Introduction Predicting Dating Violence Victimization among College Women

Exposure to Violence and Acceptance of Dating Violence Dating

violence is recognized as a major issue in today’s society, with college students reporting

rates of physical violence between 17% and 45% across 31 universities around the world

(Straus, 2004). In the 1970s, due to the women’s movement, physical violence involving

intimate partners, termed domestic violence, became a more recognizable societal

problem worldwide (Tjaden, 2005). Initially domestic violence was viewed as a criminal

justice issue, but since the 1990s it is considered more of a public health concern. Today,

some people argue that it is also a human rights issue (Tjaden, 2005).

The term domestic violence was first used by researchers and grassroots activists

in order to refer to husbands who physically abused their wives. Lenore Walker (1979)

was one of the first clinical researchers to highlight this issue. More recently the term

intimate partner violence (IPV) has gained recognition as the preferred term which is

used to refer to violence between partners since it allows for the inclusion of different

types of violence (e.g., sexual and psychological) in addition to physical violence and

expands the definition to include additional types of intimate relationships besides

marriage, including dating and same sex relationships. Makepeace (1981) was the first

researcher to focus more specifically on dating violence, reporting that 21% of college

students had experienced dating violence. The focus of this study was to evaluate the

impact of prior experiences with violence and attitudes, such as acceptance of violence, in

predicting the risk of physical dating violence victimization among college women.

In general, when studying dating violence there are several different aspects that

are important to consider when interpreting research findings. Three major areas of

consideration are sample characteristics, types of dating violence, and types of

measurements and definitional criteria. Research studies have mainly focused on either

high school or college student populations (ages 13-25) and the majority of the student

respondents have been Caucasian and heterosexual. Moreover, dating violence can

include physical, psychological, and/or sexual violence. Even though the rates of

violence reported vary across studies, researchers have found the highest rates of dating

violence for psychological violence (82% for women and 87% for men; Harned, 2002)

followed by sexual (76.9% for females and 67.4% for males; Jackson, Cram, & Seymour,

2000), and physical violence (47%; Katz, Kuffel, & Coblentz, 2002). Lastly, the types of

measurements (as well as study-specific definitional criteria) employed in a particular

study can influence the research findings. Straus, Hamby, Boney-McCoy, and

Sugarman’s (1996) Conflict Tactics Scale (CTS2) is the most widely used measure for

assessing IPV (Jackson, 1999). Nevertheless, even when different studies employ the

same measure to assess dating violence, researchers often operationalize victimization

and perpetration criteria differently, which further complicates interpreting and

generalizing findings across studies. One area of controversy that highlights the

variability of findings across studies is the issue of the relationship between victimization

and perpetration and how this may differ between the sexes.

Comparing Dating Violence in Females and Males: Perpetration and Victimization

There is an on-going debate among scholars when it comes to IPV patterns and

sex differences in the high school and college dating cohorts; this pattern appears to differ

from the pattern typically reported in the adult domestic violence literature. Research has

found that females are more likely to report being both victims and perpetrators of dating

violence compared to males (Follingstad, Wright, Lloyd, & Sebastian, 1991; Gover,

Kaukinen, & Fox, 2008; Lichter & McCloskey, 2004; Whitaker, Haileyesus, Swahn, &

Saltzman, 2007). Because of this co-occurrence of victimization and perpetration roles in

the high school and college samples, it is important to incorporate research on

perpetration when discussing victimization.

In the adult domestic violence literature males are more likely than females to use

physical violence in a relationship (Dobash, Dobash, Wilson, & Daly, 1992). However,

other studies have shown that the opposite is true, with high school and college female

samples reporting higher levels of perpetration (Chan & Straus, 2008; Follingstad, et al.,

1991; Foo and Margolin, 1995; Gover et al., 2008; Luthra & Gidycz, 2006; Makepeace,

1986; Shook, Gerrity, Jurich, & Segrist, 2000; Straus, 2004). Moreover, some researchers

report equal levels of perpetration for males and females (Bookwala, Frieze, Smith, &

Ryan, 1992; Katz, et al., 2002; O’Keefe & Treister, 1998; Riggs & O’Leary, 1996). For

example, Straus (2008) found that college students at 32 different universities around the

world reported equal levels of perpetration between partners (67%) as the most common

form of perpetration pattern of dating violence, followed by female-only perpetration

(20%) and male-only perpetration (10%).

When investigating prevalence of perpetration of dating violence and sex

differences, research findings indicate that it is important to consider the overall

prevalence rates, the severity and the frequency of the violence, and the injury caused by

the violence. In Chan and Straus’s (2008) college study, women in Hong Kong and the

United States were more likely to use physical violence in dating relationships (49.3%

versus 36.2% in Hong Kong and 29.9% versus 28.5% in the U.S), even though the

difference was not significant in the U.S. sample. In their study, the severity of the

violence did not differ between male and female perpetrators, as women reported

perpetrating both severe (23.3% of females versus 16.3% of males in Hong Kong and

10.7% of females versus 7.7% of males in the U.S.) and minor (18.7% of females versus

14.5% males in Hong Kong and 20.1% of females versus 19.3% of males in the U.S.)

physical violence more often than men. However, other researchers have found results

contradictory to those reported in the Chan and Straus study. Foo and Margolin (1995)

reported that college men perpetrated more severe violence, such as beating and

threatening and using knife or weapon against partner, while women perpetrated more

moderate forms of violence, such as pushing, grabbing, shoving, slapping, kicking,

biting, etc.

The issue of sex differences in perpetration becomes even more complex when

considering the frequency of the violence and the amount of injury caused by the

violence. In Chan and Straus’s (2008) study men were more likely to report a higher

frequency of perpetrating physical violence compared to women (21.2% versus 16.3% in

Hong Kong and 14.5% versus 7.6% in the U.S.). In this study a significant difference of

perpetration rates between the sexes was found only in the U.S. sample. The researchers

also found that men were more likely to frequently cause injury to their partners than

were women (26% versus 5.5% in Hong Kong and 14.2% versus 6.3% in the U.S.). In

another study by Straus (2004), men were also more likely to cause more severe injury to

their partners compared to women. This pattern occurred at 78% of the 27 different

universities around the world that reported severe injury due to dating violence.

However, earlier Makepeace (1986) found no difference between male and female

perpetrators in terms of how much injury they reported causing their partners. The only

exception was that males reported causing more moderate physical injury (1.4%) on their

partners than females (0%). None of the students reported inflicting severe injury on

their partner.

In summary, the research on perpetration rates and sex differences for the high

school and college aged cohorts suggests that overall more young women report

perpetrating dating violence than men. However, young men report a higher frequency of

violence perpetration and cause more injury than women. Even though more women

report perpetrating dating violence than men, the violence tends to be more mild forms

(Teten, Ball, Valle, Noonan, & Rosenbluth, 2009). This pattern differs from the adult

domestic violence literature, in which studies report violence to be more severe and

perpetrators are predominantly male (Dobash et al., 1992). Therefore, it is still unclear if

women or men in a college cohort are more likely to perpetrate IPV. More women than

men admit to perpetrating violence, albeit acts that represent mild forms; while men cause

more injury and perpetrate violence at a higher frequency.

When investigating victimization rates, research findings suggest that it is

important to consider the same variables, which affect perpetration rates (e.g., overall

prevalence rates; severity, injury) as studies have found sex differences in victimization

rates depending on those variables. First, in terms of overall victimization rates, studies

have found that women report higher levels of victimization compared to men

(Follingstad et al, 1991; Gover et al., 2008; Lichter & McCloskey, 2004; Whitaker et al.,

2007), much as they endorsed greater rates of perpetration. Severity of violence

experienced also differs between female and male victims, as Makepeace (1986) found that

college women were more likely to report being victims of severe violence, such as being

“struck by an object” and “beaten up” (p. 384), while the men were more likely to report

being victims of less severe violence, such as being pushed, slapped, kicked, bit, punched,

and having objects thrown at them. Likewise, in Molidor and Tolman’s (1998) high school

sample, a greater number of women reported being victims of more severe violence

(27.1%) compared to men (16.5%), while men reported being victims of more mild to

moderate violence in greater numbers (34.7%) compared to the women (23.3%). The

researchers found similar results when examining the students’ current or most recent

relationships.

Moreover, research on victimization and injury suggests that women report more

injury than men as a result of dating violence. In Makepeace’s study (1986) college

women who reported being victims of dating violence also reported higher rates of mild

(45.4% females versus 15.7% males), moderate (5.3% females versus 2.2% males), and

severe injury (2.4% females versus 0% males) as a result of the dating violence compared

to men. Similarly, Molidor and Tolman (1998) reported that women were more likely to

report that the violence caused them harm (47.8% females versus 3.8% males) and

physical injury (33.6% females versus 4.8% males) compared to the men.

In summary, research on victimization and sex differences indicates that young

women are more likely to report being victims of dating violence, report more severe

violence, and more injury compared to men. This is consistent with both the dating

violence literature and the domestic violence literature that report that women are more

likely to be victims of IPV and report more injury compared to men. Moreover, college

women are more likely to report being both victims and perpetrators of dating violence

and some studies report that dating violence tends to be bidirectional; therefore, it is also

important to consider relationships where partners are mutually combative, both

perpetrators and victims, when discussing victimization and perpetration rates.

Researchers have found sex differences in prevalence rates when considering

bidirectionality of dating violence. As mentioned earlier, Straus (2008) found that

mutually combative dating violence is the most common form of perpetration pattern

reported by college students. He also found that most of the severe violence was reported

in relationships where people reported being both perpetrators and victims of dating

violence.

The frequency of the violence and the amount of injury caused by the violence

further distinguish bidirectional from unidirectional violent relationships. In one study

with middle school and high school students, adolescents who were currently in or had

been in mutually abusive relationships reported a higher frequency of violence, both

victimization and perpetration, compared to adolescents who reported only victimization

or perpetration experiences (Gray & Foshee, 1997). Moreover, researchers have found

that adolescent females and males report higher levels of injury in relationships where

both partners are mutually abusive (Gray & Foshee, 1997; Whitaker et al., 2007).

The discrepancy in victimization and perpetration rates can be influenced by several

factors. Perpetration rates seem, in part, to be dependent on whether researchers focus

on acts of violence versus consequences of violence. Archer’s (2000) metaanalysis

examined sex differences and dating violence; Archer concluded that “when measures

were based on specific acts, women were significantly more likely than men to have

used physical aggression toward their partners and to have used it more frequently” (p.

664). However, “when measures were based on the physical consequences of

aggression (visible injuries or injuries requiring medical treatment), men were more

likely than women to have injured their partners” (p. 664). Nevertheless, the effect sizes

found in the meta-analysis were small (d ranged from .05 to .15).

There are several additional factors that might contribute to the contradictory

findings in regards to perpetration and victimization rates among men and women. One

such factor is self-defense, and many researchers have argued that women are more likely

to use violence as a means of self-defense. Self-defense could be a factor which

contributes to the finding of women reporting higher or equal levels of perpetration

compared to men (Makepeace, 1986; Molidor & Tolman, 1998). Makepeace (1986)

found that women were more likely to report using violence in self-defense compared to

males, who were more likely to report using it to intimidate their partners. Furthermore,

he found that more women than men reported perpetrating violence with the intent to

physically harm their partner, which he interpreted being related to women using violence

in self-defense. In fact Makepeace found that 70% of the women who reported

perpetrating violence to physically hurt their partners also reported that they perpetrated

the violence in self-defense. Likewise, Molidor and Tolman (1998) found that girls from

their high school sample more often reported that they “fought back” (35.9%) compared

to the boys (12.8%) (p. 186). In their study, both boys and girls reported that the boys

were more likely to initiate the violence (i.e., 70% of the girls versus 27% of the boys

stated their partner initiated the abuse). O’Keefe and Treister (1998) also found that high

school males and females both reported that males were more likely than females to

initiate violence; however, 41% of the females and 48% of the males reported that both

partners were responsible for initiating the violence. Moreover, it is sometimes unclear

how participants interpret initiation of violence. It can be interpreted as who hit first,

who provoked the other partner first, or who started a verbal argument.

Another factor to consider in determining true rates of partner abuse is that males

might not want to report being perpetrators of dating violence since male-to-female

violence is less accepted than female-to-male violence within society (Forbes, Jobe,

White, Bloesch, & Adams-Curtis, 2005; Merten, 2008; O’Keefe & Treister, 1998).

Because male-to-female perpetration is less acceptable, men might not want to identify

themselves with something that is undesirable by society. There is evidence that men

underreport domestic violence perpetration (Dutton & Nicholls, 2005). Furthermore,

some studies have found that women report higher levels of perpetration and

victimization (Follingstad et al, 1991; Gover et al., 2008; Whitaker et al., 2007), which

indicates that women might be more willing to report violence in general compared to

men.

