Análisis de regresión múltiple

Análisis'de'regresión'múl/ple' Málaga,'26'de'abril'de'2016' Jesús'López'Megías' Departamento'de'Psicología'Experimental' Universidad'de'Granada' • Regresión'vs'ANOVA' • Análisis'de'interacciones'(moderaciones)'en' ARM'y'representación'gráfica:'' – Interacción'entre'una'variable'dicotómica'y'una' con/nua' – Interacción'entre'variables'con/nuas' – Interacción'en'la'que'hay'una'variable' categórica'con'más'de'dos'niveles' • Mediación'con'PROCESS' • Regresión'vs'ANOVA' – Ventajas'del'ARM:'' • Variables'con/nuas.'Peligros'de'dicotomizar:' pérdida'de'potencia'estadís/ca'(equivalente'a' perder'1/3[2/3'de'la'muestra)'y'efectos'espurios' (cuando'más'de'una'VI)' • Variables'correlacionadas:'ver'el'efecto'de'unas' controlando'el'de'las'demás'(p.e.'SH'y'SB)' – ARM'simultáneo.'Equivalencia'con'ANOVA.'' – ARM'jerárquico'(secuencial).'¿Qué'aporta'algo' nuevo'a'la'explicación'de'la'variabilidad'de'la' VD?' ' • Análisis'de'interacciones'(moderaciones)'en' ARM'y'representación'gráfica:'' – Interacción'entre'una'variable'dicotómica'y'una' con/nua' – Interacción'entre'variables'con/nuas' – Interacción'en'la'que'hay'una'variable' categórica'con'más'de'dos'niveles' • Análisis'de'Mediación'con'PROCESS' Journal of Interpersonal Violence 1 –17 © The Author(s) 2016 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0886260515624212 jiv.sagepub.com Article Male Peer Support to Hostile Sexist Attitudes Influences Rape Proclivity Mercedes Durán,1 Jesús L. Megías,2 and Miguel Moya2 Abstract Sexual assault affects a large proportion of women in the world. Although most rapes are committed by one man, the act itself may be influenced by many (e.g., the peer group). Hostile sexism (HS) has repeatedly been associated with men’s rape proclivity, but the influence exerted by the HS of the peer group on rape proclivity has not been investigated. In this study, we explored the impact of perceived male peer support to HS on participants’ rape proclivity. A sample of Spanish undergraduate students from a university in the south of Spain (N = 134) completed the Ambivalent Sexism Inventory. Immediately afterwards, they received feedback on the supposed sexist responses of a peer group (high vs. low in HS); we kept the benevolent sexism (BS) of the peer group at medium levels. Next, we assessed participants’ rape proclivity using acquaintance rape scenarios. Results showed an interaction between participants’ own levels of HS and information about the HS of the peer group. Men high in HS reported higher rape proclivity in the high-HS peer-group condition than in the low-HS peer-group condition. By contrast, information on the peer group did not affect self-reported rape proclivity of men low in HS. Results also corroborated the relationship between participants’ levels of HS and rape proclivity, and expanded the literature by revealing an unexpected influence of participants’ BS on rape proclivity. 1Universidad de Sevilla, Spain 2Universidad de Granada, Spain Corresponding Author: Mercedes Durán, Facultad de Psicología, Departamento de Psicología Social, Universidad de Sevilla, Camilo José Cela, S/N, 41018, Sevilla, Spain. Email: [email protected] 624212JIVXXX10.1177/0886260515624212Journal of Interpersonal ViolenceDurán et al. research-article2016 at Biblioteca Universitaria de Granada on January 17, 2016jiv.sagepub.comDownloaded from Interacción'entre'dicotómica'y'con/nua' 4 Journal of Interpersonal Violence In relation to rape proclivity, it has been found that acceptance of rape myths is one of the most robust factors that contributes to rape proclivity (Bohner, Eyssel, Pina, Siebler, & Viki, 2009; Bohner, Jarvis, Eyssel, & Siebler, 2005). In this regard, previous studies have provided evidence of the influence of acceptance of rape myths by peers on self-reported rape proclivity. Bohner, Siebler, and Schmelcher (2006) provided participants with false information about levels of rape myth acceptance among peers. Results showed that participants with high rape myth acceptance who had been informed that their peers also endorsed such myths reported the highest rape proclivity. Thus, Bohner et al. consolidated the idea that rape myth acceptance can play a causal role in the occurrence of sexual violence against women. Prior literature has provided evidence of the relationship between rape myth acceptance and sexist attitudes (Chapleau, Oswald, & Russell, 2007). However, the possible influence of the sexist attitudes of the peer group on individuals’ rape proclivity has not been empirically studied yet. In this sense, studying HS in addition to rape myths is particularly important because HS represents a prejudice whose content expresses hostility toward women who represent a threat for men, but its