Annotated bibliography

MAJOR ARTICLE

An application of the theory of normative social behavior to bystander intervention for sexual assault

Tobias Reynolds-Tylus, PhDa�, Kaylee M. Lukacena, MAb, and Brian L. Quick, PhDc,d aSchool of Communication Studies, James Madison University, Harrisonburg, Virginia, USA; bDepartment of Communication, University of Kentucky, Lexington, Kentucky, USA; cDepartment of Communication, University of Illinois at Urbana-Champaign, Champaign, Illinois, USA; dCollege of Medicine, University of Illinois at Urbana-Champaign, Champaign, Illinois, USA

ABSTRACT Objective: Given the high prevalence of sexual assault on U.S. college campuses, the cur- rent study examines predictors of college students’ intentions to intervene to prevent sexual assault through the lens of the theory of normative social behavior (TNSB). Participants: One hundred eighty-six undergraduate students age 18–25 were recruited from an introduc- tory course at a large Midwestern university. Methods: Data were collected through an online survey during the 2015–2016 academic year. Results: Results indicated that descrip- tive norms, injunctive norms, and outcome expectations had direct positive effects on behavioral intention. However, no direct effect of group identity on intention was found. In addition to these main effects, an interaction between descriptive and injunctive norms was also observed. Conclusions: The results of the current study speak to theoretical questions surrounding the nature of TNSB variables, as well as several practical implications for coordi- nated efforts to promote bystander intervention on college campuses.

ARTICLE HISTORY Received 21 July 2017 Revised 18 May 2018 Accepted 9 July 2018

KEYWORDS Bystander intervention; descriptive norms; injunctive norms; sexual assault; theory of normative social behavior

An estimated 300,000 people in the United States are sexually assaulted each year.1 Among this population, young adults attending postsecondary educational institutions are at an increased risk for sexual victim- ization. While pursuing their degree at a college or university, it is estimated that one in five females and one in 16 males will be sexually assaulted.2 Of note, between the years of 1995 and 2013, women age 18–24 had the highest rate of sexual assault compared to any other age group.3 The consequences of sexual assault are severe, as survivors of sexual violence are at increased risk for a variety of harmful physical, social, and mental health outcomes.4 Given both the prevalence of sexual assault on college campuses and the severity of the consequences for survivors, a crit- ical area of inquiry is identifying effective communica- tion strategies for sexual assault prevention efforts.

The current investigation stands to move the litera- ture forward in two primary ways. First, by examining college students’ beliefs towards bystander intervention for sexual assault through the theoretical lens of the theory of normative social behavior (TNSB),5 this investigation holds the potential to shed light on effective message strategies for promoting bystander

intervention among college student populations. In short, bystander intervention programs focus on reframing sexual assault as a community issue, empow- ering community members to be active bystanders who stand up to prevent sexual assault, and seek to change social norms that contribute to the prevalence of sexual violence.6 Second, the current study seeks to move the scholarly literature forward by applying the TNSB5 in an underexamined context. To date, the TNSB has principally been applied within the context of college student drinking.7 In their call for future work on the theory, Rimal and Lapinski7 have challenged scholars to extend the TNSB literature by examining additional behavioral contexts and to continue gathering empirical data to test the propositional logical of the theory. The current study seeks to answer this call. We begin with an overview of the problem of sexual assault on college campuses

Sexual assault on college campuses

The United States Department of Justice2 defines sexual assault as “any type of sexual contact or behavior occurring without the explicit consent of the recipient.”

CONTACT Tobias Reynolds-Tylus [email protected] School of Communication Studies, James Madison University, 800 S Main St, Harrisonburg, VA 22807, USA. � 2018 Taylor & Francis Group, LLC

JOURNAL OF AMERICAN COLLEGE HEALTH 2019, VOL. 67, NO. 6, 551–559 https://doi.org/10.1080/07448481.2018.1499648

This definition includes any sexual coercive contact or behavior including fondling, incest, sexual intercourse, and attempted rape. Central to this definition of sexual assault is the concept of consent. At its most basic meaning, consent is an agreement between individuals to engage in sexual activity.8 In order to provide con- sent, however, an individual needs to be cognizant of the situation, know what he or she desires to do, and have the ability to communicate this desire.9 Situations in which an individual is incapable of giving consent include, but are not limited to: (1) instances in which an individual is threatened, forced, coerced, or manipulated into a sexual act against his or her will, (2) situations where an individual is unable to provide consent (ie, incapacitated due to drugs or alcohol), (3) circumstances where the individual does not have the mental capacity to give consent (ie, due to illness or disability), and (4) all cases where the individual is legally a minor.9

