Abstract
Most Americans view intimate partner violence as wrong. Less is known, however, about how the general population evaluates threats from romantic partners. When do third parties support interventions such as police involvement, restraining orders, or prohibiting the abuser from owning a gun? Through a survey-based experiment, participants reacted to a separated dating relationship scenario in which three elements were manipulated: the race of the couple, the medium of communication between the perpetrator and the victim, and whether the male character referenced a gun. Using a structural equation model, the authors find that the inclusion of a gun dramatically increases concern, which in turn fosters support for interventions. However, participants’ race and gender and the race of the couple shape these effects. When the victims in the separated dating scenario are Black, participants were less likely to call for the abuser to be prohibited from owning a gun, even when they have expressed concern about the situation. This suggests that although a gun has a clear and strong effect, racial and gender effects are more complex.
Intimate partner violence (IPV) is physical violence, sexual violence, stalking and psychological aggression by a current or former intimate partner (Breiding et al. 2015). Historically, IPV was viewed as a private family matter to be addressed by male heads of household (Carlson and Worden 2005; Fleming and Franklin 2021). Today, society has given increased attention the IPV as a social problem rather than an individual issue or private family matter (Davis 1987; Dobash and Dobash 2003; Klein et al. 1997; Leisenring 2006). Despite this shift, less is known about the public’s support for interventions to address IPV, especially within dating relationships.
This study highlights a nexus of overlooked areas of public opinion: when do third parties support interventions such as police involvement, restraining orders, or prohibiting the abuser from owning a gun for a violence scenario? We expect most Americans to express concern when reading about a high-risk dating scenario. But will that concern translate into support for formal intervention to protect victims? Moreover, we ask if these perceptions will change if the male perpetrator mentions a gun. Will people be more or less concerned when the victim receives a threat in a text message or Facebook message compared with in person? Or will support for IPV intervention be driven by one’s social position?
Prior scholarship has examined the attitudinal acceptance of IPV in the United States (Nabors and Jasinski 2009; Nguyen et al. 2013). This body of work has examined the role public attitudes play in shaping institutional responses to victims, perpetration of violence, and victim’s response to the victimization (see Flood and Pease 2009; Nabors, Dietz, and Jasinski 2006). But to date, there has been limited investigation of how the general public evaluates threats victims receive when separating from their abusive partners and how these perceptions shape support for IPV interventions. Similarly, although IPV risk factors have been extensively studied, how the public perceives these risk factors remains overlooked.
When victims attempt to separate from abusive partners, they often experience threats (Dutton and Goodman 2005; Logan and Walker 2004; Stark 2007) because separation challenges the abuser’s power and control (Campbell et al. 2007; Wells and DeLeon-Granados 2004). Research shows that separation (Campbell et al. 2003; Johnson and Hotton 2003) and abusers’ access to a gun (Cooper and Smith 2011; Lynch and Logan 2018; Spencer and Stith 2020) are a dire combination for lethal violence, especially for female victims. This gap in the literature is further compounded when considering underinvestigated topics of dating violence and the use of technology in abusive relationships.
In this study we examine how gender, race, and the mention of a gun influence threat perceptions and hypothesized interventions. Participants were presented with a vignette of threatened violence within a heterosexual dating relationship in which we manipulated how the man in the relationship threatened the female character, the race of the couple, and whether a gun was involved. We then asked participants questions about their levels of concern over the situation, their perceptions of the relationship as abusive, and their support for three intervention responses: contacting the police, issuing a restraining order, and banning firearms for the perpetrator. We find that some specific victims and specific situations are disproportionately treated as unworthy of IPV intervention, even when Americans express concern. We find that this is particularly true for situations involving Black women as victims of IPV and for situations involving threats delivered through electronic forms of communication.
How People Perceive IPV and When They Demand Interventions
Historically, IPV was viewed as a private family matter (Carlson and Worden 2005; Fleming and Franklin 2021; Pleck 1987). In the 1970s, social movements pushed to reframe domestic violence as a social and public safety issue requiring legal responses to protect victims (Dunn 2004; Renzetti, Follingstad, and Coker 2017). This push led to significant changes in legislation, funding, and victim services (Berns 1999; Bush 1992). The passage of the Violence against Women Act in 1994 signaled a cultural and legal shift in attitudes toward IPV, defining it as a public problem worthy of laws to protect victims and funding programs for programs and services (Barrick and Worsham 2020; Messing et al. 2015). Now, in the United States, the general public believes that IPV is a problem and that acts of abuse are wrong (Frieze, Newhill, and Fusco 2020; Pierotti 2013). Less is known about how the general population evaluates risk factors of dating violence and when they support interventions to address IPV. Yet antecedents to the general public’s support for IPV interventions remain underexamined (for exceptions, see Montanez and Donley 2021).