content does not directly refer to sexual assault or to sexually assaulting women. If we obtain similar results than Bohner et al. (2006), it means that HS of the peer group have the power to facilitate men’s behavioral tendencies to sexual aggression. Men who endorse hostile sexist precepts may be more permeable to the influence of the HS of their peer group and exhibit the highest levels of selfreported rape proclivity. That is, as has been shown by previous studies (Bohner et al., 2006; Durán et al., 2010), the interaction between variables of the perceivers themselves and of the context or situation related to the assault may provide the best explanation of men’s sexually aggressive tendencies. Improving our knowledge of this would be particularly important to be able to design and develop campaigns for the prevention of violence against women (Berkowitz, 2010) aimed not only at eliminating the sexist attitudes of young or adolescent men but also at involving their closest peers (classmates, friends, etc.). The Current Research The aim of the present research was to study the impact of perceived HS as a peer-group variable on men’s self-reported rape proclivity. We analyzed whether men’s self-reported proclivity toward sexual violence, which is affected by their own hostile sexist beliefs, is also influenced by perceptions of other men’s support to HS. In other words, this study adds to the literature at Biblioteca Universitaria de Granada on January 17, 2016jiv.sagepub.comDownloaded from Durán et al. 9 revealed that the provision of information about the HS of the peer group influenced participants high in HS, β = .32, t = 2.86, p < .01, not those low in HS, β = .05, t = .47, n.s. As regards the interaction between participants’ BS and information about the HS of the peer group, as shown in Figure 2, we found a totally different effect to that described in Figure 1. Tests of simple slopes (Aiken & West, 1991) revealed that information on the peer group did not affect selfreported rape proclivity of men high in BS, β = .04, t = .35, n.s.; by contrast, such information influenced the self-reported rape proclivity of men low in BS, β = .37, t = 3.07, p < .01. That is, as shown in Figure 2, participants who least endorsed benevolent sexist precepts and had received information about the high HS of their peer group reported the highest rape proclivity. In short, participants high in HS and those low in BS showed the highest levels of self-reported rape proclivity when they had received information on high HS in their peer group (compared with information on low HS in such group). Discussion This study addresses the question of how male participants’ ambivalent sexist attitudes as well as perceived information about the peer group’s support of HS influence participants’ self-reported rape proclivity. That is, the aim of this study was to determine the influence of perceived HS levels of the peer group on men’s self-reported rape proclivity and gain further insight on the relationship between men’s HS and their rape proclivity. Replicating previous findings, HS was found to be related to men’s sexually aggressive tendencies. This supported Hypothesis 1. Our results were Table 2. Hierarchical Regression Equations With Rape Proclivity as the Dependent Variable. Variable ∆R2 Adjusted ∆F SE βt Step 1 .23 12.83*** .89 HS .07 .45 5.48*** BS .09 −.26 −3.09** Peer-group HS .15 .19 2.48* Step 2 .04 2.49* .88 HS × Peer-group HS .15 .24 2.20* BS × Peer-group HS .18 −.27 −2.04* Note. HS = hostile sexism; BS = benevolent sexism. *p < .05. **p < .01. ***p < .001. at Biblioteca Universitaria de Granada on January 17, 2016jiv.sagepub.comDownloaded from 10 Journal of Interpersonal Violence Figure 2. Self-reported rape proclivity as a function of information about the HS of the peer group and the BS of the participant. Note. Values represent participants’ self-reported rape proclivity scores resulting from the mean of the last two items of the rape proclivity measure proposed by Bohner et al. (1998). Maximum value = 5. HS = hostile sexism; BS = benevolent sexism. Figure 1. Self-reported rape proclivity as a function of information about the HS of the peer group and the HS of the participant. Note. Values represent participants’ self-reported rape proclivity scores resulting from the mean of the last two items of the rape proclivity measure proposed by Bohner et al. (1998). Maximum value = 5. HS = hostile sexism. at Biblioteca Universitaria de Granada on January 17, 2016jiv.sagepub.comDownloaded from 10 Journal of Interpersonal Violence Figure 2. Self-reported rape proclivity as a function of information about the HS of the peer group and the BS of the participant. Note. Values represent participants’ self-reported