Abstract
Intimate partner violence (IPV) is a prevalent problem worldwide. Friends of perpetrators may be in a unique position to support or affect change. However, little is known about the influence of friendship with a perpetrator on responses to IPV. Social identity theory describes an ingroup bias whereby ingroup perpetrators of violence are viewed as less personally responsible than outgroup perpetrators. This bias has been consistently found for impersonal ingroup relationships, but there is limited research in relation to friends of perpetrators. Drawing on social psychological theories, this study aimed to explore the impact of friendship with a perpetrator on responses to IPV—specifically, on attributions of causality and social rejection. A fictional vignette depicting IPV perpetrated by either a friend or a stranger was presented to 174 university students, who then completed a questionnaire on attributions and social rejection. Results indicated that participants attributed high blame to the perpetrator regardless of their relationship, but friends of the perpetrator were significantly more likely than strangers to attribute the cause of the violence to external factors. Friends of perpetrators were likely to continue the friendship, though social rejection was significantly more likely when the perpetrator was attributed high blame and internal causality. Ingroup bias was not consistently present across all outcomes, demonstrating the complexity of social relationships and IPV. The findings suggest expectancy based on past behavior may influence attributions for violence in existing relationships. The combination of high blame, external attributions, and low social rejection was discussed in relation to opportunities for friends to intervene to prevent IPV. The multifaceted influence of friendship on responses to IPV perpetration suggests the need to consider relationship factors when designing violence prevention campaigns and bystander intervention programs.
Keywords
Intimate partner violence (IPV) is a form of abuse considered to be a significant global health problem by the World Health Organization (2013). IPV is defined as a multifaceted system of abuse by a romantic partner, including physical, sexual, emotional, financial, and psychological violence that crosses socio-economic groups, ages, and cultures (Phillips & Vandenbroek, 2014). The consequences of IPV can be devastating, often leading to severe mental health problems and, in many cases, death (Dunkley & Phillips, 2015).
Recent approaches to violence prevention focus on the role that community members can play in shifting culture and responding to incidents, with evidence suggesting that social systems can effectively reduce levels of IPV (Banyard, 2011; Browning, 2002). While perpetrators of IPV are unlikely to seek professional help to address their behavior, their close social supports including friends and family may be aware of the situation (Ashley & Foshee, 2005; Emery, Wu, Kim, Pyun, & Chin, 2017). Studies have suggested that informal social control by family members may be a protective factor in reducing severity of IPV, and that perceived neighborhood social control reduces the likelihood that a perpetrator will respond to a relationship problem with violence (Emery et al., 2017; Emery, Yang, Kim, Arenas, & Astray, 2016).
An online search for “my friend is abusing his or her boyfriend/girlfriend” reveals thousands of people giving and seeking advice. Friends are potentially in a unique position to influence peers through intervention and through setting social norms and acceptance of abusive behavior (Berkowitz, 2003). Despite the awareness that acquaintances play a key role in responding to IPV, little is known about how close relationships with perpetrators affect responses to IPV. Past research suggests ingroup relationships may influence perceptions of IPV perpetrators (e.g., Harrison & Abrishami, 2004). This article used social identity theory to explore the influence of friendship on responses to perpetrators of IPV.
Ingroup Bias and Attributions for Violence
Ingroup bias may influence perceptions of violent acts perpetrated by friends. Tajfel and Turner’s (1979) social identity theory proposed that social relationships are important in developing a healthy self-image. As such, individuals are motivated to protect and enhance self-esteem through positive group membership. The need to maintain this positive social identity produces a tendency to favor ingroup over outgroup members in a variety of contexts.
Ingroup bias is often measured through examination of attributions, the process of generating causal explanations for behavior and events (Weiner, 1995). Pettigrew (1979) labeled the bias in attribution the ultimate attribution error, and found it was evident when explaining negative events. A review of the literature found that ingroup members who performed a negative behavior were perceived as less influenced by internal, dispositional factors than an outgroup member who carried out the same action (Hewstone, 1990). This pattern of attributions can serve to reduce personal responsibility for negative behavior.
