DEP Paper

https://doi.org/10.1177/0886260518794011

Journal of Interpersonal Violence 2021, Vol. 36(9-10) NP4850 –NP4873

© The Author(s) 2018 Article reuse guidelines:

sagepub.com/journals-permissions DOI: 10.1177/0886260518794011

journals.sagepub.com/home/jiv

Original Research

The Bystander Behavior (For Friends) Scale: Factor Structure and Correlation With Prior Victimization

Michele Cascardi,1 Alison Krauss,2 K. Daniel O’Leary,3 Katie Lee Loatman,4 Kelli Sargent,2 John Grych,5 and Ernest N. Jouriles2

Abstract The Bystander Behavior (for Friends) Scale (BBS) offers a promising method of studying prosocial bystander behavior in the context of sexual assault and intimate partner violence. The underlying structure of the BBS has only been studied in the development sample, which was predominantly White and from one university in the Northeast region of the United States. This single sample raises questions about the replicability and generalizability of the factor structure. In addition, confirmatory factor analytic (CFA) methods, which are favored for binary data, were not used in the developmental sample. There also is limited research on individual characteristics that may relate to engagement in different types of bystander behavior. The primary aims of the current study were to (a) use CFA to evaluate the factor structure of the BBS in a sample of university

1William Paterson University, Wayne, NJ, USA 2Southern Methodist University, Dallas, TX, USA 3Stony Brook University, Stony Brook, NY, USA 4Harvard Medical School, Boston, MA, USA 5Marquette University, Milwaukee, WI, USA

Corresponding Author: Michele Cascardi, Department of Psychology, William Paterson University, 300 Pompton Road, Wayne, NJ, 07470, USA. Email: [email protected]

794011 JIVXXXXXX10.1177/088626051879401110.1177/0886260518794011Journal of Interpersonal ViolenceCascardi et al. research-article20182018

Cascardi et al. NP4851

undergraduates recruited from four universities and (b) test associations between prior victimization (general and family-specific) and BBS factors. University undergraduates (n = 556) from four U.S. universities comprised the sample. Weighted least squares CFA confirmed the original four-factor model of the BBS, namely, Risky Situations, Accessing Resources, Proactive Behaviors, and Party Safety. The Proactive Behaviors factor was positively associated with both general and family-specific prior victimization. The Risky Situations and Party Safety factors were positively associated with general prior victimization but were not associated with family-specific prior victimization. The Accessing Resources factor was not associated with either general or family-specific prior victimization. The BBS is multidimensional, and the factor structure is robust. The different associations between certain types of bystander behavior and prior victimization highlight the potential value in considering the BBS factors separately.

Keywords intimate partner violence, sexual assault, measurement, factor analysis

An estimated 19% to 25% of women experience sexual assault (SA) while in college (Fisher, Cullen, & Turner, 2000; Krebs, Lindquist, Warner, Fisher, & Martin, 2009). Annual rates of physical intimate partner violence (IPV) among college students range considerably from 16% to 50% (Edwards et al., 2015; Nabors & Jasinski, 2009; Shook, Gerrity, Jurich, & Segrist, 2000; Straus, 2004). College campuses have increased efforts to reduce SA and IPV by delivering bystander intervention programs to their students, which aim to increase different types of prosocial behaviors that can prevent sexual and physical assaults. This might include mobilizing students to become more knowledgeable about SA and IPV, engage in self-protective actions before SA or IPV occurs (e.g., party safety behaviors), interrupt high-risk situations for SA or IPV to help protect others, and provide support and resources to victims in the aftermath (McMahon & Banyard, 2012). The most comprehen- sive attempt to measure multiple types of prosocial bystander actions is the Bystander Behavior (for Friends) Scale (BBS), a 49-item self-report inven- tory that asks respondents to indicate whether they engaged in a variety of prosocial actions to assist friends in situations involving SA and IPV (Banyard, Moynihan, Cares, & Warner, 2014). The primary purposes of the current study are to (a) evaluate the factor structure of the BBS in a sample of university undergraduates recruited from four universities and (b) test asso- ciations between prior victimization and BBS factors.

NP4852 Journal of Interpersonal Violence 36(9-10)

The development of the BBS was guided by a conceptual framework of bystander behavior that focuses on three distinct aspects of bystander context (McMahon & Banyard, 2012). The first aspect addressed by the BBS is the differentiation between situations where SA and IPV tend to occur. The for- mer may be more likely to happen at parties where one or both individuals are intoxicated, and the latter may occur in the context of ongoing relationships where bystanders may observe repeated instances of insults, coercion, or other aggression. The second aspect of bystander context examined by the BBS is when bystander behavior happens: prior to, during, or after SA or IPV. For instance, bystanders may engage in behavior to prevent SA or IPV before it occurs, interrupt high-risk situations while they are occurring, and provide support and resources in the aftermath of SA or IPV. The third aspect of the model defines bystander behaviors according to the level of risk to which a victim is exposed: no risk, low risk, and high risk (McMahon & Banyard, 2012). For example, no risk is involved when students learn about resources that could help potential victims, whereas interrupting a physical assault may expose a bystander to a high risk of harm. Using this framework, the BBS accommodates the complexity and range of bystander opportunities associ- ated with SA and IPV (Banyard et al., 2014).

In the development sample, principal component analysis (PCA) yielded four factors of the BBS, indicating that bystander behavior represents a mul- tidimensional construct with distinct opportunities for helpful bystander intervention (Banyard et al., 2014). The first factor, Risky Situations, repre- sents opportunities to confront problem behaviors, potential perpetrators, and interrupt situations where SA or IPV might involve a risk of potential harm to a victim (e.g., I told a friend if I thought their drink may have been spiked with a drug). Supporting a victim in the aftermath of SA or IPV (e.g., I expressed concern to a friend who had unexplained bruises that may be signs of abuse in their relationship) is also included on this first factor. The second factor, Accessing Resources, comprises opportunities for seeking help (e.g., I called 911 because of suspicion that a friend had been drugged; I accompa- nied a friend to a local crisis center). Proactive Behaviors, the third factor, includes opportunities to obtain more knowledge or training to prevent SA and IPV (e.g., I educated myself about SA and/or IPV and what I can do about it) as well as spreading awareness about SA and IPV (e.g., I talked with a friend about SA and/or IPV as an issue for our community). The fourth factor, Party Safety, includes planning strategies to protect friends before risky situ- ations present themselves in the first place (e.g., I talked with a friend about going to parties together, staying together, and leaving together.).