rape proclivity scores resulting from the mean of the last two items of the rape proclivity measure proposed by Bohner et al. (1998). Maximum value = 5. HS = hostile sexism; BS = benevolent sexism. Figure 1. Self-reported rape proclivity as a function of information about the HS of the peer group and the HS of the participant. Note. Values represent participants’ self-reported rape proclivity scores resulting from the mean of the last two items of the rape proclivity measure proposed by Bohner et al. (1998). Maximum value = 5. HS = hostile sexism. at Biblioteca Universitaria de Granada on January 17, 2016jiv.sagepub.comDownloaded from Journal of Interpersonal Violence 1 –20 © The Author(s) 2015 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0886260515586366 jiv.sagepub.com Article More Than a Magazine: Exploring the Links Between Lads’ Mags, Rape Myth Acceptance, and Rape Proclivity Mónica Romero-Sánchez,1 Virginia Toro-García,1 Miranda A. H. Horvath,2 and Jesús L. Megías1 Abstract Exposure to some magazines aimed at young male readers—lads’ mags—has recently been associated with behaviors and attitudes that are derogatory toward women, including sexual violence. In the present study, a group of Spanish adult men was exposed to the covers of a lads’ mag while a second group was exposed to the covers of a neutral magazine. Results showed that, compared with participants in the second group, participants who were exposed to covers of lads’ mags who also showed high rape myth acceptance and legitimized the consumption of such magazines reported higher rape proclivity in a hypothetical situation. These findings suggest the need to further explore the possible detrimental effects of some sexualized media that are widely accepted in many Western countries. Keywords sexualization of the media, lads’ mags, rape myths, rape proclivity 1University of Granada, Spain 2Middlesex University, London, UK Corresponding Author: Virginia Toro-García, Faculty of Psychology, Department of Social Psychology, Research Center “Mente, Cerebro y Comportamiento” (CIMCYC), University of Granada, Campus Cartuja, s/n 18071, Granada, Spain. Email: [email protected] 586366JIVXXX10.1177/0886260515586366Journal of Interpersonal ViolenceRomero-Sánchez et al. research-article2015 by guest on June 7, 2015jiv.sagepub.comDownloaded from Romero-Sánchez et al. 9 Design and Procedure The variable “type of magazine” (Interviú vs. Muy Interesante) was experimentally manipulated between groups. We also measured two variables of individual differences: RMA and perceived legitimacy of lads’ mags. We used a convenience sampling design, inviting any students in the various libraries of the University of Granada to participate voluntarily in a study on the opinions and attitudes of college students on various issues. Instructions guaranteed that responses would be anonymous and confidential and only used for research purposes. Participants who agreed to cooperate were randomly assigned to one of the two experimental conditions: the lads’ mags condition (Interviú) or the neutral condition (Muy Interesante). Each of the questionnaires handed to participants included—in this order—the instructions, the Spanish short version of the AMMSA scale, the experimental manipulation, the manipulation check item, the self-report measure of rape proclivity, a question on participants’ reading habits regarding lads’ mags, the measure of perceived legitimacy of lads’ mags, and finally, the items assessing demographic data. The questionnaire was collected once participants had responded to all the questions. Participants were thanked for their cooperation and debriefed on the aims of the study and the way of accessing the final results. Results Preliminary Analyses First, we conducted a principal components analysis (PCA) with Varimax rotation to determine the factor structure of the four-item measure of the Perceived Legitimacy of Lads’ Mags scale. The Kaiser-Meyer-Olkin measure of sampling adequacy was found to be .73 (“adequate,” according to Field, 2009). Results of Bartlett’s test of sphericity, χ2(6) = 77.73, p < .001, confirmed that the correlation matrix was also appropriate for analysis. The PCA revealed one single factor with an eigenvalue of 2.41 that explained 60.38% of the variance. Therefore, we calculated the mean score of the four items as a measure of legitimacy of lads’ mags according to each participant. Correlations Among Variables Table 1 shows the correlations among participants’ scores on the AMMSA scale and the measures of perceived legitimacy of lads’ mags, reading frequency of lads’ mags, and self-reported rape proclivity. It also contains the means and SDs for the total sample and for control condition and experimental