Given that college students are at a heightened risk for sexual victimization,3 identifying strategies for reducing rates of sexual assault on college campuses is critical considering the unwarranted health and social consequences arising from incidents of sexual assault. These include, but are not limited to physical, social, and mental health concerns.4 Physical health issues resulting from sexual assault include unintended preg- nancy and gynecological trauma, increased risk for sexually transmitted infections (STIs), including HIV, as well as an increase in high risk behaviors such as drug and alcohol use following the assault.4,10 In add- ition to these behavioral health consequences, there are also many serious consequences for survivors’ mental health. A recent meta-analysis11 of 40 years of research on the relationship between sexual assault and psycho- pathology found that survivors of sexual assault have significantly worse psychopathology than people who have not been assaulted. In particular, sexual assault survivors face an increased risk for obsessive-compul- sive conditions, suicidality, trauma- and stressor-related conditions, bipolar conditions, depression, anxiety, dis- ordered eating, and substance abuse/dependence.11 In respect to negative social outcomes, sexual violence negatively affects survivors’ relationships with their friends, family, coworkers, and romantic partners.12

For instance, many survivors often must endure victim blaming from their peers.13 Given the serious conse- quences of sexual assault for survivors, their friends, family, and society, identifying strategies for preventing sexual assault is an issue of national importance.2

As a strategy for reducing the alarming rates of sexual assault, many U.S. colleges and universities are

implementing bystander intervention programs.6

Federal law14 requires that all educational institutions receiving Title IX funding must educate students, fac- ulty, and staff on sexual assault, sexual violence, and rape prevention. Bystander intervention programs aim to reduce rates of sexual assault by fostering a sense of shared responsibility within the campus community for the prevention of sexual assault and dating vio- lence, and seek to change social norms that help to perpetuate sexual violence.6 Central to bystander intervention programs is the idea that everyone has the potential to be an active bystander to ensure all members of the community are safe, and it is critical to speak out or take action when one witnesses a harmful situation.15 Though sexual assault is com- monly associated with private settings, research shows that nearly one-third of sexually violent situations occur within the presence of bystanders.16 Thus, given the strong focus of bystander intervention programs on changing campus norms, we now turn to the TNSB to help elucidate concepts critical for informing bystander intervention efforts on college campuses.

The theory of normative social behavior

The TNSB5 was developed to better delineate the con- ditions under which social norms impact individuals’ behavior. Building off the work of Cialdini et al,17 the TNSB distinguishes between descriptive norms, which represent an individual’s perceptions about what is commonly done by others, and injunctive norms, which represent an individual’s perceptions of what is commonly approved of or disapproved of by others.5,17 The TNSB also specifies the underlying cognitive mechanisms believed to moderate the influ- ence of descriptive norms on individuals’ intentions and behavior.5 Therefore, the TNSB proposes that the effect of descriptive norms on behavior must be understood in the context of meaningful moderators. The core of the TNSB is the relationship between descriptive norms and three variables hypothesized to moderate this relationship: (1) injunctive norms, (2) outcome expectations, and (3) group identity. The current study extends TNSB work by testing its pre- dictions in the context of bystander intervention for sexual assault. As previous work has been somewhat inconsistent in supporting TNSB predictions,7 the cur- rent investigation provides an opportunity to critically evaluate the central propositions of the TNSB in a novel context. Below, we introduce the major theoret- ical constructs of the TNSB and critically discuss empirical work supporting TNSB predictions.

552 T. REYNOLDS-TYLUS ET AL.

Descriptive norms

Descriptive norms are conceptualized as individuals’ perceptions about the prevalence of a behavior among members of a referent group.17,18 Descriptive norms are thought to influence individuals’ behavior by providing information about what is adaptive for a particular situ- ation, as well as by serving as a decisional shortcut or heuristic cue for action.17,18 Empirical investigations across a wide range of health contexts5,19–21 have shown that descriptive norms are positively associated with greater behavioral intentions. Therefore,

H1: Descriptive norms will be positively associated with intention to intervene to prevent a sexual assault.

Injunctive norms

Injunctive norms represent individuals’ perceptions of what constitutes socially appropriate behavior in a par- ticular context.17,18 In contrast to descriptive norms, injunctive norms are thought to influence individuals’ behavior by providing social approval for compliance with the norm, and social sanctions for noncompli- ance.17,18 According to the TNSB, injunctive norms are thought to have both a direct positive effect on inten- tion, as well as an interactive effect on intention.5,18

Specifically, perceiving that many others engage in a behavior (ie, high descriptive norm) is more likely to motivate individuals to engage in the behavior if they also believe there is strong social pressure to conform (ie, high injunctive norm). Conversely, if both descrip- tive and injunctive norms are low, individuals are hypothesized to be less likely to engage in the behavior than if the two norms are divergent.5,18 However, empirical evidence in support of the hypothesized inter- active effect of injunctive and descriptive norms on behavior has been mixed. Some work has supported this interaction,22–24 whereas other work has found only dir- ect effects of injunctive norms on intention.5,25

Therefore, given remaining questions about the role of injunctive norms within the TNSB,

H2: Injunctive norms will be positively associated with intention to intervene to prevent a sexual assault.