Johnson (2005; Kelly & Johnson, 2008) suggests that the sex differences in

victimization and perpetration rates can be explained if researchers consider type of

violence. He proposes three types of violence: intimate terrorism, violent resistance, and

situational couple violence. According to Johnson, intimate terrorism, also referred to as

coercive controlling violence, is the most severe type of violence and is predominately

perpetrated by men; it is oftentimes the type of IPV found in police records, courtrooms,

domestic violence shelters, and medical centers. Intimate terrorism is based on the model

of the power and control wheel, that explains how abusers use violence to control and

have power over their partners (Domestic Abuse Intervention Project). By contrast,

violent resistance is the response to intimate terrorism and is typically perpetrated by

female victims in response to their partners’ violence. Lastly, situational couple violence,

the most common type of violence encountered among intimate partners, typically results

in less severe violence. Situational couple violence is mostly seen in dating relationships

among adolescents and young adults and is not based on the power and control wheel

causal model of IPV. Males and females tend to report equal levels of perpetration and

victimization for situational couple violence. By focusing on the intent of the violence

instead of the acts of violence, Johnson’s typology of relationship violence may partly

explain the gender discrepancy reported in perpetration levels. In order to

better understand young adults’ experiences with dating violence, and because of the

previously discussed sex differences in perpetration and victimization rates, this study

will focus on female victims in an attempt to more clearly explain one aspect of this

complex experience. For this study, dating violence will be defined as physical violence

that has occurred in the past year in a heterosexual dating relationship. This study is

explicitly interested in situational couples violence among college students as defined by

Johnson (2005; Kelly & Johnson, 2008). The current study tried to predict female

victimization, but before discussing predictor variables, it is important to understand the

negative outcomes of experiencing dating violence. The high prevalence rates of dating

violence in combination with the high rates of injury from the violence among high

school and college students raise a concern about the potential consequences of the

violence both short- and long-term. Few studies have followed adolescents and college

students into adulthood to investigate the effects of dating violence, and little is known

about how experiences of dating violence affect later functioning. However, several

studies have reported the negative effects of IPV on both physical and mental health

among adults (Campbell, 2002; Coker et al., 2002; Plichta, 2004; Tjaden & Thoennes,

2000).

Negative Outcomes of Dating Violence

In contrast to the adult IPV literature, few studies have looked at outcomes of

dating violence among adolescents and young adults (Teten et al., 2009). Amar and

Gennaro (2005) compared victims and non-victims of dating violence in terms of mental

health symptoms and physical injury with a college sample of females. They found that

victims reported more mental health symptoms, a higher level of psychological distress,

and a greater risk for developing a psychiatric disorder compared to non-victims. Most of

the victims reported less severe injuries (89%), such as bruises, scratches, bite marks, and

lacerations, and about 40% of the injured victims did seek healthcare due to their injuries.

Moreover, the females who were victims of more than one incident of dating violence

reported higher mental health scores (endorsed more mental health symptoms), greater

psychological distress, a heightened risk for psychiatric disorders, and more injuries

compared to single-incident victims.

However there is some evidence that outcome for victims of dating violence is

variable. One variable affecting victim outcome is victim sex. A study of college dating

violence sampling both male and female college students found that female college

students reported feeling scared or anxious, emotionally hurt, and had a need to escape

when reporting physical violence victimization (Follingstad et al., 1991). While male

victims also reported negative effects of dating violence, female victims reported

“significantly more negative effects” (p. 56).

Furthermore, a sample of college students showed that men and women reported

equal levels of depression, anxiety, and posttraumatic stress symptoms related to being

victims of physical and psychological violence (Harned, 2002). However, males reported

a stable level of positive affect unaffected by frequency of violence while females’ ratings

of positive affect decreased as the frequency of the violence increased. Females reported

higher levels of positive affect at low frequency compared to males. Because the

females’ ratings of positive affect decreased with an increase in frequency, both males

and females reported equal levels of positive affect at the higher frequency of violence.

Studies using high school samples have found results similar to those of Follingstad’s et

al. (1991) college sample. O’Keefe and Treister (1998) found that females reported

“feeling emotionally hurt and fearful” because of the dating violence while males

reported that “they thought it was funny” or they “felt angry” (p. 211). In another high

school sample, males “were significantly more likely to report little or no physical

consequences of the violence they experienced,” while nearly half the females

“reported serious harm” and one third of the females reported physical injury (Molidor &

Tolman, 1998, p. 186). Female students were also more likely to report that their

relationships had worsened and they were more likely to end their violent relationships,

while male students were more likely to report that their relationships had improved

following the violence.

Moreover, in addition to considering the sex differences in terms of negative

effects of dating violence victimization, there is also some evidence for a more severe

negative impact when multiple types of violence were experienced. Katz, Moore, and

May (2008) compared the effects of experiencing both physical and sexual violence

versus one or the other in a sample of college women. They reported that women who

experienced both physical and sexual abuse in their relationships reported “significantly

less satisfaction, more arguing, and more psychological aggression” than women who

experienced physical or sexual violence (p. 969).

There are also some findings indicating that perpetrators seem to understand the

negative outcomes the violence has on their partner (Follingstad et al., 1991). Arguably

especially the male perpetrators seem to understand the negative outcomes of the

violence, as they reported that the violence they perpetrated “caused fear and anxiety,

sadness and depression, and a feeling in their female victims that they wanted to get away

to protect themselves” (Follingstad et al., 1991, p. 53). Female perpetrators, however,

reported that their victimized partners were likely to justify and feel guilty about the

females’ violent behavior.

Overall, there are numerous negative outcomes related to dating violence.

However, the answer to the question of why violence occurs in an intimate relationship

remains unclear. One possibility that researchers have suggested is that attitudes and

acceptance of IPV may in part explain why dating violence occurs, as some studies have

found a relation between acceptance of violence and negative outcomes. Kaura and

Lohman (2007) investigated how college students’ acceptance of violence influenced

their relationship satisfaction as well as mental health. They found that mental health and

relationship satisfaction were affected by dating violence for both men and women. The

women reported significantly more mental health symptoms (e.g., depression, anxiety,

and somatization) compared to the men. The findings of the study showed that women

who were victims of dating violence and were more accepting of men using violence in a

relationship were also more satisfied in their relationships than women who were victims

but were not as accepting of dating violence. Men who were victims of dating violence

reported less mental health problems if they were accepting of women using violence in a

relationship compared to men who were less accepting of dating violence. Acceptance of

male-to-female violence did not moderate the relationship between victimization and

mental health problems for women.

Kaura and Lohman’s (2007) findings suggest that greater acceptance of dating

violence could protect people from the negative effects of the violence. However, it is

unclear why some people are more accepting of dating violence than others. It is possible

that previous exposure to partner violence, such as witnessing interparental violence or

previous experiences of dating violence, might normalize the violence. The

normalization of dating violence might result in young adults being more likely to

experience violence in their relationships.

Since research has reported negative outcomes of IPV, more research is needed to

understand young adults’ perceptions of dating violence and how that relates to their

experiences of the violence. Due to the lack of research in this area, the current study

focused on female college students’ attitudes, especially acceptance, of dating violence

and how acceptance affects conflicts in their relationships. Consequently it is important

to understand how acceptance of dating violence is related to young adults’ experiences

of relationship violence in general. Even though the focus of this study is victimization,

perceptions of both victimization and perpetration will be included in the discussion to

provide a broader idea of how attitudes and acceptance of both dimensions of dating

violence are related to victimization experiences.

Attitudes and Acceptance of Dating Violence

In general, research has found that women tend to be less accepting of dating

violence than men (Bookwala et al., 1992; Cauffman, Feldman, Jensen, & Arnett, 2000;

Forbes et al., 2005; Josephson & Proulx, 2008; Kaura & Lohman, 2007; Merten, 2008;

O’Keefe & Treister, 1998; Price & Byers, 1999). There is also a trend for high school

and college students to be more accepting of women using violence against a partner than

men using violence against a partner (Cauffman et al., 2000; Forbes et al., 2005; Merten,

2008; O’Keefe & Treister, 1998; Price & Byers, 1999). Moreover, Price and Byers’

(1999) cross-sectional study with 823 high school students reported differences in

acceptance rates depending on type of violence and sex of perpetrator. Sexual violence

was rated as least acceptable independent of perpetrator’s sex. When females were

perpetrators, the students rated physical violence as more acceptable than psychological

violence. When males were perpetrators, the students were more accepting of

psychological violence compared to physical violence.

Furthermore, there is some evidence that adolescents and young adults are more

accepting of dating violence if they report being perpetrators and/or victims themselves.

Price and Byers (1999) found that students were more accepting of a specific type of

dating violence if they also reported perpetrating that type of violence. In other words,

students who reported that they had perpetrated physical dating violence were also more

accepting of physical dating violence. This was true for both women and men as women

perpetrators were more accepting of female-to-male specific violence and vice versa.

However, the correlation between female sexual perpetration and acceptance of femaleto-

male sexual violence was not significant because of the low number of female sexual

perpetrators. Moreover, perpetrators were also more accepting of at least one additional

type of violence, with sexual violence being the least common one.

Additional studies also suggest that acceptance of violence is associated with

victimization and perpetration. In Gray and Foshee’s (1997) cross-sectional study, they

found that adolescents who reported being both victims and perpetrators of dating

violence were more accepting of dating violence compared to students who reported

being only victims. They found no difference in acceptance levels when comparing those

who were both victims and perpetrators of dating violence and those who were only

perpetrators. These findings were based on a subsample of 77 students (out of a total of

185) who reported dating violence exposure. In Josephson and Proulx’s (2008)

crosssectional study with 290 adolescents, they found that the students’ acceptance of

dating violence was related to an increased risk of being perpetrators of physical and

psychological violence. Moreover, a greater acceptance of female-to-male violence was

related to an increased likelihood of perpetrating physical violence for both females and

males. However, the direct relationship between acceptance of male-to-female violence

and perpetration was not significant for either sex. There was evidence for an indirect

effect as acceptance of male-to-female violence increased the risk for perpetrating

psychological violence, which increased the risk for perpetrating physical violence. This

was true for both males and females. The researchers did not include victimization in

their study.

O’Keefe (1998) compared 232 high school students who had witnessed high

levels of interparental violence as children (i.e., at or above the 75th percentile) and been

involved in violent relationships to those who had not been in violent relationships.

Acceptance of dating violence predicted dating violence victimization and perpetration

for males but not for females. Malik, Sorenson, and Aneshensel (1997) recruited 719

high school males and females who participated in group-administered surveys. They

found that acceptance of using “verbal, physical, or weapon-related violence” increased

the odds of being a perpetrator and victim of physical dating violence, with higher odds

for perpetration (2.33) than victimization (1.42; p. 294). Another cross-sectional study

with 290 males and females college students found that acceptance of physical

perpetration predicted perpetration for both males and females (Foo & Margolin, 1995).

The acceptance measure was related to acceptance of using violence in self-defense or

because of a humiliating situation, with only the latter predicting physical perpetration.

Victimization was not included in their study.

In another high school sample of 1,013 adolescents, the researchers administered

surveys at baseline and 1-year after prevention program targeting dating violence

(Foshee, Linder, MacDougall, & Bangdiwala, 2001). Participants were randomly

assigned to prevention program or control group. The researchers controlled for

experiment condition when determining risk factors for perpetration of physical violence

both cross-sectional (i.e., at baseline) and longitudinal (i.e., 1 year follow up). At

baseline, both males and females were more likely to report being perpetrators if they had

friends who were perpetrators. Females were also more likely to perpetrate violence if

they had been in physical fights with female friends and if they were more accepting of

physical dating violence. At the 1-year follow-up, the only significant predictors were

having friends who were victims (for female perpetration) and acceptance of physical

dating violence (for male perpetration). Lichter and McCloskey (2004) also used a

longitudinal design to investigate predictors of physical and sexual dating violence

perpetration and victimization among male and female adolescents. The 208 adolescents

who participated in the study were interviewed twice approximately 8 years apart.

Acceptance of male-to-female dating violence was a significant predictor for perpetration

but not victimization.

Moreover, certain characteristics of the victim and the perpetrator as well as the

situation influence acceptance ratings. Findings from one study showed that college

students who read vignettes about dating violence were more likely to blame the victim

more and the perpetrator less when the victim “was verbally aggressive prior to the

violence compared to when she was not” (Witte, Schroeder, & Lohr, 2006, p. 662).

College students also report higher acceptance of violence when it is used in self-defense

(Cauffman et al., 2000; Foo & Margolin, 1995). Cauffman and colleagues (2000) found

that college students were more accepting of violence when used in self-defense, to be

playful, and as revenge. Overall, research has shown that perpetration is more likely to

be justified in situations where victims are perceived as provoking their partners and

when the violence is based on retaliation or playfulness (e.g., “how acceptable is it if Paul

was just being playful and it went too far,” p. 670).

Research on common beliefs about dating violence can further inform how

adolescents and college students’ attitudes relate to their victimization and perpetration.

Few studies have investigated attitudes of dating violence from the victims’ point of view.

