Victim–Offender Relationship and the Emotional, Social, and Physical Consequences of Violent Victimization

Abstract

Introduction:

Research is equivocal about how the social relationship between victims and offenders is linked to the emotional, social, and physical consequences of violence. This study examines the association of victim–offender relationship with the adverse outcomes reported by injured and uninjured victims of violence.

Methods:

The study analyzed 16,723 violent victimizations recorded by the National Crime Victimization Survey from 2008 to 2018. Multivariable quasi-Poisson models estimated the associations between the victim–offender relationship and victims’ emotional distress, social distress, and physical and emotional symptoms. These models also estimated a statistical interaction between victim–offender relationship and violent injury to examine how this association differed for injured and uninjured victims. The analyses occurred during 2020 and 2021.

Results:

Uninjured victims were more likely to report emotional distress (risk ratio=1.41, 95% CI=1.33, 1.50), social distress (risk ratio=3.12, 95% CI=2.78, 3.51), more physical symptoms (symptom frequency ratio=1.68, 95% CI=1.51, 1.87), and more emotional symptoms (symptom frequency ratio=1.13, 95% CI=1.08, 1.18) in family member/intimate partner violence than stranger violence. Victims also reported worse outcomes after acquaintance violence than after stranger violence. For injured victims, these differences narrowed—but were still significant—in emotional and social distress models. However, the number of emotional and physical symptoms reported by injured victims did not significantly vary across victim–offender relationships.

Conclusions:

Relational closeness between victims and offenders is a risk factor for adverse outcomes after violent victimization, and it is more strongly associated with these outcomes for uninjured victims than injured victims.

INTRODUCTION

Family and intimate partner violence (FIPV) have a pervasive and deleterious impact on victims’ well-being. Attacks by loved ones can place victims at higher risk for chronic physical disability and pain,¹ psychiatric disorders,² and suicide.³ Such attacks can also be socially isolating,⁴ reduce school and work productivity,⁵ and have a high probability of reoccurring.⁶ Understanding both the nature and extent of the consequences of FIPV continues to be a public health concern.

Current research is equivocal about the health burdens caused by FIPV compared with those caused by other types of violence. Some studies find that victims report worse health outcomes when attacked by a stranger than by someone they know. For example, analyses of police data indicate stranger offenders are more likely than known offenders to injure victims, particularly when co-offending with others.⁷ Child victims of physical and sexual abuse also report more significant psychological trauma when the offender is a non-caregiver than a parent or guardian.⁸

Other research finds that victims report worse health outcomes when they are attacked by someone they know than by a stranger. National surveys suggest that victims are more likely to sustain injuries in FIPV than stranger violence⁹ and report worse socioemotional outcomes the closer they are to their offender.¹⁰ In medical settings, sexual violence victims display more psychological symptoms if they knew the assailant.¹¹ Psychological theory also posits that traumatic events are most harmful to a person’s mental health when caused by someone in that person’s social network.¹²

Overall, the literature on how the victim and offender’s social relationship (hereafter victim–offender relationship) shapes victims’ health outcomes faces 2 limitations. First, research is equivocal because studies considerably vary in the types of violence, outcomes, and data they examine. For example, studies of child abuse often yield different results than studies of adult victims of sexual violence. Conclusions also vary between studies that examine psychological or social outcomes¹⁰ and studies that examine physical injury.⁷ Furthermore, the association between the victim–offender relationship and health outcomes often differs between studies using victimization surveys⁹ and studies using police data.⁷ By definition, police data exclude violence not reported to the police, which can introduce sample selection bias and limit the generalizability of findings.

Second, even though victimization surveys may address sample selection bias, studies rarely leverage these data¹⁰ to test how the victim–offender relationship predicts the health outcomes of different victim populations, such as those injured versus those not injured by violence. Injured victims are of particular interest to healthcare providers because they are more likely than uninjured victims to use healthcare services.¹³ Conceptually, how victims know their offenders may have different health consequences depending on the severity of violence. For example, a threat of violence from a loved one may elicit greater fear than a threat from a stranger because victims will likely see the loved one again.¹⁴ On the other hand, victims may be less impacted by the social relationship with their offender when they have to cope with violent traumatic injuries.