H3: Injunctive norms will interact with descriptive norms such that behavioral intention will be strongest when both are high, and weakest when both are low.

Group identity

Group identity, a concept grounded in the social identity perspective,26 refers to feelings of affinity or

connectedness towards a particular social group.5 Studies have long shown the strong role that individuals’ social networks play in initiating and reinforcing negative27 and positive28 health behaviors. Furthermore, stronger identi- fication with a social group has been shown to enhance the likelihood that one will be influenced by ingroup members’ behavior.29 Accordingly, the TNSB proposes both a direct positive effect of group identity on inten- tion, as well as an interactive effect between descriptive norms and group identity.5,18,19 The hypothesized inter- action is such that when individuals perceive the preva- lence of a behavior among a referent group to be widespread, and they see themselves as similar to the members of the referent group, they are more likely to engage in the behavior themselves. In the absence of a strong identity with the referent group, however, there is no reason to expect that a strong descriptive norm should influence an individual’s behavior.5,22 Empirical support for the moderating role of group identity as specified by the TNSB has been found across various health con- texts.5,19,22,23 However, other work has failed to support this hypothesized interaction,20,21 and in general these findings have consistently been rather weak, often adding less than a 1% increase in additional variance explained to the model.22 Therefore, the current study seeks to revisit the role of group identity within the TNSB,

H4: Group identity will be positively associated with intention to intervene to prevent a sexual assault.

H5: Group identity will interact with descriptive norms such that when both group identity and descriptive norms are high, individuals will have the strongest behavioral intention.

Outcome expectations

Outcome expectations refer to individuals’ assessments of the benefits of taking action relative to the costs associ- ated with the action.30,31 Quite simply, individuals are more likely to engage in behaviors they believe will result in desirable outcomes for themselves.5 Therefore, accord- ing to the TNSB outcome expectations are expected to have a direct positive effect on intention, and are also hypothesized to moderate the relationship between descriptive norms and intention such that when a high descriptive norm is accompanied by beliefs that the behavior results in significant benefits (ie, high outcome expectations), individuals are more likely to engage in the behavior.5,18 Empirical work has broadly supported a dir- ect effect of outcome expectations on behavioral inten- tion,5,22 though empirical support for the moderating role of outcome expectations as proposed by the TNSB

JOURNAL OF AMERICAN COLLEGE HEALTH 553

has been somewhat mixed. Some work has fully support- ing the predicted interaction,19,22 whereas other work has found mixed support20,32 or no support for this predic- tion.5 Therefore, given remaining questions about the role of outcome expectations within the TNSB,

H6: Outcome expectations will be positively associated with intention to intervene to prevent a sexual assault.

H7: Outcome expectations will interact with descriptive norms such that when both outcome expectations and descriptive norms are high, individuals will have the strongest behavioral intention.

Methods

Participants and procedures

Following IRB approval, undergraduates (N ¼ 186) were recruited through an introductory undergraduate class at a large Midwestern university to complete an online survey for extra credit. Participants ranged in age from 18 to 25 (M ¼ 20.24, SD ¼ 1.40). The major- ity of the participants who completed the survey were female (n ¼ 134, 72.04%). In respect to race/ethnicity, most participants identified as White/Caucasian (n ¼ 93, 50.00%), followed by Asian (n ¼ 49, 26.34%), Black/African American (n ¼ 26, 13.98%), and Hispanic/Latino (n ¼ 19, 10.21%). Seven participants (3.76%) identified as some other race or ethnicity. In terms of the representativeness of the university at large, our sample closely mirrored the university in terms of racial/ethnic representation1, but was under- represented in terms of males.2 Participants were nearly evenly split by grade level (Freshman [n ¼ 45, 24.19%], Sophomore [n ¼ 55, 29.67%], Junior [n ¼ 47, 25.27%], Senior [n ¼ 38, 20.43%]). Approximately half of the participants (n ¼ 83, 44.62%) reported being involved in a Greek-letter fraternity or sorority. Fourteen participants (7.53%) reported that they were members of a campus collegiate sports team. In respect to knowledge about sexual assault, the vast majority of participants (n ¼ 171, 91.94%) indicated they had attended at least one education course on sexual assault and rape. Approximately half of the participants (n ¼ 95, 51.08%) reported personally knowing someone who was a survivor of sexual assault or rape. Approximately 12% of participants (n ¼ 22) reported they themselves had personally experienced sexual assault or rape. Nearly 16%

(n ¼ 29) of participants reported they had previously intervened at one point in order to prevent a sexual assault from occurring.

