Action and Reaction: The Impact of Consequences of Intervening in Situations of Interpersonal Violence

Abstract

Previous research has examined barriers to helping in situations involving interpersonal violence, though little has investigated the outcomes of interventions that actually do take place. The purpose of the present study was to explore how consequences that bystanders experienced in helping situations varied by characteristics of the incident, including the type of interpersonal violence (i.e., harassing comment, dating violence, unwanted sexual advances, and controlling behavior) and the bystander’s relationship to the victim and perpetrator. We also examined whether these outcomes impacted the likelihood that bystanders would help again. Participants (n = 1,391) were recruited from a university psychology subject pool and Amazon Mechanical Turk during fall 2016. They completed online or in-person surveys consisting of quantitative measures. Descriptive statistics showed that bystanders experienced both positive (e.g., positive reactions from the victim) and negative consequences (e.g., negative reactions from others) after helping. Analyses of variance revealed that helping in instances of dating violence was associated with more negative consequences, while helping in situations of unwanted sexual advances was associated with more positive consequences. Regression analyses showed that bystanders were more likely to help again when they experienced more positive and less negative feelings about their actions. Analyses of variance demonstrated that when the victim was a close friend, bystanders reported more positive consequences and desire to help again. However, bystanders reported more negative reactions from the perpetrator when the perpetrator was a stranger. Implications for adapting intervention programming to promote the likelihood that bystanders will help in future situations are discussed.

Keywords

intervention sexual assault dating violence harassment

Interpersonal violence, including sexual and dating violence, controlling behavior, and harassment, has been highlighted as a nationwide public health problem (Black et al., 2011; Krebs, Lindquist, Warner, Fisher, & Martin, 2007). Previous work documented that acts of unwanted sexual contact, dating violence, and controlling behavior frequently occur in the presence of bystanders, or witnesses to high-risk situations who are not directly involved in the situation as the victim or perpetrator (Hamby, Weber, Grych, & Banyard, 2016; Latané & Darley, 1970; Planty, 2002). Bystanders can respond before, during, and after these situations. Specifically, bystanders can help the victim, support the perpetrator, or ignore the situation (Burn, 2009). To date, much of the bystander research has explored factors that influence the likelihood that bystanders will intervene (Latané & Darley, 1970). However, there is only minimal research about what happens after a bystander takes action. The purpose of the current study was to explore how consequences of bystander action varied based on characteristics of the helping situation, including the type of interpersonal violence (i.e., harassing comment, unwanted sexual advances, dating violence, and controlling behavior) and the bystander’s relationship with the victim and perpetrator. We also examined how these outcomes impacted the bystander’s desire to help again. Exploring what happens after bystanders intervene allows both researchers and practitioners to better prepare bystanders to handle potential consequences of their actions and to promote future bystander action.

Review of the Literature

Research exploring bystander consequences in relation to interpersonal violence (e.g., sexual and dating violence) has largely focused on what influences bystanders’ decisions to offer help. This work has examined how bystanders’ perceptions of what might happen after helping influence their decision to take action (Banyard, Moynihan, & Plante, 2007). Latané and Darley’s (1970) original situational model of bystander intervention addressed the weighing of costs and benefits of helping. This work documented that individuals reported greater intent to help when the imagined benefits outweighed the potential costs.

More recent work has adapted this model to incorporate instances of sexual and dating violence (Bennett, Banyard, & Garnhart, 2013; Burn, 2009). For example, research has found that the fear of negative evaluations from others served as a barrier to intervention in sexual assault situations through processes such as audience inhibition and evaluation apprehension (Burn, 2009; Latané & Darley, 1970). In this study, potential bystanders reported decreased desire to help when they perceived that their peers would react poorly to their actions. Furthermore, Ryan and Wessel (2012) documented that individuals reported higher levels of intent to help in instances of sexual harassment when they perceived greater potential benefits, such as social approval. In addition, perceived safety has been found to influence potential bystanders’ intent to help in situations involving sexual and dating violence. Specifically, individuals reported less intent to help when they perceived they would be met with physical consequences (Bennett & Banyard, 2016; Edwards, Rodenhizer-Stämpfli, & Eckstein, 2015). While perceived consequences of intervening impact bystanders’ decisions to help, we know little about how much bystanders actually experience these consequences (e.g., bystander safety, negative feedback) after helping. More importantly, we do not know how such consequences influence future action. It is possible that if a bystander successfully and safely intervened (e.g., stopped the incident and everyone was safe) in a situation and received positive reactions from others (e.g., the victim expressed gratitude), he or she may be more likely to help again in the future.

Recent work has attempted to extend this model to incorporate consequences of bystander action after an intervention. The Action Coils model of bystander behavior suggests that bystander intervention functions as a feedback loop, where the outcomes of helping have ramifications for future action (Banyard, 2015). After taking action, bystanders can experience a range of consequences, such as situational outcomes (e.g., bystander safety) and positive or negative reactions from the parties involved (i.e., victim, perpetrator, and other bystanders who were present). This model theorizes that what happens after bystanders help impacts their decision to help again in future situations. The Action Coils model also proposes that these consequences (e.g., bystander safety, reactions of the parties involved) shape how bystanders feel about their own behavior, which may contribute to whether or not they intend to offer help in the future. Intervention programs that aim to increase bystander behavior often address the barriers outlined by Latané and Darley’s (1970) situational model. Yet, the research on bystander consequences has been underdone, which may limit the abilities of these programs to educate and prepare bystanders to take action and handle the repercussions of their actions. Thus, as suggested by the Action Coils model, there is a need to understand what bystanders experience after taking action and how these outcomes impact future helping behavior.

To date, only few studies have actually examined what happens after bystanders take action in the context of interpersonal violence (Hamby et al., 2016; Moschella, Bennett, & Banyard, 2018; Witte, Casper, Hackman, & Mulla, 2017). One such study focused on reactions of the victim and perpetrator as reported by bystanders who intervened in situations involving sexual assault. In this exploratory study, bystanders reported that victims tended to react positively (e.g., appreciated bystander’s help) and perpetrators tended to react negatively (e.g., became angry with the bystander) toward the bystander’s actions (Moschella et al., 2018). Moschella and colleagues (2018) also found that positive and negative reactions varied by the type of bystander behavior (e.g., distancing the victim, distracting the perpetrator). However, this study only inquired about situations involving sexual violence; thus, the current study expands upon this research by considering additional forms of interpersonal violence (i.e., harassing comments, dating violence, and controlling behavior).