Given the limitations of prior research, this study addresses 2 research questions: (1) What is the association between victim–offender relationship and victims’ adverse emotional, social, and physical health outcomes? (2) How does violent injury moderate the association between the victim–offender relationship and victims’ adverse emotional, social, and physical health outcomes? To address these questions, the current study uses national survey data on violent victimization to estimate victims’ emotional distress, social distress, and physical and emotional symptoms based on their social relationships with the offenders. Additionally, it examines how the victim–offender relationship—as a risk factor for these adverse outcomes—differed between injured and uninjured victims of violence.

METHODS

Study Sample

The primary sample was a cross-section of violent victimizations in the U.S. collected by the Bureau of Justice Statistics’ National Crime Victimization Survey (NCVS) from 2008 to 2018.¹⁵ The NCVS is the leading national-level household survey on victimization among individuals aged ≥12 years. The survey asks household respondents to recall victimizations experienced in the previous 6 months. As these data are all anonymous, they do not meet the criteria for human subjects research and thus are IRB exempt. The study was restricted to the 2008–2018 period because the survey did not begin measuring the health consequences of violence, except injury, until 2008.

The primary sample included 12,493 respondents who reported 16,723 violent victimizations: 10,712 simple assaults, 3,079 aggravated assaults, 1,838 robberies, and 1,094 rapes and sexual assaults. Additionally, the study examined a subsample of 8,675 violent victimizations because the NCVS asks victims about their long-term physical and emotional symptoms only when the victims reported emotional or social distress.

Measures

The study estimated 5 adverse outcomes reported by victims: injury, emotional distress, social distress, and long-term (defined as lasting ≥1 month) physical and emotional symptoms (Appendix Table 1). Injury was a dichotomous measure of whether the victim experienced cuts, bruises, swelling, chipped teeth, a gunshot wound, stab wound, internal injuries, unconsciousness, broken bones, or sexual violence injuries.

The NCVS asks respondents to rate their emotional distress following their victimization from 1 (no distress) to 4 (severe distress). Emotional distress was categorized into a dichotomous measure of whether the victim reported moderate-to-severe distress (3–4), as opposed to no distress or mild distress (1–2). Social distress was a dichotomous indicator of whether the violence contributed to problems at job/school, in relationships with work colleagues, or with family members or friends.

When victims reported emotional or social distress, they listed the physical and emotional health symptoms they experienced for ≥1 month. Long-term physical symptoms measured the number of physical symptoms, including experiencing headaches, trouble sleeping, changes in eating or drinking habits, stomachaches, fatigue, high blood pressure, muscle tension, and “some other physical problem.” Victims could report 0–8 physical symptoms. Long-term emotional symptoms measured the number of emotional symptoms, including feeling worried/anxious, angry, sad/depressed, vulnerable, violated, distrust in people, unsafe, and “some other way.” Victims could report 0–8 emotional symptoms.

The exposure—victim–offender relationship—had 3 categories: stranger violence, acquaintance violence, or FIPV. Acquaintance violence included friends, roommates, schoolmates, neighbors, customers/clients, patients, supervisors, employees, coworkers, teachers, and other nonrelatives. FIPV included family members and intimate partners. For violence involving multiple offenders, the victim–offender relationship indicated the offender closest to the victim, with FIPV treated as the closest relationship.

Statistical Analysis

The outcomes were modeled using a quasi-Poisson distribution with robust clustered SEs. Emotional and social distress models used the total sample of violent victimizations (n=16,723). Owing to skip patterns in the survey, models of physical and emotional symptoms used only the subsample of violent victimizations where victims reported emotional or social distress (n=8,675). Robust clustered SEs accounted for: (1) inflated Poisson SEs in the binomial outcomes¹⁶ and (2) dependence across observations from NCVS respondents reporting >1 victimization. These models estimated risk ratios (RRs) in the probability of reporting emotional and social distress and symptom frequency ratios (SFRs) in the number of physical and emotional symptoms between victim–offender relationship types. The SFR indicates the percentage difference between victim–offender relationship types in the number of reported symptoms after violence.

