nevertheless, in other research also conducted in France, the targets who expressed high versus low BJW-G were judged more favorably (Gangloff & Mazilescu, 2015). Furthermore, Gangloff (2008) asked 37 recruiters to select the items of Rubin and Peplau's (1975) scale they expected a good candidate for a (unspecific) job would choose. His results showed that participants chose more items in the just versus the unjust world direction (for similar results, see also Soudan & Gangloff, 2013, Study 1).
In sum, in France and Portugal at least, the expression of BJW-P has clear social value. The social value, or normative character, of the expression of BJW-G seems less strong, and could be ambivalent according to Testé and Perrin (2013). There is, however, an important limitation across the aforementioned studies -context Running head: THE SOCIAL VALUE OF BJW IN CONTEXT 8 vagueness. Indeed, in most studies participants were asked to either self-present to an unspecified audience ("someone"/"other people") or to judge targets in an undefined context; furthermore, none of the studies manipulated contexts. Yet, what individuals say to convey various images of themselves, and the way others judge them is generally highly context dependent (Goffman, 1959). This implies that one given statement or discourse can be more or less valued depending on the context it is enacted. The hypothesized normative ambivalence of BJW-G expression could be due to its higher context sensitivity than that of BJW-P, as Alves and Correia (2010b) suggested, but did not test.

The Current Research: Organizational Contexts Used and General Hypotheses
We conducted two experimental studies in France, Germany and Portugal, and a third study only in Portugal to investigate the impact of context in the social value of BJW expression. We used four organizational contexts: a bank, a trade union, a nongovernmental organization that fights for human rights (henceforth, "NGO") and an employment assistance institution promoting employability of people with intellectual disability (henceforth, "institution"). Next, we explain why we chose these four kinds of organizations, and indicate their probable impacts on the social value of BJW-G and BJW-P expression. We begin by addressing the case of BJW-G expression, and will move on to that of BJW-P afterwards.
In Western societies the BJW has been identified as a mechanism that legitimizes the status quo (e.g., Abarri, Gangloff, & Fares, 2016;Jost & Hunyady, 2005), which in turn is anchored on economic (neo)liberalism, its associated meritocratic ideology and evaluative practices (de Gaulejac, 2005;Dubois, 1994). In this economic system, both successes and failures are judged as natural and deserved. In this highly competitive, quasi-darwinist context the BJW discourse targets individuals, Running head: THE SOCIAL VALUE OF BJW IN CONTEXT 9 organizations and countries: winners naturally -thus, justly-succeed; losers naturallythus justly -fail (e.g., de Gaulejac, 2005). Alves and Correia (2010b) suggested that expressing high degrees of BJW should be especially central in this context because it is a discourse that legitimates that system. For these reasons, we chose banks as one of the contexts for our research. We thus predict that the expression of high versus low BJW-G has higher social value there.
In democratic societies at least, whereas banks are organizations that sustain the (neo)liberal economic system, trade unions serve as a counter-power to that system.
Trade unions aim to protect and advance the interests of workers, namely their working conditions, payment, work and rest hours. They do so through protest, injustice claims and negotiation (Roca, 2016). As regards NGOs, their actions focus on diverse kinds of victims, which may intersect (e.g., war refugees, famine victims, persecutions for political dissent or/and sexual orientation). Like trade unions, their actions also center on addressing injustice. These actions may target working conditions and payment, especially in developing countries, but they typically involve a wider range of domains (e.g., freedom, nourishment, health). Finally, institutions aim at promoting the employability of individuals who are stereotypically perceived as incompetent (Fiske, Cuddy, Glick, & Xu, 2002). They can be considered victims in the sense that they are usually discriminated against for not having valued characteristics, and are seldom given opportunities to develop their potential. For these reasons, we assume that a discourse focused on general injustice, not justice, should be judged as normative in trade unions, NGOs and institutions. Such discourse justifies and legitimates their existence and goals. We thus predict that the expression of low versus high BJW-G has higher social value in these three contexts. In relation to the expression of high versus low BJW-P, there are reasons to expect that it has more social value in the four contexts.
First, people are generally frown upon when they complain about their lives. For instance, Kaiser and Major (2006) reviewed research showing that individuals disliked targets who complained about sexism or racism. This occurred even when participants acknowledged that those targets had actually been victims of such prejudice. Also, Alves et al. (2015) showed that victims of a car accident who had become permanently disfigured and maimed were less liked when they expressed low versus high BJW-P.
This occurred regardless of whether the victims had been driving carelessly (high speed and drunk) or not.
Second, being an active member in any of these organizations requires mental (and physical) stamina to withstand stress. Individuals who score higher on BJW-P measures also score higher in various measures of psychological well-being and coping with adversity (Dzuka & Dalbert, 2002. It is possible that participants may intuitively infer that the expression of higher BJW-P is associated with psychological adjustment and stamina, which seem crucial to resist the inherent stress in those contexts. This reasoning is in line with the finding that the expression of BJW-P seems to be more associated with other judgment norms in Western societies, namely the norm of internality (i.e., individuals' are supposed to take responsibility for their outcomes; e.g., Dubois & Beauvois, 2005;Jellison & Green, 1981), the norm of self-sufficiency (i.e., individuals are supposed to be autonomous entities; e.g., Dubois, 2005;Dubois & Beauvois, 2005). It has been shown that expressing higher degrees of BJW-P can even be used as a self-enhancement strategy (Alves & Correia, 2010b), which is expected in job application settings in Western societies (Meyer, 2014;Stevens & Kristof, 1995).
Third, past research has found that the social value of BJW-P is consistently stronger than that of BJW-G (Alves & Correia, 2010a, b;Testé & Perrin, 2013).
Although that research has not manipulated contexts, the consistent pattern may indicate that the social value of BJW-P is relatively resistant to contextual shifts, which results in its expression being expected in contexts with various identities and goals. We thus may expect that the expression of high versus low BJW-P has higher social value across all four contexts.
On the other hand, however, in union, NGO and institution contexts the ultimate goal is to help different kinds of "victims". As such, there is also the admittedly smaller possibility that referring to one's life as generally just may be seen as a sort of bragging, that is, unadjusted/ excessive self-enhancement, which is counter-normative even in job application settings (Holtgraves & Srull, 1989;Leary, 1995). If that is the case, we could expect that the expression of high versus low BJW-P has lower social value in these contexts than in the bank.