Attitudes toward IPV
There is a growing body of literature that examines public attitudes and beliefs about IPV. Scholarship has found consensus of concern over anonymized victim’s description of abuse from a domestic violence protection order case, reflecting the view of IPV as a widely recognized social problem (Groggel 2023). Other work examines how several factors such as gender, race, and other demographic characteristics shape these attitudes. A recent analysis of articles comparing attitudes toward IPV among two or more cultural groups showed that college students in the United States generally showed less accepting attitudes than students in other countries (Zark and Satyen 2022).
Prior vignette research demonstrates gender differences in undergraduate college students’ perceptions of marital rape and domestic abuse, with women being more sympathetic to victims and more likely to attribute responsibility for the incident to the abuser than men (Langhinrichsen-Rohling et al. 2004; Locke and Richman 1999; Pierce and Harris 1993). Similarly, women have been found to perceive domestic abuse situations as more severe or dangerous, are more likely to report a greater necessity for intervention in IPV scenarios and more likely than men to encourage a victim to seek help from professionals (Hamby and Jackson 2010; Herzog 2007; Seelau, Seelau, and Poorman 2003; Sylaska and Walters 2014). Men are more likely than women to hold a victim responsible for violence, and men are more accepting of IPV than women (Bryant and Spencer 2003; Nabors et al. 2006).
Research demonstrates that the race of the victim and perpetrator are key predictors of attitudes toward IPV and sexual violence. In vignette research, a victim’s race has been shown to affect whether participants view the victim as culpable or blameworthy. For example, George and Martínez (2002) found that participants were less likely to define an incident as rape when the victim was a woman of color and the perpetrator was White. Studies involving Black victims have shown particularly strong effects, with participants being less likely to characterize incidents involving Black women as domestic abuse (Peterson-Lewis et al. 1988), blaming female Black IPV victims for their abuse (Willis, Hallinan, and Melby 1996), or considering Black women to be more culpable than their White female counterparts in domestic violence situations (Harrison and Esqueda 2000).
In addition to participant’s race and gender, we would expect well-known risk factors to affect concern for a dating violence scenario. In the following sections, we outline key risk factors within abusive relationships and consider three formal interventions: contacting the police, victims seeking civil protection orders, and the removal of firearms from abusers. In the remainder of the article, we use the term formal interventions to distinguish such interventions from informal interventions involving family, friends, or other informal support structures.
IPV Risk Factors
Newer conceptions of dating violence and IPV now include both technology-facilitated and in-person emotional, physical, and sexual abuse and stalking behaviors (Temple et al. 2016; Woodlock 2017). Violence in dating relationships been identified and studied as an alarming social problem (Borrajo, Gámez-Guadix, and Calvete 2015; Duval, Lanning, and Patterson 2020). A growing body of scholarship has documented risk factors of IPV, victimization, and even intimate partner homicide (Campbell et al. 2017; Sheehan et al. 2015; Spencer, Cafferky, and Stith 2016). In the following, we highlight several key risk factors associated with intimate partner homicide and severe victimization.
There is a dire link between IPV and gun violence in the United States. Guns are often used to intimidate or threaten victims, and the presence of a gun is associated with the severity of nonfatal IPV (Joshi and Sorenson 2010; Kafka et al. 2021; Lynch and Logan 2018; Small, Sorenson, and Berk 2019; Sorenson and Wiebe 2004; Zeoli, Malinski, and Turchan 2016). Access to firearms increases the risk for intimate partner femicide (Campbell et al. 2007; Cooper and Smith 2011; Lynch and Logan 2018; Siegel and Rothman 2016; Sorenson and Wiebe 2004; Spencer and Stith 2020), and femicide-suicide, in which an individual commits suicide after committing a homicide (Bourget, Gagné, and Moamai 2000; Koziol-McLain et al. 2006).