condition separately. As expected, significant correlations were found by guest on June 7, 2015jiv.sagepub.comDownloaded from Interacción'entre'variables'con/nuas' 12 Journal of Interpersonal Violence 1 1,5 2 2,5 3 3,5 4 4,5 5 Muy InteresanteInterviu Rape Proclivity (1) High RMA, High Legitimacy (2) High RMA, Low Legitimacy (3) Low RMA, High Legitimacy (4) Low RMA, Low Legitimacy Figure 1. Rape proclivity as a function of level of RMA and scores in perceived legitimacy of lads’ mags in participants exposed to a general magazine (Muy Interesante) versus a lads’ mag (Interviú). Note. RMA = rape myth acceptance. men’s self-reported rape proclivity, but also interacted with the type of magazine and perceived legitimacy of lads’ mags. More specifically, when exposed Table 2. Rape Proclivity Self-Reported as a Function of Type of Magazine, RMA, and Legitimacy of Lads’ Mags. Rape Proclivity Coefficient SE t Intercept i11.68*** .09 17.86 Magazine (X’)b10.06 .13 0.49 RMA (M’) b20.28** .09 3.19 Legitimacy lads’ mags (W)b3−0.03 .10 −0.37 Magazine × RMA (X’M’) b4−0.06 .12 −0.49 Magazine × Legitimacy lads’ mags (X’W’) b50.40** .14 2.76 RMA × Legitimacy lads’ mags (M’W’) b6−0.07 .09 −0.71 Magazine × RMA × Legitimacy lads’ mags (X’M’W’) b70.25* .12 2.04 R2 = .38, F(7, 106) = 9.408, p < .001 Note. Only significant results are reported. Type of magazine: 0 = Muy Interesante; 1 = Interviú. RMA = rape myth acceptance. *p < .05. **p < .01. ***p < .001. by guest on June 7, 2015jiv.sagepub.comDownloaded from Cuando'la'categórica'/ene'más'de'dos'niveles' ORIGINAL ARTICLE Social Perception of Rape Victims in Dating and Married Relationships: The Role of Perpetrator’s Benevolent Sexism Mercedes Durán &Miguel Moya &Jesús L. Megías & G. Tendayi Viki Published online: 3 July 2009 #Springer Science + Business Media, LLC 2009 Abstract Two hypothetical scenario studies examined how situational, perpetrator, and observer factors affect blame towards rape victims. In Study 1, Spanish high school students (N=206) read about a rape committed by a boyfriend or husband who was described as benevolently sexist or not. Study 2 (N=201 British college students) replicated and extended Study 1 by adding a condition in which the rapist was described as a hostile sexist. In both studies, participants’benevolent sexism scores predicted more victim blame when the rapist was described as a husband (but not a boyfriend) who held benevolently sexist attitudes. Study 2 showed that participants’hostile sexism scores predicted more victim blame when the rapist was described as a hostile sexist. Keywords Ambivalent sexism .Intimate relationships . Rape .Victim blame Introduction Research on the social perception of rape has shown a frequent pattern of blaming the victim and exonerating the perpetrator, especially when the aggression is perpetrated by a person who is close to the woman (i.e. an acquaintance or intimate partner such as a boyfriend or husband). It has been shown in studies conducted in different countries, for instance United States (e.g., Bell et al. 1994; Yamawaki 2007); England (e.g., Abrams et al. 2003; Viki and Abrams 2002); Australia (e.g., Newcombe et al. 2008), and Spain (e.g., Durán and Moya 2007; Frese et al. 2004). Many factors seem to influence this distorted perception of the crime and the negative reactions directed at the victim. These can be categorized as observer or situation related factors. In this paper, two studies using different samples (Spanish high school students and British college students) investigated the role of one observer variable (ambivalent sexist ideology) and two rape-situational factors (perpetrator’s sexist ideology and type of intimate relationship between victim and perpetrator) in victim blaming. A particular focus was to study the effects of describing the male perpetrator of a rape occurring within a romantic relationship as having benevolently sexist attitudes (versus no information about his sexist attitudes). We conducted our studies in Spain and England to see whether similar results would be obtained and generalized to samples from other nations. We also used university students (UK) and adolescents (Spain) to see if our result would generalize across different age groups. In “Study 1a”, after reading a hypothetical marital-rape scenario or a boyfriend rape scenario, we tested the M. Durán Department of Social Psychology, Universidad de Granada, Granada, Spain M. Moya Department of Social Psychology, Universidad de Granada, Granada, Spain J. L. Megías Department of Experimental Psychology, Universidad de Granada, Granada, Spain G. T. Viki Department of Psychology, University of Kent at Canterbury, Canterbury, England M. Moya (*) Department of Social Psychology, Facultad de Psicología, Campus