Study 1
This study aimed to test for the first time whether individuals express different degrees of BJW-P and/or BJW-G across contexts, specifically at a bank, a union, a NGO, and an institution. In order to achieve that goal we used the self-presentation paradigm. French, German and Portuguese participants were thus asked to imagine that they were applying for a job in those four kinds of organizations. They were instructed to fill out either the BJW-P scale or the BJW-G scale in such a way that they would convey a positive and a negative image of themselves in each of those contexts.
First, we expected that participants would use higher scores of BJW-P and BJW-G to convey a positive versus a negative image of themselves when they applied for a job at a bank (Hypothesis 1). Reflecting the stronger social value of BJW-P expression  (Alves & Correia, 2010b;Testé & Perrin, 2013), we predicted that when participants self-presented at a bank, the score differences between images would be higher for BJW-P than BJW-G (Hypothesis 2). We expected that participants would use lower scores of BJW-G to convey a positive versus a negative image of themselves (thus showing that BJW-G expression is counter-normative) when they applied for a job at a trade union, a NGO and an institution (Hypothesis 3). Reflecting the overall higher social value of BJW-P, we expected that the participants would express higher BJW-P to convey a positive rather than a negative image across contexts (Hypothesis 4).
We explored whether these patterns differed with political orientation (Left-/Right-Wing). On the one hand, we think it is relevant to take this variable into account, given that organizations can be judged differently across the political spectrum (for the case of banks, see Bennett & Kottasz, 2012). In this view, we should expect differences involving political orientation. On the other hand, our research investigates individuals social knowledge about what is normative to express in different contexts, not their attitudes towards those contexts. In this view, we should not expect differences involving political orientation (see Dittmar & Dickinson, 1993).
Given that this research was conducted in three Western European countries, which have shared a common economic space for decades, we do not have a priori reasons to expect differences in the result patterns. Furthermore, this was a completely convenience sample of countries, not one based on cultural criteria 1 . Nevertheless, we took the countries into account in our (preliminary) analyses.

Method
Participants.
Our sample consists of 283 university students comprising 200 females and 81 males, two unreported (France: 76 females, 18 males; Germany: 74 females, 20 males; This study has a 2 (BJW spheres: personal/ general) X 2 (self-presentation goals: to convey a positive/ negative image) X 4 (contexts: bank/ trade union/ NGO/ institution) X 3 (country: France/Germany/ Portugal) mixed design, with the former and the latter factors as between-participants, and self-presentation and contexts as betweenparticipants.