Measures

All items for the current study were measured on a 5- point Likert scale (1 ¼ strongly disagree to 5 ¼ strongly agree). All scales were computed by averaging the items. See Table 1 for a zero-order correlation matrix of study variables, means and standard deviations. See Appendix for the wording of all survey items.

Descriptive norms were assessed with three items33

(eg, “Most [university name] students would intervene to prevent sexual assault,” M ¼ 3.77, SD ¼ 0.73, a ¼ .77). Injunctive norms were assessed with five items33 (eg, “Most of my [university name] friends think that I should intervene to prevent sexual assault,” M ¼ 3.70, SD ¼ 0.71, a ¼ .84). Group identity was assessed with four items5 (eg, “I want to be like other [university name] students,” M ¼ 3.02, SD ¼ 0.82, a ¼ .82). Outcome expectations were assessed with five items5 (eg, “Intervening to prevent sexual assault could stop someone from getting hurt,” M ¼ 4.40, SD ¼ 0.66, a ¼ .92). Intention to intervene to prevent a sexual assault served as the primary dependent measure for this investigation. Behavioral intention was measured with four items34 (eg, “I intend to intervene in the future if I see a sexual assault, M ¼ 4.27, SD ¼ 0.64, a ¼ .94).

Results

To examine study hypotheses, a hierarchical regres- sion analysis was employed with biological sex, race, Greek-letter organization affiliation, collegiate sports affiliation, previous sexual assault and rape education experience, personally knowing a survivor of a sexual assault, being a victim of sexual assault, and previ- ously intervening to prevent a sexual assault initially included in the regression as covariates. As none of the demographic variables were significant predictors of intention, F (9, 174) ¼ 1.04, p > .05, R2 ¼ .00, these

Table 1. Zero-order correlation of study variables, means and standard deviations. Variables 1 2 3 4 5

1. Descriptive norms — 2. Injunctive norms .48�� — 3. Group identity .14 .30�� — 4. Outcome expectations .39�� .39�� .08 — 5. Intention .45�� .46�� .03 .61�� — Mean 3.77 3.70 3.02 4.40 4.27 SD 0.73 0.71 0.82 0.66 0.64

Note. �p < .05, ��p < .01.

1The university population where data were collected is approximately 53.8% White/Caucasian, 21.7% Asian, 13.5% Hispanic/Latino, 7.1% Black/ African American, and 3.7% multiracial. 2The university population is 54.6% male and 45.4% female.

554 T. REYNOLDS-TYLUS ET AL.

variables were dropped from further analyses. See Table 2 for the full results of the hierarch- ical regression.

H1 predicted that descriptive norms would be posi- tively associated with intention to intervene to prevent a sexual assault. To test H1, descriptive norms were entered into block 1 of the hierarchical regression. Consistent with our expectations, descriptive norms (b ¼ 0.45, p < .001) were positively associated with intention, F (1, 184) ¼ 45.69, p < .001, R2 ¼ .20. H2, H4, and H6 examined the direct effect of injunctive norms, group identity, and outcome expectations on intention. Specifically, injunctive norms (H2), group identity (H4), and outcome expectations (H6) were hypothesized to be positively associated with inten- tion. To test H2, H4, and H6, injunctive norms, group identity, and outcome expectations were entered into block 2 of the hierarchical regression. In block 2, injunctive norms (b ¼ .22, p< .001) and outcome expectations (b ¼ .46, p< .001) were positively associated with intention, F (4, 180)¼ 37.49, p< .001, R2 ¼ .44. Group identity3 was not associated with intention.

H3, H5, and H7 examined interactions consistent with the predictions of the TNSB.5 Specifically, H3 proposed that descriptive norms and injunctive would interact such that when both were concordant, their strength on intention would be maximized. H5 pro- posed that the effect of a high descriptive norm would be maximized when paired with a strong group iden- tity. H7 proposed that when outcome expectations and descriptive norms were both high, behavioral intention would be strengthened. In line with recom- mendations from Aiken and West,35 all predictor vari- ables were centered prior to moderation analysis. After centering the predictors, interaction terms were calculated by taking the product of both variables. In block 3, controlling for the main effects of all predic- tors, only the interaction between injunctive norms and descriptive norms (b ¼ �.13, p < .05) arose as a

significant interaction term, F (7, 177) ¼ 22.19, p < .001, R2 ¼ .45. As can be seen in Figure 1, a decomposition of this interaction demonstrates that when both injunctive norms and descriptive norms were low (�1 SD), participants had the lowest inten- tion overall.