Furthermore, Witte and colleagues (2017) documented that intervening in instances of sexual and intimate partner violence was associated with positive feelings, including pride and empowerment (Witte et al., 2017). This study also showed that bystanders experienced negative consequences, such as traumatic stress symptoms, after helping (Witte et al., 2017). These negative outcomes are likely to impact the bystander’s desire to help again in the future, as suggested by the Action Coils model (Banyard, 2015). However, this question has been neglected in the current literature. In addition, this study did not examine how outcomes differ by the type of interpersonal violence (i.e., sexual or dating violence). Therefore, the current study attempted to fill the gaps about type of violence (i.e., harassing comments, dating violence, unwanted sexual advances, and controlling behavior) and the desire to help again.

Another study also inquired about victims’ perspective on outcomes of bystander action. For example, victims reported greater well-being if they perceived that bystanders were helpful and unharmed (Hamby et al., 2016). This was true for a variety of violence types (i.e., peer-perpetrated violence, adult-perpetrator violence), except for sexual violence (Hamby et al., 2016). Given that victims’ perceptions of sexual violence were different from other types of violence, it is important to continue exploring how outcomes differ for a variety of types of interpersonal violence (i.e., harassing comments, dating violence, unwanted sexual advances, and controlling behavior). Thus, the present study goes beyond victim perceptions by considering the perspectives of bystanders.

In addition, we know from research on bystander intervention that characteristics of the situation, such as the bystander’s relationship to the victim and perpetrator, impact action. Recent research has explored how intent to help may differ based on these relationships. For example, in a vignette study, potential bystanders expressed greater intent to help victims if they knew the victim (Bennett & Banyard, 2016; Bennett, Banyard, & Edwards, 2017). Potential bystanders were also more likely to confront perpetrators who were their friends (Bennett & Banyard, 2016; Bennett et al., 2017; Nicksa, 2014). However, there may be increased costs associated with helping friends if the bystander’s actions are unwanted or considered unhelpful. These findings suggest that the potential benefits or damage to the bystander’s relationship with the victim and perpetrator may influence future bystander behavior. To date, however, few studies have asked questions about the impact of bystander actions on bystanders, victims, and perpetrators.

The Current Study

The purpose of the present study was to investigate how consequences of bystander action differed based on characteristics of the helping situation, including the type of interpersonal violence (i.e., harassing comment, unwanted sexual advances, dating violence, and controlling behavior) and the bystander’s relationship to the parties involved (i.e., victim and perpetrator). We also examined how outcomes were related to bystanders’ self-reported desire to help again. The following research aims were explored:

Aim 1: We aimed to examine how the outcomes that bystanders experience varied by type of interpersonal violence (i.e., harassing comment, unwanted sexual advances, dating violence, and controlling behavior).

Aim 2: We aimed to explore the relationship between types of consequences (i.e., positive and negative) and bystanders’ desire to help again.

Aim 3: We aimed to examine the relationship between the desire to help again and the type of interpersonal violence (i.e., harassing comment, unwanted sexual advances, dating violence, and controlling behavior).

Aim 4: We aimed to explore how outcomes and the desire to help again varied by the bystander’s relationship with the victim and perpetrator (e.g., close friend, stranger).

Method

Participants

Two methods of data collection were used to increase the diversity of our sample of young adults between the ages of 18 and 24 years. Participants were recruited through Amazon Mechanical Turk (MTurk; n = 4,322) and a university psychology subject pool (n = 674). We chose to recruit participants between the ages of 18 and 24 years given that individuals (both students and nonstudents) in this age range are at the highest risk for sexual and dating violence (Rennison, 2001; Sinozich & Langton, 2014). To increase generalizability and capture a broader range of experiences, both college students and nonstudents were included in all analyses.

MTurk sample

Of the 4,322 MTurk participants, only 717 individuals passed the initial screening survey and completed the full survey. The participants (n = 3,605) who were screened out did not meet the requirement for age (i.e., did not fall between the ages of 18 and 24 years), country of residence (i.e., they lived outside of the United States), or helping behavior (i.e., reported not helping in a situation involving harassment, dating violence, unwanted sexual advances, and controlling behavior). The average age was 22.25 years (SD = 1.57). This sample consisted of 356 males (49.7%) and 359 females (50.1%); two participants identified as “other.” Most identified as heterosexual (83.1%). More than half the sample identified as Caucasian (57.4%), followed by Asian (18.3%), Hispanic (7.8%), African American (7.7%), American Indian (4.6%), Pacific Islander (0.7%), and Other (0.6%); several participants (2.9%) indicated two or more races. The majority of this sample had at least some college experience, an associate’s degree, or a bachelor’s degree (81.9%).

Psychology subject pool

For those who were recruited from the psychology subject pool, 129 participants (19.1%) were from in-person and 545 participants (80.9%) from online. The average age was 19.07 years (SD = 1.28). This sample included 479 females (71.8%) and 186 males (27.9%); there were two participants who identified as “other.” The majority of this sample was Caucasian (89.6%) and identified as heterosexual (95.3%). Most participants were also in their freshman or sophomore year at university (81.9%). Of these participants, only 354 reported helping in an instance of interpersonal violence. There were some differences between participants who reported helping and participants who did not report helping. Individuals who helped were less likely to be heterosexual, were less socially desirable, had more opportunity to help, and had higher scores on bystander efficacy and intent to help.

Sample comparison

The psychology subject pool and MTurk samples differed significantly in terms of their gender (MTurk included less females), age (psychology subject pool was younger), ethnicity (psychology subject pool had less racial diversity), and sexual orientation (MTurk included less heterosexual participants). We examined the impact of these demographic factors in another paper (Banyard, Moschella, Grych, & Jouriles, under review).