Procedure.
University students were invited to take part in the study on a voluntary basis. In France data were collected during classes at the University of Rouen, in Germany they were collected in the campus of Martin Luther University of Halle-Wittenberg, and in Portugal they were collected in various universities in Lisbon. In the latter two cases the experimenter (a male in Germany and a female in Portugal) approached participants and asked them whether they were interested in taking part in a study.
After participants agreed to participate, they were given a stapled block of sheets of paper that contained the instructions and the study itself. Participants were asked to fill out either the BJW-P scale or the BJW-G scale as if they were applying for a job at various organizations that were presented in generic terms: a bank, a trade union, an institution, and a NGO. We briefly summarized the missions of the organizations and what they required from the successful candidate. For instance, in the bank condition, participants read: "Imagine that you are applying to work at a bank as a financial consultant. If selected, you would be responsible for the development of client investment strategies and the supervision of their current investments." We counterbalanced the order of the contexts across participants by using a Latin square. Finally, participants indicated their sex, age and political orientation. Afterwards the experimenter debriefed participants and answered any questions they had.

Measures.
Expressed general belief in a just world. Participants in the BJW-G conditions responded to the six items of the BJW-G scale (Dalbert et al., 1987; e.g., "I think basically the world is a just place."; "I think people try to be fair when making important decisions"). Participants responded on 6-point Likert-type scales (1 = completely disagree; 6 = completely agree). Cronbach alphas varied between .83 and .88 in France, between .75 and .90 in Germany, and between .74 and .85 in Portugal.
Expressed personal belief in a just world. Participants in the BJW-P conditions responded to the seven items of the BJW-P scale (Dalbert, 1999; e.g., "I believe that, by and large, I deserve what happens to me"; "I think that important decisions that are made concerning me are usually just"). Participants responded on 6point Likert-type scales (1 = completely disagree; 6 = completely agree). Cronbach Political orientation. Participants responded to an item: "In politics one usually refers to left-wing or right-wing. How would you indicate your political position on this scale, in which "a" represents the position most to the Left, and "k" the position most to the Right?". Participants thus responded on an 11-point scale (a = left-wing; k = rightwing). We used letters that were converted into numbers afterwards (a = 1, b = 2, ... k = 11), so that participants made no implicit judgments about the value of left-wing or right-wing.

Results
In order to simplify the presentation of results, we calculated the differences of scores between the positive and the negative images within each context (henceforth, "images within contexts"). Scores could thus vary between -5 and + 5. Positive mean values indicate that the BJW scores in the positive image were higher than in the negative image, and negative mean values indicate the reverse. Scores equivalent to zero indicate that participants used similar degrees of BJW to convey both kinds of images. The interested reader can consult the score means and standard deviations by experimental condition in Table 1.

Preliminary analyses.
A one-way ANOVA with country as the factor and political orientation as the dependent variable indicated there were differences in political orientation according to the participants nationality, F(1, 263) = 11.65, p < .001, η 2 p =.08. Tukey post-hoc tests  = 30.80, p < .001, η 2 =.11. We decided to collapse the data across countries for three reasons: 1) as seen, this interaction only involved the strength of the differences among the countries, not pattern differences (Tukey post-hoc tests did not indicate any significant differences among the countries within each images within context index, all p's ≥ .46); 2) the strength of the differences involved the same contexts, and 3) BJW spheres did not qualify this two-way interaction, F(6, 813) = 1.33, p = .24, η 2 = .01. We conducted a 2 (BJW spheres: Personal/ General) X 4 (images within contexts: bank/ trade union/ NGO/ institution) mixed measures ANOVA, with the former factor as between-participants and the latter factor as within-participants. We then analyzed how participants conveyed a positive and a negative image through BJW-P when they applied for a job at a NGO, an institution, and a union. As

Discussion
This is the first study to address the role of context in the expression of BJW-G and BJW-P. As Alves and Correia (2010b) suggested, the social value of BJW-G versus BJW-P expression is more prone to changes across contexts. Indeed, expressing higher BJW-G is seen as normative in one of the presented contexts (a bank). In the other cases -a trade union, a NGO and an institution -the opposite was true. In contrast, the expression of BJW-P was clearly seen as normative in three organizational contexts -a bank, a NGO and an institution. In the case of the bank, the expression of BJW-P was seen as even more normative than that of BJW-G. Surprisingly, however, in the case of the trade union, participants indicated they would express equivalent BJW-P to convey both a positive and a negative image of themselves when applying for a job there. These patterns occurred in the three countries (but more strongly in France and Germany than in Portugal). Although political orientation differed among countries, it did not significantly predict participant self-presentation strategies.