The following studies suggest that victims tend to report jealousy, control, and retaliation

as common motivations for the violence. In O’Keefe and Treister’s (1998) cross-

sectional study of 939 high school students, women reported jealousy and anger as the

most common motivations for violence while men reported jealousy and retaliation as the

most frequent motivations. Moreover, in Molidor and Tolman’s (1998) crosssectional

study of 635 high school students’ IPV, both genders reported jealousy as a reason for

violence; however, males were more likely to report being victimized because their

partners were jealous of their interactions with the opposite sex (49% of males versus

25% of females) as well as because the males themselves were jealous regarding their

partners’ interactions with the opposite sex (21% of males versus 10% of females).

Another cross-sectional study on victimization reported that, among a college sample of

primarily freshmen students (n = 495), female victims of physical violence believed their

partner used violence to gain control over them or to strike back as a response to the

women hitting first (Follingstad et al., 1991). Male victims, on the other hand, thought

their partner used violence to express their anger or to show that they felt “emotionally

hurt or mistreated” (p. 53). Both women and men reported the strongest motive for their

partner’s use of physical violence was “to get control or get their own way”; this was

followed by, “retaliation for being emotionally hurt”, jealousy, or “to show how angry

they were” (p. 53).

Most studies of attitudes about dating violence have focused on perpetration.

Follingstad and collaborators (1991) found that female perpetrators reported using

physical violence as a response to feeling emotionally hurt or wanting to express their

anger, and male perpetrators reported using physical violence in response to the woman

hitting first or because of jealousy. Both genders reported their strongest motive for using

violence to be “feeling emotionally hurt or mistreated” or “a way of expressing anger” (p.

53). In a later study, Follingstad, Bradley, Laughlin, and Burke (1999) found the largest

effect sizes for the variable of “efforts to control partner,” followed by jealousy when

asking college students for the main reasons they had perpetrated physical dating violence

(p. 371). The researchers also reported that male perpetrators had more issues with

alcohol use and with controlling their anger compared to female perpetrators. The

researcher used a stratified sampling method to recruit 617 students who represented the

general student population (by race and sex) for their cross-sectional study. In

another college sample of 493 women the participants were asked an openended question

regarding why they had perpetrated violence against their partners (Hettrich & O’Leary,

2007). The 11 most common responses in this cross-sectional study were: anger (the

most common response), escalation of verbal argument, frustration, emotionally hurt,

retaliation, poor communication, to show seriousness, external act by the boyfriend,

boyfriend lied, externally cued, and preventing boyfriend from leaving argument (least

common).

Moreover, Foshee, Bauman, Linder, Rice, and Wilcher (2007) interviewed 116

adolescents about their dating violence experiences as part of an evaluation of a dating

violence prevention program in which participants were randomly assigned to prevention

or control group. Foshee and colleagues found different patterns of perpetration among

males and females, excluding adolescents who reported dating violence that was

accidental or seen as playful. The researchers reported four female perpetration types,

with both the most common type and the least common type reporting perpetrating

violence as a response to their boyfriends’ perpetration. The first type of female

perpetrators reported being victims of ongoing violence from their boyfriends while the

fourth type reported experiencing it for the first time. Both types described self-defense

as one motive for the violence in addition to the first type reporting being “fed up” and

the fourth type reporting retaliation as a motive (p. 506). The second and the third type of

female perpetrators also reported similar reasons for perpetrating violence, which were

mostly connected to infidelity or jealousy. However, the second most common type of

female perpetrator reported anger as the main motive, while the third most common type

reported wanting to let their boyfriends know they had done something wrong. The

researchers were only able to find one male perpetrator type, which was described as

using violence to prevent their girlfriends’ perpetration from escalating. Moreover, the

majority of the males reported that their girlfriends’ use of violence was a response to

them finding out the boyfriends had cheated on them or because of jealousy. Some

researchers have speculated on whether men and women could have different reasons for

using violence in a relationship. Again the focus of the studies is on perpetration as most

researchers are trying to understand why adolescents and college students would

perpetrate violence against a partner. Gover and colleagues (2008) argue in their

discussion of their empirical study that women might be more likely to use violence in

relationships because of the lower level of commitment in college dating relationships,

which creates jealousy and frustration and might result in violence. Katz and

collaborators (2002) speculate in the discussion section of their empirical study that

gender roles can influence when college women and men use violence. Since women are

socialized to be more emotional and to have closeness to people they might be more

likely to perpetrate violence when they feel like their partner is being passive or

withdrawing (Katz et al., 2002). Men on the other hand are socialized to be very

independent and therefore they might be perpetrating violence when their partners are

criticizing them or they are demanding (Katz et al., 2002). However, neither Gover and

colleagues nor Katz and collaborators explicitly tested their ideas in their current studies.

In summary, research has shown that dating violence is a major societal concern based on

the high prevalence rates (Harned, 2002; Jackson et al., 2000; Katz et al., 2002) and

multiple negative physical and psychological consequences (Amar & Gennaro, 2005;

Follingstad et al., 1991; Harned, 2002; Molidor & Tolman, 1998; O’Keefe & Treister,

1998). Moreover, dating violence appears to be related to attitudes and acceptance of

IPV, which is of major focus in this study. Few theories have been developed which

predict or explain why dating violence occur, but the following section will discuss two

theories that are relevant for this study.

Theories of Dating Violence

Most theories of the etiology of dating violence focus on predicting perpetration.

This project focused on predicting victimization risk, recognizing the lack of research in

this area. More specifically, this study attempted to add to our understanding of how

attitudes, such as acceptance, of dating violence, can contribute to women’s vulnerability

to or put them at greater risk for victimization. Even though few theories or models exist

regarding dating violence, and most of them are focused on perpetration, they can provide

a framework for understanding risk factors for dating violence in general.

One of the explanatory theories for dating violence is the intergenerational

transmission of violence theory, which is based on Bandura’s social learning theory

(1977). The intergenerational transmission of violence theory suggests that people who

are abused and/or witnessed abuse as children are more likely to become victims and/or

perpetrators as adults than children who have not been exposed to violence. Both

vicarious (e.g., learning to be violent by observing others) and instrumental learning (e.g.,

learning to be violent by the consequences that follows) are assumed to play a role.

Multiple studies of high school and college students have offered empirical support for

this theory.

Gover and colleagues (2008) distributed surveys to groups of college students

(total of 2,541 men and women). They found that women who had been physically

abused as children were more likely to become victims of both physical and

psychological dating violence than the non-abused college students. For males, they were

only more likely to be victims of psychological violence if they had been physically

abused as children. Women who reported witnessing their fathers abuse their mothers

were more likely to be victims of physical dating violence. Witnessing interparental

abuse did not predict dating violence vicitmization for males. Gover and colleagues also

found that both male and female college students who had been physically abused as

children were more likely to perpetrate physical violence in relationships than students

with no abuse history. Women were also more likely to perpetrate psychological violence

if they had been physically abused as children. Witnessing interparental abuse did not

predict psychological dating violence victimization or perpetration. Another cross-

sectional study focused on 232 high school students who had witnessed high levels of

interparental violence as children (O’Keefe, 1998). Even though O’Keefe excluded

students who had not witnessed parental violence or had witnessed lower levels of

violence from participating in her study, her findings give some insight to risk factors as

well as protective factors for adolescents who witnessed interparental abuse as children.

She found that school performance was a protective factor for females and self-esteem

was a protective factor for males. Even though they had witnessed parental violence as

children, females who did well in school were less likely to perpetrate and be victims of

dating violence and males with higher self-esteem were less likely to perpetrate violence.

O’Keefe also found that for females being abused as children increased the likelihood for

them to report being both victims and perpetrators of dating violence. Males who were or

had been in violent dating relationships were more likely to be of lower socioeconomic

status and be more accepting of dating violence than those who had not experienced

dating violence. Moreover, both male and female perpetrators were more likely to have

been exposed to community and school violence. In another high school sample of 719

males and females, they found that both males and females were more likely to perpetrate

physical violence if they had witnessed their mother abuse their father, but the opposite

was not a significant predictor (Malik et al., 1997). Moreover, they were more likely to

be victims of physical violence if they reported being victims of child physical abuse.

Similar results were found in two different college student samples. Foo and

Margolin’s (1995) cross-sectional study with 290 college males and females showed that

witnessing interparental abuse predicted physical perpetration for males while child

sexual abuse predicted physical perpetration for females. Another cross-sectional study

found that students were more likely to be victims of dating violence if they had

witnessed both parents abuse each other (Jankowski, Leitenberg, Henning, & Coffey,

1999). This was not supported for those witnessing one of their parents abuse the other

(either same sex parent or opposite sex parent). The authors argue that this finding might

be more related to acceptance of violence (attitudinal influence) than parental modeling

of violence. The students were more likely to perpetrate violence in relationships if they

had witnessed the same sex parent abusing their partner or both parents abusing each

other. The effect was strongest for witnessing the same sex parent abusing their partner.

There was no effect for witnessing the opposite sex parent abusing their partner

predicting perpetration. The researchers recruited 1,576 college students (women and

men) for their study.

A cross-sectional study by Hendy and colleagues (2003) found that the greater the

number of role models of violence that college students had been exposed to in their lives

the more likely they were to be victims and perpetrators of dating violence. They

recruited 608 men and women who were in heterosexual dating relationships. The

predictor variables in the model included in this study were two types of interparental

violence (mother abusing father and/or father abusing mother), two types of childhood

abuse (being abused by father and/or mother), one type of past relationship violence

(being victimized by past partners), and two types of current relationship violence

(victimization and perpetration). Hendy et al. found that those who reported being

abused by their mothers as children were more likely to be victims of dating violence.

Females were also more likely to be victims if they had witnessed their mother abuse

their father and males were more likely to be victims if they had been victims in past

relationships. Being victimized by a current partner was only relevant for predicting

perpetration of violence. Both males and females were more likely to perpetrate violence

if their current partners were violent. For females, witnessing their mother abusing their

father increased the likelihood of them perpetrating violence against their partners. For

males, past abuse from their mothers and previous partners also predicted perpetration.

An extension to the intergenerational transmission of violence theory is the background-

situational model (Riggs & O’Leary, 1989, 1996), which is based on social learning

theory and conflict theory. Bandura’s (1977) social learning theory states that violence is

learned through observations and imitations (e.g., instrumental or vicarious learning);

these behaviors are then reinforced by different factors. Conflict theory, on the other

hand, “assumes that conflict is an inevitable part of human association, whereas violence

as a tactic to deal with conflict is not” (Straus et al., 1996, p. 284).

The background-situational model includes contextual and situational factors that

are related to dating violence. The background or contextual factors include: models of

aggression in intimate relationships, parental aggression toward the child, acceptance of

aggression as an appropriate response to conflict, frustration, or threat, and prior use of

aggression (Riggs & O’Leary, 1989). The situational factors include: the expectation of

a positive outcome of the aggression, the partner’s use of aggression, and relationship

conflict (Riggs & O’Leary, 1989). The contextual factors are more distal while the

situational are more proximal (O’Keefe & Treister, 1998). According to Riggs and

O’Leary (1989, 1996), people’s early experiences of violence make them more likely to

accept violence and later use it in different situations, especially in intimate relationships.

The contextual factors predict who will be violent and the situational factors predict when

or in what situations someone will be violent (Riggs & O’Leary, 1989).

In Luthra and Gidycz’s (2006) study, where the background-situational model was

used to predict perpetration, they further emphasize that, according to the model,

“attitudinal acceptance of aggression increases the likelihood that one will behave

aggressively, both in general relationships and in dating relationships” (p. 719). Also,

past aggressive behaviors towards partners and non-partners predict future use of

violence.

Riggs and O’Leary (1996) tested their model of dating violence perpetration by

using a sample of 375 college students who were all currently involved in a heterosexual

dating relationship. The researchers included four contextual/background variables

(interparental violence, childhood abuse, acceptance of relationship violence, and

previous violent behavior) and one situational variable (relationship conflict) to predict

physical and psychological/verbal dating violence. The situational factor in their study

consisted of two questions that they combined into one variable. The questions were

related to number of arguments in the past month and frequency of conflict. The

researchers decided to only include relationship conflict as a situational factor because

they believe that it is very central to their model and that the background factors they

included, especially acceptance of violence, increases the likelihood of IPV. They used a

structural equation analysis to test their hypothesized model.

Riggs and O’Leary (1996) found that their model was better at predicting

perpetration of dating violence by males than by females, as it explained 62% of the

variance for men but only 32% for women. Females and males who witnessed their

parents abuse each other were more likely to report being victims of childhood abuse.

Being a victim of childhood abuse predicted aggressive behavior such as arguing and

fighting with peers for both genders, while witnessing interparental abuse predicted

aggression against peers and acceptance of violence only for females. Childhood abuse

did not predict acceptance of violence for either sex. Moreover, acceptance of violence

predicted perpetration of dating violence for both females and males, as did previous

violent behavior against peers. Acceptance did not predict relationship conflict but

previous violent behavior did for both genders. Finally, relationship conflict predicted

perpetration of dating violence for males and females.

