The legitimation of wife abuse among women: The impact of belief in a just and

Past research has found that when victims are ingroup members, observers’ social identification interacts with general belief in a just world (GBJW) to predict judgments about those victims. In this correlational study (N= 284 women, ages from 18 to 80) we aimed to test whether and how women’s explicit endorsement of BJW, both personal belief in a just world (PBJW), and GBJW, interacts with their identification as women to predict wife abuse legitimization. We predicted and found that the interaction between PBJW and social identification predicted legitimization of wife abuse. Specifically, for highly identified women, PBJW was positively associated with wife abuse legitimization, for less identified women, PBJW was not associated with wife abuse legitimization. This interaction was significant above and beyond other variables associated with this phenomenon: hostile and benevolent sexism, empathy (cognitive and emotional), and social desirability. On the contrary, the interaction between GBJW and social identification was a nonsignificant predictor of legitimization of wife abuse. These results contribute to reconceptualize the role of PBJW and GBJW on judgments about victims and to highlight the importance of considering the victimization situations in the social context and the social groups in which they actually occur.


The legitimation of wife abuse among women: The impact of belief in a just world and gender identification
Wife abuse, defined as physical and/or sexual violence against women by their male partners, is a very common problem around the globe involving severe consequences for victims, their families and society as a whole. For example, studies conducted in ten countries (Bangladesh, Brazil, Ethiopia, Japan, Namibia, Peru, Samoa, Serbia and Montenegro, Thailand and the United Republic of Tanzania) estimated that lifetime prevalence rates of physical or/and sexual partner violence varied from 15% to 71% (Garcia-Moreno, Jansen, Ellsberg, Heise, & Watts, 2006). Other studies estimated such rates as varying between 17.4% and 25.5% in the USA (Malley-Morrison & Hines, 2004), and between 10% and 50% in Europe (Machado & Dias, 2008). However shockingly high these figures are, most authors sustain they may be underestimations of the actual rates (Machado, Dias, & Coelho, 2010).
In most Western countries wife abuse is a crime, thus being officially illegitimate. Nevertheless very often wife abuse is given some degree of unofficial legitimation in various social interactions (for a review, see Baker, Cook, & Norris, 2003). For instance, individuals recurrently justify wife abuse by attributing it to the victims' presumed negative actions or/and their bad character. In fact, the media very often consider abused wives as ultimately responsible for both being in that situation and for putting an end to it (Berns, 1999). Furthermore, these victims may also expect these unsupportive reactions from the formal or informal systems that are supposed to help them, for instance their families, the clergy, the police, the welfare, the shelters, the justice system, the courts, the helping professionals, medical doctors and nurses, and even other women (for a review see Machado, Dias, & Coelho, 2010). Instead, these victims often meet a decrease in (or even to the absence of) social support. Given that COMPARING PERSONAL AND GENERAL BJW 4 social support is crucial for individuals' physical and psychological well-being in general, this state of affairs is especially deleterious for victims (Cohen & Wills, 1985).
In this paper our goal is to deepen our understanding of how just world and gender identification processes can contribute to explain women's judgements of wife abuse legitimation.