Study 2
The main goal of this study was to replicate Study 1 by using the judge paradigm. Participants read low or high BJW-P or BJW-G responses of bogus candidates for jobs at the same organizations as in Study 1. Afterwards, participants judged the social value of candidates.
We first expected that targets who expressed high versus low BJW (either personal or general) would be judged as having more social value when they applied for a job at a bank (Hypothesis 1). We also expected that this effect would be stronger for the expression of BJW-P than BJW-G, such that the target expressing high BJW-P would be judged as having even more social value than the target expressing high BJW-G, and/or the target expressing low BJW-P would be judged as having even less social value than the target expressing low BJW-G (Hypothesis 2). As regards the other contexts, we expected that the targets who expressed high versus low BJW-G would be judged as having less social value (Hypothesis 3). Finally, we expected that targets who expressed high versus low BJW-P would be judged as having higher social value when they applied for a job at a NGO or an institution (Hypothesis 4). As regards the expression of BJW-P at a union, we do not present a hypothesis. On the one hand, if we replicate Study 1, there will not be a significant difference judgments of target social value. On the other hand, given that the judge paradigm is more powerful in detecting the normativity of an object (Gilibert & Cambon, 2003), there could be such a difference. Taking into account the assumed higher resistance of BJW-P to contextual changes, we reasoned that if we found such a difference, it would likely be in the same direction as in the other contexts.

Method
Participants.
Our Design.
This study has a 2 (BJW spheres: Personal/ General) X 2 (degree of BJW expressed: low/high) X 4 (context: bank/ trade union/ NGO/ institution) X 3 (country: France/Germany/ Portugal) mixed design, with context as within-participants and the remaining three factors as between-participants. The procedure used to collect data was similar to that in Study 1. In this study, however, participants had to imagine that they already worked at the same four organizations, and had been asked to give their opinions on the hireability of a candidate.
To minimize fatigue, participants were randomly presented with either high or low BJW answers across organizations. Also to minimize fatigue, the bogus candidates' Political orientation. We used the same item as in Study 1.

Results
Manipulation check.

Preliminary Analyses
In order to check whether political orientation differed among the participants from the three countries, we conducted a one-way ANOVA with country as the factor Portuguese participants (M = 6.71, SD = 2.05; p < .001).
As in Study 1, we also checked whether we could collapse the data across countries and not take political orientation into account as a covariate in our main analyses. We thus ran a 2 (BJW spheres) X 2 (degree of BJW expressed) X 4 (contexts) There was a three-way interaction among BJW spheres, contexts and country, F(6, 1356) = 2.30, p = .03, η 2 p = .01, which was qualified by a four-way interaction, F(6, 1356) = 2.71, p = .01, η 2 p = .01. Tukey post-hoc tests identified several significant differences, but none was theoretically relevant. As an example, these tests indicated that German participants judged the target who expressed high BJW-P at a bank as having more social value than the target who expressed high BJW-G at a NGO. There were no significant differences among the countries regarding each degree of BJW expressed within each BJW sphere and context (all ps ≥ .40). For these reasons, we decided to collapse data among countries. Also, when we reran analyses without the covariate, the patterns did not change. We thus decided to drop political orientation, We conducted a 2 (BJW spheres) X 2 (degree of BJW expressed) X 4 (contexts) mixed-measures ANOVA, with the former two factors as between-participants and the latter factor as within-participants.