Although femicide in adulthood has been marked as a critical issue, this topic has been gaining prominence concerning adolescents (Leen et al. 2013). Gun violence victimization among adolescents is common in the United States compared with other wealthy, industrialized nations (Olufajo et al. 2020). The risk for violence and femicide is also heightened for IPV victims in the period after separation or attempted dissolution of a relationship (Campbell et al. 2003; Elisha et al. 2010; Garcia, Soria, and Hurwitz 2007; Johnson and Hotton 2003; Logan and Walker 2004). Victims’ efforts to separate from an abusive partner can challenge an abuser’s sense of control and result in a backlash effect to regain the perceived dominance lost over his former partner (Campbell et al. 2007; Wells and DeLeon-Granados 2004). Recent work suggests that coercive control, jealousy, and a desire to control female partners are risk factors associated with IPV and intimate partner homicide (Elmquist et al. 2014; Harris 2003; Koziol-McLain et al. 2006).
Abusers can now use technologies and social media to maintain control over their victims (Dimond, Fiesler, and Bruckman 2011), monitor and stalk their intimate partners (Messing et al. 2020; Southworth et al. 2005), and harass them (Borrajo et al. 2015). Risk factors such as jealous behaviors and stalking are observed and evidenced through a perpetrator’s use of technology (McLachlan and Harris 2022). However, knowledge about the role of mediated communication has lagged. Professionals who provide services to survivors of IPV report feeling as though they lack the adequate expertise to identify or cope with technology-enabled IPV (Freed et al. 2017). This oversight is critical given that digital technologies and social media are increasingly integral to the dynamics of IPV, allowing abusers to establish and enforce coercion and control over partners before and after separation (Belknap, Chu, and DePrince 2011; Dragiewicz et al. 2018).
Police
One key formal intervention is victims reporting IPV to law enforcement (Akers and Kaukinen 2009; Felson, Ackerman, and Gallagher 2005). Reporting incidents of violence to law enforcement can help victims receive other support and assistance, and such reporting is often a requirement for judicial responses (Dowling et al. 2018). Yet only a small fraction of IPV victims call the police (Felson et al. 2002). White victims are less likely to report domestic violence than other victims (Ackerman and Love 2014; Flicker et al. 2011), and female victims are more likely to report incidents to the police than male victims (Dugan 2003; Kingsnorth and Macintosh 2004). Many victims do not contact the police because of fears of repercussions, cultural perceptions that the matter is private or not serious enough to involve police, negative previous experiences with police response, or concern over mandatory arrest policies (Gover et al. 2013; Novisky and Peralta 2015; Wolf et al. 2003).
Protection Orders
Domestic violence protection orders, also called restraining orders, are one of the most widely used interventions by victims of IPV (Adhia et al. 2020; Durfee and Messing 2012; Richards, Tudor, and Gover 2018). Civil protection orders are binding court orders intended to constrain abusers who threaten or harass victims, through provisions such as exclusion from the shared residence, granting temporary custody of minor children to victims, temporary child support, and requiring the removal of weapons such as firearms (Holt 2004; Logan et al. 2006; Zeoli et al. 2019). Violators of these orders are subject to criminal and civil penalties and prosecution including fines, contempt penalties, and criminal charges (Groggel 2021; McFarlane et al. 2004). Each state in the United States has enacted statutes authorizing civil protection orders to provide victims immediate relief from abusive partners, but eligibility criteria may differ between states (DeJong and Burgess-Proctor 2006).
Prohibiting Guns
The removal of guns is a particularly important intervention in the context of IPV. Protection order provisions can prohibit a perpetrator from possessing a firearm or remove the perpetrator’s firearm for the duration of a temporary or permanent order (Lyons et al. 2021; Zeoli et al. 2019). A recent report shows that 36 states have laws that prohibit those who have civil protection orders issued against them from buying or possessing firearms or ammunition (Bejinariu, Troshynski, and Miethe. 2021). Statutes vary by state, with some explicitly requiring those subject to an order to surrender their firearms or include firearms provisions in temporary protection orders in addition to permanent orders.
Beyond existing civil protection order provisions, several states have enacted risk-based temporary firearm removal laws called extreme risk protection orders (Bonnie and Swanson 2018; Swanson 2019). At a federal level, Congress passed the Domestic Violence Offender Gun Ban (the “Lautenberg amendment”), whereby people convicted of domestic violence misdemeanors or who had restraining orders for domestic abuse against them are restricted from purchasing or possessing a firearm or ammunition (18 U.S.C. 921) (Barboza 2019; Corbin 2015; O’Connor 2010). However, these laws have a “boyfriend loophole” whereby only perpetrators of domestic abuse who have been married, lived with, or had a child with their partner are prohibited from purchasing guns (Logan and Lynch 2022). This excludes victims who are noncohabitating dating partners, even though nearly half of intimate partner homicides are perpetrated by unmarried partners (Sorenson and Spear 2018). Given that abusers’ access to firearms is a risk factor for femicide, understanding support for the removal of firearms upon the issuance of a protection order is critically important (Groggel 2023). Prohibiting abusers’ possession of firearms through protection orders has been shown to lower rates of IPV homicide (Vigdor and Mercy 2006; Zeoli and Webster 2010; Zeoli et al. 2018, 2019).