Belief in a just world, social identification and legitimation of victimization
Many types of innocent victims face negative reactions from other people as if their suffering is fair and therefore legitimate (e.g., Lerner & Simmons, 1966). Just world theory (e.g., Lerner, 1980) has offered an explanation for this surprising and apparently perverse phenomenon. According to this theory, individuals legitimize the suffering of innocent victims in order to preserve their illusory but fundamental perception that the world is a just place where everyone gets what they deserve (Lerner, 1980). This may be accomplished through various ways, such as derogating or/and blaming the victim. By so doing, individuals are able to have confidence in their "fundamental delusion" that unjust events will be unlikely in their lives (Lerner, 1980). This pattern tends to be more visible among individuals endorsing a higher degree of BJW, and it occurs even when the participants themselves are the victims (e.g., Choma, Hafer, Crosby, & Foster, 2012;Hafer & Olson, 1989; for a review, see Hafer & Bègue, 2005). According to just world theory, this assimilation of injustices happening to either the self (Dalbert, 2001) or to other people (Lerner, 1980) derives from the threat that innocent victimization poses to individuals' BJW. This threat should be especially felt by high believers in a just world, who thus need to defend such worldview to a higher extent than low believers.
Just world research has also found evidence that social identity and social identification are important factors explaining the threat that innocent victims pose to individuals' BJW. Sharing a common identity with the victim is a potential cause of threat to one's BJW because ingroup members are more relevant than outgroup members in indicating what may happen to the self (Aguiar, Vala, Correia, & Pereira, 2008;Lerner & Miller, 1978;Novak & Lerner, 1968). This finding is in line with social identity theory (Tajfel & Turner, 1979). According to this theory when people categorize themselves as members of social groups, they define themselves in terms of their social identities rather than in terms of their unique personal characteristics.
Furthermore, self-categorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987) conceptualizes social identification as readiness to categorize the self as a member of a particular group in a certain context. This categorization accentuates intragroup similarities and intergroup differences. In turn, highly (versus low) identified individuals perceive other ingroup members in a more depersonalized way and as more interchangeable entities.
Applying the aforementioned reasoning to victimization cases, if individuals know that someone from their ingroup (versus an outgroup) suffers innocently, highly (versus low) identified members are more likely to believe that the same might occur to them. In fact, Correia et al. (2012) found that the interaction of participants' explicit endorsement of BJW and their identification with the identities shared with the victims predicted derogation and psychological distancing. Specifically, the positive relationship between BJW and those judgements was significant for strongly identified participants but nonsignificant for weakly identified participants.

The present study
In the current study female participants judged wife abuse legitimation. Based on Correia et al.'s (2012) findings, we may expect a joint effect of BJW and social identification on wife abuse legitimization. In this research we extend on previous work by making a distinction between the general belief in a just world (GBJW; Dalbert, Montada, & Schmitt, 1987) and the belief in a personal just world (PBJW; Dalbert, 1999) when examining the interaction of just world beliefs with social identification.
According to this distinction, GBJW indicates the degree to which individuals believe that people in general get what they deserve; whilst PBJW indicates the degree to which individuals believe that they themselves get what they deserve. The literature has shown that GBJW better predicts how individuals assimilate injustices happening to other people than PBJW (e.g., Bègue & Bastounis, 2003), and that PBJW better predicts how individuals react to injustices happening to themselves than GBJW (Correia & Dalbert, 2007;Dalbert 2001).
It is also important to notice that there is a crucial difference between the situations in Correia et al. (2012) and the one in this study. In fact, in Correia et al.
(2012) the identity of the victim and the victimization cases are not necessarily related (e.g., a university student that was run over by a car). On the contrary, in the case of wife abuse there is an intrinsic relation between being a women and being victim of wife abuse: the victimization situation is more likely to affect members of the perceivers' ingroup than members of an outgroup. Therefore, in the present study we measured the predictive value of GBJW and that of PBJW on the legitimation of wife abuse. In this study we aimed to test whether and how women's explicit endorsement of BJW (both PBJW and GBJW) interacts with their identification (with being a woman) to predict legitimization of wife abuse. More specifically, we predicted that for highly identified women, BJW would be positively associated with wife abuse legitimization. For those who were less identified, we expected that BJW would not be associated with wife abuse legitimization. Furthermore, in order to isolate these predicted effects from the effects of other significant variables in the processes of legitimization of wife abuse, we controlled for a number of relevant variables. First, we controlled for sexism as previous research has shown that this variable is a significant predictor of attitudes legitimizing wife abuse (Glick, Sakalli-Ugurlu, Ferreira, & Souza, 2002). Second, previous research suggests that empathy reduces victim blaming (Aderman, Brehm, & Katz, 1974;Haegerich & Bottoms, 2000) and we thus controlled for this variable. And finally, we also assessed and controlled for social desirability given that in such sensitive topics it is advisable to control for social desirability response bias (Saunders et al., 1987).

Method
Participants. Two hundred and eighty-four female participants voluntarily participated in this study (ages between 18 and 80, M = 35.93, SD = 15.47). They held a variety of occupations/ professions (students, teachers, managers, nurses, lawyers, accountants, social service professionals, commercial workers). Their highest level of education varied between 3 years of total education to holding a PhD degree (M years of schooling = 12.92, SD = 3.06). About 16% had at least partially completed the 9 th grade, about 32% had at least partially completed the 12 th grade, and about 52% had at least received a certain amount of higher education, including BAs and MAs.