The models controlled only for pre-exposure covariates (Appendix Table 2), including year fixed effects, victim characteristics at the time of the interview, the characteristics of the offenders, and the victim’s history of violent victimization. Models also included statistical interactions between the victim and offenders’ age (victim age X offender age) and gender (victim gender X offender gender) because how victims know their offenders and the severity of violence is associated with victims and offenders being of the opposite (versus same) gender or age group.^17,18

All models also included an interaction term between the victim–offender relationship and injury (i.e., victim–offender relationship X injury). The estimates of these interaction terms (Appendix Tables 3 and 4) were used to calculate whether differences in adverse outcomes between victim–offender relationship types varied between injured and uninjured victims.

Overall, 31% of cases had missing data on ≥1 measure in the study. Measures that had the most missing data were the victim’s emotional and social distress (14.3% missing), the offender characteristics (12.5% missing), and the victim’s household income (12.5% missing). Multiple imputation with chained equations was used to handle missing data, using all covariates and some additional information about the victimization incident (Appendix Table 2). This approach assumes that the data are missing at random, conditional on the observed measures in the multiple imputation procedure.

All statistics were estimated using Stata, version 15.1 poisson, margins, and mi packages and NCVS victimization sampling weights.

RESULTS

Table 1 presents selected descriptive statistics across victim–offender relationship categories. In the total sample, victims of violence knew their offenders in more than half of the violent victimizations. Victims reported injuries (26%) less often than emotional distress (53%) and social distress (28%). Victims’ probabilities of injury, emotional distress, and social distress tended to be highest in FIPV and lowest in stranger violence. Among victimizations involving emotional or social distress, victims reported—on average—2.10 physical symptoms and 4.46 emotional symptoms lasting for ≥1 month. Additionally, the number of long-term physical and emotional symptoms was also highest in FIPV and lowest in stranger violence.

Table 1.

Proportions, Means, and SDs of Selected Study Variables by Victim‒Offender Relationship, Total Sample NCVS (2008‒2018)

		Victim-offender relationship
Health outcomes	Study sample (n=16,723)	FIPV (n=3,630)	Acquaintance (n=5,629)	Stranger (n=7,463)
Injury	0.26	0.42	0.23	0.20
Emotional distress	0.53	0.73	0.52	0.45
Social distress	0.28	0.49	0.31	0.15
# Long-term physical symptomsc	2.10 (2.62)	2.79 (2.73)	2.02 (2.62)	1.59 (2.45)
# Long-term emotional symptomsc	4.46 (2.69)	5.01 (2.44)	4.34 (2.78)	4.11 (2.78)

Notes: Complete descriptive statistics provided in Appendix Table 2.

^aMeans of long-term physical and emotional problems are available only for victims who reported emotional or social distress (n=8,675).

FIPV, family/intimate partner violence; NCVS, National Crime Victimization Survey.

Figure 1 presents the probabilities that uninjured and injured victims reported emotional and social distress across victim–offender relationships. Uninjured victims of FIPV were more likely than victims of acquaintance violence to report emotional and social distress, and uninjured victims of acquaintance violence were more likely than uninjured victims of stranger violence to report emotional and social distress.

Figure 1.

Emotional distress and social distress across victim-offender relationships, by victim injury status.

FIPV, Family/intimate partner violence.