Evidence for the ingroup bias is present in several studies concerning perceptions of violence. Hunter, Stringer and Watson (1991) examined how group membership impacted on attributions for violent behavior. They identified an ingroup bias when Catholic and Protestant students in Ireland viewed a violent act performed by a person of the same or other religion. Violent behavior by an ingroup member was attributed to external, situational factors like self-defense. In contrast, outgroup violence was attributed to internal, dispositional factors like psychopathic tendencies. This pattern of attributions has been replicated in research with ingroups defined by political affiliation (Communist or Fascist; Schruijer et al., 1994) and race (Black or White; Duncan, 1976).
Harrison and Abrishami (2004) found further support for ingroup bias in the context of IPV. In their study, group membership was defined by attendance at the same or a rival university. They found that IPV offenders from the same university were attributed less guilt for their offense, with the cause of the event attributed more to external factors. The influence of ingroup membership was associated with a small effect size; however, the authors predicted that more cohesive ingroup relationships may produce a larger effect.
The Influence of Friendship on Responses to Violence
Friendship is likely to be a more reciprocal and entitative ingroup relationship than general category-based ingroups explored in past research (e.g., university membership, Harrison & Abrishami, 2004; Tajfel & Turner, 1979). The influence of friendship on attributions for violence has not yet been considered. However, friendship has been explored in bystander intervention research on campus sexual assault and IPV.
Relationship with a perpetrator has been found to have an influence on intentions to respond to IPV, although outcomes were mixed. Burn (2009) found students were more likely to intervene to prevent a friend than a stranger from committing a sexual assault. Bennett, Banyard and Edwards (2017) found that knowing the perpetrator of a sexual assault (rather than knowing the victim or knowing neither) was associated with lower intentions to help the victim but higher likelihood of confronting the perpetrator.
In contrast, Palmer, Nicksa, and McMahon (2018) found knowing the perpetrator in a hypothetical IPV scenario was related to selecting indirect interventions such as causing a distraction, rather than directly addressing the problem by confronting the perpetrator or seeking assistance. Branch, Richards and Dretsch (2013) found that while most college students would intervene to prevent IPV if their friend was a victim or a perpetrator, far fewer friends of perpetrators would report the crime to authorities. Bennett and Banyard (2016) found that a friendship with the perpetrator decreased perceptions of severity of a sexual violence scenario, but increased perceptions of safety to intervene.
Overall, the findings are mixed, with some suggesting friendship increased the likelihood of appropriate responses and others suggesting it may present a barrier to effective intervention. It is clear that further research is required to disentangle the complex influence of friendship on responses to IPV.
Social Rejection
Past research has considered numerous aspects of responses to ingroup perpetrators of IPV. A hitherto unexplored area relates to the effect that IPV perpetration may have on an individual’s friendship with a perpetrator. The knowledge that a friend is a perpetrator of IPV may change the dynamics of the relationship, depending on the perceived acceptability of the behavior. Ingroup bias predicts that people are likely to maintain a friendship despite negative behavior, using cognitive distortions to resolve cognitive dissonance (Bennett & Banyard, 2016; Festinger, 1962). However, in cases where the behavior is judged as an unacceptable violation of group norms, ingroup members may be rejected. This exception to ingroup bias is known as the black sheep effect (Marques, Yzerbyt, & Leyens, 1988).
The possible interventions available to the bystander in the short and long term may be influenced by whether they terminate or continue the friendship. This information is relevant to bystander education programs designed to support effective interventions for IPV. Research is required to address the likelihood of social rejection among friends of IPV perpetrators.
The Current Study
The current study explored the influence of friendship with an IPV perpetrator on causal attributions and social rejection. It was designed to address gaps and clarify inconsistencies in the current literature. Attributions of causality were investigated in line with past research on ingroup bias and violence, with the aim of adding further to the knowledge on how friendship influences responses to IPV. Social rejection was investigated to generate practical information to use when designing bystander intervention programs.
It is noted that gender of the observer, perpetrator, and victim has been found to significantly influence attributions for IPV (for a discussion of gender effects, see Worthen & Varnado-Sullivan, 2005). As such, the current study controlled for the influence of gender in all analyses so as to focus on the role of the relationship.