Measures such as the BBS assess different types of bystander behaviors that are consistent with current conceptualizations of both the complexity and

Cascardi et al. NP4853

breadth of bystander behavior (Banyard et al., 2014). In addition, multidi- mensional measures can prompt research on the correlates of different types of bystander behavior, and such research has the potential to lead to a more comprehensive understanding of how to motivate individuals to engage in these different types of bystander behavior (i.e., different strategies might be necessary for different types of bystander behavior). However, before such research can be conducted, it is necessary to first confirm the underlying fac- tor structure of the measures. To date, the only test of the underlying structure of the BBS has been conducted in a single sample in the development study. Thus, the replicability and generalizability of the BBS four-factor structure is in question. In addition, there is debate about the most appropriate method of analysis for binary response options, which are used on the BBS. The BBS was analyzed with PCA in the development sample, but PCA analyses may lead to inaccurate conclusions about the underlying dimensions of opportu- nity for bystander behavior (Bollen & Barb, 1981; Mislevy, 1986). A more rigorous approach is confirmatory factor analysis (CFA), which provides estimation methods suited to binary data and tests the fit of the hypothesized measurement model based on theory of the underlying factor structure and prior analyses of related data (e.g., Muthén & Muthén, 2012).

Although there are multiple types of bystander behavior, with few excep- tions (Banyard & Moynihan, 2011; Banyard et al., 2014), bystander behavior is typically treated as a unidimensional construct in evaluation studies (Banyard, Moynihan, & Plante, 2007; Coker et al., 2011; Jouriles et al., 2016; Jouriles et al., 2020; Kleinsasser, Jouriles, McDonald, & Rosenfield, 2015; McMahon et al., 2015) and research on correlates of bystander behavior (Banyard, 2008; Bennett, Banyard, & Garnhart, 2014). Research on distinct correlates and determinants of specific bystander behaviors may lead to more personalized intervention approaches and improve the design and effective- ness of bystander education programs (Banyard & Moynihan, 2011; McMahon & Banyard, 2012). For instance, strategies to mobilize students to interrupt risky situations or assist friends who have been assaulted access resources may vary depending on individual characteristics, such as prior victimization experiences.

Prior Victimization and Bystander Behavior

Prior victimization experiences are theorized to increase bystander action by increasing feelings of responsibility and motivation to intervene to prevent violence (e.g., Casey, Lindhorst, & Storer, 2017; Frazier & Berman, 2008; Staub, 2003). In addition, those who have been previously victimized may be more empathic to a potential victim’s experiences (Casey et al., 2017), and

NP4854 Journal of Interpersonal Violence 36(9-10)

empathy is associated with increased bystander intervention (e.g., Gini, Albiero, Benelli, & Altoè, 2008). However, theory and past research suggest that prior victimization may be associated with some types of bystander behaviors, but not others, which would highlight the potential value in con- sidering the BBS factors separately.

Research on the relation between prior victimization and helpful bystander behavior is scant (Labhardt, Holdsworth, Brown, & Howat, 2017). Palmer (2016) found that past SA victimization was associated with helping some- one after an assault had already taken place. Specifically, those with a victim- ization history were more likely to assist a victim of SA or IPV or confront someone endorsing rape myths. Similarly, Nabi and Horner (2001) found that previously abused women were more likely to report having performed a variety of IPV-related bystander behaviors, including confronting a man about his abusive actions and calling 911, as compared with nonabused women. The bystander actions described in these two studies align most strongly with the BBS dimensions of Risky Situations and Accessing Resources. Prior victimization experiences may therefore be associated most strongly with the Risky Situations and Accessing Resources factors of the BBS, which include actions that interrupt high-risk situations to protect indi- viduals from SA and IPV as well as actions to support victims in the after- math of SA and IPV. It is less clear whether prior victimization experiences will be associated with the other dimensions of bystander behavior measured by the BBS, specifically Proactive Behavior and Party Safety.

There is also reason to believe that certain types of victimization may relate differently to specific dimensions of bystander behavior. For instance, victimization by interpersonal violence across a wide range of settings, compared to victimization only in the family, may heighten awareness of risk in potentially dangerous situations (Dalgeish, Moradi, Taghavi, Neshat- Doost, & Yule, 2001; Kimble et al., 2014; Kimble, Fleming, & Bennion, 2013). Thus, individuals with more varied prior victimization experiences may readily anticipate situations where risks for SA and IPV may be pres- ent. As a result, those with prior victimization experiences by either family or nonfamily members (i.e., general) may be more likely to engage in safety planning to avoid risky situations, that is, bystander action aligning with the BBS dimension Party Safety, rather than those with family-specific victim- ization experiences.

The Current Study

The primary aim of the current study was to evaluate the factor structure of the BBS in a college student sample recruited from four universities that are

Cascardi et al. NP4855

different than the one used in the Banyard et al. (2014) development sample. The first hypothesis is that the four-factor model of the BBS identified by Banyard et al. (2014), Risky Situations, Accessing Resources, Proactive Behaviors, and Party Safety, will achieve acceptable levels of model fit using CFA appropriate for binary data. In addition, we hypothesize that the four- factor model will fit the data better than a single factor model consisting of all 49 items. We reasoned that a comparison of the four-factor model to a one- factor model would be valuable, as investigators often treat bystander behav- ior as a unidimensional construct. A secondary aim is to examine prior victimization as a correlate of the different BBS dimensions. Total prior vic- timization (general + family-specific) is hypothesized to be more strongly associated with the Risky Situations and Accessing Resources factors than the Proactive Behaviors and Party Safety factors. We also explore whether gen- eral and family-specific prior victimization experiences relate to some BBS dimensions, but not others.

Method

Participants

University undergraduates (n = 556) were recruited from psychology courses in the beginning of the 2015 Spring semester from four U.S. universities: two private universities (in the Southwest and Midwest) and two public universi- ties, both in the Northeast. The sample was predominantly female (77.4%). The racial and ethnic characteristics of the total sample are comparable with national rates of postsecondary minority enrollment (National Center for Educational Statistics, 2013): 12.7% Hispanic, 59.6% White, 22.6% Asian, 7.9% Black, 2.7% as Bi- or Multi-racial, 0.2% as American Indian/Alaska Native, 0.4% as Native Hawaiian/Pacific Islander, and 6.3% as another race. The average age of the sample was 20.15 years (SD = 2.98, range 17-54).

Procedures

Participants were recruited for a study examining the effects of a bystander intervention program (Jouriles et al., 2020); only baseline data were used in the current study. During the baseline assessment, participants completed self-report questionnaires using a web-based survey in Qualtrics. Survey completion was supervised by research assistants who answered participants’ questions and ensured quality control. Participants received extra credit in their psychology course for participation. Although all universities had exist- ing campus activities addressing SA (i.e., social media campaigns, events

NP4856 Journal of Interpersonal Violence 36(9-10)

during SA awareness month), students had not been exposed to any bystander intervention programs at college prior to their baseline assessment. The Institutional Review Board at each university granted study approval.