Discussion
With this study we replicated the results of Study 1 while using a different experimental paradigm. Indeed, the social value of BJW-G vs. BJW-P expression was again more prone to changes across contexts. Specifically, in the four contexts presented expressing high BJW-G was seen as normative when applying for a job at a bank but not at a union, a NGO and an institution. In contrast, the expression of BJW-P was again clearly normative at a NGO, an institution and a bank. In the latter case, the expression of high BJW-P was judged as having even more social value than that of high BJW-G, whilst the expression of low BJW-P was judged as having even less social value than that of low BJW-G. This puts into evidence that the expression of high BJW-P is especially valued and that of low BJW-P is especially aversive in a context that is central to the (neo)liberal status quo. In contrast, the expression of BJW-P at a union was again judged as neither normative nor counter-normative, which will be addressed in Study 3. These patterns were present among the participants of the three countries. In the previous two studies, the expression of BJW-P at a union was judged as neither normative nor counter-normative. It is as if the participants are in a conflict between the societal norm of expressing PBJW and the protest function of unions, the only context in which economic liberalism can be questioned. Again using the judge paradigm, the main goal of this study was to tackle this unexpected pattern. Next, we present various possibilities that may account for it.
First, it is possible that opposing processes simultaneously pull up and down the social value of low and high BJW-P respectively, thus resulting in a null difference. In this study we tested whether perceived target stamina and excessive self-enhancement could be such intervening processes, for the reasons presented in the introduction. On the one hand, the expression of high vs. low BJW-P should be associated with higher target stamina and higher social value, with target stamina also positively predicting target social value. In this view, expressing high PBJW should be aligned with discursive norms. On the other hand, the expression of high vs. low BJW-P could be associated with target excessive self-enhancement with the latter negatively predicting target social value. In this view, expressing low, not high, PBJW should be aligned with discursive norms. In our rationale these two opposing patterns could thus explain the null difference when participants were asked to self-present (Study 1) and to judge the social value of targets (Study 2). In sum, we put forward that when participants think in terms of target stamina, they may attribute higher social value to PBJW expression than when they think in terms of target self-enhancement BJW-P expression by having six out of seven items marked as "agree" ["disagree"]. We thus may have unwittingly operationalized BJW-P expression in a way deemed as extreme for a trade union context. In Study 3 we thus operationalized BJW-P expression in a way that is facially less extreme.
A third possibility is that our dependent measure in Study 2 was not detailed enough to capture the social value of targets in trade unions. In Study 3 we thus not only included the five items used in Study 2, but also added others that capture more specific aspects of the trade union domain.
Finally, the null differences between high and low BJW-P social value may indicate that an intermediate degree is judged as the most normative in this context.
Indeed, in Correia (2008, 2010a), in which no contexts were used, the expression of moderate BJW was as valued as that of high BJW. Furthermore, Alves and Correia (2010a) showed that the target who expressed moderate BJW-P was judged as the most realistic, which we assume to be a valued characteristic in trade unionists. In this study we thus explored whether in the trade union context the expression of moderate BJW-P is judged as the most normative degree.

Method
Participants.
One hundred and seventy-five university students took part in this study.
Seventeen participants were excluded from analyses either because they were not Portuguese (Brazilian or Cape Verdean; n = 14) or because they did not respond to several measures (n = 3). Our final sample thus consists of 161 Portuguese university Design.
This study has a unifactorial between-participants experimental design with three levels of BJW-P expression: low versus moderate versus high BJW-P. The bogus responses of the targets were presented on the original format of PBJW scale (Dalbert, 1999), that is, on 6-point Likert-type scales (1 = totally agree; 6 = totally disagree).
Following Alves and Correia (2008, Study 2), we operationalized the expression of: high BJW-P by marking four items as "completely agree" and three as "agree" (points 1 and 2); moderate BJW-P by marking four items as "somewhat agree" and three as "somewhat disagree" (points 3 and 4); low BJW-P by marking four items as "disagree" and three as "completely disagree" (points 5 and 6). Procedure.
The procedure was the same as that used to collect data for the previous studies in Portugal. As far as the study itself is concerned, we again used the judge paradigm but included some changes already presented in the introduction to this study (e.g., focus on BJW-P at a trade union, between-participants design, inclusion of measured intervening variables). Measures.
Manipulation check. In order to check whether participants interpreted our manipulation of BJW-P expression as intended, they responded to the same item as in Study 2.
Targets social value (α = .96). Participants judged target social value by responding to 17 items on 7-point Likert-type scales. Besides the five items used in Study 2, participants judged each target by responding to 12 other items (e.g., "S/he will Running head: THE SOCIAL VALUE OF BJW IN CONTEXT 28 be able to contribute to make the world a more just place."; "S/he will be able to seek fair solutions to labor problems"; "S/he is expeditious, shows initiative"; "S/he often protests against the injustices s/he witnesses"; "If s/he verbalizes what s/he answered, s/he increases chances of being successful at negotiations."; "This is someone who tries to motivate those around him/her to behave fairly").
Political orientation. We used the same item as in Studies 1 and 2.
All items, except the one gauging political orientation, had a 7-point Likert-type format. The anchors of the new items were: 1 = very much disagree and 7 = very much agree.

Manipulation Check
A one-way ANOVA on the manipulation check item revealed that participants interpreted our manipulation of BJW-P as intended