The following hypotheses and research questions were addressed:
Method
Participants
Undergraduate students were recruited by convenience sampling from the University of Canberra, a mid-sized Australian university. The sample consisted of 174 adults (35 male, 124 female). Fifteen participants were excluded on the basis of incompletion of the survey. Students from all faculties were invited to participate; however, most respondents (82%) were introductory psychology students. The age of the sample was positively skewed, consistent with an undergraduate population (M = 23 years, SD = 7.40, range = 18-73 years). Of the final sample, 94.3% were Australian citizens; in an open question, 51% described their ethnicity as Australian, 28% as Caucasian, 4% as Aboriginal/Torres Strait Islander, 15% as other ethnicities, and 2% did not indicate their ethnicity.
Design
A cross-sectional survey-based experimental design was used with participants randomly allocated to one of two relationship conditions: friends with or stranger to the perpetrator. Participants were also randomly allocated to either male-to-female or female-to-male violence conditions to control for the influence of gender of both perpetrator and victim.
Materials
Demographics
Participants completed questions about their gender, age, ethnicity, and citizenship.
Friendship questionnaire
Participants in the friend condition were asked to provide the initials of a friend with whom they felt close and shared personal information, and who supported them during hard times. As ingroup bias is more apparent when the relationship is made salient (Mullen, Brown, & Smith, 1992), these participants were asked to consider the duration of the friendship, and which adjectives from a list of 12 best described their friend.
Vignettes
Four short vignettes were developed by the author (see Appendix A) based on real-life stories of IPV incidents (Domestic Violence Resource Centre Victoria, 2013). All the vignettes depicted the same event within a culturally nonspecific heterosexual relationship, consisting of shouting, verbal abuse, pushing, and slapping directed at the victim. Although IPV also occurs in same-sex relationships (Pitts, Smith, Mitchell, & Patel, 2006), this was beyond the scope of this research. The level of violence described in the vignettes did not require urgent medical attention, as responses to emergencies tend to become homogeneous as severity increases (Fischer et al., 2011).
The four vignettes differed by relationship to the perpetrator (friend, stranger) and character gender. The stranger scenario described witnessing the neighbors fighting soon after moving into a new neighborhood. The friend scenario described moving next door to a close friend and then witnessing the violent event. In the friend condition, the initials of the friend (supplied earlier by the participant) were substituted into the vignette as the perpetrator’s name. Genuine ingroup relationships were used because they produce stronger, more realistic effects than contrived relationships (Mullen et al., 1992), consistent with the design used by Bennett and Banyard (2016).
To ensure that the vignette was realistic and had adequate face validity, all participants were asked to indicate on a 5-point scale (1 = no, 5 = yes) “Do you think things like this happen to other people in Australia?”
Manipulation check
To ensure the vignette manipulations were successful, participants were asked to indicate what happened in the scene, identify the name and gender of the aggressor, and describe their relationship to the people in the scene.
Causal attributions
The Revised Causal Dimension Scale (CDSII) is a 15-item questionnaire measuring attributions of causality and has been revised to apply to IPV perpetrators’ behavior (Lehmann & Santilli, 1996). Of the measure’s five subscales, the two subscales used to address the current study’s research questions were locus of causality and blame (see Appendix B). They achieved satisfactory internal reliability in the current sample (Blame: α = .91; Locus of Causality: α = .78), consistent with past research (Lehmann & Santilli, 1996). The entire scale (15 items) was presented to participants as per Lehmann and Santilli’s design (1996). Where required, the perpetrator was referred to as Matthew, Elizabeth, or “your friend,” depending on the experimental condition.
The Locus of Causality subscale consisted of three items assessing the extent to which the cause of the event was attributed to internal factors, along a 9-point continuum. An example item asked the extent to which the cause of the violence reflected an aspect of the situation or an aspect of the perpetrator. Higher scores on this variable reflected greater attributions of internal causality, and low scores indicated attributions of external causality.