Measures

Bystander behaviors. Students completed the 49-item BBS (Banyard et al., 2014) to assess their engagement in a variety of bystander behaviors in the past month. As noted previously, the BBS comprises four subscales: Risky Situations, Accessing Resources, Proactive Behaviors, and Party Safety. The BBS is scored by counting the number of “yes” responses for each subscale, and the number of “yes” responses to all items for a total score. Five items on the BBS (two regarding party safety and three regarding proactive confronta- tion) cross-loaded in the prior factor analysis. In the current study, these five items were assigned to their conceptual dimensions (i.e., Party Safety and Proactive Behaviors).

Juvenile victimization. The Juvenile Victimization Questionnaire–Short Form (JVQ; Hamby, Finkelhor, Ormrod, & Turner, 2004) is a 17-item measure of interpersonal victimization in five domains. These include the following: wit- nessing violence (two items, e.g., “At any time in your life, in real life, did you SEE anyone get attacked or hit on purpose WITHOUT using a stick, rock, gun, knife, or something that would hurt?”); peer victimization (six items, e.g., “At any time in your life, in real life, did you get scared or feel really bad because kids were calling you names, saying mean things to you, or saying they didn’t want you around?”); assault (two items, e.g., “At any time in your life, did someone make you do sexual things when you didn’t want to?”); exposure to family violence (five items, e.g., “At any time in your life, did one of your parents get hit or pushed by another parent?”); and par- ent-child dysfunction (two items, e.g., “Not including spanking on your bot- tom, has a grown-up in your life ever hit you?”). Participants indicated whether they had experienced each of the 17 forms of victimization. The JVQ is scored by summing the number of “yes” responses. In the current study, three composite scores were computed: general prior victimization, including items related to witnessing violence, peer victimization, and assault; family- specific prior victimization, including items related to exposure to family violence and parent–child dysfunction; and total prior victimization, includ- ing all items. The JVQ has demonstrated internal reliability and validity (Hamby et al., 2004). Cronbach’s alpha was α = .70 for general prior victim- ization, α = .77 for family-specific prior victimization, and α = .77 for total prior victimization.

Cascardi et al. NP4857

Data Analysis

CFA models extract latent factors from a covariance or correlation matrix, which assume linear associations among scale items (i.e., indicators) and between an indicator and its underlying factor(s). However, when indicators are binary, cor- relations can be attenuated due to the restricted range or inflated due to the simi- larities in the distributional shape of two indicators (Bollen & Barb, 1981; Mislevy, 1986). In addition, the assumption of multivariate normality required for maximum likelihood estimation is not met. These problems may produce biased estimates (Hutchinson & Olmos, 1998), underestimate factor loadings and factor correlations (Wang, Fan, & Willson, 1996), and underestimate stan- dard errors, which results in misleading significance tests and confidence inter- vals (CIs; West, Finch, & Curran, 1995). To overcome these limitations, weighted least squares estimation with tetrachoric correlations (Flora & Curran, 2004; Muthén, 1983, 1984) estimates factor structure and model fit without vio- lating the linearity or the multivariate normality assumptions of the classical factor analysis. With this method, parameter estimates and associated standard errors are obtained using the estimated asymptotic covariance matrices of the tetrachoric correlation and threshold estimates and the weighted least squares fit function (Bollen, 1989; Jöreskog, 2005; Li, 2016).

The hypothesized measurement models were tested using MPlus version 7.31 (Muthén & Muthén, 2012). Model fit was assessed with several com- monly used indices in addition to the robust chi-square test of model fit: the comparative fit index (CFI; Bentler, 1990; Hu & Bentler, 1999) and Tucker– Lewis index (TLI; Tucker & Lewis, 1973). These indices estimate the pro- portionate improvement in fit by comparing the hypothesized model with the baseline model, typically the null model (Hoyle & Panter, 1995; Hu & Bentler, 1999). The fit estimates are scaled to an approximate range of 0 and 1. By convention, values of .90 or higher are considered acceptable values of fit and values above .95 indicate good fit (Bentler & Bonett, 1980). The root mean square error adjustment (RMSEA) is also reported; values less than .08 indicate a reasonable error of approximation (Browne & Cudeck, 1993). The one- and four-factor models were compared using the likelihood ratio test DIFFTEST procedure, which provides a corrected Chi-Square difference test for nested models when weighted least squares estimators are used (Muthén & Muthén, 2012). The adjusted Bayesian information criterion (BIC) was also used to compare model fit.

Pearson product–moment correlation coefficients were used to evaluate the association between prior victimization and BBS dimensions. Differences in the magnitude of correlation coefficients were evaluated with Fisher’s r-to-z transformations for dependent samples (Lee & Preacher, 2013).

NP4858 Journal of Interpersonal Violence 36(9-10)

Results

Participants’ endorsement of specific bystander behaviors ranged from 4.3% to 65.6% of behaviors included on the BBS (rates per item are available from first author). The specific bystander behavior that most participants (65.6%) endorsed was “Talked with a friend about going to parties together,” whereas the specific bystander behavior that the fewest participants (4.3%) endorsed was “Called 911 because of suspicion that a friend had been drugged.” In general, the highest rates of endorsement were for protecting friends before, during, and after parties (e.g., I talked to a friend about going to parties together, staying together, and leaving together). The lowest rates of endorse- ment were for accessing resources to protect a victim of SA or IPV (e.g., I called a crisis center or community resource when a friend told me they expe- rienced sexual or relationship abuse).

In the first measurement model, a single bystander behavior factor was rep- resented by all 49 indicator variables (i.e., items; α =.94 [male] and α =.95 [female]). The measurement model was an acceptable fit to the data, χ2, df = 1,127, 4,513.97, p < .001, TLI = .905, CFI = .901, and RMSEA = .074 (90% CI [.071, .076]). All loadings were statistically significant at p < .001 (see Table 1). Next, the hypothesized four-factor model was tested. The factor Risky Situations was represented by 19 indicator variables (items 1-19; α = .94 [male] and α = .93 [female]), Accessing Resources was represented by eight indicator vari- ables (items 20-27; α = .95 [male and female]), Proactive Behaviors was rep- resented by 14 indicator variables (items 28-38, 45, 46, 49; α = .85 [male] and α = .86 [female]), and Party Safety was represented by eight indicator variables (items 39-44, 47, 48; α = .89 [male] and α = .91 [female]). The measurement model was a good fit to the data, χ2, df = 1,121, 1,759.94, p < .001, TLI = .982, CFI = .981, and RMSEA = .032 (90% CI [.029, .035]). All loadings were sta- tistically significant at p < .001 (see Table 1). The χ2 likelihood ratio DIFFTEST procedure was used to evaluate whether the one- or four-factor model was a better fit to the data. In this test, each model is compared against the null model, and the adjusted difference in χ2 is computed. Results showed that the four- factor model was a significantly better model fit than the one-factor model (likelihood ratio test χ2, df = 6, 4804.20, p < .001). Adjusted BIC values were also examined. Smaller values indicate better fit, providing additional support for the four-factor model: adjusted BIC, one-factor model = 15,275.51 and adjusted BIC, four-factor model = 10,490.15.