Procedures

The survey procedures slightly differed based on recruitment method (i.e., MTurk and psychology subject pool). Participants from the psychology subject pool were first given a consent form that described the study and the potential risks. These procedures were the same regardless of in-person or online survey administration. Participants then answered questions on demographic information and bystander behavior. If participants reported offering help in a situation involving a harassing comment, dating violence, unwanted sexual advances, or controlling behavior, they were asked questions on outcomes they experienced as a result of helping in their most recent incident. Participants who did not report helping were instructed to skip ahead (for the in-person survey) or were redirected (for the online survey) to other bystander measures (e.g., bystander efficacy, intent to help). Upon survey completion, psychology subject pool participants received partial course credit and were provided with a debriefing form that included a list of resources to contact for questions about interpersonal violence.

MTurk participants were given an informed consent form and then directed to several screening questions (i.e., age, country of residence, helping behavior) in exchange for US$0.10. If participants qualified for the survey, they were directed to another online survey that inquired about demographic information and bystander behavior (the social desirability, readiness to help, and decisional balance scales were not included in the MTurk survey to shorten it). Participants then answered questions on outcomes they experienced in response to the most recent situation they helped in (i.e., harassing comment, dating violence, unwanted sexual advances, or controlling behavior), along with other bystander measures (e.g., bystander efficacy, intent to help). Participants received an additional US$2.50 for completing the second survey and were provided with a debriefing form and a list of resources upon survey completion.

Measures

Banyard and colleagues (under review) created a series of new scales to measure actual outcomes and reactions reported by bystanders who intervened in situations involving interpersonal violence (i.e., harassing comment, unwanted sexual advances, dating violence, and controlling behavior).

These new measures were given to participants who indicated that they had helped someone in a situation related to interpersonal violence in the past 6 months. They were then asked to think about the most recent time they were in a situation where they chose to help and answered several sets of questions (see below) about that specific incident.

Type of interpersonal violence

Participants were asked to think about the most recent time they did something to try to help in a situation involving interpersonal violence. They were then presented with a list of four types of interpersonal violence and asked to choose which type described the most recent situation they helped in. Each type of violence was broadly described in several sentences, which are summarized below. Harassing comments included catcalling or street harassment, telling sexist, racist, homophobic, or rape jokes, and blaming a victim of sexual assault or relationship abuse. This was the most frequent type of interpersonal violence reported by bystanders (n = 344; 35.6%). Controlling behavior was described as any situation where they observed someone receiving or sending incessant texts or phone calls, demanding to know or being asked where he or she is or what he or she is doing, and someone’s phone being monitored. This was reported by more than a quarter of bystanders (n = 261; 27.0%). Unwanted sexual advances included any situation where they observed behaviors that are unwanted by the recipient and include remarks about physical appearance, persistent sexual advances that are undesired by the recipient, and unwanted touching and unwanted oral, anal, or vaginal penetration. This was reported by less than 20% of bystanders (n = 181; 18.7%). Dating violence was described as any situation where they observed emotional, psychological, or physical abuse between two relationship partners, including use of physical force or threats of force against a partner; any situation where one or both relationship partners were arguing, physically fighting, name calling, insulting, or intimidating one another. This was the least common type of interpersonal violence reported (n = 180; 18.6%).

Bystander relationship and desire to help again

Two questions were asked about the bystander’s relationship to the people involved in the situation. Participants reported that the person who needed help (i.e., victim) was either an acquaintance (n = 131; 15.4%), stranger (n = 164; 19.3%), close friend (n = 485; 57.1%), or other (n = 70; 8.2%). Participants reported that the person who was harming (i.e., perpetrator) was either an acquaintance (n = 330; 38.8%), stranger (n = 298; 35.1%), close friend (n = 173; 20.4%), or other (n = 49; 5.8%). Participants were asked how likely it is that they would help again if they were given the opportunity, which was rated on a 5-point scale ranging from not likely to very likely (M = 4.46, SD = 0.96).

Incident-specific consequences

Participants were asked if they had experienced various positive or negative outcomes in response to a specific situation where they intervened (see Table 1 for items). They answered “Yes” or “No” (scored as 1 and 0) to indicate their response. Factor analysis yielded three constructs. Positive outcomes consisted of three items (e.g., stopping the incident and people were okay). Cronbach’s alpha for positive outcomes was .35 (M = 1.47, SD = 0.94). Intangible negative outcomes included five items that describe verbal threats and harassment (e.g., being harassed or verbally threatened). Tangible negative outcomes consisted of five items that include physical acts and consequences (e.g., cost a lot of time, got in trouble as a result of my actions). Cronbach’s alpha for negative intangible outcomes was .79 (M = 0.66, SD = 1.19), while Cronbach’s alpha for negative tangible outcomes was .71 (M = 0.63, SD = 1.12). We did not present the factor analysis in the results below because this measure was not the focus of the article.

Table 1.

Incident-Specific Outcome Descriptives.

	Opportunity (%)	Yes (%)
What I did stopped the incident and people were okay.	79.8	62.6
I saw positive posts on social media about what I did.	71.2	15.2
People said positive things about me to others because of what I did.	74.2	35.0
I was harassed because of what I did.	77.0	9.7
People said negative things about me to others because of what I did.	76.8	9.4
I was physically hurt because of what I did.	78.3	7.1
I was threatened physically because of what I did.	77.8	9.4
I was threatened verbally because of what I did.	78.1	14.9
I got in trouble as a result of my actions.	77.5	6.3
My involvement ended up costing me a lot of time.	78.2	12.2
What I did made things worse.	78.5	5.9
What I did stopped the incident, but I learned things got bad later.	75.9	11.3
I was interviewed as a witness because of what I did.	75.4	9.7

Bystander feelings

Bystander feelings is a new measure including 16 items that inquire about positive, negative, and unsure feelings after engaging in bystander behavior (Banyard et al., under review). Participants indicated how much they felt a certain way on a 4-point scale (ranging from not at all to a lot). The Positive subscale includes six items (e.g., “Felt good about doing something”) and Cronbach’s alpha was .91 (M = 3.28, SD = 0.71). The Negative subscale includes seven items (e.g., “Wished I hadn’t stepped in”). Cronbach’s alpha on the Negative subscale was .92 (M = 1.54, SD = 0.78). The Unsure subscale includes three items (e.g., “Felt confused about what to do”) and Cronbach’s alpha was .59 (M = 2.12, SD = 0.83).