The Present Study
Faced with a scenario involving these type of risk factors, how do Americans assess the situation? Under what conditions do they become concerned? How does concern translate, or not, into support for formal interventions? And finally, what do these assessments tell us about future directions for studying and intervening in situations involving IPV? To explore concern over a victim of dating violence and support for intervention among members of the American public, we designed an online survey-based experiment that accounts for multiple paths of influence among participant characteristics, communication of threats, and different types of evaluation of threatening situations.
Our study design incorporates several well-established risk factors for IPV by having a separated heterosexual dating relationship scenario in which an ex-boyfriend threatens his partner. The design allowed us to examine subjects’ reactions to a hypothetical scenario while controlling for variables by randomly assigning different versions of the vignette (Aviram 2012). We examined how participants assessed a given vignette scenario, variations of which involved the inclusion or absence of a gun, how the threatening message was delivered, and whether the couple in the vignette appeared to be Black or White. In this way, our study builds upon previous scholarship involving IPV vignettes (Harrison and Esqueda 2000; Parker et al. 2022; Seelau and Seelau 2005; Seelau et al. 2003; Sylaska and Walters 2014).This research design provides visibility not only into elements that affected participants’ levels of concern, but also onto factors of support for interventions in response to a given abusive dating encounter. We propose a two-stage model in which factors that may directly influence calls for intervention, such as type of violence (e.g., guns), means of communication, gender, and race, are additionally mediated through “concern.” By manipulating vignette conditions and using structural equation modeling (SEM) to analyze the results, we were able to account for well-known risk factors as well as variations in race and gender among participants and vignette characters.
Sample
We use a national sample of 1,136 participants. The survey instrument was hosted by Ipsos, an independent market research company, and delivered through an Omnibus survey. Participants were compensated with iSay points, which can be redeemed for cash or prizes, such as Amazon purchases. The points earned through participation in a typical Ipsos survey are equivalent to $1 to $3 per survey, and panelists are entered into a lottery for a larger cash prize. Individuals were recruited by e-mail and completed the survey online. Ipsos uses an online “opt-in” method whereby the panel of survey respondents is developed to closely mimic the demographic composition of the United States. Furthermore, Ipsos surveys include quality checks to ensure survey panelist attention. Institutional review board approval was obtained for the study.
Table 1 describes our sample. Forty-nine percent of participants are women, 23 percent have bachelor’s degrees, and approximately 68 percent self-identify as White. Approximately 48 percent of participants in the sample were married. The average age of participants was 43 years.
Survey Respondent Demographics (n = 1,136).
Vignettes and Questionnaire
Each participant read a single randomly assigned vignette of an ex–intimate partner violence scenario. The scenario draws on existing scholarship of both patterns of IPV relationships, and the specific text is based on actual victims’ descriptions of incidents of abuse, which come from an online justice database containing the text of restraining order requests. Although some elements of the scenario were manipulated for individual vignettes, the scenario consistently included instances of threatening or intimidating language, such as “If we’re not together you can’t be with anyone. You better watch out.”
Our study had a total of 16 possible vignettes. Within each version, the characters’ names were manipulated to imply race (White couple or Black couple), the medium of communication between the characters was manipulated to imply different levels of contact (in-person, voicemail, text message, or private Facebook message), and the comments of the ex-boyfriend were manipulated to indicate whether he had a gun. A full list of the versions is provided in the Appendix. Two versions of the scenario are presented below, with bracketed text indicating the characteristics manipulated across versions: [Shanice/Emily] ran into her ex-boyfriend, [DeShawn/Cody], and he stated, “If we’re not together you can’t be with anyone. [You know I have a gun.] You better watch out.” [Shanice/Emily] received a [text message/voicemail/private Facebook message] from her ex-boyfriend, [DeShawn/Cody], stating, “If we’re not together you can’t be with anyone. [You know I have a gun.] You better watch out.”