Procedure and Measures.
When recruiting this sample we aimed at reaching a wide range of occupations and age groups in order to reflect the nuances in perspectives in society. Therefore, partly based on Glick and Fiske (1996, Study 6), in exchange for credit for a course, university students who volunteered for this study were invited to recruit around 6 adult females from their close circles. It was explicitly mentioned that they could include family members and friends. Apart from the demographics described above, no other data was recorded and, for confidentiality purposes, there is no information of the students' relationship with the sample.
The research was presented to women as a study aiming to validate new measures and at the end of the survey the participants were thanked and debriefed.
Personal belief in a just world. Personal BJW was measured with the PBJW Scale (Dalbert, 1999). The scale comprises seven items (e.g.: "I am usually treated fairly"; α= .86).
General belief in a just world. We measured this construct with the 6-item General Belief in a Just World Scale (Dalbert et al., 1987) (e.g., "I think basically the world is a just place"; α= .74).
Group identification. We used the 14 items of ingroup identification scale adapted from Leach et al.'s (2008;e.g., "I often think about the fact that I am a woman"; α = .86).
Hostile and Benevolent Sexism. We measured these constructs with the Ambivalent Sexism Inventory (Glick & Fiske, 1996). Hostile Sexism was measured with 11 items (e.g., "Women seek to gain power by getting control over men", α = .70); Benevolent Sexism was measured with 11 items (e.g. "Many women have a quality of purity that few men possess; α = .77).
Cognitive and emotional empathy. We measured these constructs with the Interpersonal Reactivity Index (Davis, 1980), a 28-item measure that has 7 items related to cognitive empathy (e.g., "I sometimes try to understand my friends better by imagining how things look from their perspectives."; α = .69), and 7 items related to emotional empathy (e.g., "I often have tender, concerned feelings for people less fortunate than me."; α = .63).
Legitimization of wife abuse. We measured this construct with thirty items from the Inventory of Beliefs about Wife Beating (Saunders, et al., 1987; e.g., "A husband has no right to beat his wife even if she breaks agreements she has made with him"; α= .90).
All measures, except social desirability, had 6-point scales ranging from 1 (totally disagree) to 6 (totally agree). We computed scores within each scale by averaging across items, with higher scores indicating stronger endorsement of the construct.

Results
First, we inspected the zero-order correlations among all variables. As can be seen in Table 1, PBJW and GBJW correlated positively and significantly. Wife abuse legitimation correlated significantly with empathy (cognitive and emotional), sexism (hostile sexism and benevolent sexism), and years of schooling correlated positively and significantly with age. Wife abuse legitimation did not correlate with social desirability. PBJW, GBJW, and gender identification alone did not correlate significantly with wife abuse legitimization.
We then tested whether identification with being a woman moderated the relationship between both BJW measures (GBJW and PBJW) and legitimization of wife abuse (the outcome variable), while controlling for the effects of both BJW measures, social identification, benevolent sexism, hostile sexism, cognitive and emotional empathy, and social desirability. As age and years of schooling correlated significantly with the main predictor variables (PBJW, GBJW and identification) as well as with the criterion variable (legitimation of wife abuse) they were also introduced in the regression.
We thus conducted a multiple regression analysis. In a first block we entered the socio-demographic (age, years of schooling) and the control variables: (benevolent sexism, hostile sexism, cognitive empathy, emotional empathy and social desirability).
In a second block, we entered GBJW, PBJW and social identification. In a third block, we entered the product between PBJW and social identification, and the product between GBJW and social identification. All variables were centered before analyses (Aiken & West, 1991).
The results are shown in Table 2. Twenty three percent of the variance in legitimization of wife abuse was explained by the main effects of age, years of schooling, emotional empathy and hostile sexism. Whereas age and hostile sexism predicted stronger legitimization of wife abuse, years of schooling and emotional empathy predicted lower legitimization of wife abuse.
Neither identification nor BJW alone (either PBJW or GBJW) predicted wife abuse legitimization. Importantly, a significant two-way interaction between PBJW and social identification, but not between GBJW and social identification, explained 2 percent of the variance in legitimization of wife abuse. This significant effect held over and above the main effects obtained in Block 1, which remained significant predictors after considering this interaction between PBJW and identification.