Riggs and O’Leary’s (1996) model was better at predicting male perpetration than

female perpetration. They suggested this might be because they did not examine the

variable of partner’s use of aggression, which they hypothesized might be a better

predictor for women than men. Findings from the Luthra and Gidycz’s (2006) study

independently supported this idea even though they did not specifically aim to test that

idea. In their perpetration study with 200 college students, they included the same

contextual factors as Riggs and O’Leary (1996) except for separating previous violent

behavior into physical and verbal fighting (Riggs and O’Leary combined the two).

However, compared to Riggs and O’Leary’s one situational factor (relationship conflict),

Luthra and Gidycz included three situational factors (substance use, dating violence

victimization, and problem-solving skills) in their study. The data was analyzed

separately for men and women using a logistic regression analysis; only predictor

variables that were significantly correlated with perpetration were included in the final

model.

Overall, Luthra and Gidycz’ model was better at predicting female perpetration

than male perpetration, as they found that their model explained 83.3% of the female

perpetrators’ behavior but only 30% of the male perpetrators’ behavior. In their study,

they also found that among the five factors that predicted female perpetration “the single

largest predictor…was a partner’s use of aggression” (p. 725). The other significant

predictors for female perpetration were alcohol use, being abused by father, neglecting

conflictual situations, and ending the relationship to resolve conflict. The model for

predicting male perpetration included three factors which were alcohol use, longer

relationships, and being victims of dating violence in their current relationship. Other

researchers have also found that having a violent partner increases the likelihood for

perpetration. Hendy and colleagues (2003) found that both females and males are more

likely to perpetrate violence if their partner is violent.

Since there are few theories of dating violence and most of them focus on

perpetration, some researchers have used perpetration risk theories to predict

victimization risk. O’Keefe and Treister (1998) were interested in using the

backgroundsituational model to predict victimization. The researchers used a sample of

939 high school students, a majority of which were Latino and of low socioeconomic

status. Six contextual factors were evaluated as predictors of victimization, including the

same three as both previously mentioned perpetration studies used (childhood abuse,

interparental violence, acceptance of violence). O’Keefe and Treister also evaluated three

additional contextual factors (exposure to community and school violence, interpersonal

control, and self-esteem). Moreover, they included a total of six situational factors

(perpetration of dating violence, relationship conflict, relationship commitment,

relationship satisfaction, length of time dating, and number of dating partners).

O’Keefe and Treister (1998) used multiple regression analyses to determine which

factors predicted victimization. They found that high school females and males were

both more likely to be victims of dating violence if they also had perpetrated dating

violence. Furthermore, women were more likely to be victims if they were accepting of

male-to-female violence, had been in more dating relationships, had lower self-esteem,

reported less interpersonal control, and had more exposure to community and school

violence. They were also more likely to report being victims of dating violence if their

relationships had high levels of conflict, were less satisfying, and more committed. For

the women, the model explained 55% of the variance. The only variable that predicted

victimization among males was their perpetration of violence; male perpetration

explained 44% of the variance.

When testing the background-situational model, Luthra and Gidycz’s model was

better at predicting female than male perpetration. Their model included partner’s use of

violence, which was the strongest predictor for perpetration among females. Other studies

support that dating violence tends to be bidirectional (Bookwala et al., 1992; Harned,

2002; Katz et al., 2002; O’Keefe & Treister, 1998; Riggs & O’Leary, 1996). The

victimization study (O’Keefe & Treister, 1998) also found support for bidirectionality, as

they reported that partners’ use of violence was the main predictor of victimization for

females.

The two perpetration studies (Luthra & Gidycz, 2006; Riggs & O’Leary, 1996)

found some evidence that exposure to violence as a child predicts later aggression,

especially for females; however, this was not supported in the victimization study

(O’Keefe & Treister, 1998). Acceptance of relationship violence predicted perpetration

for both females and males and witnessing interparental abuse predicted acceptance of

violence for females but not for males (Riggs & O’Leary, 1996). O’Keefe and Treister

(1998) found that acceptance of male-to-female violence predicted victimization among

females.

Rationale for Current Study

In summary, according to the intergenerational transmission of violence theory,

early experiences with violence can lead to victimization and perpetration of dating

violence later in life; this finding has been supported by research (Gover et al., 2008;

Jankowski et al., 1999; O’Keefe, 1998). In the background-situational model, Riggs and

O’Leary (1989, 1996), while including early exposure to violence as an important factor

in their model, expand the model to include more background factors, as well as specific

situational factors, that can further predict risk of dating violence. By including both

distal and proximal factors in their model, they hypothesized that several background

factors can predict who will be violent and that certain situational factors can further

predict when someone will perpetrate violence. Riggs and O’Leary state that previous

experiences with violence can make people more accepting of violence, which can

subsequently lead to perpetration or victimization of dating violence.

Even though other studies have investigated previous exposure of violence and

acceptance as contextual factors (Luthra & Gidycz, 2006; O’Keefe & Treister, 1998;

Riggs & O’Leary, 1996) none of them have investigated the additional predictive value of

acceptance after accounting for exposure. Previous exposure to violence might put

college students at risk for becoming victims of dating violence. Potentially they have

become tolerant of the violence and believe that it is a normal way of handling

relationship conflict.

Some researchers have emphasized the need to understand people’s perception of

dating violence (Flynn & Graham, 2010; Lichter & McCloskey, 2004); however, current

findings on perceptions as they relate to victimization are limited. Therefore, by focusing

on women only, this study attempted to clarify how attitudes regarding acceptance of IPV

predicted victimization. Many models, including the background-situational model, were

developed to predict IPV perpetration. By using variables from the

backgroundsituational model to predict female victimization risk, this study aimed at

understanding how women who have been exposed to violence might hold attitudes

towards dating violence, which might increase their acceptance of the violence as well as

their tolerance for the violence. If that is the case, then an attitude which is indicative of

acceptance of dating violence might be an important factor to include in dating violence

prevention and intervention programs. Hypotheses

This study aimed at replicating and extending the O’Keefe and Treister’s (1998)

model of victimization by focusing on certain components of the model. The focus was

on the predictive value of previous exposure of violence (i.e., witnessing interparental

violence and childhood abuse) and acceptance of violence for dating violence

victimization among college women. This study more specifically predicted that college

students who had been exposed to violence and reported being more accepting of the

violence would be at a greater risk to become victims of dating violence.

Hypothesis 1:

Previous exposure to violence (witnessing interparental violence and/or childhood

abuse) and acceptance of dating violence would predict dating violence victimization.

Acceptance of dating violence was expected to account for significantly more of the

variance in victimization after considering previous exposure.

Hypothesis 2:

Acceptance of dating violence would mediate the relationship between previous

exposure and victimization.

Hypothesis 3:

Based on previous research, this study predicted that victims were also likely to

report being perpetrators of dating violence. Therefore, among the victims, a positive

correlation was predicted between the victimization and perpetration indicating that

higher levels of victimization were related to higher levels of perpetration.

Method Participants

Undergraduate female students enrolled in introductory classes in psychology at

the University of Arkansas were recruited for this study. Two hundred women consented

to participate in this study. However, 11 of them did not answer any of the questionnaires

and they were therefore excluded from this study. This resulted in a final sample of 189

participants.

Measures

Demographics and relationship status. This questionnaire included information

about respondent demographics (e.g., age, sex, year in school, ethnicity, and sexuality)

and relationship status (e.g., current and past dating experiences).

Witnessing interparental abuse. A total of four questions were created to

address witnessing father physically abuse mother and witnessing mother physically

abuse father. There were two questions for each parent as a perpetrator to allow for

inclusion of both minor (e.g., pushing, shoving, grabbing, slapping, or throwing

something at partner) and severe (e.g., choking, beating up, burning, scolding, kicking, or

using a knife or gun on partner) violent acts as used in the CTS2 (Straus et al., 1996).

Participants were coded as having witnessed interparental abuse if they answered yes to

one or more of the four questions. Witnessing interparental abuse was scored as 1 and

not witnessing was scored as 0.

Childhood abuse. Selected items from Briere’s (1992) Childhood Maltreatment

Interview Schedule Short Form (CMIS-SF) were used to determine participants’

experiences of childhood abuse. The questions included physical, sexual, and

psychological abuse by parent or caregiver before the age of 17. According to Briere

there are no studies on the psychometrics of the CMIS-SF, partly because the measure

does not have a total score that is used for a clinical cutoff to determine abuse status.

There are no scores for two of the subscales (physical and sexual abuse).

This study identified participants as being abused as children if they answered yes

to one or more of the questions regarding physical and/or sexual abuse experiences. For

sexual abuse, the perpetrator also had to have been either a family member and/or at least

five years older than the participant. Presence of childhood abuse was scored as 1 and

absence of it was scored as 0.

Predictions for psychological childhood abuse were not initially included in the

hypotheses for this study but data regarding such abuse were used in the post-hoc

analyses. Using Briere’s CMIS-SF, participants were asked if any of the seven items

included as descriptive of psychological abuse ever happened to them before the age of

16. The scale used for each item ranges from never (0) to over 20 times a year (6).

Respondents are asked to endorse an item “if their parents, stepparents, foster parents, or

other adult in charge of them as a child ever did any of the following to them” (See

Appendix E). Examples of items are “criticize you,” “ridicule or humiliate you,” and

“make you feel like you were a bad person.” A total score was calculated by summing the

responses to the seven items. Participants were considered psychologically abused as

children if they had a score of 21 or higher (i.e., at or above the 75th percentile).

Acceptance of violence. The Attitudes Towards Dating Violence Scales was used

to determine college students’ acceptance of dating violence (Price & Byers, 1999). The

measure has two scales: Attitudes Towards Male Dating Violence (AMDV) and Attitudes

Towards Female Dating Violence (AFDV). Both AMDV and AFDV have three subscales

of attitudes towards males’ and females’ use of psychological (AMVDPsyc and AFVD-

Psyc), physical (AMVD-Phys and AFVD-Phys), and sexual violence (AMVD-Sex and

AFVD-Sex). The AMVD consists of 39 items and the AFVD consists of 37 items. The

physical and the sexual violence scales each have 12 items while the psychological

violence scale has 15 for the AMVD and 13 for the AFVD. Participants are asked to

endorse their agreement or disagreement with 76 items. Items are scored using a likert

scale ranging from strongly disagree (1) to strongly agree (5); “higher scores indicate a

greater acceptance of abusive behavior” (p. 359).

Examples of items referring to acceptance of male and female perpetrated physical

violence are “a guy usually does not slap his girlfriend unless she deserves it” (p. 358)

and “some guys deserve to be slapped by their girlfriends” (p.360). For sexual violence,

examples are “a guy should not touch his girlfriend unless she wants to be touched” (p.

359) and “a guy who goes into a girl's bedroom is agreeing to sex” (p. 361). Lastly, two

examples of male and female perpetrated psychological violence are “a guy should not

insult his girlfriend” (p. 358) and “girls have a right to tell their boyfriends what to do”

(p. 360).

Internal consistencies for the AMVD range between α = .76 (girls’ responses to

the AMVD-Psyc) to α = .88 (boys’ responses to the AMVD-Sex) and between α = .72

(boys’ responses to the AFVD-Psyc) and α = .87 (boys’ responses to the AFVD-Sex) for

the AFVD. The measure shows evidence for convergent validity as it correlates with

AWSA (Attitudes Toward Women Scale for Adolescents; Galambos, Peterson, Richards,

& Gitelson 1985), which is a measure of traditional gender role beliefs (higher scores

indicating being more accepting of traditional gender roles). Price and Byers (1999)

reported evidence for criterion-related validity of their measure as well. Boys and girls

who were more accepting of violence were also more likely to perpetrate violence, and

boys were more accepting of violence if they reported having a male friend who is a

perpetrator; nevertheless, this was not true for physical violence.

Participants’ acceptance for male-to-female violence was determined by calculating

their total score on the AMDV. Similarly, participants’ acceptance for female-to-male

violence was determined by calculating their total score on the AFDV. After reverse

coding some of the items, a higher score indicates greater acceptance. The AMDV and

AFDV scores were also combined into a total acceptance score.

Victimization and perpetration rates. The Revised Conflict Tactics Scale

(CTS2; Straus et al., 1996) was used in order to determine perpetration and victimization

rates of dating violence. The CTS2 and its previous versions are one of the most

common measures used to establish rates of IPV. According to Straus and colleagues, the

measure is designed to measure “concrete acts and events” and “is not intended to

measure attitudes about conflict or violence nor the causes or consequences of using

different tactics” (1996, p. 284). The measure has a total of 78 questions (39 items asked

twice; once about the participant and once about his or her partner). For each item,

participants are asked how often the described event has occurred in the past year and the

options range from 0 (this has never happened) to 7 (not in the past year, but happened

before) with values in between ranging from 1 (once in the past year) to 6 (more than 20

times in the past year). The measure has five scales with two subscales each: Physical

assault (minor, severe), sexual coercion (minor, severe), psychological aggression (minor,

severe), negotiation (emotional, cognitive), and injury (minor, severe). The physical

assault scale determines acts of physical violence, while the psychological aggression

scale determines verbal and non-verbal psychological/emotional violence such as

“insulted or swore at my partner” and “threatened to hit or throw something at my

partner” (p. 308). Straus and colleagues define sexual coercion as “behavior that is

intended to compel the partner to engage in unwanted sexual activity” (p. 290) such as

“made my partner have sex without a condom” (p. 309) and “used threats to make my

partner have sex” (p. 309). Negotiation is defined as “actions taken to settle a

disagreement through discussion” (p. 289); for example, “showed respect for my

partner’s feelings about an issue” (example of emotional negotiation; p. 308) and

“suggested a compromise to a disagreement” (example of cognitive negotiation; p. 308).