Comment

Given that college students are at a heightened risk for sexual victimization,3 a broader theoretical under- standing of the antecedents to bystander intervention is critical for informing campaigns aimed at reducing the number of sexual assaults on college campuses. Two major objectives guided the current study. First, by examining predictors of college students’ bystander intervention intentions by utilizing the theoretical constructs of the TNSB,5 we sought to provide prag- matic evidence of use for coordinated efforts to pro- mote bystander intervention among college student populations. Our second major objective was to extend the literature on the TNSB by examining cen- tral questions about the role of TNSB constructs and behavioral intention more broadly. In doing so, we met the call of TNSB scholars who have advocated for more empirical work examining the propositional logic of the theory.7,22 In discussing our findings, we focus on the practical and theoretical contributions of the current study.

Overall, we found broad support for the utility of the TNSB in predicting bystander intervention inten- tions among college students. In particular, the inclu- sion of the TNSB variables into the regression model was able to explain approximately 47% of the variance in intention. The large variance explained by these variables lends support to the utility of the TNSB in the context of bystander intervention, and suggests the important of strategically emphasizing these con- structs in promotional efforts. In particular, our results revealed that descriptive norms (b ¼ .45) and outcome expectations (b ¼ .46) were quite strongly related to behavioral intention (see Table 2). From a practical standpoint, campaign designers may benefit by strategically emphasizing these constructs in cam- paign messages and materials aimed at promoting bystander intervention. For instance, promotional messages targeting descriptive norms may emphasize the large number of students who have intervened to help a peer, or the number of who have formally pledged to prevent sexual assault on their campus (see www.itsonus.org). Likewise, campaign messages emphasizing the positive outcomes associated with

3Based on the suggestion of one reviewer, we conducted subgroup analyses examining the role of Greek life identity on study outcomes. Specifically, given that we had a sizeable number of Greek life-affiliated students in our sample (44.6%, n ¼ 86), we conducted separate hierarchical regressions for Greek life and non-Greek life affiliated students. Our results showed a similar pattern of findings in respect to main effects among each sub-group, with some subtle differences. Overall, both regressions explained approximately the same amount of variance (R2 ¼ .48 for non-Greeks; R2 ¼ .52 for Greeks). For both Greek and non-Greek students, group identity was not associated with intention. For Greek students, outcome expectations were most strongly associated with intention (b ¼ .58, p < .001), followed by descriptive norms (b ¼ .40, p < .001). Interestingly, among Greek students, injunctive norm was not associated with intention. For non-Greek students, descriptive norms were the strongest predictor of intention (b ¼ .50, p < .001), followed by outcome expectations (b ¼ .34, p < .001), and injunctive norms (b ¼ .33, p < .001).

JOURNAL OF AMERICAN COLLEGE HEALTH 555

bystander intervention (such as personal profiles of university students describing the benefits of bystander intervention) provide just one instance of how campaigns can strategically emphasize outcome expectations within their outreach efforts.

Another finding with both theoretical and practical implications was the observed interaction between descriptive and injunctive norms. A decomposition of this interaction (see Figure 1) demonstrated that when injunctive norms and descriptive norms were both low (�1 SD), individuals had the lowest intention to intervene to prevent sexual assault. This finding is consistent with what would be predicted by the TNSB,5 thus lending further empirical support for the hypothesized interaction between injunctive norms and descriptive norms.5,18 Beyond the theoretical implications of this finding, this interaction between descriptive and injunctive norms seems likely to be of particular interest to campaign efforts. From a prac- tical standpoint, these findings suggest that injunctive norms may be a critical construct for targeting within campaign messages. Namely, the results from the cur- rent investigation revealed that when both injunctive and descriptive norms were low, individuals’ inten- tions were the lowest overall. Therefore, these findings suggest that strategic efforts to strengthen injunctive norms, in addition to descriptive norms, may be par- ticularly consequential in this context.

In line with previous work,5,36 we found direct effects of both injunctive norms and outcome expecta- tions on intention (see Table 2). In contrast to previ- ous work,19,22 however, our results did not support the hypothesized interaction between outcome expect- ations and descriptive norms. Though the original formulation of the TNSB5 proposed an interactive effect of outcome expectations and descriptive norms,

previous work has failed to consistently support this finding.5,20,32 This has led some scholars to suggest a reconsideration of the role outcome expectations play within the TNSB.18,22 Our findings provide further evidence that outcome expectations may influence intentions in a direct, rather than moderated, manner.

Interestingly, we found that group identity was completely unrelated to intention (see Tables 1 and 2). Though some previous work has found a direct effect of group identity on intention,20 our results did not support this finding. Nor did we find an inter- active effect between descriptive norms and group identity as would be predicted by the TNSB.5

Considering these findings, a discussion of referent groups seems relevant. In the current study, the refer- ent group utilized was the general student population at the university where the data were collected (eg, “other [university name] students”). A growing body of evidence suggests that norms emanating from more proximal referent groups (eg, same-race, same-sex) hold a stronger influence on behavior than those norms emanating from more distal referent groups (eg, a “typical” student on campus).18,25,36 A boundary condition of the hypothesized interactive effect of group identity and descriptive norm within the pur- view of the TNSB is that individuals must have a strong identity with the referent group in order for the descriptive norms of that group to influence their behavior. In the current study, group identification with university students was moderate (M ¼ 3.02, SD ¼ 0.82; see Table 1). Therefore, future work may benefit from examining referent groups with higher group identification, such as Greek letter organiza- tions, as these organizations hold the potential to be powerful partners in promoting bystander interven- tion efforts.