Similar patterns were present in violence involving injury, but with 2 exceptions. First, the relative differences in the probabilities of emotional and social distress between victim–offender relationships were larger for uninjured victims than injured victims. Uninjured victims of FIPV were 1.63 (95% CI=1.54, 1.73) and 3.68 times (95% CI=3.30, 4.11) more likely than uninjured victims of stranger violence to report emotional and social distress, respectively. By contrast, injured victims of FIPV were only 1.32 (95% CI=1.24, 1.40) and 1.98 (95% CI=1.75, 2.23) times more likely than injured victims of stranger violence to report emotional and social distress,. Second, there was no difference in the probabilities of emotional distress between victims injured by acquaintances and victims injured by strangers.

Although injured victims generally were more likely than uninjured victims to report emotional and social distress, uninjured victims were more likely to report emotional and social distress than injured victims in certain cases. For example, the probabilities of emotional and social distress were lower for injured victims of stranger violence (61% emotional distress and 27% social distress) than uninjured victims of FIPV (68% emotional distress and 45% social distress). These differences were statistically significant for both emotional distress (p<0.001) and social distress (p<0.001).

Figure 2 summarizes the number of physical symptoms and emotional symptoms reported by victims who experienced emotional or social distress, across victim–offender relationships and victim injury. Uninjured victims of FIPV violence reported 34% more physical symptoms (SFR=1.34, 95% CI=1.21, 1.48) and 11% more emotional symptoms (SFR=1.11, 95% CI=1.06, 1.16) than uninjured victims of acquaintance violence. Uninjured victims of acquaintance violence reported 47% more physical symptoms (SFR=1.47, 95% CI=1.31, 1.65) and 9% more emotional symptoms (SFR=1.09, 95% CI=1.04, 1.15) than uninjured victims of stranger violence. For injured victims, however, some of these differences were not statistically significant. Injured victims of FIPV—on average—reported 30% more physical symptoms (SFR=1.30, 95% CI=1.17, 1.43) and 16% more emotional symptoms (SFR=1.16, 95% CI=1.20, 1.46) than injured victims of acquaintance violence. Yet, there were no significant differences between injured victims of acquaintance and stranger violence in the incidence of physical symptoms (SFR=1.02, 95% CI=0.91, 1.14) or emotional symptoms (SFR=0.98, 95% CI=0.93, 1.04).

Figure 2.

Long-term physical and emotional symptoms across victim-offender relationships, by victim injury status.

FIPV, family/intimate partner violence.

In general, injured victims reported more physical and emotional symptoms than uninjured victims. However, the mean number of physical symptoms did not differ significantly between uninjured victims of FIPV and injured victims of acquaintance violence (p=0.485) or stranger violence (p=0.270). Moreover, the mean number of emotional symptoms did not differ significantly between uninjured victims of FIPV violence and injured victims of acquaintance (p=0.551) or stranger violence (p=0.910).

Figure 3 illustrates a forest plot of the logged RRs of emotional and social distress and the logged SFRs of long-term physical and emotional symptoms between victim–offender relationships, by victim’s injury status (Appendix Figure 3 shows plot of absolute differences). These estimates were adjusted for all covariates (Appendix Tables 3 and 4 provide complete models). Logged estimates allow for more accurate visual comparisons of effect sizes. A list of all effect sizes, including their CIs, is available in Appendix Table 5.

Figure 3.

Adjusted logged relative differences in victims’ adverse health outcomes between victim-offender relationships, by victim injury status.

RR, risk ratio; SFR, symptom frequency ratio; FIPV, family/intimate partner violence.

Even after accounting for covariates, the relational closeness between victims and offenders was a significant risk factor in victims’ adverse outcomes; however, the victim–offender relationship was more strongly associated with uninjured victims’ than injured victims’ adverse outcomes. This pattern was most evident when comparing between victims of FIPV versus victims of stranger violence. Uninjured victims were more likely to report emotional distress (RR=1.41, 95% CI=1.33, 1.50), social distress (RR=3.12, 95% CI=2.78, 3.51), more physical symptoms (SFR=1.68, 95% CI=1.51, 1.87), and more emotional symptoms (SFR=1.13, 95% CI=1.08, 1.18) after FIPV than after stranger violence. When victims experienced injury, however, these differences significantly narrowed in models of emotional distress (SFR=1.10, 95% CI=1.02, 1.17) and social distress (SFR=1.63, 95% CI=1.43, 1.86). Additionally, there were no significant differences between victim–offender relationships in the number of emotional and physical symptoms reported by injured victims.