Bystander action response–victim (BAR-V)

This measure is a new series of nine items inquiring about both positive and negative responses of the victim after engaging in bystander behavior (Banyard et al., under review). Participants answered “Yes” or “No” (scored as 1 and 0) to indicate whether they had experienced various positive or negative outcomes in response to a specific situation where they intervened. The Positive subscale includes five items (e.g., “The action strengthened my friendship with the person I tried to help”) and Cronbach’s alpha was .83 (M = 3.36, SD = 1.85). The Negative subscale includes four items. An example item is “Person I tried to help got mad or upset at me.” Cronbach’s alpha on the Negative subscale was .73 (M = 0.50, SD = 0.96).

Bystander action response–perpetrator (BAR-P)

This measure is a new series of four items inquiring about the negative responses of the perpetrator after bystanders have taken action (Banyard et al., under review). Participants answered “Yes” or “No” (scored as 1 and 0) to indicate whether they had experienced a range of negative outcomes in response to a specific situation where they intervened (e.g., “Person who was harming told me to mind my own business”). Cronbach’s alpha on for this measure was .66 (M = 1.36, SD = 1.31).

Bystander action responses–other (BAR-O)

This measure is a new measure including 14 items that inquire about positive, negative, and apathetic responses of other bystanders after engaging in bystander behavior (Banyard et al., under review). Participants indicated “Yes” or “No” (scored as 1 and 0) whether they had experienced various outcomes in response to a specific situation where they intervened. The Positive subscale includes four items (e.g., “Others thanked me for what I did”) and Cronbach’s alpha was .78 (M = 2.00, SD = 1.55). The Negative subscale includes six items (e.g., “Others said I did the wrong thing”) and Cronbach’s alpha was .79 (M = 0.77, SD = 1.36). The Apathy subscale includes four items (e.g., “Others acted like nothing happened [ignored what I did]”) and Cronbach’s alpha was .78 (M = 1.32, SD = 1.44).

Data Analysis Plan

Only participants who reported helping in a situation involving a harassing comment, dating violence, unwanted sexual advances, or controlling behavior were included in analyses. Thus, the sample for the current analyses included 966 bystanders. Mean scores were created for each subscale. For the dichotomous outcome subscales (i.e., victim, perpetrator, and other bystanders), mean scores were multiplied by the number of questions in each subscale. This procedure allowed us to interpret the mean scores as the total number of positive, negative, or apathetic items participants experienced (as opposed to a 0 to 1 scale). These average scores were used in all analyses (including mean scores given above).

Descriptive statistics were calculated for incident-specific outcomes. To explore the first research aim (i.e., how outcomes varied by type of interpersonal violence), a series of ANOVAs were performed. Furthermore, to examine the second research aim (i.e., explore the relationship between the desire to help again and consequences of bystander action), correlation and linear regression analyses were conducted. To explore the third research aim (i.e., examine the relationship between the desire to help again and types of interpersonal violence), one-way ANOVAs were conducted. In addition, to examine the fourth research aim (i.e., how consequences and the desire to help again differed by the relationship between the bystander and the victim and perpetrator), one-way ANOVAs were performed.

Results

Incident-Specific Outcomes

As shown in Table 1, the most frequently reported positive outcome was “What I did stopped the incident and people were okay” (62.6%, n = 605), followed by “People said positive things about me to others because of what I did” (35.0%, n = 338). Participants reported negative outcomes less often. The most frequent intangible negative outcome was “I was threatened verbally because of what I did” (14.9%, n = 144), whereas the most frequent tangible negative outcome was “My involvement ended up costing me a lot of time” (12.2%, n = 118).

Type of Interpersonal Violence and Consequences

To explore our first research aim, a series of one-way ANOVAs were performed to test whether consequences of bystander action varied by the type of interpersonal violence (i.e., harassing comment, dating violence, unwanted sexual advances, and controlling behavior). As shown in Table 2, the levels of unsure feelings significantly differed between types of interpersonal violence, F(3, 845) = 13.24, p < .001, η² = .05. Post hoc tests using the Bonferroni correction (p < .0083) indicated that unsure feelings toward dating violence were significantly higher than all other types of violence. Means and standard errors can be found in Table 2. Levels of negative feelings also significantly differed across types of interpersonal violence, F(3, 840) = 13.42, p < .001, η² = .05. Bonferroni post hoc tests indicated that negative feelings toward dating violence were significantly higher than all other types of violence. Furthermore, levels of positive feelings significantly differed between types of violence, F(3, 786) = 5.16, p < .01, η² = .02. Post hoc tests using the Bonferroni correction indicated that positive feelings toward unwanted sexual advances were significantly higher than controlling behavior.

Table 2.

ANOVA Type of Interpersonal Violence and Consequences.

Consequences	Harassing comment			Dating Violence			Unwanted Sexual Advances			Controlling Behavior			F	η²
Consequences	n	M	SE	n	M	SE	n	M	SE	n	M	SE	F	η²
Feelings—unsure	302	1.99	.05	167	2.44^a,b,c	.06	160	1.96	.06	220	2.16	.06	13.24***	.045
Feelings—negative	301	1.40	.04	165	1.86^a,b,c	.06	160	1.48	.06	218	1.55	.05	13.42***	.046
Feelings—positive	283	3.27	.04	156	3.23	.06	146	3.49^c	.06	205	3.21	.05	5.16**	.019
Victim—negative	266	0.25^c,d	.06	158	0.87^b	.07	149	0.34	.08	209	0.64	.06	17.61***	.064
Victim—positive	259	3.32	.11	160	3.02	.15	150	3.92^c,d	.15	209	3.27	.13	6.73***	.025
Perpetrator—negative	279	1.07^c,d	.08	153	1.71	.11	143	1.40	.11	200	1.48	.09	9.02***	.034
Others—negative	261	0.63	.08	145	1.17^a,b	.11	144	0.66	.11	180	0.74	.10	5.60**	.023
Others—apathy	249	1.32	.09	142	1.29	.12	139	1.66^c	.12	172	1.06	.11	4.52**	.019
Help again	302	4.56	.05	167	4.20^a,b	.07	160	4.63	.08	221	4.38	.06	7.82***	.027

Note. Others – positive is not included due to lack of significance.