From a message design standpoint, the nature of distal versus proximal referent groups in respect to campaign messages presents an interesting dilemma.

Table 2. Hierarchical regression predicting intention to inter- vene to prevent a sexual assault. Predictors b DR2 Total R2 F-value

Block 1 .20��� .20 45.69��� Descriptive norms (DN) .45���

Block 2 .26��� .45 37.49��� Injunctive norms .22��� Group identity �.10 Outcome expectations .46���

Block 3 .01 .47 22.19��� DN x injunctive norms �.13� DN x group identity .06 DN x outcome expectations .03

Note. �p < .05, ��p < .01, ���p < .001.

Figure 1. Behavioral intention as a function of descriptive and injunctive norms. Note. Low ¼ M � 1 SD, High ¼ M þ 1 SD.

556 T. REYNOLDS-TYLUS ET AL.

Clearly, a growing body of research suggests that emphasizing closer referent groups in promotional messages would likely constitute a fruitful strategy for developing effective persuasive messages.25,36

However, despite evidence showing that norms result- ing from more proximal referent groups are more impactful on individuals’ behavior than those emanat- ing from more distal referent groups,25,36 practical considerations constrain this approach in respect to its implementation. Though it may be quite simple to manipulate normative perceptions of a more distal referent group (eg, “typical” college student at a uni- versity) in campaign messages, this becomes more dif- ficult in the case of closer referents (eg, friends, family). Specifically, individuals have greater access to normative information to closer referents than they do from more distal referents (eg, “typical” college student), and therefore these message manipulations run the risk of being hampered by believability con- cerns. Clearly, this is an area ripe for further inquiry.

Limitations and future directions

Despite the aforementioned theoretical and practical implications, our study is not without limitations. First, our data were cross sectional, and as such we cannot make claims about causality. Future work may benefit from the use of panel designs that can assess behavioral outcomes rather than simply behavioral intention. Second, as our data were collected at a sin- gle large public university in the Midwest, we caution the reader in generalizing these results to other col- leges and universities. A final limitation is related to the diversity of our sample in respect to biological sex. As our sample was predominantly female, the results reported here primarily reflect the beliefs of female undergraduate students. Though our sample was representative of the university population in most respects (eg, race/ethnicity, grade-level), our sample was underrepresented in terms of males. Future work would benefit by recruiting more male participants in order to better represent their voices on this important issue.

Our data were collected in the spring of 2016. On October 15, 2017 actor Alyssa Milano encouraged women on Twitter to tweet #MeToo if they had been sexually harassed or assaulted. This single tweet brought viral attention to conversations about sexual assault and harassment. Though our data predate the #MeToo movement, both bystander intervention efforts and the #MeToo movement are similar in that both efforts are focused on reducing the sheer

prevalence of unwanted sexual acts. In respect to nor- mative message framing, however, the #MeToo move- ment and bystander intervention efforts differ in interesting ways. Bystander intervention efforts typic- ally provide normative information that is “positively valanced”37 in that the focus of the norm is to direct individuals to act in a particular way (ie, prescribing that one should intervene in situations where consent has not been given; see www.itsonus.org). In contrast, the #MeToo movement typically focuses on norms that are “negatively valanced”37 in that the normative information being communicated highlights the preva- lence of socially undesirable behavior (ie, people shar- ing their #MeToo stories on social media highlights the pervasiveness of sexual harassment and assault). Future research may benefit by examining how undergraduates respond to positively vs. negatively framed norms messages advocating for bystander intervention.

Conclusion

The current investigation examined predictors of col- lege students’ bystander intervention intentions through the lens of the TNSB.5 In doing so, our results speak to several theoretical questions surrounding the nature of TNSB variables. Our results revealed direct effects for descriptive norms, injunctive norms, and outcome expectations on intention, yet no support for a direct effect of group identity on intention. An interaction between descriptive and injunctive norms consistent with TNSB predictions was also observed. However, our results did not reveal interactions between descriptive norms and outcome expectations, nor between descrip- tive norms and group identity, as would be predicted by the TNSB. The results of our study also speak to several practical implications for coordinated efforts to pro- mote bystander intervention on college campuses. Namely, descriptive norms, outcome expectations, and injunctive norms seem to be particularly consequential determinants of intention, and as such seem to be choice constructs for targeting via strategic message design. The promises and pitfalls of targeting close ver- sus distal referent groups within normative messages were also discussed. It is our hope that practitioners, educators, and government officials will consider these findings when developing future bystander interven- tion campaigns.