Main Analyses
We ran three ANOVAs which indicated that target judgments differed depending on the degree of BJW-P they expressed. As regards social value, participants judged the targets who expressed low or moderate BJW-P equally less favorably (Ms To determine whether judged target stamina and self-enhancement statistically mediated the relation between degree of BJW-P expressed and target social value we conducted a bootstrapped parallel mediation analysis for a multicategorical independent variable using the indicator coding (Process 2.16, Model 4 with 10,000 resamples and 95% Confidence Intervals [CI]; Hayes & Preacher, 2014). We first calculated an indirect effect with low BJW-P as the referent group (which compares low BJW-P with the other two degrees), and another with moderate BJW-P as the referent group to obtain the comparison between moderate and high BJW. Table 3 shows the coefficients and standard errors of paths a (from the independent variable to the measured mediators, corresponding to the ANOVAs) and b (from the mediators to dependent variable). The An indirect effect is significant if zero is not included in the CI.
As can be seen in Table 3, paths a and b involving judged target stamina are positive in the three cases. In other words, the higher the BJW-P expressed, the higher participants judged target stamina. In turn, the higher participants judged target stamina, the higher they judged target social value. The indirect effects were significant: point As regards judged target self-enhancement, paths a and b were negative. When comparing the expression of low vs. moderate and low vs. high BJW-P, the higher the BJW-P expressed, the lower participants judged the targets as self-enhancing. In turn, the lower participants judged the targets as self-enhancing, the higher they judged target social value. In these cases, the indirect effects were significant: point estimateLowvsModerate = 0.13 SE = .07, CI95%= 0.02, 0.30; point estimateLowvsHigh = 0.13 SE = .07, BCa CIs = 0.02, 0.31. When comparing the expression of moderate vs. high BJW-P, path a was nonsignificant (as already shown by the ANOVA), and so was the indirect effect, point estimate = 0.00, SE = .03, CI95% = -0.07, 0.07.

Discussion
With this study we intended to test possible reasons for the lack of difference in social value of BJW-P expression at a union found in the previous studies. We put forward several possibilities to account for this surprising result: intervening variables (perceived target stamina and excessive self-enhancement), the operationalization of BJW-P expression and the measure of target social in Study 2. We also explored In this study the target who expressed high BJW-P was judged as having higher social value than the targets who expressed low and moderate BJW-P. The higher the BJW-P expression, the more stamina the targets were judged to have and, as expected, the higher their predicted social value. Also as expected, the higher the targets were judged to self-enhance, the less their judged social value. Surprisingly, however, participants judged the target expressing low BJW-P as engaging the most in selfenhancement. In relation to the expression of moderate BJW-P, this study indicates that it is as devalued as the expression of low BJW-P in the trade union at least. This contrasts with previous research not using specific contexts, which found that the expression of moderate BJW-P was judged as valued as and more realistic than the expression of high BJW-P (e.g., Alves & Correia, 2010a). In the general discussion we will address this matter.
Again, political orientation did not predict judgments significantly.