de Cartuja S/N, 18011 Granada, Spain e-mail: [email protected] Sex Roles (2010) 62:505–519 DOI 10.1007/s11199-009-9676-7 Results and Discussion We extracted principal components for our three items and their criteria consistently suggested a single-factor solution that explained 83.58% of the total variance. This scale also showed high internal consistency, α=.90. Accordingly, these items were averaged to provide a perceived positivity score for each participant. In order to see how the information given to participants affected the perceived positivity of men, we ran a one-way ANOVA with type of man (benevolent, hostile or control) as the independent variable and the perceived positivity scale as the dependent variable. This analysis showed a significant main effect of type of man, F(2, 71)=28.47; p<.001. Post hoc Tukey tests revealed that the perceived positivity ratings were significantly higher for the ordinary man (M= 10.7), than for the BS man (M=8.9) and for the HS man (M=5.52) (all p s <.05). These results demonstrated that a BS man was indeed perceived more negatively than an ordinary man although more positively than an HS man. Thus, we can conclude that the specific characteristics of BS information, rather than its general positivity, underlie the main result found in “Study 1a”. Study 2 In “Study 1a”, we obtained an interesting (but unexpected) main effect of HS participant ideology on victim blame. Study 2 was designed to explore if participants’HS could also influence participants’judgments when it was presented as a situational variable, that is, as a perpetrator characteristic. We examined this question so that we could expand our knowledge about the influence of perpetrator’s sexist ideology information on victim blaming. In “Study 2”, we also tried to replicate the main results of Study 1a. Therefore, we analyzed victim blaming when the perpetrator was described as a BS perpetrator, as an HS perpetrator and in a control condition (no information given about perpetrator’s sexist ideology). We also differentiated between a perpetrator who was the victim’s husband and one who was the victim’s boyfriend. Finally, we conducted Study 2 in England with college students to see if Study 1a’s results (with Spanish high school students) would generalize to a sample from another nation and of a different age range. Method Participants In this study, 201 British psychology students participated (men = 97 and women = 104) from the University of Kent at Canterbury. Participants ranged from 18 to 66 years of age, (men: M=27.09 and SD=11.3; women: M=22.6; SD=6.92). Design This study used a similar between-subjects design to that in “Study 1a”, the only difference being the new condition with HS information about the perpetrator (perpetrator-HS). Thus, the independent variables were gender, information about perpetrator’s ideology (perpetrator-BS vs. perpetrator-HS vs. no ideological information –control-) and type of relationship between victim and perpetrator (boyfriend/girlfriend vs. married). Participant’s attribution of victim blame was the dependent variable. Each questionnaire incorporated one of the six rape scenarios designed to insert experimental manipulations. We employed the same four scenarios used in “Study 1a”, translated into English, and added two more vignettes including the information about the perpetrator’sHS ideology (see “Appendix B”). In order to describe the perpetrator as an HS man, we used the eleven items of the HS subscale of the ASI (Glick and Fiske 1996). Also, the main characters’names were replaced by English names. Materials We used the background questionnaire and measures used in “Study 1a”. In this study, we obtained a Cronbach’s alpha of .94 for the entire ASI (Glick and Fiske 1996), similar to that obtained in “Study 1a”. Cronbach’s alpha for HS = .92 and for BS = .90. We also used the same victim blame scale as in Study 1. We presented the items in a 7point, Likert-type rating scale with response options ranging from 1 (strongly disagree) to 7 (strongly agree). This scale showed high internal consistency, Cronbach’s alpha of .86, similar to that obtained in “Study 1a”. Procedure All participants were approached while they were studying in a library and asked if they would answer different questions concerning relationships between men and women. Participants completed the questionnaire, in approximately fifteen minutes. Instructions ensured that their responses would be both anonymous and confidential and used only for research purposes. Each questionnaire incorporated one of the six rape scenarios designed to insert experimental manipulations. Firstly, participants answered the ASI. Then, they read a hypothetical rape scenario in which a