After reading the vignette, participants answered questions about whether the ex-couple had an abusive relationship, and the extent to which the female character should seek formal interventions.
Dependent Variables
Victims can seek formal intervention from the legal system by filing a petition for a protection order, filing or trying to file criminal charges, or calling the police (Fugate et al. 2005; Goodman et al. 2003). Three statements measured support for intervention: (1) “[Shanice/Emily] should contact the police,” (2) “[Shanice/Emily] should be given a restraining order,” and (3) “Her ex-boyfriend, [Cody/DeShawn], should be prohibited from owning a gun.” Participants were asked to indicate the extent to which they agreed with the statement on a seven-point Likert-type scale ranging from “strongly disagree” to “strongly agree.”
The questionnaire also contained items to measure perceptions concerning the threat from the character’s ex-boyfriend. Three statements measured concern: (1) “The incident was threatening,” (2) “[Shanice/Emily] is in danger,” and (3) “I would be concerned if a close friend or family member received a message like this.” From these three questions measured on a seven-point Likert-type scale, we created a continuous concern scale (Cronbach’s α = .80), with higher scores indicating greater concern over the IPV scenario.
Independent Variables
The independent variables are the vignette conditions of the IPV scenario, including the race of the couple (Black couple or White couple), the medium of communication (in-person, voicemail, text message, or private Facebook message), and whether the abuser referenced a gun (gun, no gun). We created a variable for the race of the couple on the basis of participants’ guesses of the race of each character (White, Black or African American, Hispanic or Latino, Native American or American Indian, Asian or Pacific Islander, and unsure). More than 87 percent of participants guessed that Shanice was Black, and 94 percent of the sample guessed that DeShawn was Black. Similarly, we had high agreement for the White couple characters, with more than 85 percent of the participants guessing that Emily was White, and 93 percent guessing that Cody was White. From these responses, we created a new binary variable for the race of couple by restricting responses only to cases in which the victim’s and perpetrator’s racial categories were correctly classified (White couple = 0, Black couple = 1).
The “medium” variable allowed us to make comparisons between versions of the vignette that manipulated the medium of communication (in person = 0, text message = 1, voicemail = 2, Facebook message = 3). Last, we created a binary variable for the manipulation of a gun being referenced in the scenario, “You know I have a gun” (0 = no, 1 = yes).
Survey participants were also asked to indicate the extent to which they agreed on a seven-point scale with the statement that “[Shanice/Emily] was in an abusive relationship.”
Participants also completed manipulation checks. This was done in the form of a question that centered on the vignette detail about how the male character contacted the female character. Approximately 71 percent of the participants correctly responded to the manipulation check by identifying the method of communication used in their version of the vignette.
Control Variables
We controlled for sociodemographic variables such as respondent’s educational attainment, marital status, household income, age of respondent, gender, and race. Participants self-identified their race by selecting from racial categories that we then used to create a single variable coded as White, Black or African American, Hispanic, Asian, or another category that includes Native Hawaiian, Pacific Islander, or multiracial. As the sample was predominantly White, it was used as the reference category.
Because individuals who have personally experienced or known someone who has experienced IPV may evaluate the vignette differently, we controlled for participants’ response to the question “Have you or a close friend or family member ever received a threat from a current or previous intimate or romantic partner?” This variable is called experienced IPV threat. We also controlled for participants’ daily hours of social media use, because the medium of communication between the victim and their perpetrator was one of our vignette manipulations. And in consideration of scholarship highlighting the association between southern states and opposition to gun-control policies, the region of the country participants reside in (South, Northeast, Midwest, West) was included in the model (O’Brien et al. 2013).
Analytic Strategy
Our analytical model is summarized in Figure 1. SEM was used to model the data to simultaneously estimate the direct and indirect effects of independent variables on dependent variables. The model incorporates pathways between vignette conditions, concern for the victims, and calls for intervention. This allowed us to fully assess how participants’ levels of concern could have a mediating effect on support for the interventions. We ran a linear structural equation model to calculate the direct, indirect, and total effect by treating the seven-point Likert-type scale measures as continuous outcomes.

Hypothesized model of support for intimate partner violence interventions for the separated dating relationship scenario.
First, we model how vignette manipulations (gun, race of couple, and medium of exchange), participants’ race and gender, and perceptions that the relationship in the scenario is abusive affect perceptions of concern. The scale for concern was set by constraining the loading of the friend variable (concern over a close friend or family member’s receiving a similar message) to one. We then examine how these factors affect support for contacting the police, issuing a restraining order, and prohibiting the male character from owning a gun directly or indirectly mediated through concern.