General Discussion
We aimed to study the role of context in the expression of both BJW-G and BJW-P, which had been absent in this line of research (Alves & Correia, 2008;Testé & Perrin, 2013). Following Alves and Correia's (2010a) suggestion, we put forward and found that the social value of BJW-G expression is more context sensitive than that of BJW-P. This pattern was consistent in three Western European societies -France, Germany and Portugal -independently of participant political orientation.
Indeed, in Study 1 we found that participants expressed higher BJW-G to convey a positive image and expressed a lower BJW-G to convey a negative image of themselves at a bank. In Study 2 the target who expressed high BJW-G was judged Running head: THE SOCIAL VALUE OF BJW IN CONTEXT 32 more favorably than the target who expressed low BJW-G there. These patterns are consistent with the idea that expressing high BJW-G is valued in a context that is intimately associated with money, the ultimate symbol of social utility in Western, economically liberal societies (Cambon, 2006). Putting it differently, if one wants to be seen as having what it takes to be successful in such societies, expressing high BJW-G can be one such strategy.
In relation to the expression of BJW-G in the other contexts -a trade union, a NGO and an employment assistance institution -the patterns were, as expected, in the opposite direction. Specifically, in Study 1 the participants expressed lower BJW-G to convey a positive image, and higher BJW-G to convey a negative image there. In Study 2 the target who expressed high BJW-G was judged less favorably than the target who expressed low BJW-G there. This is consistent with the idea that it is necessary to adopt a discourse that does not support the status quo to achieve social/societal change.
Expressing low BJW-G can thus be a confrontational discourse used to promote change in the quest for more justice. In contexts which aim to achieve such change the expression of low, not high, BJW-G is valued.
These findings are relevant because they specify where expressing BJW-G can be valued or devalued, thus showing that the expression of BJW-G, not only that of BJW-P, may be used strategically (Alves & Correia, 2010b). These results are consistent with the idea presented in our footnote 1, according to which although it is not possible to ascertain whether nor why participants in different studies by Testé and Perrin (2013) evoked different contexts when responding, that seems a plausible explanation for the fact that the social value of BJW-G seemed ambivalent 6 . Indeed, in hindsight it seems surprising that in research conducted in Portugal, which did not specify contexts either, the expression of BJW-G was socially valued, not ambivalent. We should stress, however, that Correia (2008, 2010a) found that the social value of BJW-G expression was less strong that that of BJW-P. Our results also support that notion in two ways.
First, as expected, the expression of high BJW-P at a bank is even more valued, and that of low BJW-P is even more devalued that the expression of their BJW-G counterparts. This is in line with the idea that discourses that have individuals as their focus points are crucial in Western societies, for they uphold their main pillareconomic liberalism (Dubois, 1994). For instance, in these societies there is a preference for accounts of personal versus situational causality ("the norm of internality"), individual autonomy versus dependency ("the norm of self-sufficiency") (Dubois, 2005;Dubois & Beauvois, 2005; see also Bay-Cheng, Fitz, Alizaga, & Zucker, 2015;Jellison & Green, 1981;Kluegel & Smith, 1981). Our results suggest that the expression of high BJW-P, even more than the expression of high BJW-G, contributes to the maintenance of economic (neo)liberalism.
Second, the expression of high BJW-P is markedly valued across most of the presented contexts. Indeed this relative insensitivity of BJW-P social value to contextual shifts could be surprising taking that in contexts such as NGOs or employment support institutions individuals work with disadvantaged people. Someone who applies to work in those organizations and says their life is just could be seen as bragging and/or implicitly blaming the disadvantaged for their misfortunes. Nevertheless, the results in our research are more consistent with the notion that these individuals are judged as having what it takes to withstand the stress associated with working there.
The case of BJW-P expression at a trade union deserves a few lines. Indeed, Studies 1 and 2 indicated that the expression of BJW-P was neither valued (i.e., normative) nor devalued (i.e., counter-normative) in the three countries. In Study 3 we introduced changes that could potentially explain that pattern, among which the same operationalization of high and low BJW-P expression as in Alves and Correia (2008).
With this less extreme operationalization in comparison to that used in Study 2, participants attributed more social value to the target who expressed high versus low (and moderate) BJW-P. Although we did not manipulate orthogonally the two operationalizations, which would allow us to directly ascertain their influence on participant social value judgments, with the data available we think that BJW-P operationalization is the most plausible explanation for the difference. Indeed, as indicated in footnote 4, the pattern found in Study 3 can be observed with different combinations of social value items, including one that only comprises the items used in Study 2. We cannot ascertain why, contrary to the other contexts, participants only distinguished the social value of targets at the trade union with a less extreme operationalization of BJW-P expression. Speculatively, however, we propose that the norm of self-interest may explain it, in part at least.
According to Miller (1999), in Western societies "self-interest both is and ought to be a powerful determinant of behavior" (p. 1053; see also Ratner & Miller, 2001).
Although pro-social motives are likely important to explain activities in trade unions, we think that self-interest issues are more inherent there than at NGOs or at employment assistance institutions, because trade unionists are workers too. Someone who applies to work at a trade union and says their life is extremely just -as it seems to have been the case in Study 2 -may be perceived as neither identifying with the working class nor having a vested interest in the outcomes that trade unions may achieve. As a result, that person may be judged as not being much motivated or even comfortable to engage in the trade union activities. If we are right in our reasoning, displaying a normative BJW-P discourse is perceived as a more difficult task there than in the other contexts, one requiring a more precise degree of high BJW-P. On the one hand, if individuals express either a more or less extreme degree of low BJW-P, they will never be much socially valued (although they may be perceived as having vested self-interest, they lack stamina). On the other hand, if individuals express too much BJW-P, they also lose social value (possibly because they are perceived as lacking self-interest, hence, motivation). Even the expression of moderate BJW-P, which in previous research was judged positively (Alves & Correia, 2010a), has as little social value in this context as the expression of low BJW-P. Possibly, participants associated the expression of moderate BJW-P at trade unions with indecisiveness, which may be taken as evidence that the person is not suited to central activities in this context. In this conundrum, the expression of "reasonably" high BJW-P may be the solution in that context, for it indicates that speakers have some vested interested in the game -thus being motivated -and stamina to play it. Although we did not measure target self-interest in our research, Study 3 shows that perceived target stamina (and self-enhancement) statistically mediates the relation between BJW-P expressed and target social value.
As expected, the target expressing high versus low BJW-P was judged as having higher stamina (with the target expressing moderate BJW-P being in the middle), and higher stamina predicted higher target social value. Also as predicted, perceived selfenhancement predicted target social value negatively. Unexpectedly, however, the target expressing low BJW-P was judged as engaging more in self-enhancement than the targets expressing either moderate or high BJW-P. In other words, in comparison to the targets who expressed moderate or high BJW-P, the target who expressed high BJW-P was perceived as engaging in less (not more) self-enhancement, which in turn predicted  (Goffman, 1959), and express the injustice they feel they are being targets of.
This can be inferred in mundane practices, such as marriage vows ("to have and to hold from this day forward, for better, for worse") or in popular sayings turned into songs (e.g., "That's what friends are for/ In good times and bad times/ I'll be on your side forever more"; Bacharach & Sager, 1982). As far as the social value of BJW-G expression in these contexts is concerned, it may be contingent on what is normative to Western societies the social value of BJW-G is more context sensitive than that of BJW-P. The expression of higher degrees of BJW-G is socially valued in a context of social utility (bank). In organizational contexts that deal with disadvantage, the expression of higher BJW-G is counter-normative. In contrast, the social value of BJW-P is relatively resistant to contextual shifts, being even more normative than BJW-G at a bank. In trade unions it seems that displaying the approved of degree of BJW-P is more difficult than in other contexts. Even in the case of trade unions, however, the expression of higher degrees of BJW-P is judged as non-normative at most, but never counter-normative.