Conflict of interest disclosure No potential conflict of interest was reported by the authors.

JOURNAL OF AMERICAN COLLEGE HEALTH 557

References

1. Rape, Abuse, & Incest National Network. Statistics. Rape, Abuse, & Incest National Network Website. https://rainn. org/statistics?gclid¼CIXQksn3kKcCFRhzgwodMFUSbw. Accessed July 20, 2017.

2. United States Department of Justice. Protecting stu- dents from sexual assault. United States Department of Justice Website. https://www.justice.gov/ovw/ protecting-students-sexual-assault. Published March 1, 2017. Accessed July 20, 2017.

3. Sinozich S, Langton L. Special Report: Rape and Sexual Assault Victimization among College-Age Females, 1995–2013. Washington, DC: United States Department of Justice; 2014. Bureau of Justice Statistics publication NCJ 248471.

4. World Health Organization. Understanding and Addressing Violence against Women: Sexual Violence. World Health Organization Web site. http://apps.who. int/iris/bitstream/10665/77434/1/WHO_RHR_12.37_ eng.pdf. Published 2012. Accessed July 20, 2017.

5. Rimal RN, Real K. How behaviors are influenced by perceived norms: a test of the theory of normative social behavior. Commun Res. 2005;32(3):389–414. doi:10.1177/0093650205275385

6. Banyard VL, Moynihan MM, Plante EG. Sexual vio- lence prevention through bystander education: an experimental evaluation. J Commun Psychol. 2007; 35(4):463–481. doi:10.1002/jcop.20159

7. Rimal RN, Lapinski MK. A re-explication of social norms, ten years later. Commun Theor. 2015; 25(4):393–409. doi:10.1111/comt.12080

8. Rape, Abuse, & Incest National Network. What con- sent looks like. Rape, Abuse, & Incest National Network Web site. https://www.rainn.org/articles/ what-is-consent. Accessed July 20, 2017.

9. United States Department of Health and Human Services. Sexual assault. United States Department of Health and Human Services Web site. https://www. womenshealth.gov/files/documents/fact-sheet-sexual- assault.pdf. Published September 18, 2015. Accessed July 20, 2017.

10. Turchik JA, Hassija CM. Female sexual victimization among college students: assault severity, health risk behaviors, and sexual functioning. J Interpers Violence. 2014;29(13):2439–2457. doi:10.1177/0886260513520230

11. Dworkin ER, Menon SV, Bystrynski J, Allen NE. Sexual assault victimization and psychopathology: a review and meta-analysis. Clin Psychol Rev. 2017; 56:65–81. doi:10.1016/j.cpr.2017.06.002

12. Langton L. Special Report: Socio-Emotional Impact of Violent Crime. Washington, DC: United States Department of Justice; 2014. Bureau of Justice Statistics publication NCJ 247076.

13. Randall M. Sexual assault law, credibility, and “ideal victims”: consent, resistance, and victim blaming. Can J Wom Law. 2010;22(2):397–433. doi:10.3138/ cjwl.22.2.397.

14. Campus Sexual Violence Elimination Act, S 128, 113th Cong, 1st Sess. 2013.

15. Banyard VL, Plante EG, Moynihan MM. Bystander education: bringing a broader community perspective

to sexual violence prevention. J Community Psychol. 2004;32(1):61–79. doi:10.1002/jcop.10078

16. Planty M. Third-Party Involvement in Violent Crime, 1993–1999. Washington, DC: United States Department of Justice; 2012. Bureau of Justice Statistics publication NCJ 247076.

17. Cialdini RB, Reno RR, Kallgren CA. A focus theory of normative conduct: recycling the concept of norms to reduce littering in public places. J Pers Soc Psychol. 1990;58(6):1015–1026. doi:10.1037/0022-3514.58.6.1015.

18. Lapinski MK, Rimal RN. An explication of social norms. Commun Theory. 2005;15(2):127–147. doi:10.1111/j.1468-2885.2005.tb00329.x.

19. Lapinski MK, Anderson J, Shugart A, Todd E. Social influence in child care centers: a test of the theory of normative social behavior. Health Commun. 2014;29(3):219–232. doi:10.1080/10410236.2012.738322.

20. Mabry A, Turner MM. Do sexual assault bystander interventions change men’s intentions? Applying the theory of normative social behavior to predicting bystander outcomes. J Health Commun. 2016;21(3): 276–292. doi:1080/10810730.2015.1058437.

21. Rimal RN, Lapinski MK, Cook RJ, Real K. Moving toward a theory of normative influences: how per- ceived benefits and similarity moderate the impact of descriptive norms on behaviors. J Health Commun. 2005;10(5):433–450. doi:10.1080/10810730591009880.