man, after having an argument with his partner, raped her. Finally, they answered manipulation check items and offered their judgments about victim 512 Sex Roles (2010) 62:505–519 responsibility. After completing the questionnaire they were thanked and debriefed. Other aspects of the procedure were similar to those described in “Study 1a”. Results Preliminary Analyses Firstly, the relationships between all variables were investigated. HS and BS were significantly positively correlated (r=.60; p<.01), similar to previous findings (Glick et al. 2000). In line with previous research (e.g., Abrams et al. 2003) and with the results of Study 1a, victim blame was positively correlated with participants’BS (r=.29; p<.01) and participants’HS (r=.46; p>.01), replicating the effects found in “Study 1a”. The results confirmed the validity of our experimental manipulations. With regard to the question, “What type of relationship do the characters of the story have? (boyfriend/girlfriend vs. marriage)”, 100% of people who read the scenario in which we described a boyfriend/ girlfriend relationship identified this correctly, and 97% of people who read a scenario in which we said that the characters maintained a marital relationship also answered this question in the correct way, χ 2 (1)=231.3, p<.001. With regard to the questions, “Do you think that the man of the story holds a positive attitude towards women?”and “Do you think that the man of the story holds a negative attitude towards women?”, the ANOVA showed that people thought the benevolent sexist perpetrator (M=5.92) held a more positive attitude towards women than the hostile sexist perpetrator (M=2.45), F(1,160)=365; p<.001; also, participants said that the hostile sexist perpetrator held more negative attitudes toward women (M=5.45) than the benevolent sexist perpetrator (M=1.81), F(1,160)=441.4; p<.001. Victim Blame Hierarchical regression analysis was performed to test the impact on victim blame of participant’s gender, perpetrators’beliefs (perpetrator-BS, perpetrator-HS and control), relationship between victim and perpetrator in the scenario (married, boy/girlfriend) and participant’s BS and HS. BS and HS variables were centered prior to the analysis (Jaccard et al. 1990). We created dummy variables to yield two contrasts: the first contrast (perpetrator-BS = 1, perpetrator-HS = −1, control = −1) compared the benevolent perpetrator condition to the two other conditions; the second contrast (perpetrator-BS = −1, perpetrator-HS = 1, control = −1) compared the hostile perpetrator condition to the two other conditions. Together, these contrasts tested the effects of the information about the perpetrators’sexist ideology. In the first step, we tested the main effects of all variables. In the second step, two-way interactions were entered, in the third step, three-way interactions, in the fourth step, four-way interactions. Because there was insufficient power to examine 5-way or 6-way interactions, we did not proceed further. In the first step, the overall equation was significant, F(6,200)=12.79, p<.001; significant main effects for Gender (β=.25, t=4.00, p<.001) and participants’HS (β=.38, t=4.84, p<.001) were obtained. Men blamed the victim more than did women (M=2.26, SD=1.07 and M= 1.68, SD=.54 respectively). The higher participants scored on HS, the more they blamed the victim. In the second model, the interaction terms Gender x participants’HS and participants’HS x perpetrator-HS had a significant effect on the victim blame measure (t(200)= 3.44, β=.43, p<.01; t(200)=2.39, β=.22, p<.05, respectively). With regard to the first interaction, Gender x participants’HS, tests of simple slopes (Aiken and West 1991) revealed that men’s HS scores were significantly related to victim blame, t(96)=6.51, β=.55, p<.001, but not women’s HS scores, t(103)=1.46, β=.14, p=.15. The second interaction involving participant’sHSand perpetrator-HS, showed that participants’HS predicted victim blame in both cases, when the perpetrator was a BS perpetrator or when no beliefs information was given, t (134)=3.44, β=.29, p<.01, and when the perpetrator was described as a HS perpetrator, t(65)=7.50, β=.68, p<.001. Nevertheless, the influence of participants’HS scores was stronger in the perpetrator-HS condition, as can be seen in Fig. 2. Finally, in the third model, the three-way perpetrator-BS x Relationship x participant’BS interaction was significant, t(200)=3.17, β=.29, p<.01, as previously found in Study 1. To interpret this interaction, we analyzed the influence of 0 0.5 1 1.5 2 2.5 Low HS High HS Participants' sexism HS partner BS/Contr partner Fig. 2 Victim blaming as a function of perpetrator’s sexist beliefs and participant’s HS. Sex Roles (2010) 62:505–519 513