The injury scale measures physical injury due to partner’s physical abuse. The

CTS2 subscales have been shown to be reliable by internal consistency values ranging

from α = .79 (psychological aggression) to α = .95 (injury) (Straus et al.,

1996). The various items have a mean correlation of α = .77 with values ranging from α

= .34 (sexual coercion item; insisted on sex without a condom) to α = .92 (injury item;

partner was cut or bleeding) (Straus et al., 2006). Moreover, the measure has evidence

for construct validity as indicated by its discriminative, convergent, and divergent

validity. According to Straus and colleagues (1996), the higher correlation for males

compared to females between the sexual coercion scale and the psychological aggression

scale (r = .66 for men and r = .25 for women) and between the sexual coercion scale and

the physical assault scale (r = .90 for men and r = .26 for women) indicates that the

sexual coercion scale has discriminative validity. Men are more likely to be sexually

coercive than women and therefore their scores should be more highly correlated with the

other two scales. Furthermore, discriminative validity is evidenced by a higher

correlation between the injury and the physical assault scales for men (r = .87) than for

women (r = .29) because men’s use of physical violence tends to result in more serious

injury. Additionally, the psychological aggression and the physical assault scales are

correlated for men (r = .71) and women (r = .67) which is evidence for convergent

validity. Lastly, the authors show a negative correlation between a measure of social

integration (the SI scale) and the physical assault scale, which indicates divergent validity

because people who use physical violence should be less socially integrated. Divergent

validity was also evidenced by low correlations between the negotiation and the sexual

coercion scales, which according to the authors should not be highly correlated since they

are two different constructs.

In this study the physical assault subscale of the CTS2 was used to determine

levels of victimization and perpetration among female college students. There is a total

of 24 physical assault items (12 for determining participant’s victimization level and 12

for determining their perpetration level). The answer options included for this study were

from 0 (this has never happened) to 6 (more than 20 times in the past year). A total score

was calculated for each participant by adding the scores together for all 12 items (max =

72). A higher score indicates higher levels of physical violence.

The sexual and psychological victimization scales were used for some of the

posthoc analyses. As for physical victimization, only answer options 0 (this has never

happened) to 6 (more than 20 times in the past year) were included in this study. The

items were summed to yield a total score, with a higher score indicating higher levels of

victimization. There are 8 items for psychological victimization and 7 items for sexual

victimization.

Procedure

Female undergraduate students at the University of Arkansas were recruited to

participate in this study through the online experiment system and they earned two credits

for their introductory psychology class for participating. First, the participants read the

consent form online, in which they were informed about the intentions of the study as

well as its potential risks. After consenting, they answered the questionnaires online.

The participants completed questionnaires about demographics and relationship status,

witnessing interparetal abuse, childhood abuse (CMIS-SF; Briere, 1992), acceptance of

dating violence (AMDV and AFDV; Price & Byers, 1999), and dating violence

victimization and perpetration (CTS2; Straus et al., 1996). Lastly, the participants were

debriefed online.

Analytic Approach

The first and second hypotheses were tested by using a hierarchical regression

analysis. Witnessing interparental abuse and childhood abuse were entered as a functional

set in the first step of the regression analysis. In the second step, acceptance of dating

violence was entered. Dating violence victimization was entered as the dependent

variable. Both exposure to violence and acceptance of violence were expected to be

significant predictors of dating violence victimization. Effect sizes were calculated using

f2 for hierarchical regression analyses, f2 = (R2AB – R2A) / (1 – R2AB). Small effect size is

.02, medium is .15, and large is .35.

Additionally, acceptance was expected to mediate the relationship between

previous exposure and victimization. Baron and Kenny’s (1986) approach to test for

mediation was utilized, which includes a series of regression analyses. First, the two

exposure variables (i.e., witnessing interparental abuse and childhood abuse) were

entered as independent variables and physical victimization was entered as the dependent

variable (e.g., path c). Next, the two exposure variables were entered as independent

variables, but this time acceptance was entered as the dependent variable (e.g., path a).

The third regression analysis was done by entering acceptance as the independent

variables and physical victimization as the dependent variable (e.g., path b). Lastly, the

two exposure variables and the acceptance variable were entered as independent variables

and physical victimization was entered as the dependent variable (e.g., path c′). The first

three regression analyses should be significant. In the last regression analysis in which

both the exposure variables and the acceptance variable were entered as independent

variables, the path should either not be significant any more (i.e., full mediation) or the

path should be less significant (i.e., partial mediation). The regression analyses including

acceptance were performed using the total acceptance score (combining AMDV and

AFDV) as well as AMDV and AFDV separately.

In addition to the Baron and Kenny (1986) approach, the Sobel test was used to

test for any indirect effects of acceptance in the relationship between exposure to violence

and victimization (Preacher, & Hayes, 2004). In the Sobel test, the unstandardized

coefficients and their standard errors were entered for path a and b to test for the indirect

effect of acceptance. The analyses were done separately for the two exposure variables to

test if there was an indirect effect of acceptance in the relationship between witnessing

interparental abuse and victimization as well as childhood abuse and victimization. This

was performed for the total acceptance score as well as for AMDV and AFDV separately.

The third hypothesis was analyzed by correlating the victims’ scores on the

physical assault victimization scale with their perpetration scores to determine if there

was a positive correlation between victimization and perpetration scores. Power

Analysis

Previous research on dating violence among college students has found medium

to large effect sizes when examining acceptance of violence based on blaming the victim

or the perpetrator (η2= 0.15-0.52; Witte et al., 2006) and less than small effect sizes when

examining motivations for violence (ε2= 0.01-0.12; Follingstad et al., 1999). Based on

the previous mentioned effect sizes, this study expected at least a small to medium effect

size of f = 0.06 and a power of 0.8, which required using 200 participants.

Post hoc power analyses were done using G* Power 3 (Erdfelder, Faul, &

Buchner, 1996) for multiple regressions (Omnibus R2 deviation from zero). The program

calculates power based on effect size f2, α - level (.05), total sample size, and number of

predictors.

Results Demographics

Mean age was 19.40 years (SD = 4.04; range 18-53 years) and 97.4% (n = 184) of

the women were younger than 25 years old. Most of the participants were freshman

(54.5%; n = 103) and sophomore (31.2%; n = 59) college students. The majority of the

women self-identified as Caucasian (84.7%; n = 160) whereas the other participants

selfidentified as Asian American (5.3%; n = 10), Hispanic (4.2%; n = 8), African

American (1.6%; n = 3), Native American (1.1%; n = 2), and other (3.2%; n = 6).

Moreover, most students reported being heterosexual (98.9%; n = 187) and two students

reported being bisexual. The self-identified bisexual women were included in the study

as there were only two of them, which did not impact the results.

The participants were on average 15.97 years old when they started dating (SD =

1.92; range 11-28). Out of the 189 participants, 59.3% (n = 112) reported currently being

in a dating relationship. Seventy-six of the daters reported being in a steady relationship

(67.8%), 20 in a casual relationship (17.9%), 9 lived with their partner (8.0%), and 6 were

engaged (5.4%). Most of the daters reported that their partner was older than them

(55.8%; n = 63) while 32.7% (n = 37) reported that their partner was the same age and

11.5% (n = 13) reported that they were dating someone younger than them. Mean age for

partners was 21.02 years (SD = 5.95; range 17-58). Three quarters of the sample had

been in one to five dating relationships (n = 143), 16.4% had been in six to ten

relationships (n = 31), and the rest of the daters had been in more than ten relationships (n

= 9). On average the women reported that they had been in 1.77 number of serious

relationships (SD = .96; range 0-5) and 40.7% of the daters (n = 74) reported that their

current relationship was also their longest relationship.

Missing data

Both childhood exposure to violence variables (witnessing interparental abuse and

childhood abuse) had less than 5% missing data. Acceptance of male perpetration

(AMDV) had 9.0% (n = 17) missing data and acceptance of female perpetration (AFDV)

had 11.1% (n = 21) missing data. When combining AMDV and AFDV, there was 18.0%

(n = 34) missing data. Acceptance for physical, sexual, and psychological dating

violence were scored by summing acceptance for male and female physical, sexual, and

psychological perpetrated violence independently (e.g., AMDV-Phys and AFDV-Phys,

AMDV-Psyc and AFDV-Psyc, and AMDV-Sex and AFDV-Sex). There was 7.9% (n =

15) missing data for acceptance of physical and sexual perpetration and there was 9.0%

(17) missing data for acceptance of psychological perpetration. There was 13.8% (n =

25) missing data for physical victimization and 13.2% (n = 26) for physical perpetration.

In terms of the additional variables that were used in the post-hoc analyses, there was less

than 5% missing data for psychological child abuse. There was 13.8% (n = 25) missing

data for sexual victimization and 19.0% (n = 36) missing data for psychological

victimization. For the perpetration variables, there was 8.5% (n = 16) missing data for

sexual perpetration and 17.5% (n = 33) for psychological perpetration.

Since both the acceptance and dating violence (victimization and perpetration)

variables had more than 5% missing data, several analyses were done to test for any

patterns to the missing data. All the acceptance, victimization, and perpetration variables

were dummy coded for missing data. Next, independent t-tests were used to test for any

differences on the dummy coded acceptance variables across the other acceptance

variables (e.g., the dummy coded variables were entered as independent variables and the

total sums of the other acceptance variables were entered as dependent variables). The

same was done for the dummy coded victimization and perpetration variables to test for

differences on the other victimization and perpetration variables (e.g., dummy coded

physical victimization variable and differences on sexual and psychological victimization

total scores). There were no significant differences; however, there was a marginal

difference for the dummy coded sexual perpetration variable and amount of

psychological perpetration, t(154) = 1.77, p = .081. The missing group reported less

psychological perpetration (M = 3.11, SD = 3.66) compared to the non-missing group (M

= 7.52, SD = 7.40).

Additionally, Chi Square and independent t-tests were used to test for patterns in

the dummy coded missing data variables across predictor and outcomes variables. The

dummy coded acceptance variables were tested for differences across the child exposure

variables as well as the victimization and perpetration variables. Likewise, the dummy

coded victimization and perpetration variables were tested for differences across the child

exposure variables and the acceptance variables. There were significant differences on

the dummy coded variable for acceptance of physical dating violence (total sum of

AMDV-Phys and AFDV-Phys) and sexual perpetration, t(171) = 2.90, p < .01. The

missing group reported more sexual perpetration (M = 2.82, SD = 3.54) compared to the

non-missing group (M = .83, SD = 2.10). There were marginal differences for dummy

coded AMDV variable and witnessing interparental abuse (χ2 = 2.93, p = .087), dummy

coded physical victimization variable and child sexual abuse (χ2 = 3.55, p = .059), and

dummy coded psychological victimization variable and psychological child abuse (χ2 =

2.87, p = .090).

Lastly, the dummy coded missing data variables were tested for differences on

several demographical variables (i.e., participant’s age, ethnicity, dating status, number of

partners, age at first relationship, and partner’s age). There were several significant

differences on participant’s age (physical and sexual victimization, p < .05; physical and

sexual perpetration, p < .01), ethnicity (acceptance of psychological violence, p < .01;

physical and sexual victimization, p < .05; sexual perpetration, p < .05), dating status

(psychological victimization, p < .01; psychological perpetration, p < .05), number of

partners (physical victimization and perpetration, p < .05), and partner’s age (physical

victimization, p < .01; physical, sexual, and psychological perpetration, p < .05). There

were also several marginal differences on dummy coded missing data variables and

participant’s age (psychological perpetration, p = .059), ethnicity (acceptance of sexual

violence, p = .088), and partner’s age (sexual victimization, p = .061). For the two

continuous variables (participant’s and partner’s age), the missing groups were older and

had older partners than the non-missing group. There were no significant differences for

missing data variables and age at first relationship.

Because several variables had more than 5% missing data, regression analyses

were run with and without missing data for physical victimization and the acceptance

variables (e.g., total acceptance score, AMDV, AFDV, acceptance of physical violence,

AMDV-Phys, and AFDV-Phys). There were no differences in the results and therefore

the analyses below include the participants with missing data in the analyses.

Assumptions

Hierarchical regression analyses are based on several assumptions. The residuals

should be normally distributed and linear. Homoscedasticity and independence of errors

are also assumed. There should not be any singularity or multicollinearity among the

predictor variables and there should not be any outliers. Outliers were determined by

Mahalanobis and Cook’s distance as well as standardized residuals values exceeding 3.3

standard deviations (above or below 0).