Significantly different from harassing comment.

Significantly different from unwanted sexual advances.

Significantly different from controlling behavior.

Significantly different from dating violence.

p < .05. **p < .01. ***p < .001.

In addition, negative victim reactions also significantly differed across types of interpersonal violence, F(3, 778) = 17.61, p < .001, η² = .06. Bonferroni post hoc tests indicated that negative victim reactions were significantly higher for dating violence and controlling behavior than for harassing comment, and higher for dating violence than unwanted sexual advances. Positive victim reactions also significantly differed between types of violence, F(3, 774) = 6.73, p < .001, η² = .03. Bonferroni post hoc tests indicated that positive victim reactions were significantly higher for unwanted sexual advances than for dating violence and controlling behavior. The negative perpetrator reactions significantly differed across types of interpersonal violence, F(3, 771) = 9.02, p < .001, η² = .03. Post hoc tests using the Bonferroni correction indicated that negative perpetrator reactions were significantly higher for dating violence and controlling behavior than for harassing comment. Furthermore, negative other reactions significantly differed between types of interpersonal violence, F(3, 726) = 5.60, p < .01, η² = .02. Bonferroni post hoc tests indicated that reactions of others were significantly higher for dating violence than for harassing comment and unwanted sexual advances. Apathetic other reactions also significantly differed across types of violence, F(3, 698) = 4.52, p < .01, η² = .02. Post hoc tests using the Bonferroni correction that apathetic other reactions were significantly higher for unwanted sexual advances than controlling behavior. There was not a significant difference between the types of interpersonal violence for positive reactions of other bystanders.

Help Again and Consequences

To explore our second research aim (i.e., examine the relationship between positive and negative consequences and the desire to help again), we performed a series of correlations and linear regressions. As shown in Table 3, higher levels of bystanders’ positive feelings were associated with an increased desire to help again in the future. Bystanders who reported more positive reactions from victims and other bystanders reported being more likely to help again. Furthermore, greater reports of bystanders’ negative feelings and feeling of uncertainty were related to less likelihood of helping again. Bystanders who reported more negative reactions from victims and from other bystanders also reported decreased desire to help again in the future. Negative reactions from the perpetrator and apathetic reactions of other bystanders were not related to the desire to help again. The regression model for desire to help again and bystander consequences was significant, R² = .44 and F(7, 632) = 70.16, p < .001 (see Table 4). Only bystanders’ negative and positive feelings were significant, β = –.59, t(639) = −12.85, p < .001 and β = .22, t(639) = 6.54, p < .001, respectively.

Table 3.

Correlations Between Consequences and Desire to Help Again.

	Help Again
Feelings—unsure	−.39***
Feelings—negative	−.59***
Feelings—positive	.34***
Victim—negative	−.32***
Victim—positive	.36***
Perpetrator—negative	−.023
Others—negative	−.30***
Others—positive	.12**
Others—apathy	.01

p < .05. **p < .01. ***p < .001.

Table 4.

Regression Analyses for Consequences and Desire to Help Again.

	B	SE	β	t	F
Help again					70.16***
Constant	4.45	.18	—;	24.13
Feelings—unsure	−0.02	.05	−.02	−0.51
Feelings—negative	−0.68	.05	−.59***	−12.85
Feelings—positive	0.31	.05	.22***	6.54
Victim—negative	0.07	.04	.07	1.73
Victim—positive	0.02	.02	.05	1.29
Others—negative	0.003	.03	.01	0.12
Others—positive	−0.01	.02	−.02	−0.62

p < .05. **p < .01. ***p < .001.

Helping Again and Type of Interpersonal Violence

To examine our third research aim (i.e., explore the relationship between the desire to help again and the type of interpersonal violence), we conducted a one-way ANOVA. As shown in Table 2, the model for helping again and type of interpersonal violence was significant, F(3, 846) = 7.82, p < .001, η² = .03. Bonferroni post hoc tests (p < .0083) indicated that bystanders were significantly less likely to help again in situations involving dating violence than a harassing comment and unwanted sexual advances.

Helping Again and Relationship Type

To explore our fourth research aim, we performed two one-way ANOVAs to test whether the desire to help again varied by the bystander’s relationship with the victim and perpetrator (see Table 5). The desire to help again significantly differed by relationship to the person who needed help, F(3, 845) = 6.86, p < .001, η² = .02. Bonferroni post hoc tests (p < .0083) indicated that bystanders reported being significantly more likely to help again when helping a close friend compared with stranger. The bystander’s relationship with the perpetrator was not related to helping again.

Table 5.

ANOVA Victim Relationship and Consequences.

Consequences	Acquaintance			Stranger			Close Friend			Other			F	η²
Consequences	n	M	SE	n	M	SE	n	M	SE	n	M	SE	F	η²
Help again	130	4.25	.08	164	4.26^a	.07	485	4.55	.04	70	4.64	.11	6.86***	.024
Feelings—unsure	131	2.17	.07	164	2.33^a	.07	484	2.03	.04	69	2.14	.10	5.61**	.020
Feelings—negative	130	1.72^a	.07	162	1.78^a,b	.06	483	1.44	.04	68	1.36	.09	11.94***	.041
Feelings—positive	122	3.29	.06	153	3.23	.06	448	3.34	.03	66	3.07	.09	3.22*	.012
Victim—positive	116	2.76	.17	150	2.95	.15	459	3.67^c,d	.09	52	3.11	.25	11.65***	.043
Others—positive	118	1.96	.14	147	1.74	.13	409	2.14	.08	56	1.64	.21	3.54*	.014

Note. Victim—negative, perpetrator—negative, others—negative, and others—apathy are not included due to lack of significance.

Significantly different from close friend.

Significantly different from other.

Significantly different from acquaintance.

Significantly different from stranger.

p < .05. **p < .01. ***p < .001.