Modeling the data in this way allows us to evaluate both predictors of concern and interventions among the American public under different conditions. The model includes all three interventions, but the direct and indirect effects of each intervention are displayed in separate figures. The model was fit with a maximum likelihood estimator, and Satorra-Bentler standard errors are reported. The model includes a covariance between the errors for the endogenous variables prohibiting a gun and restraining order.
Results
Using SEM, we examine the pathways through which the vignette manipulations and participants’ social demographics affect concern for the victim and support for the three interventions measures. Our model demonstrates a strong fit according to multiple SEM fit statistics: the root mean square error of approximation of 0.04 was less than 0.05, and the comparative fit index was greater than 0.95 (Acock 2013; Hu and Bentler 1999). Because most of the variables are observed, and in many cases categorical, unstandardized estimates are reported.
Concern Scale
Higher concern scale scores indicate greater concern over the separated dating scenario by aggregating agreement with statements about whether the incident was threatening or whether a respondent would be concerned if a close friend or family member received a message such as the one in the vignette. Figure 2 displays the unstandardized coefficients of the direct effects vignette conditions and participants’ race and gender on concern. We find only partial support that vignette conditions will significantly affect participants’ concern over the scenario. Although the race of the couple and the medium of communication did not significantly affect concern scale scores, when a gun was included in the IPV scenario, participants’ concern scale scores increased (b = 0.27, SE = 0.03, p < .001). Figure 2 shows that we also find that women have higher concern scale scores (b = 0.16, SE = 0.05, p < .01). The figure also shows that participants who viewed the relationship as abusive were also more concerned over the scenario (b = 0.41, SE = 0.03, p < .001).

Direct effects on concern over the separated dating relationship scenario.
Gun Prohibition
Figure 3 displays the direct and the indirect effects of variables as they operate through the concern scale on support for prohibiting the man making threats to his ex-girlfriend from owning a gun. We find that concern over the scenario increased support for the prohibiting the man from owning a gun (b = 1.14, SE = 0.12, p < .001).

Direct and indirect effects on support for prohibiting a gun.
We find that the vignette manipulations of the race of the couple and the medium of the exchange have significant direct effects on support for restricting the abuser’s access to guns. When the scenario involved the ex-boyfriend leaving the woman a threatening voicemail, participants were more likely to agree that he should be prohibited from owning a gun compared with when the exchange occurred in person (b = 0.27, SE = 0.14, p < .05). The couple being perceived as Black rather than White lowers support for prohibiting the male character from owning a gun (b = −0.36, SE = 0.09, p < .001). The total effects of perceived race of the couple and medium of exchange remain statistically significant. Including a gun in the scenario has a significant and positive indirect effect on support for prohibiting the male character from having a gun by increasing concern (b = 0.31, SE = 0.06, p < .001). However, the total effect of the gun vignette manipulation lost significance.
Similarly, perceptions that the relationship was abusive increased support for this intervention indirectly through concern (b = 0.47, SE = 0.05, p < .001). Figure 3 also shows that women and Black participants expressed greater support for restricting the abuser’s gun ownership. Women were significantly more likely to agree with the intervention of prohibiting a gun indirectly through the pathway of concern over the scenario (b = 0.18, SE = 0.06, p < .01). Black participants compared with White participants were more likely to support the prohibition (b = 0.41, SE = 0.13, p < .001). The total effects of participants’ perceptions of the relationship as abusive and their own demographic characteristics remained statistically significant.
Restraining Order
Figure 4 displays the predictors of support for the victim’s receiving a restraining order. We find that concern over the scenario increased agreement that the woman should be granted a restraining order (b = 1.58, SE = 0.14, p < .001). Only one vignette condition significantly shaped support for issuing a restraining order: a gun. A gun reference increased support for the woman’s receiving a restraining order indirectly by increasing concern (b = 0.43, SE = 0.08, p < .001). The total effect of the man in the vignette stating “You know I have a gun” remained significant (b = 0.50, SE = 0.09, p < .001; not shown in Figure 4).

Direct and indirect effects on support for a restraining order.