The Blame subscale consisted of three items assessing the level of blame and responsibility attributed to the perpetrator, along a 9-point continuum. An example item asked the extent to which the cause of the event was the perpetrator’s fault or not the perpetrator’s fault. The composite score for blame was reverse coded such that higher scores on this variable indicated that the perpetrator was assigned more blame.
Social rejection
Participants were asked one question to establish the influence of IPV perpetration on friendship with the perpetrator. The appropriate names were substituted in to the item depending on experimental condition. Those in the friend condition were asked to indicate agreement on a 5-point Likert-type scale (1 = definitely agree; 5 = strongly disagree) with the statement, “I plan on remaining friends with [name].” Those in the stranger condition were similarly asked to indicate agreement with the statement, “I would be friends with [name].” Higher scores on this variable indicated stronger intentions to reject the friendship.
Debrief
Participants in the friend condition answered three questions about positive aspects of their friendship to highlight the positive characteristics of their friendship and ensure there were no negative after-effects of participation. All participants were fully debriefed and advised about available counseling services.
Procedure
Students were recruited from the university during 2016. The study was voluntary, and students who participated were offered partial course credit and the chance to win a gift voucher. The study was advertised as “Perceptions of Couple Conflict,” with manipulation of experimental variables concealed until debriefing as this may have influenced responses.
Participants completed an online survey consisting of demographic information, a vignette, and questionnaire. Participants were randomly allocated to the friend or stranger condition and perpetrator gender. To achieve gender randomization in the friend condition, participants were randomly allocated based on their gender to receive instructions to describe a friend of the same or other gender. They were also able to indicate if they did not have a close friend in the category they were assigned to, at which point they were redirected to the other option. No participants indicated that they did not have any close friends.
Those allocated to the friend condition answered a series of questions about a real-life friendship, before reading a short vignette describing a scene of IPV with their friend in the role of the perpetrator. Participants allocated to the stranger condition read a short vignette describing a scene of IPV using fictional characters. All participants then completed a questionnaire about their responses to the vignette, and were fully debriefed. The study took an average of 15 min to complete (M = 14.87 min).
Results
Preliminary Statistical Analyses
Data were screened prior to analysis and there were no missing data. Seven cases were removed because they failed the manipulation check by incorrectly identifying the perpetrator. The majority of participants (97.5%) agreed that the scenario would happen to other people in Australia, indicating the vignette had adequate face validity. The average duration of friendship reported by those in the friend condition was 2.63 years, with a range of 4 months to 27 years; 34% of friendships were 2 to 4 years long, 30% were 5 to 9 years long, and 23% were over 10 years. All participants in the friend group nominated positive adjectives to describe their friend; 27% also considered them short-tempered and 8% considered them aggressive.
Descriptive statistics are reported in Table 1. Statistical assumptions were met for all analyses, and a critical alpha level of .05 was used.
Descriptive Statistics for Attribution Variables and Social Rejection by Relationship to Perpetrator.
Attributions of Causality
The first hypothesis proposed that the relationship with the perpetrator would have a significant effect on causal attributions. One-way ANCOVAs were conducted to determine the influence of the relationship with the perpetrator (friend, stranger) on locus of causality and blame, controlling for the influence of participant gender and perpetrator gender.
The first ANCOVA tested the influence of relationship with the perpetrator on locus of causality attributions. Consistent with the hypothesis, there was a statistically significant effect of relationship on locus of causality, F(1, 150) = 26.68, p < .001, with a large effect size, ηp2 = .151. Participants attributed the violence to internal factors significantly more when the perpetrator was a stranger (adjusted M = 6.98, SE = 0.19), rather than a friend (adjusted M = 5.69, SE = 0.18).
The second ANCOVA was conducted to investigate the influence of relationship on attributions of blame. Contrary to prediction, the level of blame attributed to friends (adjusted M = 3.18, SE = 0.19) was not significantly different to blame attributed to strangers (adjusted M = 2.86, SE = 0.19), F(1, 150) = 1.44, p = .232. Relationship to the perpetrator was not found to influence attributions of blame.