Descriptive statistics for the four BBS factors, total BBS score, prior vic- timization (total), as well as general and family-specific victimization are presented in Table 2. The BBS Accessing Resources factor showed signifi- cant skew and kurtosis (Curran, West, & Finch, 1996), so it was square root

Cascardi et al. NP4859

Table 1. Standardized Model Results.

BBS item

Single factor solution

Factor 1: risky

situations

Factor 2: accessing resources

Factor 3: proactive behaviors

Factor 4: party safety

λ z test λ z test λ z test λ z test λ z test

1. I saw a friend and their partner in a heated argument. I asked if everything was ok.

.61 14.95* .66 15.25*

2. If a friend was being shoved or yelled at by their partner, I asked if they needed help.

.80 27.65* .84 28.64*

3. I suspected that a friend had been sexually assaulted and I let that friend know that I was available for help and support.

.88 41.25* .90 43.72*

4. I approached a friend if I thought they were in an abusive relationship to let them know that I was there to help.

.85 39.84* .89 43.81*

5. When a friend who looked very intoxicated was being taken to someone’s room at a party or home with someone they just met, I stopped and checked in with the friend.

.76 26.14* .81 25.86*

6. If I saw a friend grabbing or pushing their partner, I said something to them.

.89 47.44* .92 51.66*

7. I expressed concern to a friend if I saw their partner being very jealous and trying to control them.

.63 17.10* .69 18.00*

8. If a friend said they had an unwanted sexual experience but they don’t call it “rape,” I expressed concern and/or offered to help.

.84 37.83* .87 40.42*

9. I expressed concern when I heard a friend talking about using physical force with their partner.

.91 52.34* .93 55.54*

10. I confronted a friend who made excuses for abusive behavior by others.

.75 23.16* .80 24.69*

11. If I saw a friend taking a very intoxicated person to their room, I said something and asked what they were doing.

.78 26.16* .82 26.87*

(continued)

NP4860 Journal of Interpersonal Violence 36(9-10)

BBS item

Single factor solution

Factor 1: risky

situations

Factor 2: accessing resources

Factor 3: proactive behaviors

Factor 4: party safety

λ z test λ z test λ z test λ z test λ z test

12. I supported a friend who wanted to report sexual assault or relationship abuse that happened to them even if others could get in trouble.

.94 72.98* .97 76.24*

13. I expressed concern when I heard a friend talking about forcing someone to have sex with them.

.94 64.42* .96 68.91*

14. I told a friend if I thought their drink may have been spiked with a drug.

.92 50.95* .94 53.00*

15. I expressed disagreement with a friend who said having sex with someone who is passed out or very intoxicated is okay.

.80 25.11* .85 26.45*

16. I saw a guy talking to a female friend. I could see she was uncomfortable. I asked her if she was okay and tried to start a conversation with her.

.62 16.28* .66 15.97*

17. If I heard sounds of yelling and fighting coming from a friend’s dorm room or other residence walls, I knocked on the door to see if everything was okay.

.65 13.90* .70 14.75*

18. If I heard a friend insulting their partner, I said something to them.

.62 15.30* .67 15.76*

19. I expressed concern to a friend who had unexplained bruises that may be signs of abuse in their relationships.

.89 34.85* .93 37.87*

20. I called 911 because of suspicion that a friend had been drugged.

.99 74.58* .99 75.57*

21. I called 911 or authorities when I heard sounds of yelling and fighting.

.95 45.42* .96 47.97*

22. I accompanied a friend to a local crisis center.

.96 48.22* .97 50.59*

23. I called a crisis center or community resource for help when a friend told me they experienced sexual or relationship abuse.

.96 53.68* .98 58.23*

Table 1. (continued)

(continued)

Cascardi et al. NP4861

BBS item

Single factor solution

Factor 1: risky

situations

Factor 2: accessing resources

Factor 3: proactive behaviors

Factor 4: party safety

λ z test λ z test λ z test λ z test λ z test

24. I called 911 or authorities because someone was yelling for help.

.97 79.57* .99 92.23*

25. I called 911 or authorities when a friend needed help because they had been hurt sexually or physically.

.96 59.01* .98 64.25*

26. When I heard that a friend was accused of sexual abuse or relationship abuse, I came forward with what I knew rather than keeping silent.

.89 35.56* .92 38.35*

27. I went with a friend to talk with someone (community resource, police, crisis center, etc.) about an unwanted sexual experience or relationship abuse.

.97 62.00* .98 66.31*

28. I got further training in skills to confront and prevent sexual abuse and relationship abuse.

.60 11.52* .71 13.61*

29. I got advice from others about how to help someone who has experienced sexual abuse or relationship abuse.

.60 14.20* .73 17.89*

30. I developed a specific plan for ways I might safely intervene as a bystander if I see sexual abuse or relationship abuse happening around me.

.56 13.09* .68 15.97*

31. I educated myself about sexual abuse and/or relationship abuse and what I can do about it.

.60 15.05* .74 18.65*

32. I thought through the pros and cons of different ways I might help if I saw an instance of sexual abuse or relationship abuse.

.58 15.03* .71 17.89*

33. I encouraged others to learn more and get involved in preventing sexual or relationship abuse.

.62 14.84* .74 18.07*

Table 1. (continued)

(continued)

NP4862 Journal of Interpersonal Violence 36(9-10)

BBS item

Single factor solution

Factor 1: risky

situations

Factor 2: accessing resources

Factor 3: proactive behaviors

Factor 4: party safety

λ z test λ z test λ z test λ z test λ z test

34. I talked with a peer about sexual and/or relationship abuse as an issue for our school.

.46 9.02* .57 10.32*

35. I tried to get others to help me before trying to do something about sexual abuse or relationship abuse that I saw going on.

.83 23.38* .96 26.78*

36. I refused to remain silent about instances of sexual abuse or intimate partner abuse I knew about.

.71 17.35* .84 19.89*

37. I talked with a friend about what makes a relationship abusive and what warning signs might be.

.60 14.29 .72 17.34

38. I shared information or resources about sexual assault or relationship abuse with a friend.

.67 16.14* .80 19.72

39. I made sure I left the party with the same people I came with.

.81 39.96* .92 53.73*

40. I made sure an intoxicated friend didn’t get left behind at a party.

.93 63.79* .99 95.66*

41. I talked with a friend about going to parties together, staying together, and leaving together.

.65 16.96* .84 26.41*

42. If a friend had too much to drink, I asked them if they needed help getting home from the party.

.85 36.98* .96 58.86*

43. I took a friend home from a party when they had too much to drink.

.78 31.78* .92 53.06*

44. I asked a friend who seemed upset if they were okay or needed help.

.49 9.01* .69 12.61*

45. I refused to remain silent when a friend asked me to keep quiet about an instance of sexual abuse or intimate partner abuse that I knew about.