Discussion
In this study we extended research on the interaction between BJW and social identification by comparing the predictive power of PBJW and GBJW on the reactions to ingroup members' suffering, in this case women's legitimization of wife abuse. We did this controlling for age, highest level of education achieved, empathy, social desirability and ambivalent sexism. The results showed that BJW in fact interacts with identification, but only with PBJW and not with GBJW. Specifically, for highly identified women, PBJW was positively associated with wife abuse legitimization, whereas for less identified women no such association was found. This indicates that for high identifiers a higher degree of PBJW increases the threat to the BJW under conditions of high interchangeability between the perceiver and the victim, as it is the case of female participants judging the legitimization of wife abuse. Importantly, the interaction between PBJW and social identification was found over and above the effect of negative attitudes towards equality of women and emotional empathy, which puts into evidence the importance of those variables.
The study presented in this paper points to the importance of considering the social position of the group of the victims and the observers. In fact, the predictive value of PBJW or GBJW may be highly dependent on this matter. Until now researchers have assumed that GBJW would be the best predictor of reactions to the victimization of other people and have not included PBJW in studies that aimed to study reaction towards victims (e.g., Montada, 1998, for a review). The case may be different when the victim is from a low-status ingroup as in this study. In fact, women perceive themselves as a dominated group (Amâncio, 1989;Lorenzi-Cioldi, 1988), which means that they perceive themselves as undifferentiated elements (Deschamps, 1982). This suggests that when the victims are from a low status ingroup, and therefore there is a high perceived interchangeability between the participant and the victim, the ingroup perceivers will react to the victimization of others as they would react to the victimization of the self.
In line with previous studies (e.g., Kristiansen & Giulietti, 1990) negative attitudes towards equality of women predicted wife abuse legitimization. The fact that hostile sexism, but not benevolent sexism, predicted wife abuse legitimization supports and extends previous results with samples from other countries (Glick, et al., 2002).
This contributes to establish cross-cultural validity of previous results.
The fact that emotional empathy, but not cognitive empathy, predicted a lower legitimization of wife abuse gives further evidence that empathy is important to reduce victim blaming (Stel, van den Bos, & Bal, 2012). It also suggests that victim blaming may be more related with the emotional reaction of observers than to their capability of estimating other people's thoughts and feelings.
However, we must not forget that the correlational design of this study limits the nature of conclusions that can be drawn about the causal and sequential relations among belief in a just world, identification and legitimization of wife abuse. Despite this limitation, our predicted causal directions were much in line with those of previous work based on experimental studies (Correia et al., 2012).
We should also note that wife legitimation scores are generally low. This may have derived from our sample characteristics, even though it is diverse in terms of age, occupations, and years of schooling. Note that our scores in Figure 1 are based on an analysis of one standard deviation above and below our sample's mean scores. It is plausible to think that with another sample that scored higher in terms of identification with other women or PBJW, scores in wife abuse legitimization would be higher.
It could be interesting if future studies compared the degree of threat to BJW (Aguiar et al., 2008;Correia, Vala, & Aguiar, 2007) produced by an innocent ingroup victim under conditions of lower and higher perceived interchangeability with the observer. We expect that threat to BJW will be higher under conditions of higher perceived interchangeability than under conditions of lower perceived interchangeability between an observer and a victim. Furthermore, it is also possible to predict that a threat to BJW could be a mediator between the degree of perceived interchangeability and the negative reaction towards ingroup victims.
It goes without saying that the evidence presented here would benefit from further research with different victimization situations affecting the ingroup, different social identities and different samples of participants.
Nevertheless, we consider this study as an important step towards the reconceptualization of the functions of PBJW and GBJW, at least in Western societies (for a reconceptualization of BJW in China, see Wu et al., 2011, Wu et al., 2013. This study also highlights the importance of considering the victimization situations in the social context and the social groups in which they actually occur. Research about victimization must not continue to ignore this.  Table 2 Regression of legitimation of wife abuse on age, years of schooling, benevolent sexism, hostile sexism, cognitive empathy, emotional empathy and social desirability (block 1), personal BJW, general BJW and identification (block 2), and interaction between PBJW and identification and interaction between GBJW and identification (block 3). For all measures, scores were computed by averaging across items, with higher scores indicating stronger endorsement of the construct. For gender, 0 indicates "male" and 1 "female." * p < .05; ** p < .01; *** p < .001. High Identification (+1 SD)