In this study there were violations of assumptions on linearity of residuals and

homoscedasticity, mostly because several of the variables were not normally distributed.

The victimization variables (physical, sexual, and psychological) as well as some of the

acceptance variables (AMDV, AFDV, and acceptance of sexual violence) that were not

normally distributed were transformed in an attempt to normalize the distributions. The

variables were transformed using three formulas: inverse, logarithm, and square root.

Even after the transformations, the variables were still not normally distributed and they

still violated the same assumptions. Therefore, the decision was made to run the

regression analyses with the non-transformed variables.

Descriptive Statistics

Exposure variables. Twenty-seven women (14.3%) reported witnessing at least

of their parents abusing their other parent. Fifteen (55.6%) of those 27 childhood victims

reported that their father was the sole abuser. A third of the childhood victims reported

witnessing their father perpetrating mild violence only (n = 9), 18.5% reported witnessing

their father perpetrate both mild and severe violence (n = 5), and 3.7% reported

witnessing their father perpetrate severe violence only (n = 1). Eight childhood victims

(29.6%) had seen both of their parents abuse each other. Five of them (18.5%) witnessed

both parents perpetrate mild violence only, 2 (7.4%) witnessed both parents perpetrate

mild violence as well as their father perpetrate severe violence, and 1 (3.7%) witnessed

both parents perpetrate mild and severe violence. The last 4 childhood victims (14.8%)

reported witnessing their mother abuse their father. Two of them (7.4%) witnessed their

mother perpetrate mild and severe violence and the other 2 witnessed their mother

perpetrate mild violence only.

For the childhood abuse variables, 11.1% (n = 21) reported physical abuse and

15.3% (n = 29) sexual abuse. The sexual abuse victims reported either being abused by

someone who was at least 5 years older than them (48.3%, n = 14) or a family member

(34.5%, n = 10). The remaining 5 sexual abuse victims (17.2%) reported that they had

been sexually abused but did not report the identity of the abuser. When combining

sexual and physical childhood abuse, 23.3% (n = 44) of the women reported being

victims of at least one form of abuse. Of those 44 women, 52.3% (n = 23) were sexually

abused, 34.1% (n = 15) physically abused, and 13.6% (n = 6) were both physically and

sexually abused.

Approximately a third of the sample (30.7%, n = 58) reported some exposure to

interpersonal violence as children. Thirty-one of these women (53.4%) had been abused,

14 (24.2%) witnessed interparental violence, and 13 (22.4%) had both been abused and

witnessed their parents abuse each other.

As mentioned previously, psychological child abuse was not part of the

hypotheses but was used for some of the post-hoc analyses. See Table 1 for means,

standard deviations, and range for the 7 items and the total score. One-fourth of the

women (24.9%, n = 47) were psychologically abused as children.

Acceptance. Table 2 shows the means, standard deviations, and range for the

acceptance variables. Paired t-test analyses showed that participants were more accepting

of psychological perpetration than physical (t(162) = 12.19, p < .001) and sexual

perpetration (t(163) = 16.50, p < .001). They were also more accepting of physical than

sexual dating violence, t(163) = 3.76, p < .001. Female perpetrated violence was rated as

more acceptable than male perpetrated violence, t(154) = 2.59, p < .05, which was true

for both physical and sexual female perpetration (physical: t(173) = 10.43, p < .001;

sexual: t(173) = 5.33, p < .001) but male perpetrated psychological violence was rated as

more acceptable than female perpetrated psychological violence, t(171) = 9.25, p < .001.

Male perpetrated psychological violence was rated as more acceptable than male

perpetrated physical (t(174) = 21.13, p < .001) and sexual violence (t(176) = 21.85, p <

.001). There was no significant difference between acceptability of physical and sexual

dating violence for male perpetration, t(176) = 1.52, p = .13. Both female perpetrated

psychological (t(171) = 6.48, p < .001) and physical (t(172) = 5.87, p < .001) violence

were rates as more acceptable than sexual violence. There was no significant difference

between acceptability of physical and psychological dating violence for female

perpetration, t(172) = .36, p = .72.

Dating violence. Table 3 shows the means, standard deviations, and range for

physical, sexual, and psychological victimization and perpetration. Almost a fourth of the

sample (23.3%; n = 44) reported at least one incident of physical victimization.

Twenty-five of the victims (56.8%) reported being victims of minor physical violence, 4

(9.1%) reported being victims of severe physical violence, and 15 (34.1%) reported being

victims of both mild and severe physical violence. Almost a third of the sample (31.2%;

n = 59) reported perpetrating physical violence at least once. Thirty-five of the

perpetrators (59.3%) reported perpetrating minor physical violence, 7 (11.9%)

perpetrating severe physical violence, and 17 (28.8%) perpetrating mild and severe

physical violence.

Overall in this sample of college females, 32.8% (n = 62) reported at least one

experience with physical dating violence; 8.1% (n = 5) reported being victims only,

32.3% (n = 20) perpetrators only, and 59.7% (n = 37) reported being both victims and

perpetrators.

Sexual victimization was reported by 27.0% (n = 51) of the sample; 90.2% (n =

46) of the sexual abuse victims reported minor abuse, 2.0% (n = 1) severe abuse, and

7.8% (n = 4) both minor and severe abuse. Sexual perpetration was reported by 20.6% of

the sample (n = 39); 94.9% (n = 37) of the sexual perpetrators reported minor abuse,

2.6% (n = 1) severe, and 2.6% (n =1) both minor and severe. Overall, 28.0% (n = 53)

reported at least one experience with sexual dating violence and 34.0% (n = 18) of them

reported being victims only, 5.7% (n = 3) perpetrators only, and 60.4% (n = 32) reported

being both victims and perpetrators.

Psychological victimization and perpetration were the most commonly reported

forms of dating violence, with 59.8% (n = 113) reporting victimization and 62.4% (n =

118) reporting perpetration of acts of psychological violence. Sixty-eight of the victims

(60.2%) reported minor acts of victimization, and 39.8% (n = 45) minor and severe acts.

Sixty-eight of the perpetrators (57.6%) reported minor acts of perpetration and 50

(42.4%) reported minor and severe acts. Overall, 61.9% (n = 117) reported at least one

experience with psychological dating violence, with 2.6% (n = 3) reporting being victims

only, 4.3% (n = 5) perpetrators only, and 93.2% (n = 109) reporting being both victims

and perpetrators.

Lastly, 70.4% (n = 133) of the sample reported some experience with dating

violence (e.g., physical, sexual, and/or psychological violence), with 48.1% (n = 64) of

those participants reporting the experience of one type of dating violence, 29.3% (n = 39)

two types, and 22.6% (n = 30) all three types.

Hypotheses

Hypothesis 1. See Table 4 for the correlations between child exposure variables,

acceptance variables, and victimization and perpetration variables. Refer to Table 5 for

the unstandardized regression coefficients (B), standard errors (SE), standardized

regression coefficients (β), and R2. In the first hierarchical regression analysis with the

total acceptance score in the second step (combining AMDV and AFDV), witnessing

interparental abuse and child abuse (physical and/or sexual) did not account for

significant amount of variance in physical victimization, R2 = .04, F(2, 134) = 2.52, p =

.084. Moreover, acceptance did not explain a significant proportion of the variance in

physical victimization, ∆R2 = .01, ∆F(1, 133) = 1.65, p = .20. None of the variables were

significant predictors of physical victimization. The effect size for these analyses was

small, f2 = .013.

In the second hierarchical regression analysis with the AMDV score in the second

step, witnessing interparental abuse and child abuse (physical and/or sexual) did not

account for significant amount of variance in physical victimization, R2 = .04, F(2, 149) =

2.80, p = .064, and acceptance of male perpetrated violence (AMDV) did not account for

significant amount of variance in physical victimization ∆R2 = .01, ∆F(1, 148) = .73, p =

.39. The first step was marginally significant and witnessing interparental was a

significant predictor, B = 2.14, t = 2.03, p < .05. Childhood abuse and acceptance of male

perpetrated violence were not significant predictors. The effect size was small, f2 = .005.

See Table 6 for the unstandardized regression coefficients (B), standard errors (SE),

standardized regression coefficients (β), and R2.

In the third analysis, with AFDV in the second step, witnessing interparental abuse

and child abuse (physical and/or sexual) did not account for significant amount of

variance in physical victimization, R2 = .04, F(2, 145) = 2.73, p = .069, and acceptance of

female perpetrated violence (AFDV) did not account for significant amount of variance in

physical victimization, ∆R2 = .02, ∆F(1, 144) = 3.68, p = .057. Both the first and the

second step were marginally significant. Witnessing interparental abuse was a significant

predictor (B = 2.14, t = 2.00, p < .05) while AFDV (B = .04, t = 1.92, p = .057) was

marginally significant. The effect size for these analyses was small, f2 = .026. See Table

7 for the unstandardized regression coefficients (B), standard errors (SE), standardized

regression coefficients (β), and R2.

Taken together, hypothesis 1 was not supported as neither exposure to childhood

violence nor acceptance were significant predictors of physical dating violence

victimization.

Hypothesis 2. Refer to Table 8 for the unstandardized regression coefficients (B),

standard errors (SE), standardized regression coefficients (β), and R2. Several regression

analyses were used to test for acceptance as a mediator. In the first analysis, the exposure

variables (witnessing interparental abuse and physical and/or sexual child abuse) did not

account for significant amount of variance in physical victimization, R2 = .04, F(2, 161) =

3.03, p = .051; however, it was marginally significant, and witnessing interparental abuse

was a significant predictor of physical victimization, B = 2.14, t = 2.11, p < .05,

indicating that those who witnessed interparental abuse as children were more likely to

report being physically victimized as adults compared to those who did not witness

interparental abuse. Child abuse was not a significant predictor. The second regression

analysis was not significant, which indicates that exposure to violence as a child did not

account for significant amount of variance in acceptance of dating violence (combining

AMDV and AFDV), R2 = .01, F(2, 152) = .95, p = .39, and none of the exposure

variables were significant predictors of acceptance. Moreover, acceptance did not

account for significant amount of variance in physical victimization, R2 = .02, F(2, 135) =

2.29, p = .13, and was therefore not a significant predictor of victimization. Lastly, there

was no difference in the results when entering the exposure variables and acceptance as

predictors of physical victimization, R2 = .05, F(2, 133) = 2.24, p = .087, as none of the

variables significantly predicted physical victimization.

Acceptance of male perpetrated violence (AMDV) did not change the previous

results as the exposure variables did not account for significant amount of variance in

AMDV, R2 = .02, F(2, 169) = 1.28, p = .28, AMDV did not explain a significant amount

of variance in physical victimization, R2 = .01, F(1, 150) = 1.25, p = .27, and all three

predictors entered at once (exposure variables and AMDV) did not explain a significant

amount of variance in physical victimization, R2 = .04, F(3, 148) = 2.11, p = .10. Refer to

Table 9 for the unstandardized regression coefficients (B), standard errors (SE),

standardized regression coefficients (β), and R2 for acceptance of male perpetrated

violence as a mediator between childhood exposure and dating violence victimization.

Similarly, the exposure variables did not account for significant amount of variance in

acceptance of female perpetrated violence (AFDV), R2 = .01, F(2, 164) = .81, p = .45.

However, AFDV did explain a significant amount of variance in physical victimization,

R2 = .03, F(1, 146) = 4.50, p < .05, and significantly predicted victimization, B = .05, t =

2.12, p < .05. Additionally, when entering the exposure variables and AFDV as

independent variables, they accounted for a significant amount of variance in physical

victimization, R2 = .06, F(3, 144) = 3.08, p < .05. None of the predictors were

significant; however, AFDV was marginally significant, B = .04, t = 1.92, p = .057. Refer

to Table 10 for the unstandardized regression coefficients (B), standard errors (SE),

standardized regression coefficients (β), and R2 for acceptance of female perpetrated

violence as a mediator between childhood exposure and dating violence victimization.

The above analyses do not support hypothesis 2 as path c (exposure variables

predicting physical victimization) was not significant and it remained non-significant

when entering the exposure variables and the acceptance variables together to predict

victimization (path c′). The last path was significant when entering the exposure

variables together with acceptance of female perpetration but this finding does not

support that acceptance mediates the relationship between exposure to violence and later

victimization.

The Sobel tests were consistent with this conclusion because none of the Sobel

tests were significant (acceptance total: witnessing interparental abuse, z = .97, SE = .17,

p = .33; child abuse, z = .31, SE = .12, p = .76; AMDV: witnessing interparental abuse, z

= .90, SE = .16, p = .37; child abuse, z = .09, SE = .08, p = .93; AFDV: witnessing

interparental abuse, z = 1.02, SE = .20, p = .31; child abuse, z = .22, SE = .15, p = .83).

Hypothesis 3. As predicted, there was a significant positive correlation between physical

dating violence victimization and perpetration (r = .69, p < .001).

Power analysis. The post hoc power analysis showed that the power was .225 for

the regression analysis using the total acceptance score, .110 for AMDV analysis, and

.425 for AFDV. These results suggest that there could be problems with Type II errors,

which is the rate of false negatives or poor sensitivity. Because of this the null

hypothesis probably should have been rejected but it was not.