Relationship Type and Consequences

To further examine our fourth research aim (i.e., how outcomes varied by the bystander’s relationship with the victim and perpetrator), we conducted two ANOVA analyses (see Tables 5 and 6). Levels of unsure feelings significantly differed by the bystander’s relationship with the person who needed help, F(3, 844) = 5.61, p < .01, η² = .02. Bonferroni post hoc tests (p < .0083) indicated that bystanders reported significantly higher levels of uncertainty when the victim was a stranger compared with a close friend. Furthermore, levels of negative feelings significantly differed by the bystander’s relationship with the person who needed help, F(3, 839) = 11.94, p < .001, η² = .04. Bonferroni post hoc tests indicated that bystanders reported significantly higher levels of negative feelings when the victim was an acquaintance and stranger compared with a close friend. The model for positive feelings and the bystander’s relationship with the person who needed help was significant, F(3, 785) = 3.22, p < .05, η² = .01. However, there were no significant differences across types of relationships using the Bonferroni correction. In addition, positive victim reactions significantly differed by the bystander’s relationship with the person who needed help, F(3, 773) = 11.65, p < .001, η² = .04. Bonferroni post hoc tests indicated that bystanders reported significantly more positive victim reactions when the victim was a close friend compared with an acquaintance and stranger. The model for positive other reactions and the bystander’s relationship with the person who needed help was also significant, F(3, 726) = 3.54, p < .05, η² = .01. However, there were no significant differences across types of relationships using the Bonferroni correction. The bystander’s relationship with the victim was not related to negative victim, perpetrator, or other reactions and apathy of others.

Table 6.

ANOVA Perpetrator Relationship and Consequences.

Consequences	Acquaintance			Stranger			Close Friend			Other			F	η²
Consequences	n	M	SE	n	M	SE	n	M	SE	n	M	SE	F	η²
Feelings—positive	306	3.37	.04	270	3.29	.04	168	3.19	.05	45	3.06	.11	3.99*	.015
Victim—positive	303	3.54	.11	280	3.51	.11	155	2.82^a,b	.15	39	2.94	.29	6.83***	.026
Perpetrator—negative	292	1.41	.08	279	1.47	.08	164	1.06^b	.10	39	1.44	.21	3.80*	.015

Note. Help again, Feelings—unsure, feelings—negative, victim—negative, others—negative, others—positive, and others—apathy are not included due to lack of significance.

Significantly different from acquaintance.

Significantly different from stranger.

p < .05. **p < .01. ***p < .001.

As shown in Table 6, the model for positive feelings and the bystander’s relationship with the person who was harming was significant, F(3, 785) = 3.99, p < .05, η² = .02. However, there were no significant differences across types of relationships using the Bonferroni correction (p < .0083). Furthermore, positive victim reactions significantly differed by the bystander’s relationship with the person who was harming, F(3, 773) = 6.83, p < .001, η² = .03. Bonferroni post hoc tests indicated that bystanders reported significantly more positive victim reactions when the perpetrator was an acquaintance and stranger compared with a close friend. In addition, negative perpetrator reactions significantly differed by the bystander’s relationship with the person who was harming, F(3, 770) = 3.80, p < .05, η² = .02. Post hoc tests using the Bonferroni correction indicated that bystanders reported significantly more negative perpetrator reactions when the perpetrator was a stranger compared with a close friend. The bystander’s relationship with the perpetrator was not related to unsure feelings, negative feeling, negative victim reactions, negative other reactions, positive other reactions, and apathy of others.

Discussion

The present study aimed to examine how consequences that bystanders reported varied by characteristics of the helping incident, including the type of interpersonal violence (i.e., harassing comment, dating violence, unwanted sexual advances, and controlling behavior) and the bystander’s relationship to the victim and perpetrator. We also explored how these outcomes impacted the likelihood that bystanders would help again. Studying consequences of bystander action helps us gain a better understanding of how to promote future bystander behavior and better prepare bystanders to handle potential consequences. Overall, bystanders reported most frequently helping in instances of harassing comments, followed by controlling behavior. Dating violence and unwanted sexual advances had lower, almost equal rates of intervention.

Bystanders may be less likely to intervene in instances of dating and sexual violence due to the unique barriers posed by these situations, such as ambiguity and severity of the situation (Bennett et al., 2013; Burn, 2009). For example, if a potential bystander is unsure whether sexual advances are consensual, they may be hesitant to help for fear of acting on an incorrect assumption (Burn, 2009). Bennett and colleagues (2013) also found that bystanders reported being less likely to help when they were unsure if the victim wanted help. Further, if an argument between two people does not meet a potential bystander’s threshold for dating violence, they will be less likely to recognize that help is needed and offer help (Edwards et al., 2015). Potential bystanders may also be concerned about getting hurt by placing themselves in the middle of an argument or a situation where they are unsure whether their help is wanted (Bennett & Banyard, 2016; Edwards et al., 2015). The possible risks associated with sexual and dating violence may be perceived as more severe than speaking out about an offensive comment or joke. These challenges highlight the need for intervention programs to educate potential bystanders about risk factors for sexual and dating violence, so they can better identify when these situations need help and feel safe doing so.

We found that bystanders reported a varying degree of consequences for different types of interpersonal violence. Our first research aim explored how outcomes varied by type of interpersonal violence. We found that dating violence was associated with the most negative consequences (e.g., negative feelings about helping, negative reactions from others). Negative reactions from victims have the potential to cause damage to or end their friendship with the bystander. The reactions from victims may differ depending on how they perceive their relationship with the perpetrator and whether they are contemplating leaving their partner. For example, victims may react more negatively to help from bystanders if they do not want their relationship with the perpetrator to end, and the bystander’s help is deemed unwelcome. Intervention programming should consider including strategies that address how to approach close friends about their unhealthy relationships to reduce conflict between friends.

In addition, bystanders reported more positive consequences (e.g., feeling positive about their actions, positive reactions from victims) in instances of unwanted sexual advances, even though these situations were identified as being more severe. It is possible that situations at the high end of the sexual assault continuum send a clear message that help is needed, which increases the bystander’s certainty about their actions and the victim’s gratitude toward the bystander. On the contrary, situations that are low risk may be viewed as safer, which makes it more likely that bystanders will offer help. Experiencing positive feelings and receiving positive reactions from victims after helping is consistent with prior research (Moschella et al., 2018; Witte et al., 2017). More work should be done to identify situational characteristics (e.g., severity) and the types of behaviors the bystanders engaged in that led to these positive responses. Previous research suggests that the types of actions bystanders take influence the outcomes they receive from the victim and perpetrator (Moschella et al., 2018). Training programs could utilize this information to promote positive helping experiences.