.85 26.31* .99 30.13*

Table 1. (continued)

(continued)

Cascardi et al. NP4863

transformed. The correlation matrix presented in Table 3 shows partial sup- port of the hypothesis that the association between total prior victimization and BBS dimensions would be different, but not in line with predictions that Risky Situations and Accessing Resources would be stronger than the associa- tion of total prior victimization and Proactive Behaviors and Party Safety. Specifically, the correlation between total (general + family-specific) prior

Table 2. Descriptive Statistics of Correlates and Bystander Behavior Subscales.

Variable M (SD) Skewness Kurtosis Range

Total victimization 6.61 (3.57) 0.31 −0.33 0-17 General victimization 4.97 (2.40) −0.23 −0.62 0-10 Family-specific victimization 1.64 (1.89) 1.09 0.31 0-7 Bystander behavior Risky situation 4.11 (5.17) 1.47 1.16 0-19 Accessing resources 0.48 (1.62) 3.80 13.63 0-8 Accessing resources (SR) 0.23 (0.65) 2.99 7.91 0-2.83 Proactive behaviors 3.46 (3.42) 1.04 0.37 0-14 Party safety 4.77 (2.99) −0.39 −1.45 0-8 Total 12.81 (10.24) 1.28 1.64 0-49

Note. Family-specific victimization = victimization in family of origin; general victimization = victimization in community and/or family or by peers; total victimization = family-specific and general victimization combined; SR = square root transformation.

BBS item

Single factor solution

Factor 1: risky

situations

Factor 2: accessing resources

Factor 3: proactive behaviors

Factor 4: party safety

λ z test λ z test λ z test λ z test λ z test

46. I indicated my displeasure when I heard sexist, racist, homophobic joke, or catcalls made by a friend.

.36 7.13* .44 7.59*

47. I watched a friend’s drink at a party.

.62 17.28* .83 29.25*

48. I made sure a friend left the party with the same people he or she came with.

.72 25.18* .88 43.35*

49. I spoke up if I heard “she deserved to be raped.”

.62 14.52* .74 15.57*

Note. BBS = Bystander Behavior (for Friends) Scale. *p < .001.

Table 1. (continued)

NP4864 Journal of Interpersonal Violence 36(9-10)

victimization and Risky Situations was significantly larger than Accessing Resources, z = 4.37, p < .001, but not Proactive Behaviors or Party Safety, z = 1.04, p = .30 and z = 1.84, p = .06, respectively. In addition, the correlation between total prior victimization and Accessing Resources was significantly smaller than the correlations between total prior victimization and both Proactive Behaviors and Party Safety, z = 11.56, p < .001 and z = 5.27, p < .001, respectively. General prior victimization was more strongly related to Party Safety than family-specific prior victimization, z = 4.04, p < .001 (see Table 3). Other comparisons of correlations between specific dimensions of the BBS and general versus family-specific prior victimization did not yield statistically significant differences.

Discussion

As predicted, the BBS demonstrated a four-factor solution, replicating Banyard et al.’s (2014) original findings in an ethnically, racially, and geo- graphically diverse sample from four universities within the United States. The four factors are Risky Situations, Accessing Resources, Proactive Behaviors, and Party Safety. While analyses also indicated that the four fac- tors could be subsumed under one overall Bystander Factor, the four-factor model had a better fit than the one-factor model. Correlations among the fac- tors ranged from .19 to .67, with Party Safety demonstrating the smallest correlations with the three other subscales. Five of the six items on the Party Safety scale measure direct preventive action one can take at a party to pre- vent SA or other unwanted behavior, whereas the other scales focus on inter- rupting high-risk situations for violence, providing support for victims of

Table 3. Correlations Among Bystander Behavior and Victimization Subscales.

Variable 1 2 3 4 5 6 7 8

1. Risky situations — 2. Accessing resources (SR) .67*** — 3. Proactive behaviors .58*** .55*** — 4. Party safety .33*** .19*** .29*** — 5. Bystander behavior total .90*** .73*** .78*** .58*** — 6. Total victimization .14** −.01 .18*** .05 .14** — 7. General victimization .13** −.01 .18*** .13** .16*** .87** — 8. Family-specific victimization .08 −.001 .11** −.06 .07 .79** .38** —

Note. Family-specific victimization = victimization in family of origin; general victimization = victimization in community and/or family or by peers; total victimization = family-specific and general victimization combined; SR = square root transformation. **p < .01. ***p < .001.

Cascardi et al. NP4865

violence, and describing education and skills training one can pursue. Because the party safety items all pertain to a specific context, it makes logical sense that they would be less highly correlated with the behaviors measured by the other three factor scales, which were highly correlated (.55 to .67). The find- ing that four bystander dimensions exist supports the idea that the ways indi- viduals may act to prevent SA and IPV are somewhat distinct and encourages research on distinct correlates of these dimensions.

Of particular importance, the current analyses were conducted with an analytic approach that is better suited to binary data than the analyses used in Banyard et al. (2014). The more rigorous CFA approach used herein also provided estimation methods to test the fit of the hypothesized measurement model based on theory of the underlying factor structure and prior analyses of related data (e.g., Muthén & Muthén, 2012). This analytic approach reduced problems associated with traditional PCA, such as attenuation of the association among items due to restricted range, which decreases the ability to identify dimensions (Bollen & Barb, 1981; Mislevy, 1986). It also reduced spurious inflation of associations due to the similarities in the distributional shape of the items, such that dimensions emerge based on high- or low-fre- quency endorsement and not theoretical constructs. Thus, the confirmation of the factor structure with this method strengthens the conclusions about the factor structure of the BBS. Also noteworthy, in Banyard and colleagues’ original factor analysis study, five items (two regarding party safety [e.g., “I watched a friend’s drink at a party”] and three regarding proactive confronta- tion [e.g., “I indicated my displeasure when I heard sexist, racist, homopho- bic jokes or catcalls made by a friend”]) did not clearly align with specific factors. In contrast, in the current study, these five items were assigned as expected, that is, Party Safety and Proactive Behaviors, with good model fit, perhaps because of the analytic approach used.