22. Rimal RN. Modeling the relationship between descriptive norms and behaviors: a test and extension of the theory of normative social behavior (TNSB). Health Commun. 2008;23(2):103–116. doi:10.1080/10410230801967791.

23. Rimal RN, Real K. Understanding the influence of perceived norms on behaviors. Commun Theory. 2003;13(2):184–203. doi:10.1111/j.1468-2885.2003. tb00288.x.

24. Smith JR, Louis WR. Do as we say and as we do: the interplay of descriptive and injunctive group norms in the attitude–behaviour relationship. Br J Soc Psychol. 2008;47(4):647–666. doi:10.1348/014466607X269748.

25. Woolf J, Rimal RN, Sripad P. Understanding the influence of proximal networks on high school athletes’ intentions to use androgenic anabolic ste- roids. J Sport Manage. 2014;28(1):8–20. doi:10.1123/ jsm.2013-0046.

26. Tajfel H. Social psychology of intergroup relations. Annu Rev Psychol. 1982;33(1):1–39. doi:10.1146/ annurev.ps.33.020182.000245.

27. Fraser MW, Hawkins JD. The social networks of opi- oid abusers. Int J Addict. 1984;19(8):903–917. doi:10.3109/10826088409061994.

28. House JS, Landis KR, Umberson D. Social relation- ships and health. Science. 1988;241(4865):540–545. doi:10.1126/science.3399889.

29. Christensen PN, Rothgerber H, Wood W, Matz DC. Social norms and identity relevance: a motivational approach to normative behavior. Pers Soc Psychol Bull. 2004;30(10):1295–1309. doi:10.1177/0146167204264480.

30. Bandura A. Social Learning Theory. Englewood Cliffs, NJ: Prentice Hall; 1977.

31. Bandura A. Social Foundations of Thought and Action: A Social Cognitive Theory. Englewood Cliffs, NJ: Prentice-Hall; 1986.

558 T. REYNOLDS-TYLUS ET AL.

32. Carcioppolo N, Jensen JD. Perceived historical drink- ing norms and current drinking behavior: using the theory of normative social behavior as a framework for assessment. Health Commun. 2012;27(8):766–775. doi:10.1080/10410236.2011.640973.

33. Park HS, Klein KA, Smith S, Martell D. Separating subjective norms, university descriptive and injunctive norms, and US descriptive and injunctive norms for drinking behavior intentions. Health Commun. 2009;24(8):746–751. doi:10.1080/10410230903265912.

34. Fishbein M, Ajzen I. Predicting and Changing Behavior: The Reasoned Action Approach. New York, NY: Psychology Press; 2010.

35. Aiken LS, West SG. Multiple Regression: Testing and Interpreting Interactions. Thousand Oaks, CA: Sage; 1991.

36. Neighbors C, LaBrie JW, Hummer JF, et al. Group iden- tification as a moderator of the relationship between per- ceived social norms and alcohol consumption. Psychol Addict Behav. 2010;24(3):522–528. doi:10.1037/ a0019944.

37. Cialdini RB, Demaine LJ, Sagarin BJ, et al. Managing social norms for persuasive impact. Soc Influence. 2006;1(1):3–15. doi:10.1080/15534510500181459

Appendix: Survey items

Descriptive norms

� “Most of my friends at [university name] would inter- vene to prevent sexual assault”

� “Most [university name] students would intervene to prevent sexual assault”

� “Most of the people important to me at [university name] would intervene to prevent sexual assault”

Injunctive norms

� “Most [university name] students think that I should intervene to prevent sexual assault”

� “Most [university name] students think that I should intervene to prevent sexual assault”

� “Most people at [university name] who are important to me think that I should intervene in a sexual assault situation”

� “My [university name] friends may judge me based on whether or not I intervened to prevent sexual assault”

� “Most [university name] students that I know would judge me based on whether or not I intervened to pre- vent sexual assault”

Group identity

� “I see myself as similar to other [university name] students”

� “I want to be like other [university name] students” � “It is important for me to do what other [university

name] students think I should do” � “What [university name] students think I should do is

important to me”

Outcome expectations

� “Intervening to prevent sexual assault sets a good example”

� “Intervening to prevent sexual assault increases others’ sense of responsibility”

� “Intervening to prevent sexual assault could stop some- one from getting hurt”

� “Intervening to prevent sexual assault could reduce the amount of sexual assault cases on college campuses”

� “My intervening to prevent sexual assault will make [university name] a safer place for students”

Intention

� “I intend to intervene in the future if I see a sex- ual assault”

� “In the near future, I will intervene if I see a sexual assault situation”

� “If I see a sexual assault about to take place, I would intervene to prevent it from happening”

� “I would be willing to intervene to prevent sex- ual assault”

JOURNAL OF AMERICAN COLLEGE HEALTH 559

Copyright of Journal of American College Health is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.