Our second research aim was to examine the relationship between the desire to help again and consequences of bystander action. We found that positive consequences were positively related to the desire to help again. It makes sense that bystanders who felt good about helping and received positive reactions were able to fondly recall their helping experience, which increased their desire to help again. Indeed, these positive reactions may work against audience inhibition, which has been found to be a barrier to taking action (Burn, 2009; Latané & Darley, 1970). According to the Action Coils model, it is likely that if bystanders are presented with another opportunity to help, they will reflect on these positive helping experiences (Banyard, 2015). Recalling these positive feelings and reactions is likely to reaffirm their decision to take action again. Bystanders may also be more likely to overcome barriers to helping, such as audience inhibition, and gain confidence in their ability to help by recalling prior positive experiences (Burn, 2009; Latané & Darley, 1970).

However, we also found that negative consequences were negatively associated with the desire to help again. Experiencing negative consequences may present an additional barrier when faced with an opportunity to intervene in the future. This suggests that barriers like audience inhibition can affect future behaviors if confirmed by negative reactions in the present. Even if bystanders are able to overcome the barriers suggested by Latané and Darley (1970) and Burn (2009), it is possible that before taking action they may compare their current helping situation with past negative helping experiences. Bystanders who may have once felt they possessed the knowledge to help may now doubt their abilities due to these negative consequences. To avoid feeling poorly about their actions and fearing negative reactions, it is likely that bystanders will choose not to help. It is also possible that bystanders who experience negative consequences experience more traumatic stress symptoms, and that these stress symptoms influence their decision to help in the future (Witte et al., 2017).

Our third research aim was to examine the relationship between the desire to help again and types of interpersonal violence. Our results showed that bystanders reported being more likely to help again in situations involving a harassing comment and unwanted sexual advances. Given that bystanders reported the most positive consequences to helping in instances of harassing comments and unwanted sexual advances, it makes sense that they would be more likely to help again in those situations. These findings provide support for the Action Coils model, and suggest that more work should be done to educate bystanders about the possible consequences they may experience as a result of their actions (Banyard, 2015). Intervention programming should not only prepare bystanders for both positive and negative reactions but also educate them on how to respond to de-escalate or diffuse the situation.

Our fourth research aim examined how consequences and the desire to help again varied by the relationship between the bystander and the victim and perpetrator. Bystanders were more likely to be close friends with victims but were more likely to be strangers or acquaintances with perpetrators. We found that bystanders reported being more likely to help again when the victim was a close friend, which is consistent with prior research (Bennett et al., 2017). Bystanders also reported more positive consequences (e.g., positive reactions from victims) when the victim was a close friend. Being close friends with the victim may facilitate helping in some instances because bystanders may be more knowledgeable about the situation and have more opportunities to discuss the situation with the victim than a stranger or acquaintance. Victims may also be more comfortable asking for help from the bystander if they are close friends, which would most likely result in more positive outcomes. In addition, bystanders reported more negative reactions from the perpetrator when the perpetrator was a stranger compared with a close friend. It is possible that bystanders may be better able to diffuse the situation or mitigate negative reactions if they have an existing relationship with the perpetrator. More research is needed to explore the nuances of the relationship between bystanders, victims, and perpetrators. Given that bystanders are likely to be with close friends when these incidences occur, intervention programming should consider training bystanders within their friend networks.

Implications for Research and Practice

To date, a host of research has examined the linear progression of stages that underlie the process bystanders undergo when deciding to take action (Burn, 2009; Latané & Darley, 1970). However, our findings support the Action Coils model, which suggests that bystander intervention functions like a feedback loop, where bystanders factor in their own feelings about their actions and reactions from the people involved (i.e., victim, perpetrator, and other bystanders) when deciding to offer help in future situations (Banyard, 2015). Future research should continue to explore how situational characteristics (e.g., severity) impact consequences experienced by bystanders, along with which strategies and actions lead to the most positive outcomes to increase the likelihood that bystanders will help again.

In addition, our findings have a multitude of implications for intervention programming. Given that our sample includes college students and nonstudents, our findings can be applied to both college- and community-level intervention programming. Recently, there has been a push to offer bystander training to young adults (i.e., aged 18 to 24 years) in a variety of contexts, not just within university settings (Banyard, Edwards, & Seibold, 2016). Statewide coalitions and crisis centers have targeted entire communities, including specific professions (e.g., staffs at bars and clubs), to receive intervention training. The #MeToo movement has also prompted conversations about the importance of bystander action in the workplace.

Furthermore, given that bystanders reported a fair amount of negative consequences after helping, intervention programming should consider educating bystanders how to respond to and process negative consequences. Teaching bystanders how to respond to negative reactions may reduce conflict and inhibit further negative responses from others. Learning how to process these feelings in a healthy way may also act as a buffer between experiencing negative consequences and not wanting to help again. It may also be beneficial for intervention programs to educate bystanders on ways to reduce the number of negative consequences they experience. There may be strategies that bystanders can use to avoid negative consequences altogether or mitigate them when they do occur. For example, prior research has found that the type of helping used by bystanders, such as creating a distraction, may result in more negative reactions from others (Moschella et al., 2018). Therefore, additional work is needed to identify specific factors that result in negative consequences and possible strategies to reduce them.

Limitations

Although the current study contributes to our knowledge of what happens after bystanders intervene, there are several limitations that should be noted. Even though we recruited participants from MTurk to increase representation, our sample still lacked racial and ethnic diversity. Future research should continue this research with more diverse populations. In addition, the list of consequences measured is not exhaustive. Although the consequences measures were generated from previous research on perceived consequences (Burn, 2009; Latané & Darley, 1970) and experts in the field, and validated through focus groups, it is possible that there are other consequences experienced by bystanders that were not examined in this paper (Banyard et al., under review). Furthermore, some of the measures, such as the incident-specific Positive Outcomes subscale and the Unsure Bystander Feelings subscale, had relatively weak alpha values. Given that bystanders frequently reported experiencing outcomes from these scales, future research should consider assessing individual items, as opposed to a composite score, when alphas are low.