The current study also provides new information on the relation between prior victimization and the willingness to intervene as a bystander. In contrast with hypotheses, the association between Risky Situations and prior victimization was stronger than the correlation between Accessing Resources and prior victimization. The lack of association between prior victimization and Accessing Resources contradicts previous research. The 1-month assessment frame in the current study may have resulted in a lower estimate of Accessing Resources due to limited opportunities. That is, a longer assessment frame may have resulted in a higher rate of endorsement of these behaviors and stronger association with prior victimization. Alternatively, those who experience victimization have been found to be more likely to support victims themselves (Beeble, Post, Bybee, & Sullivan, 2008). Results of the current study suggest that prior victimization may

NP4866 Journal of Interpersonal Violence 36(9-10)

inhibit one’s willingness to intervene directly to support a victim of SA or IPV by contacting authorities or accompanying a friend seeking assistance. Although reasons for this are unclear, bystander education has been found to be more effective in increasing helpful bystander behavior among indi- viduals who perceived their college campus as responsive to disclosures of SA or IPV (Jouriles et al., 2020). In the context of the current study, it is plausible that those with prior victimization experiences had negative expe- riences accessing resources in the past, and this may lessen their willing- ness to seek support on behalf of other victims. It is also possible that the lack of association between prior victimization and Accessing Resources was due to restriction of range of scores on this BBS factor. Specifically, helping a victim of SA or IPV obtain medical, legal, or counseling support was uncommon: the mean score was 0.48 on an 8-point scale, and fewer than 10% of participants endorsed Accessing Resources items.

In contrast with hypotheses, there was no difference in the strength of association between total prior victimization (i.e., both general and family- specific) and Risky Situations and Proactive Behaviors. In particular, a wider assortment of victimization experiences, such as witnessing violence in the community (i.e., witnessing someone being physically assaulted) or being victimized by others (i.e., being physically assaulted), may be similar to the types of situations college students encounter on campus (i.e., suspecting a friend is in a physically abusive relationship). These similarities between the types of interpersonal victimization may heighten awareness about harm and increase knowledge about what to do when in situations that may lead to SA or IPV, which in turn may increase willingness to intervene in Risky Situations and engage in Proactive Behaviors. The present results are also consistent with the idea “altruism born of suffering” (Staub, 2003; Staub & Vollhardt, 2008), whereby victims show empathy and compassion toward the victimiza- tion of others. Thus, those with prior victimization may be more engaged bystanders who recognize risk, interrupt situations that may lead to SA or IPV, seek knowledge to prevent SA and IPV, and encourage others to speak out against these forms of violence.

Even though total prior victimization was not significantly associated with Party Safety, this BBS dimension was a significant correlate of general vic- timization and was associated with general victimization more strongly than victimization that occurred in the family. This finding suggests that experi- encing victimization by one’s peers, dating partners, or in one’s community may specifically heighten awareness about risks in situations involving peers and community gatherings, such as parties at college. It is particularly encour- aging that those who have been previously victimized take proactive steps to plan for safety at social gatherings to avoid future victimization.

Cascardi et al. NP4867

Limitations

The current study has several limitations that should be noted. The sample was not large enough to allow for reliable comparisons of bystander behavior and factor structure by gender, geographic region, or racial or ethnic group. Major findings about the factor structure from Banyard et al.’s (2014) original study were replicated in a college sample that includes a larger number of racial and ethnic minorities, but it remains unclear whether the components or rates of bystander behavior differ by gender, region, or racial or ethnic groups. Another limitation is the measurement of prior victimization, which was evaluated with a series of yes–no items. The intensity, duration, and recency of specific victimization experiences may relate differentially to BBS dimensions in more nuanced ways than what was found in the current study.

Implications and Future Directions

Results showed that different types of bystander behaviors had different cor- relations with prior victimization. Understanding the mechanisms by which prior victimization increases engagement in prosocial bystander behavior warrants additional study. Moreover, it is important to examine whether prior victimization moderates the efficacy of bystander education. It is conceivable that there may be a ceiling effect for those with prior victimization, such that they already understand how to be an effective bystander and do not benefit from general training.

Future research would also benefit from continued examination of the dif- ferent facets of bystander behavior. Results from the current study showed that there are distinct ways in which college students may help their peers protect each other at parties, intervene when they perceive risk for SA and IPV, and/or support their peers in the aftermath of an assault. Identifying cor- relates of the variable circumstances linked to prosocial bystander behavior may lead to more personalized intervention approaches and improve the design and effectiveness of bystander education programs (Banyard & Moynihan, 2011; McMahon & Banyard, 2012).

It is also important to recognize that, like many behavioral self-report measures, the BBS has limitations, most notably, the conflation of opportu- nity with action (e.g., McMahon et al., 2014; McMahon, Palmer, Banyard, Murphy, & Gidycz, 2017). It would be important to investigate whether those with prior victimization have both greater opportunity to intervene and whether they are also more likely to take action. For instance, the potential factors that drive a prior victim’s willingness to be helpful (e.g., greater awareness of risk, empathy) may also be related to noticing risky situations, which creates more opportunities to intervene.

NP4868 Journal of Interpersonal Violence 36(9-10)

In general, increasing the helpful behavior of bystanders who witness situ- ations involving SA and IPV may have an important role in reducing victim- ization on college campuses (Coker et al., 2015; Jouriles, Krauss, Vu, Banyard, & McDonald, 2018; Katz & Moore, 2013). Arguably, reductions in SA and IPV would, in turn, reduce or prevent a wide range of adverse psy- chological difficulties (e.g., posttraumatic stress disorder, depression, anxi- ety, and substance abuse), which are commonly experienced in the aftermath of victimization. From a policy standpoint, the strong support for distinct dimensions of helpful bystander action indicates that universities have mul- tiple avenues to increase community responsiveness to mobilize students to take action in an effort to prevent SA and IPV. Although interrupting risky situations is often viewed as the most immediate strategy for reducing victim- ization risk, results from the factor analysis support the idea that there are several different types of prosocial action that may be used to combat SA and IPV. Expressing disapproval of community norms that encourage perpetrator behaviors, spreading community awareness about SA and IPV, supporting victims who have already been assaulted, and developing proactive plans to stay safe at parties all may incrementally contribute to the community-wide effort to combat SA and IPV. Although the effect of bystander education on preventing SA and IPV has not yet been demonstrated in college samples, future research should examine whether specific types of bystander behavior differentially relate to decreases in SA and IPV.

The findings that prior victimization was related to Risky Situations, Proactive Behaviors, and Party Safety suggests that victims may have heightened awareness or sensitivity to the need to protect others in large gatherings that are common for college students. From a policy standpoint, exploring ways that individuals with prior victimization, especially expo- sure to multiple types of interpersonal violence outside of the family, may be engaged to educate their peers of the dangers of bystander apathy may open a new avenue for bystander education. There has been ample anec- dotal evidence that individuals with prior victimization may be in a pivotal position to share their stories as a means for building peer empathy and compassion in the larger community through raising public awareness and education. Developing sensitive strategies for reaching out to college stu- dents who have prior victimization experiences and tailoring programs that build on their pre-existing motivation to prevent SA and IPV may also enhance the effectiveness of bystander education.

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.

Cascardi et al. NP4869

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by Grants R21 HD075585 and R21 HD085063 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development.