Compliance with Ethical Standards
Ethical approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Footnotes 1 Indeed, as reviewed, Testé and Perrin's (2013) results involving the expression of BJW-P and Gangloff and Duchon's (2010) results involving the expression of BJW-G in France are consistent with Alves and Correia's (2008;2010a) in Portugal. The divergent patterns involving the expression of BJW-G exist not only between countries, but also within France. Possibly, while responding, participants evoked different contexts for reasons we cannot ascertain now. If, as Alves and Correia (2010b) suggested, and we intend to show here, the expression of BJW-G is indeed contextsensitive, the various contexts that participants evoked may have influenced their responses differently. Although it is not possible to ascertain whether nor why participants in different studies evoked different contexts, that possibility seems more plausible than an explanation based on hypothetical "cultural differences" between France and Portugal (which would also have to predict deep cultural differences within  we intended to measure target social utility, that is the characteristics that Western, economically liberal societies evaluate as essential if their members are to become successful (e.g., autonomous, industrious, entrepreneurial), thus having market value (Beauvois & Dépret, 2008;Cambon, 2006) . With item 5 we intended to have a more behavioural measure. Exploratory factor analysis with Varimax rotation, however, .001, η 2 p = .12. When the targets applied for a job at a bank, those who expressed high vs. low BJW were judged as having higher social value (M = 4.55, SD = 1.21 vs. M = 3.16, SD = 1.16; p <.001). When they applied for a job at a union the pattern was reversed (M = 3.75, SD = 1.47 vs. M = 4.14, SD = 1.37; p = .03). There were no significant differences regarding the Human Rights NGO (p = .20) or the institution (p = .48). 5 We first conducted a factorial analysis with varimax rotation including all items (except the distractors). This resulted in a six-factor solution with eigenvalues higher than 1 explaining 69.70% of variance. We excluded: a) "If someone responds to the Running head: THE SOCIAL VALUE OF BJW IN CONTEXT 51 questionnaire like this in a trade union, they will convey a bad image of themselves."; "Responses similar to those of this person diminish the bargaining power of trade unionists.", which comprised (an uninterpretable) Factor 6 (with the latter item also having a high loading in Factor 3); and b) "In order to improve the labour situation, it is necessary to say something similar to what the person answered in the questionnaire", which had a high loading in Factor 5 only.
We then ran another factorial analysis with varimax rotation, which indicated a four-factor solution explaining 66.30% of variance. We aggregated Factors 1 and 3 (the latter comprising the five items used in Study 2) to calculate our measure of target social value. This aggregation is justified statistically by the fact that Factors 1 and 3 were highly correlated (r = .76) with three items having high loadings on both factors (≥ .49). Importantly, in terms of meaning, the items of both factors refer to the social value of the targets and of what they expressed. Finally, exploratory analyses comparing our social value measure versus a measure without Factor 3 items versus a measure comprising only Factor 3 items indicated similar results. 6 One could also argue that the ambivalence found in Testé and Perrin (2013) could be the result of their using Lipkus et al. (1996) BJW-G scale. We think this is a very unlikely possibility. Indeed, Alves and Correia (2008, 2010a used Dalbert et al's (1987) scale, which comprises very similarly worded items, and found no such ambivalence. Also, research that used the very differently worded scale by Rubin and Peplau (1975) arrived at conclusions that are in line with GBJW expression being socially valued, not ambivalent (e.g., Duchon & Gangloff, 2008;Gangloff, 2008).