Figure 4 also shows that viewing the relationship as abusive had a significant direct effect on support for the woman’s receiving a restraining order. Viewing the relationship as abusive lowered support for issuing a restraining order (b = −0.13, SE = 0.06, p < .05) but held a positive and significant effect indirectly through concern (b = 0.65, SE = 0.06, p < .001). Although these mixed effects warrant further investigation, we find that the total effect of viewing the relationship as abusive significantly and positively increases support for issuing a restraining order (b = 0.51, SE = 0.03, p < .001; not shown).
The gender of participants also had opposing effects. The gender of participants had a direct effect on this intervention, with women less likely to support the restraining order (b = −0.26, SE = 0.09, p < .001). However, the indirect effect operated in the opposite direction. Through the indirect pathway of concern over the scenario, women had increased agreement with receiving a restraining order. This led to an insignificant total effect of gender of participants on support for granting Shanice or Emily a restraining order.
Contact the Police
Figure 5 displays the direct and indirect effects of the demographics of participants and vignette manipulations on support for the woman’s calling the police. We find that the greater concern expressed over the scenario, the greater the support for the woman contacting the police (b = 1.86 SE = 0.14, p < .001).

Direct and indirect effects on support for calling the police.
The vignette variables of the gun and medium of communication also significantly affect participants’ support for this intervention. Medium of exchange had a direct effect on support for the woman’s calling the police after a threatening exchange with her ex-boyfriend. Support for contacting the police was significantly higher for participants who read the scenario in which the ex-boyfriend left a voicemail compared with when the scenario was in person (b = 0.26, SE = 0.10, p < .05). Additionally, Figure 5 shows that the mention of a gun increased support for contacting the police indirectly by increasing concern (b = 0.50, SE = 0.09, p < .001). We find that both these vignette conditions had significant total effects, with both the threat being delivered over voicemail or the mention of a gun increasing support for the victim’s contacting the police.
Figure 5 shows that although women participants were less supportive of the woman’s calling the police, the indirect effect of gender was in the opposite direction. Women increased support for this intervention indirectly by expressing greater concern over the scenario. Similarly, we find that although viewing the relationship as abusive has a significant negative direct effect on support for contacting the police (b = −0.18, SE = 0.06, p < .01), the indirect effect through concern operated in the opposite direction (b = 0.77, SE = 0.06, p < .001). This leads to a total positive effect of perceptions of the relationship as abusive on support for the victim’s calling the police (b = 0.59, SE = 0.03, p < .001).
Discussion
Participants in our study overwhelmingly viewed the scenario as concerning. Americans within our sample evaluated the scenario as threatening, that the woman was in danger, and that they would be concerned if a close friend or family member received a message such as this. We do find slight variations in concern, with participants who viewed the relationship as abusive, read the vignette version that mentioned a gun, and were women had greater concern over the separated dating scenario. These results suggests that societal concern over IPV has expanded to include threats within dating relationships. We also find that support for IPV interventions was filtered through concern and assessments about guns, race, gender, and communication. Despite Americans expressing concern, this filtering led some victims, and some specific threats, to be disproportionately treated as unworthy of intervention.
Another key finding is that when the ex-boyfriend perpetrator stated, “You know I have a gun,” support for all three interventions increased, mediated through the concern over the separated dating scenario. Thus, a gun in the scenario made participants more supportive of the woman’s receiving a restraining or contacting the police or prohibiting gun ownership indirectly through concern. In short, the presence of a gun was the most straightforward factor that connects concern to intervention for participants. These differences hold meaningful implications for exposing potential inequalities and understanding under what conditions the public will seek to protect victims.
But even without the gun, the separated dating scenario included several high-risk factors. This is noteworthy. The ex-boyfriend stated “If we’re not together you can’t be with anyone. You better watch out.” This reflects two significant predictors of violence and femicide: jealousy and a separated relationship (Harris 2003; Spencer and Stith 2020). Perhaps participants held views consistent with the underlying logic of “boyfriend loopholes” that leaves victims who do not live with their abusers unprotected despite facing similar firearm-related risk (Logan and Lynch 2022). Perceptions of dating violence matter, given that adolescent dating violence is a predictor of marital violence in adulthood (Meneghel and Portella 2017). Only recently has scholarship focused on dating violence, with prior work on earlier legal recognitions of IPV focused on marriage contract and domestic partnership in the United States (Barner and Carney 2011). Future research should further explore how different patterns of romantic relationships affect the assessment of, and support for, IPV interventions.