ORCID iD

Kelli Sargent https://orcid.org/0000-0001-9316-3797

Supplemental Material

Supplemental material for this article is available online.

References

Banyard, V. L. (2008). Measurement and correlates of prosocial bystander behavior: The case of interpersonal violence. Violence and Victims, 23, 83-97.

Banyard, V. L., & Moynihan, M. M. (2011). Variation in bystander behavior related to sexual and intimate partner violence prevention: Correlates in a sample of col- lege students. Psychology of Violence, 1, 287-301.

Banyard, V. L., Moynihan, M. M., Cares, A. C., & Warner, R. (2014). How do we know if it works? Measuring outcomes in bystander-focused abuse prevention on campuses. Psychology of Violence, 4, 101-115.

Banyard, V. L., Moynihan, M. M., & Plante, E. G. (2007). Sexual violence prevention through bystander education: An experimental evaluation. Journal of Community Psychology, 35, 463-481.

Beeble, M. L., Post, L. A., Bybee, D., & Sullivan, C. M. (2008). Factors related to will- ingness to help survivors of intimate partner violence. Journal of Interpersonal Violence, 23, 1713-1729.

Bennett, S., Banyard, V. L., & Garnhart, L. (2014). To act or not to act, that is the ques- tion? Barriers and facilitators of bystander intervention. Journal of Interpersonal Violence, 29, 476-496.

Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246.

Bentler, P. M., & Bonett, D. G. (1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88, 588-606.

Bollen, K. A. (1989). A new incremental fit index for general structural equation models. Sociological Methods & Research, 17, 303-316.

Bollen, K. A., & Barb, K. H. (1981). Pearson’s r and coarsely categorized measures. American Sociological Review, 46, 232-239.

Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K.A. Bollen & J.S. Long (Eds.), Testing structural equation models (pp. 136- 162). Newbury Park, CA: Sage.

Casey, E. A., Lindhorst, T., & Storer, H. L. (2017). The situational-cognitive model of adolescent bystander behavior: Modeling bystander decision-making in the context of bullying and teen dating violence. Psychology of Violence, 7, 33-44.

NP4870 Journal of Interpersonal Violence 36(9-10)

Coker, A. L., Cook-Craig, P. G., Williams, C. M., Fisher, B. S., Clear, E. R., Garcia, L. S., & Hegge, L. M. (2011). Evaluation of Green Dot: An active bystander intervention to reduce sexual violence on college campuses. Violence Against Women, 17, 777-796.

Coker, A. L., Fisher, B. S., Bush, H. M., Swan, S. C., Williams, C. M., Clear, E. R., & DeGue, S. (2015). Evaluation of the Green Dot bystander intervention to reduce interpersonal violence among college students across three campuses. Violence Against Women, 21, 1507-1527.

Curran, P. J., West, S. G., & Finch, J. F. (1996). The robustness of test statistics to non- normality and specification error in confirmatory factor analysis. Psychological Methods, 1, 16-29.

Dalgeish, T., Moradi, A. R., Taghavi, M. R., Neshat-Doost, H. T., & Yule, W. (2001). An experimental investigation of hypervigilance for threat in child and adoles- cents with post-traumatic stress disorder. Psychological Medicine, 31, 541-547.

Edwards, K. M., Sylaska, K. M., Barry, J. E., Moynihan, M. M., Banyard, V. L., Cohn, E. S., . . . Ward, S. K. (2015). Physical dating violence, sexual violence, and unwanted pursuit victimization: A comparison of incidence rates among sexual-minority and heterosexual college students. Journal of Interpersonal Violence, 30, 580-600.

Fisher, B. S., Cullen, F. T., & Turner, M. G. (2000). The sexual victimization of col- lege women (Research Report). Washington, DC: Department of Justice, Bureau of Justice Statistics.

Flora, D. B., & Curran, P. J. (2004). An empirical evaluation of alternative methods of estimation for confirmatory factor analysis with ordinal data. Psychological Methods, 9, 466-491.

Frazier, P. A., & Berman, M. I. (2008). Posttraumatic growth following sexual assault. In S. Joseph & Linley P. A. (Eds.), Trauma, recovery, and growth: Positive psy- chological perspectives on posttraumatic stress (pp. 161-181). Hoboken, NY: John Wiley.

Gini, G., Albiero, P., Benelli, B., & Altoè, G. (2008). Determinants of adoles- cents’ active defending and passive bystanding behavior in bullying. Journal of Adolescence, 31, 93-105.

Hamby, S. L., Finkelhor, D., Ormrod, R. K., & Turner, H. A. (2004). The Juvenile Victimization Questionnaire (JVQ): Administration and scoring manual. Durham, NH: Crimes Against Children Research Center.

Hoyle, R. H., & Panter, A. (1995). Structural equation modeling: Concepts, issues, and applications. Thousand Oaks, CA: Sage.

Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance struc- ture analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6, 1-55.

Hutchinson, S. R., & Olmos, A. (1998). Behavior of descriptive fit indexes in con- firmatory factor analysis using ordered categorical data. Structural Equation Modeling: A Multidisciplinary Journal, 5, 344-364.

Jöreskog, K. G. (2005). Structural equation modeling with ordinal variables using LISREL (Technical report). Lincolnwood, IL: Scientific Software International, Inc.

Cascardi et al. NP4871

Jouriles, E. N., Krauss, A., Vu, N. L., Banyard, V. L., & McDonald, R. (2018). Bystander programs addressing sexual violence on college campuses: A sys- tematic review and meta-analysis of program outcomes and delivery methods. Journal of American College Health., 66(6), 457–466.

Jouriles, E. N., McDonald, R., Rosenfield, D., Levy, N., Sargent, K. S., Caiozzo, C., & Grych, J. H. (2016). TakeCARE, a video bystander program to help prevent sexual violence on college campuses: Results of two randomized controlled tri- als. Psychology of Violence, 6, 410-420.

Jouriles, E. N., Sargent, K. S., Salis, K. L., Caiozzo, C., Rosenfield, D., Cascardi, M., . . . McDonald, R. (2020). TakeCARE, a video to promote bystander behavior on college campuses: Replication and extension. Journal of Interpersonal Violence, 35(23-24), 5652–5675.

Katz, J., & Moore, J. (2013). Bystander education training for campus sexual assault prevention: An initial meta-analysis. Perspectives on College Sexual Assault: Perpetrator, Victim, and Bystander, 28, 1054-1067.

Kimble, M. O., Boxwala, M., Bean, W., Maletsky, K., Halper, J., Spollen, K., & Fleming, K. (2014). The impact of hypervigilance: Evidence for a forward feed- back loop. Journal of Anxiety Disorders, 28, 241-245.

Kimble, M. O., Fleming, K., & Bennion, K. A. (2013). Contributors to hypervigilance in a military and civilian sample. Journal of Interpersonal Violence, 28, 1672-1692.