DISCUSSION

This study investigated 2 questions: (1) how the victim–offender relationship is linked with the emotional, social, and physical sequelae of violence and (2) how these associations differ between injured and uninjured victims. On average, victims of violence who had a closer relationship with their offender(s) reported worse emotional, social, and physical health outcomes. More specifically, victims of FIPV reported worse outcomes than victims of acquaintance violence, and victims of acquaintance violence reported worse outcomes than victims of stranger violence. Further analyses revealed that these associations were stronger for uninjured victims than injured victims.

Victims often rely on their social network to cope with violent trauma,¹⁹ and deleterious social ties can undermine this effort.²⁰ For those attacked by someone they know or love, social ties become points of conflict that can promote stress and hinder recovery. For example, victims of marital abuse upend their lives when leaving their partners, as they are often emotionally, socially, and economically dependent on their attackers.²¹ For youth, bullying can be particularly distressful when it damages or ends the child’s friendships.²² Thus, the social fallout from violence by known offenders may be an important mechanism to explain the associations between victim–offender relationship and victims’ outcomes.

Yet, the victim–offender relationship may have a stronger link to uninjured victims’ health outcomes than injured victims’ health outcomes. A recent study found that whether people with violent injuries are treated by healthcare providers depended on the severity of injury, how the person knew their attacker(s), and the interaction between these factors.¹³ This evidence suggests that the consequences of violence may be linked to the degree and context of violence. For example, a victim may be more distressed if threatened by an intimate partner—as opposed to a complete stranger—because the threat carries greater weight and may be used as a tool for coercive control.²³ Conversely, the victim–offender relationship may not be as important in determining longer-term outcomes when victims have to cope with the traumatic consequences of violent injury. Further research is necessary to determine why the presence of injury moderates the association between victim–offender relationship and victims’ outcomes.

Limitations

The current study has 2 empirical limitations. First, the NCVS instrument does not include clinically evaluated health measures, such as scales for post-traumatic stress disorder (e.g, Posttraumatic Stress Disorder Checklist for DSM-5 [PCL-5])²⁴ or depression (e.g., Center for Epidemiologic Studies Depression [CES-D]).²⁵ Still, the NCVS provides unique insight into the emotional, social, and physical impact of victimization from a large, nationally representative sample of crime. The results were also remarkably consistent across adverse outcomes and aligned with prior research finding individuals score higher on the PCL-5 and CES-D when a traumatic event is caused by someone close to them.²⁶ Second, the study did not test mediators that may explain the association between the victim–offender relationship and victims’ adverse outcomes. For example, the nature and type of violence significantly differ between attacks involving stranger and attacks involving known offenders.^7,14 Future research that tests such mediators would be a valuable extension to the current study.

CONCLUSIONS

When attacked by someone they know or love, victims must cope with the physical trauma, the emotional trauma, and the betrayal caused by the offender’s actions. Overall, the current study provides evidence that the consequences of violence are linked to how well victims know their offenders, and this link is partly attenuated for victims who report violent injury. Yet, more than two thirds of violent crime in the U.S. does not involve victim injury,²⁷ and even in the absence of injury, victims feel the socioemotional impact of violence on their well-being.^28,29 This impact seems particularly harmful when the attacker is embedded in the victim’s social network.

In treating victims of violence, discerning the attacker’s identity assists patient care, which is why hospitals and health agencies invest substantial resources into identifying FIPV. Identifying whether the attacker is in the victim’s social network, such as whether they were acquaintances of the victim, may also reveal critical points for intervention. Such efforts may pay dividends particularly for victims who experience unseen emotional abuse or violent threats but do not present with physical injury.