In addition, bystanders self-reported their experiences. Thus, the consequences that were measured are the bystander’s own perceptions of the reactions of the victims, perpetrators, and others. While we asked bystanders to report their most recent incident, we do not know how much time passed between when the incident occurred and when the bystander completed the survey. The descriptions of interpersonal violence (e.g., dating violence, unwanted sexual advances) also included multiple different behaviors that bystanders could have witnessed; therefore, we only broadly know which type of interpersonal violence was happening and not which specific behaviors prodded the bystander to intervene. This information may tell us more information situation characteristics, such as situation severity. Furthermore, the behavior that bystanders engaged in was not examined. The type of behavior that bystanders engaged in, such as calling 911, may elicit different responses from the parties involved (i.e., victim, perpetrator, and other bystanders).

Conclusion

The current study described how positive and negative experiences after helping are linked to characteristics of the situation (i.e., the type of interpersonal violence and the bystander’s relationship to parties involved). Our findings support and extend those from previous work and show that bystanders receive a variety of responses to their actions in situations of interpersonal violence (Hamby et al., 2016; Moschella et al., 2018; Witte et al., 2017). These findings also more fully describe the nuances of how bystanders think about the reactions they receive. Furthermore, in support of the Action Coils model, these experiences influence bystanders’ desire to help again. More work should be done to further understand how to optimize bystanders’ experience when helping in situations involving interpersonal violence.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Elizabeth A. Moschella

Author Biographies

Elizabeth A. Moschella is a doctoral candidate in the Department of Psychology at the University of New Hampshire. She has a MA in justice studies from the University of New Hampshire. Her current research focuses on bystander intervention in response to risk for interpersonal violence. She is also interested in researching the ability of survivors to experience growth after victimization and protective factors against revictimization.

Victoria L. Banyard is a professor in the Department of Psychology at the University of New Hampshire. She is also part of the Prevention Innovations Research Center. She conducts research on the long-term consequences of interpersonal violence including factors related to resilience and recovery. Her work also centers on bringing broader community approaches to sexual and intimate partner violence prevention with a particular focus on bystander intervention.

References

Banyard

V. L.

(2015). Criminology—Toward the next generation of bystander prevention of sexual and relationship violence: Action coils to engage communities. doi:10.1007/978-3-319-23171-6_4

Banyard

V. L.

Edwards

K. M.

Seibold

(2016). Involving community in sexual violence prevention: Engaging bystanders. In Sturmey

R. P.

(Ed.), Handbook of violence and aggression, Volume 3: Societal interventions (p. 1-13). Malden, MA: John Wiley.

Banyard

V. L.

Moschella

E. A.

Grych

Jouriles

. (under review). What happened next? Measuring consequences of bystander actions to prevent interpersonal violence.

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. doi:10.1002/jcop.20159

Bennett

Banyard

V. L.

(2016). Do friends really help friends? The effect of relational factors and perceived severity on bystander perception of sexual violence. Psychology of Violence, 6, 64-72. doi:10.1037/a0037708

Bennett

Banyard

V. L.

Edwards

(2017). The impact of the bystander’s relationship with the victim and the perpetrator on intent to help in situations involving sexual violence. Journal of Interpersonal Violence, 32, 682-702. doi:10.1177/0886260515586373

Bennett

Banyard

V. L.

Garnhart

(2013). To act or not to act, that is the question? Barriers and facilitators of bystander intervention. Journal of Interpersonal Violence, 29, 476-496. doi:10.1177/0886260513505210

Black

M. C.

Basile

K. C.

Breiding

M. J.

Smith

S. G.

Walters

M. L.

Merrick

M. T.

. . . Stevens

M. R.

(2011). The National Intimate Partner and Sexual Violence Survey (NISVS): 2010 summary report. Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention.

Burn

S. M.

(2009). A situational model of sexual assault prevention through bystander intervention. Sex Roles, 60, 779-792. doi:10.1007/s11199-008-9581-5

10.

Edwards

K. M.

Rodenhizer-Stämpfli

K. A.

Eckstein

R. P.

(2015). Bystander action in situations of dating and sexual aggression: A mixed methodological study of high school youth. Journal of Youth and Adolescence, 44, 2321-2336. doi:10.1007/s10964-015-0307-z

11.

Hamby

Weber

M. C.

Grych

Banyard

V. L.

(2016). What difference do bystanders make? The association of victim outcomes with bystander involvement in a community sample. Psychology of Violence, 6, 91-102. doi:10.1037/a0039073

12.

Krebs

C. P.

Lindquist

C. H.

Warner

T. D.

Fisher

B. S.

Martin

S. L.

(2007). The Campus Sexual Assault (CSA) Study. Washington, DC: U.S. National Institute of Justice.

13.

Latané

Darley

J. M.

(1970). The unresponsive bystander: Why doesn’t he help. New York, NY: Appleton-Century-Crofts.

14.

Moschella

E. A.

Bennett

Banyard

V. L.

(2018). Beyond the situational model: Bystander action consequences to intervening in situations involving sexual violence. Journal of Interpersonal Violence, 33(20), 3211-3231.

15.

Nicksa

S. C.

(2014). Bystander’s willingness to report theft, physical assault, and sexual assault: The impact of gender, anonymity, and relationship with the offender. Journal of Interpersonal Violence, 29, 217-236. doi:10.1177/0886260513505146

16.

Planty

(2002). Third-party involvement in violent crime, 1993–1999 (Bureau of Justice Statistics Special Report). Washington, DC: U.S. Department of Justice.

17.

Rennison

C. M.

(2001). Intimate partner violence and age of victim, 1993-99. Washington, DC: Bureau of Justice Statistics, U.S. Department of Justice.

18.

Ryan

Wessel

(2012). Sexual orientation harassment in the workplace: When do observers intervene? Journal of Organizational Behavior, 33, 488-509.

19.

Sinozich

Langton

(2014). Rape and sexual victimization among college-aged females, 1995-2013. Washington, DC: Bureau of Justice Statistics, U.S. Department of Justice.

20.

Witte

T. H.

Casper

D. M.

Hackman

C. L.

Mulla

M. M.

(2017). Bystander interventions for sexual assault and dating violence on college campuses: Are we putting bystanders in harm’s way? Journal of American College Health, 65, 149-157.