Social Rejection
A descriptive research question was posed to establish whether participants who were friends with a perpetrator would continue the friendship. A frequency analysis was conducted on participants in the friend condition only. Those in the stranger condition were excluded, as they did not have an initial friendship with the perpetrator. The majority of participants (65.8%) somewhat or definitely agreed that they would remain friends with the perpetrator, while 17.8% indicated that they would be likely to reject the perpetrator. Table 2 displays a summary of responses.
Frequency of Friends’ Agreement to Stay Friends With Perpetrator.
The Black Sheep Effect
The final hypothesis predicted that there would be a significant relationship between attributions of blame, locus of causality, and social rejection such that those who attributed high blame and internal causality to the perpetrator would be more likely to reject the friendship. A hierarchical regression was employed to test this prediction, controlling for gender of the participant and perpetrator at Step 1. Multicollinearity was deemed unlikely to be a problem given that all correlations between variables were <.5 (see Table 3).
Correlations Between Attribution Measures and Social Rejection for Friends of the Perpetrator.
p < .05. **p < .01.
In Step 1 of the regression, gender of the participant and perpetrator significantly explained 13.6% of the variance in social rejection, R2 = .14, F(2, 74) = 5.82, p = .004. The addition of the attribution variables at Step 2 significantly explained a further 13.1%, ∆R2 = .13, ∆F(2, 72) = 6.41, p = .003. The final model significantly explained a total of 26.7% of the variance in social rejection, R2 = .27, adjusted R2 = .23, F(4, 72) = 6.54, p < .001.
Consistent with Hypothesis 2, the attribution variables were significantly, positively related to social rejection such that high blame and internal causal attributions were associated with high social rejection. A summary of the regression model is presented in Table 4. The significant attribution predictors after controlling for gender in the final model were locus of causality, t(72) = 2.45, p = .017, and blame, t(72) = 2.46, p = .016, explaining 6.1% and 6.2% unique variance, respectively.
Summary of Hierarchical Regression Predicting Social Rejection Using Gender and Attributions (N = 79).
Note. CI = confidence interval.
Males as reference group.
p < .05. **p < .01.
Discussion
This study produced a unique insight into the role of relationship with a perpetrator on perceptions of IPV. It was found that friends of perpetrators were more likely than strangers to attribute the cause of the violence to external factors, but were equally likely to attribute high blame to the perpetrator. The majority of friends indicated they would maintain their relationship with the offender, and social rejection was significantly more likely when perpetrators were attributed higher blame and internal causality.
The first hypothesis predicted that friendship would influence attributions. This was partially supported in that the relationship to the perpetrator significantly influenced locus of causality. Participants were less likely to attribute the violence to internal causes when the perpetrator was a friend rather than a stranger. This means that they were less likely to believe that their friend who committed IPV was driven by dispositional factors, and more likely to consider external, situational factors. The finding was associated with a large effect size, indicating that the relationship manipulation explained a noticeable portion of variance in attributions. These results were consistent with research on the ultimate attribution error (e.g., Harrison & Abrishami, 2004; Hunter et al., 1991). In their study on ingroup bias and dating violence, Harrison and Abrishami (2004) predicted that the ingroup bias effect would be larger as group entitativity increased. The current findings confirmed that friendship was more influential than weaker ingroup relationships, such as university affiliation, used in past research.
Unexpectedly, the second element of the first hypothesis was unsupported. Attributions of blame did not differ significantly between friends and strangers. Theory and past research on ingroup bias indicated that strangers would be blamed more than friends (e.g., Harrison & Abrishami, 2004; Hunter et al., 1991). However, both groups consistently attributed high responsibility for the violence to the perpetrator regardless of their relationship.
This finding could be explained in the light of recent media exposure condemning perpetrators of IPV. It may reflect genuine attitude change or a social desirability bias, although the anonymity of the survey aimed to reduce the impact of social desirability. It is possible the severity of the violence in the current study was too great for participants to excuse, particularly given the increased public awareness. In this case, a difference may be detectable in the context of a less violent scenario, given that responses to violence become homogeneous as severity increases (Fischer et al., 2011). The finding was promising in the context of encouraging bystander intervention: friends were able to recognize this form of IPV and attributed fault to the perpetrator despite the close relationship, which is considered an important stage in bystander intervention (Latané & Darley, 1970).