Post-hoc Analyses

Several studies have found a link between previous exposure to violence and

subsequent dating violence (Gover et al., 2008; Hendy et al., 2003; Jankowski et al.,

1999; O’Keefe, 1998; Riggs & O’Leary, 1996); however, this was not found in this study.

A series of post-hoc analyses were done in an attempt to understand why this effect was

not supported. Moreover, the post-hoc analyses were used to further explore any impact

acceptance of dating violence might play in the relationship between childhood exposure

to violence and dating violence victimization.

Hierarchical regression analyses were used to test if exposure to a specific type of

violence was related to an increased acceptance for that violence, which would predict

being a victim of that violence. Witnessing interparental abuse and physical child abuse

were entered in the first step and acceptance of physical dating violence was entered in

the second step to predict physical dating violence victimization. The same analysis was

done for acceptance of male physical perpetration (AMDV-Phys) and acceptance of

female physical perpetration (AFDV-Phys). Similarly, for sexual victimization as the

dependent variable, witnessing interparental abuse and sexual child abuse were entered in

the first step and acceptance of sexual dating violence was entered in the second step.

Separate analyses were done with AMDV-Sex and AFDV-Sex. Lastly, witnessing

interparental abuse and psychological child abuse were entered in the first step and

acceptance of psychological dating violence was entered in the second step to predict

psychological victimization. Separate analyses were done with AMDV-Psyc and

AFDVPsyc. The mediation analyses were also done by separating exposure, acceptance,

and victimization variables by type of violence.

Hierarchical regression analyses. See Table 11 for the unstandardized

regression coefficients (B), standard errors (SE), standardized regression coefficients (β),

and R2 for childhood exposure to physical violence, acceptance of physical violence, and

physical victimization. Witnessing interparental violence and physical child abuse did

not account for significant amount of variance in physical victimization, R2 = .04, F(2,

151) = 2.88, p = .059 and acceptance of physical dating violence did not explain a

significant proportion of the variance in physical victimization, ∆R2 = .02, ∆F(1, 150) =

2.85, p = .093. The first step was marginally significant and witnessing was a significant

predictor of physical victimization, B = 2.64, t = 2.27, p < .05. Exposure to physical child

abuse was not a significant predictor.

Similar results were found for acceptance of male perpetrated physical violence as

the exposure variables did not explain a significant amount of variance in physical

victimization, R2 = .04, F(2, 155) = 2.96, p = .055 and acceptance did not explain a

significant proportion of the variance in physical victimization, ∆R2 = .00, ∆F(1, 154) =

.00, p = .99. The first step was marginally significant and witnessing interparental abuse

was again a significant predictor, B = 2.64, t = 2.43, p < .05. When acceptance of female

perpetrated physical violence was entered in the second step, the exposure variables

marginally accounted for a significant amount of variance in physical victimization, R2 =

.04, F(2, 157) = 3.00, p = .053, and acceptance accounted for additional variance in

physical victimization above and beyond the main effects of exposure variables, ∆R2 =

.04, ∆F(1, 156) = 6.48, p < .05. Both witnessing and AFDV-Phys significantly predicted

victimization, B = 2.64, t = 2.44, p < .05 and B = .11, t = 2.55, p < .05 respectively. The

effect sizes for these analyses were small, f2 = .019 (acceptance of physical violence) and

f2 = .041 (AFDV-Phys). There was no effect size for AMDV since the second step did

not account for any variance. Refer to Tables 12-13 for the unstandardized regression

coefficients (B), standard errors (SE), standardized regression coefficients (β), and R2 for

acceptance of male and female perpetrated physical violence as predictors of physical

dating violence victimization.

See Table 14 for the unstandardized regression coefficients (B), standard errors

(SE), standardized regression coefficients (β), and R2 for exposure to sexual violence as a

child, acceptance of sexual violence, and sexual victimization. Witnessing interparental

violence and child sexual abuse accounted for significant amount of variance in sexual

victimization, R2 = .11, F(2, 148) = 8.79, p < .001, and both exposure variables were

significant predictors, B = 2.36, t = 2.88, p < .01 (witnessing), B = 2.24, t = 2.82, p < .01

(child sexual abuse). Acceptance of sexual dating violence did not explain additional

variance in sexual victimization after accounting for the exposure variables, ∆R2 = .01,

∆F(1, 147) = 1.21, p = .27.

The exposure variables also accounted for significant amount of variance when

acceptance of male perpetrated sexual violence was entered in the second step, R2 = .11,

F(2, 157) = 9.34, p < .001, and acceptance explained additional variance above and

beyond the main effects of witnessing interparental abuse and child sexual abuse ∆R2 =

.02, ∆F(1, 156) = 3.93, p < .05. All three predictors were significant, B = 2.36, t = 2.97, p

< .01 for witnessing, B = 2.24, t = 2.90, p < .01 for child sexual abuse, and B = .13, t =

1.98, p < .05 for acceptance of male perpetrated sexual violence. The first step remained

significant when acceptance of female perpetrated sexual violence was entered in the

second step, R2 = .11, F(2, 151) = 8.97, p < .001, but the second step was not significant,

∆R2 = .00, ∆F(1, 150) = .39, p = .53. Both exposure variables remained significant

predictors, B = 2.36, t = 2.91, p < .01 for witnessing, B = 2.24, t = 2.85, p < .01 for child

sexual abuse. The effect sizes range from less than small (f2 = .008 for acceptance of

sexual violence) to small (f2 = .025 for AMDV-Sex). The effect size for AFDV-Sex was

close to 0 (f2 = .002). See Tables 15-16 for the unstandardized regression coefficients

(B), standard errors (SE), standardized regression coefficients (β), and R2 for acceptance

of male and female perpetrated sexual violence as predictors of sexual dating violence

victimization.

See Table 17 for the unstandardized regression coefficients (B), standard errors

(SE), standardized regression coefficients (β), and R2 for exposure to psychological

violence as a child, acceptance of psychological violence, and psychological

victimization. Witnessing interparental violence and child psychological abuse accounted

for significant amount of variance in psychological victimization, R2 = .09, F(2, 140) =

6.75, p < .01, and acceptance of psychological dating violence explained additional

variance in victimization above and beyond the main effects of exposure variables, ∆R2 =

.06, ∆F(1, 139) = 9.05, p < .01. Both child abuse and acceptance were significant

predictors of psychological victimization, B = 5.55, t = 3.54, p < .01 and B = .16, t = 3.01,

p < .01 respectively.

Similar results were found for acceptance of male and female perpetration of

psychological violence separately. The exposure variables accounted for significant

amount of variance in victimization, R2 = .09, F(2, 146) = 7.04, p < .01 for AMDV-Psyc,

and R2 = .09, F(2, 143) = 6.90, p < .01 for AFDV-Psyc. Acceptance accounted for

significant amount of variance above and beyond the main effects of the exposure

variables, ∆R2 = .06, ∆F(1, 145) = 10.17, p < .01 for AMDV-Psyc and ∆R2 = .03, ∆F(1,

142) = 5.16, p < .05 for AFDV-Psyc. Again, both child abuse and acceptance were

significant predictors of psychological victimization, B = 5.55, t = 3.62, p < .001 and B =

.29, t = 3.19, p < .01 for AMDV-Psyc, B = 5.55, t = 3.58, p < .001 and B = .22, t = 2.27, p

< .05 for AFDV-Psyc. The effect sizes range from small (f2 = .036 for AFDV-Psyc) to

small-to-medium (f2 = .065 for acceptance of psychological violence and f2 = .070 for

AMDV-Psyc). Refer to Tables 18-19 for the unstandardized regression coefficients (B),

standard errors (SE), standardized regression coefficients (β), and R2 for acceptance of

male and female perpetrated psychological violence as predictors of psychological dating

violence victimization.

Mediation analyses. Several regression analyses were used to test for acceptance

of physical dating violence as a mediator between childhood exposure to physical

violence and later physical victimization. Refer to Table 20 for the unstandardized

regression coefficients (B), standard errors (SE), standardized regression coefficients (β),

and R2. In the first analysis, the exposure variables (witnessing interparental abuse and

physical child abuse) accounted for significant amount of variance in physical

victimization, R2 = .04, F(2, 161) = 3.07, p < .05. Witnessing interparental abuse was a

significant predictor of physical victimization, B = 2.64, t = 2.47, p < .05, but child abuse

was not a significant predictor. Second regression analysis was not significant, which

indicates that exposure to physical violence as a child did not account for significant

amount of variance in acceptance of physical dating violence, R2 = .01, F(2, 170) = .62, p

= .54. None of the exposure variables were significant predictors of acceptance.

Moreover, acceptance did not account for significant amount of variance in physical

victimization, R2 = .02, F(1, 152) = 3.45, p = .065, and was therefore not a significant

predictor of victimization. Lastly, childhood exposure to physical violence and

acceptance of physical violence accounted for significant amount of variance in physical

victimization, R2 = .06, F(3, 150) = 2.90, p < .05, and witnessing abuse as a child

remained a significant predictor, B = 2.49, t = 2.27, p < .05.

The results were similar when acceptance of male perpetrated physical violence

was used as a predictor. The exposure variables did not account for significant amount of

variance in acceptance of male perpetrated physical violence, R2 = .03, F(2, 176) = 2.34,

p = .10, and acceptance did not explain a significant amount of variance in physical

victimization, R2 = .00, F(1, 156) = .14, p = .71. This time entering the exposure

variables and acceptance were not significant predictors of victimization, R2 = .04, F(3,

154) = 1.96, p = .12. See Table 21 for the unstandardized regression coefficients (B),

standard errors (SE), standardized regression coefficients (β), and R2.

Exposure variables were not significant predictor of acceptance of female

perpetrated physical violence, R2 = .00, F(2, 176) = .26, p = .77; however, in contrast with

acceptance of male perpetrated violence, female perpetrated violence accounted for

significant amount of variance in physical victimization, R2 = .04, F(1, 158) = 6.71, p <

.05, and acceptance was a significant predictor, B = .12, t = 2.59, p < .05. Moreover, the

exposure variables and acceptance of female perpetrated physical violence explained a

significant amount of variance in physical victimization, R2 = .08, F(3, 156) = 4.23, p <

.01, and both witnessing interparental abuse and acceptance were significant predictors, B

= 2.56, t = 2.41, p < .05 and B = .11, t = 2.55, p < .05 respectively. See Table 22 for the

unstandardized regression coefficients (B), standard errors (SE), standardized regression

coefficients (β), and R2.

Refer to Table 23 for the unstandardized regression coefficients (B), standard

errors (SE), standardized regression coefficients (β), and R2 for acceptance of sexual

dating violence as a mediator between childhood exposure (witnessing interparental

abuse and child sexual abuse) and subsequent sexual victimization. The exposure

variables explained significant amount of variance in sexual victimization, R2 = .11, F(2,

159) = 9.45, p < .001, and both witnessing abuse and child sexual abuse were significant

predictors, B = 2.36, t = 2.99, p < .01 and B = 2.24, t = 2.92, p < .01 respectively. The

exposure variables did not account for significant amount of variance in acceptance of

sexual violence, R2 = .01, F(2, 170) = .62, p = .54, and acceptance did not explain a

significant amount of variance in sexual victimization, R2 = .01, F(1, 149) = 1.79, p = .18.

When entering the exposure variables and acceptance as predictors of sexual

victimization, the variables accounted for significant amount of variance, R2 = .11, F(3,

147) = 6.27, p < .001. Again, both of the exposure variables were significant predictors,

B = 2.34, t = 2.86, p < .01 for witnessing and B = 2.17, t = 2.72, p < .01 for child sexual

abuse.

The results were similar for acceptance of female sexual perpetration as the

exposure variables did not predict acceptance, R2 = .01, F(2, 173) = .92, p = .40, and

acceptance did not predict sexual victimization, R2 = .01 F(1, 152) = .87, p = .35. As in

previous results, the exposure variables and acceptance of sexual violence explained a

significant amount of variance in sexual victimization, R2 = .11, F(3, 150) = 6.09, p < .01,

and both the exposure variables were significant predictors, B = 2.35, t = 2.90, p < .01 for

witnessing and B = 2.19, t = 2.76, p < .01 for child sexual abuse. Refer to Tables 24 for

the unstandardized regression coefficients (B), standard errors (SE), standardized

regression coefficients (β), and R2.

For acceptance of male perpetrated sexual violence, the exposure variables did not

predict acceptance, R2 = .00, F(2, 179) = .35, p = .70, but acceptance accounted for

significant amount of variance in sexual victimization, R2 = .03, F(1, 158) = 4.59, p < .05.

Acceptance of male perpetrated violence was a significant predictor, B = .14, t = 2.14, p <

.05. The last regression analysis was also significant indicating that the exposure

variables and acceptance accounted for significant amount of variance in sexual

victimization, R2 = .13, F(3, 156) = 7.65, p < .001, and all three predictors were

significant, B = 2.28, t = 2.89, p < .01 for witnessing, B = 2.19, t = 2.86, p < .01 for

sexual child abuse, and B = .13, t = 1.98, p < .05 for acceptance of male perpetrated

sexual violence. See Table 25 for the unstandardized regression coefficients (B), standard

errors (SE), standardized regression coefficients (β), and R2.