Kleinsasser, A., Jouriles, E. N., McDonald, R., & Rosenfield, D. (2015). An online bystander intervention program for the prevention of sexual violence. Psychology of Violence, 5, 227-235.

Krebs, C. P., Lindquist, C. H., Warner, T. D., Fisher, B. S., & Martin, S. L. (2009). College women’s experiences with physically forced, alcohol- or other drug- enabled, and drug-facilitated sexual assault before and since entering college. Journal of American College Health, 57, 639-647.

Labhardt, D., Holdsworth, E., Brown, S., & Howat, D. (2017). You see but you do not observe: A review of bystander intervention and sexual assault on university campuses. Aggression and Violent Behavior, 35, 13-25.

Lee, I. A., & Preacher, K. J. (2013, September). Calculation for the test of the dif- ference between two dependent correlations with one variable in common [Computer Software]. Available from http://quantpsy.org

Li, C. H. (2016). Confirmatory factor analysis with ordinal data: Comparing robust maximum likelihood and diagonally weighted least squares. Behavior Research Methods, 48, 939-949.

McMahon, S., Allen, C. T., Postmus, J. L., McMahon, S. M., Peterson, N. A., & Lowe Hoffman, M. (2014). Measuring bystander attitudes and behavior to prevent sex- ual violence. Journal of American College Health, 62, 58-66.

McMahon, S., Palmer, J. E., Banyard, V., Murphy, M., & Gidycz, C. A. (2017). Measuring bystander behavior in the context of sexual violence prevention: Lessons learned and new directions. Journal of Interpersonal Violence, 32, 2396- 2418.

NP4872 Journal of Interpersonal Violence 36(9-10)

McMahon, S., Peterson, N. A., Winter, S. C., Palmer, J. E., Postmus, J. L., & Koenick, R. A. (2015). Predicting bystander behavior to prevent sexual assault on college campuses: The role of self-efficacy and intent. American Journal of Community Psychology, 56, 46-56.

McMahon, S., & Banyard, V. L. (2012). When can I help? A conceptual framework for the prevention of sexual violence through bystander intervention. Trauma, Violence, & Abuse, 13, 3-14.

Mislevy, R. J. (1986). Bayes modal estimation in item response models. Psychometrika, 51, 177-195.

Muthén, B. (1983). Latent variable structural equation modeling with categorical data. Journal of Econometrics, 22, 43-65.

Muthén, B. (1984). A general structural equation model with dichotomous, ordered cat- egorical, and continuous latent variable indicators. Psychometrika, 49, 115-132.

Muthén, L. K., & Muthén, B. O. (2012). Mplus version 7 user’s guide. Los Angeles, CA: Author.

Nabi, R. L., & Horner, J. R. (2001). Victims with voices: How abused women con- ceptualize the problem of spousal abuse and implications for intervention and prevention. Journal of Family Violence, 16, 237-253.

Nabors, E. L., & Jasinski, J. L. (2009). Intimate partner violence perpetration among college students: The role of gender role and gendered violence attitudes. Feminist Criminology, 4, 57-82.

National Center for Education Statistics. (2013). Percentage of 18- to 24-year-olds enrolled in degree-granting institutions, by level of institution and sex and race/ ethnicity of student: 1967 through 2012. Retrieved from http://nces.ed.gov/pro- grams/digest/d13/tables/dt13_302.60.asp

Palmer, J. E. (2016). Recognizing the continuum of opportunities for third parties to prevent and respond to sexual assault and dating violence on a college campus. Crime Prevention & Community Safety, 18, 1-18.

Shook, N. J., Gerrity, D. A., Jurich, J., & Segrist, A. E. (2000). Courtship violence among college students: A comparison of verbally and physically abusive cou- ples. Journal of Family Violence, 15, 1-22.

Staub, E. (2003). The psychology of good and evil: Why children, adults, and groups help and harm others. New York, NY: Cambridge University Press.

Staub, E., & Vollhardt, J. (2008). Altruism born of suffering: The roots of car- ing and helping after victimization and other trauma. American Journal of Orthopsychiatry, 78, 267-280.

Straus, M. A. (2004). Prevalence of violence against dating partners by male and female university students worldwide. Violence Against Women, 10, 790-811.

Tucker, L. R., & Lewis, C. (1973). A reliability coefficient for maximum likelihood factor analysis. Psychometrika, 38, 1-10.

Wang, L., Fan, X., & Willson, V. L. (1996). Effects of nonnormal data on param- eter estimates and fit indices for a model with latent and manifest variables: An empirical study. Structural Equation Modeling: A Multidisciplinary Journal, 3, 228-247.

Cascardi et al. NP4873

West, S. G., Finch, J. F., & Curran, P. J. (1995). Structural equation models with non- normal variables: Problems and remedies. In R. H. Hoyle (Ed.), Structural equa- tion modeling: Concepts, issues, and applications (pp. 56-75). Newbury Park, CA: Sage.

Author Biographies

Michele Cascardi is a professor of psychology and graduate program director of the William Paterson University PsyD program in clinical psychology. Her work focuses on measurement of intimate partner violence in adolescents and young adults, the link between childhood trauma and intimate partner violence, and prevention of sexual assault and intimate partner violence.

Alison Krauss is a doctoral student in clinical psychology. Her research interests include factors contributing to sexual assault perpetration, as well as sexual assault and antiracism intervention evaluation on college campuses.

K. Daniel O’Leary is distinguished professor of psychology. His work focuses on etiology, prevention, and treatment of psychological and physical aggression in inti- mate relationships; multivariate models (biological, psychological, and social) of inti- mate partner aggression; the bidirectional role of marital problems and depression; marital and dyad based treatments for clinical depression; prevalence and correlates of intense love.

Katie Lee Loatman is a licensed psychologist who has extensive training in the treat- ment of mood and anxiety disorders (such as depression, OCD, and social anxiety) as well as substance use disorders in adults and adolescents. She has specialized training in the delivery of cognitive behavioral therapy (CBT) approaches including exposure and response prevention (ERP) therapy.

Kelli Sargent is a doctoral student in clinical psychology. Her research interests include adolescent adjustment problems, interpersonal violence, and intervention evaluation for at-risk youth.

John Grych is a professor of psychology. His primary research interests focus on resil- ience in children and adolescents, the potential for bystander intervention to reduce sex- ual and physical violence, and aggression in family and romantic relationships.

Ernest N. Jouriles is a Dedman Family distinguished professor of psychology. He has two overlapping programs of research. One focuses on violence in adolescent romantic relationships. He is attempting to better understand risk factors for sexual and relationship violence among adolescents, and to use this knowledge to develop and evaluate interventions for preventing such violence. A second research program focuses on children’s exposure to interparental conflict and violence.