The results indicated that friends who perpetrated IPV were attributed full responsibility for their actions, but were given some leeway considering the cause of the violence. This conflicted with expectations based on social identity theory, which explains ingroup bias as motivated by a need to maintain a positive social identity (Tajfel & Turner, 1979). Accordingly, attributions of blame and locus of causality were both expected to be more lenient for ingroup members. An alternative theoretical perspective to explain the contradictory findings is expectancy theory (Jones & McGillis, 1976).
Jones and McGillis (1976) found that attributions are influenced by an individual’s expectations of how a person will behave. Negative behaviors that are incongruent with expectations based on past actions or group membership tend to be attributed to situational influences (Shaver, 1985). Past research on IPV indicated that lenient attributions were associated with first-time offenders but repeat offenders were judged more harshly (Witte, Schroeder, & Lohr, 2006). Given the low statistical occurrence of IPV in the community, it is likely that the friends nominated by participants in this study did not have a history of relationship violence, although future research should address this point empirically. The friend-perpetrators were assigned full blame for the violence because the vignette clearly described them as the perpetrator. However, given the violent behavior was unexpected based on past interactions, the behavior was considered unlikely to be caused by internal, dispositional characteristics. The difference in attributions between friends and strangers could be explained by a cognitive effort to reconcile conflicting facts about a friend’s behavior as well as a desire to protect one’s social identity. As such, the results from this study may only be generalizable to first-time IPV offenders and situations where friends have no prior knowledge of IPV behaviors.
The second focus of the study explored whether participants would stay friends with perpetrators of IPV and proposed that social rejection would be predicted by attributions. Results indicated the majority of participants would not end an existing friendship with a perpetrator of IPV. This indicated friends were likely to remain in a position where they could influence the behavior of the perpetrator, either through appropriate intervention or lack of action. It was noted that this finding may only apply to first-time perpetrators of IPV, based on research indicating that repeat offenders were not afforded the same ingroup favoritism (Harrison & Abrishami, 2004) and in accordance with expectancy theory.
Consistent with the final hypothesis, rejection of the friendship was associated with high blame and internal attributions. This supported the black sheep effect (Marques et al., 1988), whereby ingroup members who were judged as deviant were rejected from the social group. Conversely, stronger endorsements of continued friendship were associated with lower blame and external attributions. This was consistent with social identity theory, whereby protection of social identity allowed participants who attributed less responsibility to the perpetrator to ignore the negative act, or alternately, indicated that the perpetrator required help rather than judgment.
Given the effect sizes associated with this finding were small, many other factors likely influenced the decision to stay friends with a perpetrator of violence. This finding did not infer causality: negative attributions may encourage rejection of friendship, but initial rejection of the friendship could allow a person to make harsh attributions. This research illuminated a tendency for people to stay friends with a perpetrator of IPV, but the mechanisms behind this require further investigation.
In sum, friends of perpetrators in this study endorsed a combination of high blame, external attributions, and an intention to continue the friendship. Although a tendency toward external attributions can be interpreted as excusing violence perpetrated by an ingroup member (e.g., Harrison & Abrishami, 2004), a meta-analysis on attributions and help-giving found individuals were more likely to receive help when their problem was attributed to external factors (Rudolph, Roesch, Greitemeyer, & Weiner, 2004). Attributional theories of helping predict that perceptions of an event being externally influenced, and therefore uncontrollable, increase the likelihood of a sympathetic, prosocial response (Rudolph et al., 2004). Attributing the cause to changeable situational determinants (e.g., mental health problems rather than dispositional aggression; Worden & Carlson, 2005) may increase the perceived likelihood that intervention can effectively change a negative behavior. In combination with recognition of the perpetrator’s responsibility, the external attributions made by participants in this study could support eventual positive interventions. This indicates that with effective training, friends could be a starting point in guiding a perpetrator toward support to change his or her behavior. More research is required to establish how this attributional pattern relates to willingness to intervene in social relationships.