We also find that the race of the couple appeared to become salient only when participants considered the prohibition of the abuser from owning a gun. In contrast, the race of the couple had no significant bearing on concern over the scenario or agreement that the woman should receive a restraining order or contacting the police. When the victim was Black, participants were less supportive of prohibiting the male character from owning a gun. Although Black female adolescents are at greater risk for dating violence victimization (Alleyne-Green, Coleman-Cowger, and Henry 2012; Debnam, Milam, and Finigan-Carr 2022), within our sample participants were less supportive of prohibiting the Black victim’s abuser from owning a gun. This suggests that threatened gun violence against Black women was more tolerated than this threatened violence against White women. It is striking that Black participants appeared more sensitive to the threat of gun violence by expressing greater support for the gun restriction with the IPV relationship compared with White participants.
Participants’ gender also consistently and significantly factored into concern and support for interventions with holding greater concern about the scenario and greater support for IPV interventions indirectly through concern. However, when considering actions such as calling the police or issuing a restraining order, gender had a negative direct effect. We find a similar pattern for the effects of the view that the relationship was abusive. Viewing the relationship as abusive increased support for the woman calling the police or receiving a restraining order through the pathway of concern but had a negative direct effect. Prior work shows that victims and researchers are apprehensive over the effectiveness of protection orders or calling the police and victims fear repercussions when using these interventions (Logan and Walker 2010; Messing et al. 2021; Wolf et al. 2003). Awareness of the limitations of these interventions would explain why women and those who viewed the relationship in the scenario as abusive had a negative direct effect while still increasing support for these interventions indirectly through concern.
A potentially surprising finding about methods of communication in situations involving IPV also emerged in the study. In our vignettes, when the threat occurred in person, participants were less likely to agree that the female character should contact the police, compared with if the threat occurred over voicemail. Recent scholarship has detailed how abusers exploit digital technologies to intimidate, threaten, or stalk victims (Freed et al. 2018), but little attention has been paid to how the general public evaluates such behaviors. As abusive behaviors shift online, evidence of threats and harassment over social media also increases. Participants’ general lack of concern over threats from a previous intimate partner made over text message or Facebook message is especially alarming in the context of IPV, given that digital technologies can arm abusers with victims’ location data. Our findings suggest that greater attention should be paid to threats over mediated communication. Understanding how and why such threats are discounted by members of the public matter given the growing number of court cases in which social media evidence is used in court (Baughman 2009; Ramirez and Lane 2019).
Understanding what factors shape support for interventions matters because IPV is a broader social issue involving consequential assessments from many people who are neither victims nor perpetrators. Members of the American public are also the neighbors who overhear a violent argument, the jurors considering evidence of abuse, and the voters shaping policy and funding priorities at the ballot box. Because individuals contribute to the ongoing social response to IPV in a variety of ways, understanding how members of the American public assess IPV is crucial for understanding it as a social problem.
Limitations
Vignette methodology has been criticized for not eliciting responses that reflect real-world situations. If participants were actual bystanders to an incident of IPV, the logic goes, they might respond differently. This is probably correct, and the present study is not intended to predict responses to such a terrifying, high-stakes situation. Our research design is intended to shed light on the many routine moments in which members of the public have opportunities to evaluate the seriousness of IPV, for example in the jury box, the voting booth, their living rooms, and online communities. The other main limitation of the study is its restriction to one specific type of relationship. Our scenario refers only to a heterosexual relationship between separated dating partners. Within the scope of this study, we are unable to make comparisons between other relationship statuses or same-sex couples.
Conclusion
Our results suggest that guns, in particular, trigger concern and calls for interventions, but participants’ concern over a dating violence scenario was also filtered in complicated ways as they take race, gender, and communication method into account. These differences result in important differences as they consider responses to IPV threats. These differences hold meaningful implications for exposing the conditions under which the public will seek to protect victims of IPV. Our results mark the ways in which programs might be most beneficial in swaying public opinion toward greater protections for victims of interpersonal violence.
Supplemental Material
sj-docx-1-srd-10.1177_23780231221149627 – Supplemental material for How Americans Assess Intimate Partner Violence: Evidence from a Survey Experiment
Supplemental material, sj-docx-1-srd-10.1177_23780231221149627 for How Americans Assess Intimate Partner Violence: Evidence from a Survey Experiment by Anne Groggel and Fabio Rojas in Socius
Footnotes
Acknowledgements
We would like to thank Clifford Young and IPSOS for their assistance in collecting this rich dataset.
Supplemental Material
Supplemental material for this article is available online.
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