Implications
The current findings have implications for programs aimed at reducing IPV in the community. They illuminate how friends may react in today’s climate of domestic violence awareness, demonstrating that ingroup bias may not be as influential as previously thought. Friends of offenders may be in a unique position to recognize their friend’s violence, offer support for behavioral change, and potentially reduce reoccurrence of violent offending (Banyard, 2011; Emery et al., 2016). It is important that friends actively voice disapproval and address IPV behaviors to avoid inferring acceptance and normalization of the violent behavior (McMahon & Banyard, 2012). The combined response of blaming the perpetrator but retaining the friendship may provide a space for this conversation to occur. Bystander intervention programs and violence prevention campaigns should consider including a specific component to address this.
Contrary to past research, ingroup bias was not consistent across attributional dimensions, nor did it induce a tendency to relieve blame for the violence. The results demonstrated that attributions alone are not sufficient to understand ingroup bias in response to complex events. Considering the mixed results found by the current study and past research on friendship and responses to IPV, it is clear that numerous factors influence cognitive and behavioral responses in this scenario. Although this research adds another piece to the puzzle, continued exploration is needed. Based on these findings, a particularly important question is raised around how this attributional pattern of high blame and external causality may relate to willingness to intervene in friends. Other influential factors requiring further research in relation to friendship include empathy, barriers and facilitators of intervention, self-efficacy, and intervention decision making (Banyard, 2008; Chabot, Gray, Makande, & Hoyt, 2016).
Limitations
The findings should be interpreted in light of the limitations. Two limitations of this study relate to the use of a convenience sample of university students. First, the nonrandom sampling strategy may have biased the testing in favor of finding statistical significance. Although the observed p values were substantially lower than the .05 cutoff, it is recommended that these findings are interpreted with caution until the study is replicated with a probability sample. Second, generalizability is limited as the sample was dominated by culturally homogeneous, female young adults enrolled in a psychology degree. To address these issues, future research would benefit from replicating the study with a randomly recruited community sample.
Some aspects of the design may be considered limitations. A reliance on self-report may have increased the chances of a social desirability bias influencing the results, although this effect should have been limited by the anonymous online format. The use of fictional vignettes was removed from the emotion and complexity of real-life responses to violence. As specified by Harrison and Abrishami (2004), a stronger effect of friendship would be expected in a genuine IPV situation.
Gender has previously been found to have a significant effect on perceptions of IPV (e.g., Worthen & Varnado-Sullivan, 2005). Although it was outside the scope of this research, future research should explicitly explore gender differences in the influence of relationship on attributions for IPV. Furthermore, this study explored responses to violence in the context of heterosexual relationships only, restricting its generalizability to violence occurring in same-sex relationships.
It is noted that the stranger-to-perpetrator relationship in this study was actually a stranger–neighbor relationship, and this may have invited an anticipation of a future relationship. Neighborhood relationships could be considered ingroup to some participants, although they are less entitative than friend relationships. To expand this research, future studies should introduce a pure stranger condition with no anticipation of future interaction. Furthermore, the use of a real-life friendship may have introduced a confound in the cases where participants had an additional relationship with their friend’s partner. Future research should establish whether participants hold multiple relationships to isolate the unique effect of friendship with the perpetrator. Finally, to disentangle ingroup bias from the role of expectancy on attributions for previously nonviolent friends, it would be of interest to repeat this study with friends and family of known IPV offenders.
Conclusion
This study applied social psychological theory and randomized experimental methodology to an important but neglected factor influencing IPV outcomes. Peers of perpetrators may have unique opportunities to intervene to prevent violence and model revised social norms (Banyard, 2011; Berkowitz, 2003). This study has contributed to knowledge on how people respond to IPV and highlights opportunities to provide critical education to reduce incidences of IPV in the community. These findings should be used to inform bystander intervention programs and contribute to the global movement against IPV.
Footnotes
Appendix A
Appendix B
Acknowledgments
The authors would like to thank the 174 participants who took part in this research and provided their opinions and honest dialogue concerning how they would respond to intimate partner violence. Their thanks also go to the lecturers and staff at the University of Canberra for allowing them access to their students.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