The last mediation analyses were done for psychological dating violence to test if

acceptance of psychological violence would mediate the relationship between exposure to

violence as a child (i.e., witnessing interparental abuse and psychological child abuse)

and subsequent psychological dating violence victimization. See Table 26 for the

unstandardized regression coefficients (B), standard errors (SE), standardized regression

coefficients (β), and R2. The exposure variables explained a significant amount of

variance in psychological victimization, R2 = .09, F(2, 150) = 7.24, p < .01, but only child

psychological abuse was a significant predictor, B = 5.55, t = 3.66, p < .001. The

exposure variables were not significant predictors of acceptance, R2 = .02, F(2, 169) =

1.86, p = .16. However, acceptance accounted for significant amount of variance in

victimization, R2 = .06, F(1, 141) = 9.03, p < .01, and acceptance was a significant

predictor, B = .17, t = 3.00, p < .01. All three predictors (exposure and acceptance)

accounted for significant amount of variance in victimization, R2 = .14, F(3, 139) = 7.78,

p < .001. Child abuse and acceptance were significant predictors, B = 5.54, t = 3.64, p <

.001 and B = .16, t = 3.00, p < .01 respectively, but witnessing was not.

Similar results were found for acceptance of male and female perpetrated

psychological violence when entered instead of the overall acceptance variable.

Exposure to violence did not predict acceptance of male or female perpetrated

psychological violence, R2 = .01, F (2, 177) = 1.15, p = .32 and R2 = .03, F(2, 173) =

2.29, p = .11 respectively. Both of the acceptance variables explained a significant

amount of variance in psychological victimization, R2 = .06, F(1, 147) = 8.49, p < .01 for

male perpetration and R2 = .04, F(1, 144) = 6.30, p < .05 for female perpetration, and they

were both significant predictors, B = .28, t = 2.91, p < .01 for male perpetration and B =

.24 t = 2.51, p < .05 for female perpetration. Lastly, exposure and acceptance explained a

significant amount of variance in victimization for acceptance of male perpetration, R2 =

.15, F(3, 145) = 8.38, p < .001, and female perpetration, R2 = .12, F(3, 142) = 6.45, p <

.001. Both child abuse, B = 5.83, t = 3.91, p < .001 for male perpetration and B = 5.35 t =

3.49, p < .01 for female perpetration and acceptance were significant predictors, B = .29, t

= 3.19, p < .01 for male perpetration and B = .22 t = 2.27, p < .05 for female perpetration.

See Tables 27-28 for the unstandardized regression coefficients (B), standard errors (SE),

standardized regression coefficients (β), and R2.

Based on the results above acceptance did not mediate the relationship between

childhood exposure to violence and later dating violence victimization. Path c (exposure

variables predicting physical victimization) was significant for physical, sexual, and

psychological violence but it also remained significant when entering the exposure

variables and the acceptance variables together to predict victimization (path c′).

The Sobel tests were consistent with this conclusion because none of the Sobel

tests were significant for physical victimization (acceptance total: witnessing

interparental abuse, z = .89, SE = .18, p = .37; child abuse, z = .03, SE = .18, p = .98;

AMDV: witnessing interparental abuse, z = .37, SE = .18, p = .72; child abuse, z = -.28,

SE = .05, p = .78; AFDV: witnessing interparental abuse, z = .58, SE = .14, p = .56; child

abuse, z = .06, SE = 1.69, p = .95), sexual victimization (acceptance total: witnessing

interparental abuse, z = .24, SE = .09, p = .81; child abuse, z = .84, SE = .11, p = .40;

AMDV: witnessing interparental abuse, z = .64, SE = .14, p = .52; child abuse, z = .44, SE

= .13, p = .66; AFDV: witnessing interparental abuse, z = .20, SE = .06, p = .84; child

abuse, z = .77, SE = .10, p = .44), or psychological victimization (acceptance total:

witnessing interparental abuse, z = 1.58, SE = .53, p = .11; child abuse, z = .00, SE = .36,

p = 1.00; AMDV: witnessing interparental abuse, z = 1.30, SE = .47, p = .19; child abuse,

z = -.77, SE = .35, p = .44; AFDV: witnessing interparental abuse, z = 1.43, SE = .45, p =

.15; child abuse, z = .73, SE = .31, p = .47).

The post-hoc analyses showed that childhood exposure to violence predicted

physical victimization when looking at specific types of violent exposure as opposed to

averaging across the different types. Witnessing interparental abuse significantly

predicted physical victimization but child physical abuse did not. The exposure variables

only accounted for marginally significant amount of variance in physical victimization.

Both of the exposure variables (i.e, witnessing interparental abuse and sexual child abuse)

predicted sexual victimization while only psychological child abuse predicted

psychological victimization. Moreover, acceptance of female-to-male perpetrated

physical violence predicted physical victimization, acceptance of male-to-female

perpetrated sexual violence predicted sexual victimization, and all three acceptance

variables for psychological violence (i.e., male-to-female, female-to-male, and the

combined score) predicted psychological victimization. None of the acceptance variables

mediated the relationship between childhood exposure and subsequent dating violence

victimization.

Discussion

Several research studies have found evidence supporting the intergenerational

transmission of violence (Gover et al., 2008; Hendy et al., 2003; Jankowski et al., 1999;

O’Keefe, 1998; Riggs & O’Leary, 1996). Children who are exposed to interpersonal

violence while growing up are at heightened risk for being in violent intimate

relationships as teenagers and adults. Most of these studies have focused on childhood

exposure predicting later perpetration. More importantly, few studies have attempted to

explain the mechanism by which childhood exposure may result in later dating violence

exposure. It is likely that one or more variables could mediate or moderate the

relationship between childhood exposure to violence and later experiences of intimate

partner violence. Some researchers have proposed that attitudes, such as acceptance of

dating violence, can affect the risk of later intimate partner violence (Flynn & Graham,

2010; Lichter & McCloskey, 2004).

The present study was specifically focused on extending the literature regarding

intergenerational transmission of violence to victimization in an attempt to better

understand possible risk factors associated with being a victim of dating violence. The

current study also attempted to determine if the trajectory of such transmission extended

to victims the same way it did perpetrators of intimate partner violence. Based on

findings from previous perpetration studies, it was expected that there would be a

relationship between general childhood exposure to interpersonal violence and

subsequent intimate partner physical violence victimization, such that exposure to any

type of violence as a child (e.g., physical or sexual) would increase the risk for being a

victim of physical dating violence. The assumption was that the dynamics that operate

for perpetration would apply to victimization. The present study also examined the role

of the mediating factor of acceptance of IPV as well as the relation between victimization

and perpetration experiences in an individual.

Discussion of the findings will be organized to first review the major findings,

followed by comparing the present findings to the existent literature, limitations,

recommendations for future research in this area of study, and general conclusions.

Hypotheses 1 and 2

Based on previous research, the first hypothesis predicted that previous exposure

to violence in childhood, either witnessing interparental abuse and/or experiencing child

abuse, and acceptance of dating violence would predict physical dating violence

victimization. Acceptance was predicted to account for additional variance in physical

victimization in dating relationships after accounting for previous exposure to violence.

The hypothesis was not supported by the data. Results were similar when examining

acceptance of male and female physical perpetration separately as well as combining

scores for the two perpetration types to yield a general acceptance of dating violence

perpetration.

The second hypothesis predicted that acceptance of dating violence would

mediate the relationship between childhood exposure and subsequent dating violence

victimization. Again, the data did not support this hypothesis. Using the Baron and

Kenny (1986) approach to analysis, all three paths between the variables (e.g., from

exposure to victimization, from exposure to acceptance, and from acceptance to

victimization) should have been significant. Furthermore, when acceptance is entered in

addition to the exposure variables, the path from exposure to victimization should have

become non-significant or decreased in significance. This was not supported by the

results. While the exposure variables did significantly predict victimization none of the

other paths were significant. Even testing for an indirect effect of acceptance in the

relationship between exposure and victimization was not supported by the results.

To possibly explain why current findings did not support a link between childhood

exposure and later dating violence victimization, as had been found in previous research,

several post-hoc analyses were conducted. The hypotheses for the current study

predicted that there would be a general effect between childhood exposure to violence

and subsequent physical victimization. In other words, it was predicted that exposure to

any type of violence would predict physical victimization. However, Price and Byers

(1999) found that students were more accepting of the type of violence they also reported

perpetrating. Consequently, post-hoc analyses were conducted to test for the effect of

exposure to specific types of violence (e.g., physical, sexual, and psychological) and

being a victim of specific types of violence.

The first post-hoc analyses explored the possibility of predicting physical

victimization by only entering physical exposure variables (witnessing interparental

abuse and physical child abuse) and acceptance of physical dating violence as predictors.

This auxiliary post-hoc hypothesis predicted that women who were exposed to physical

violence as children might be more likely to be accepting of physical violence and

therefore more likely to be victims of physical dating violence.

The exposure variables accounted for marginally significant amount of variance in

physical victimization, and witnessing interparental abuse was a significant predictor for

physical victimization. This suggests that previous exposure, more specifically

witnessing interparental abuse, predicts physical victimization. However, the effect sizes

were small. Additionally, acceptance of female perpetrated physical violence was a

significant predictor and accounted for additional variance above and beyond the

exposure variables. Again, the effect size was small.

Next, acceptance was investigated as a mediator between childhood exposure to

physical violence and later physical dating violence victimization (hypothesis 2). Neither

the regression analyses nor the Sobel test supported the prediction of acceptance as a

mediator for exposure to physical violence as a child and subsequent physical

victimization.

Post hoc analyses were also conducted for both sexual and psychological violence

to further explore the relationships between exposure, acceptance, and victimization with

greater specificity of exposure and abuse type. Witnessing interparental abuse and child

sexual abuse were significant predictors of sexual victimization, and child psychological

abuse was a significant predictor of psychological victimization. For sexual

victimization, only acceptance of male perpetrated sexual violence was a significant

predictor. For psychological victimization all the acceptance variables were significant

predictors (acceptance of psychological violence, acceptance of male perpetrated

psychological violence, and acceptance of female perpetrated violence). The effect sizes

remained small on average.

Lastly, acceptance of dating violence was investigated as a mediator between

childhood exposure and dating violence victimization (hypothesis 2) with separate

analyses for sexual and psychological violence. As was found for physical violence,

there was no support for acceptance as a mediator between exposure to violence as a

child and subsequent dating violence victimization.

Overall, the findings did not support a global impact of childhood exposure to any

type of familial violence (e.g., witnessing interparental abuse, physical and sexual child

abuse) on later dating violence victimization. Instead the data supported a specific effect,

such that being exposed to physical violence as a child increased the risk for being a

victim of physical dating violence.

Hypothesis 3

The third hypothesis predicted a significant positive relation between physical

victimization and physical perpetration of dating physical violence in females; this was

supported by the data. This is consistent with previous studies showing that victims are

also likely to be perpetrators (Bookwala et al., 1992; Harned, 2002; Katz et al., 2002;

Luthra & Gidycz, 2006; O’Keefe & Treister, 1998; Riggs & O’Leary, 1996). There could

be several different explanations for the bidirectionality of dating violence. Females tend

to perpetrate milder types of violence (Teten et al., 2009) and in general dating violence

most frequently involves milder violence. Females might also perpetrate violence as self-

defense, which several research studies have found (Makepeace, 1986; Molidor &

Tolman, 1998; O’Keefe & Treister, 1998). It is also more socially acceptable for women

to report perpetrating violence (Forbes et al., 2005; Merten, 2008; O’Keefe & Treister,

1998) and therefore they might be more likely to report acts of violence than men.

Studies have shown that college females are more likely to report being both victims and

perpetrators of dating violence compared to males (Follingstad et al, 1991; Gover et al.,

2008; Lichter & McCloskey, 2004; Whitaker et al., 2007).

In previous studies the highest rates of dating violence have been reported for

psychological violence followed by sexual and lowest rates for physical violence. These

patterns have been found for both perpetration (Jackson et al., 2000) and victimization

rates (Harned, 2002; Katz et al., 2002). The same pattern was found in the current study

as 23.3% reported at least one incident of physical victimization, 27.0% reported sexual

victimization, and 59.8% reported psychological victimization. Most of the women

reported being victims of minor acts of violence (56.8% of victims of physical violence,

90.2% for sexual violence, and 60.2% for psychological violence).

While victimization and perpetration were correlated, the perpetration rates were

somewhat higher than the victimization rates for physical (31.2% vs. 23.3%) and

psychological violence (62.4% vs. 59.8%), with the sexual perpetration rate somewhat

lower (20.6% vs. 27.0%). Overall, 67.2% of the sample reported being a victim of dating

violence at least once (e.g., physical, sexual, and/or psychological) and 69.3% of the

sample reported perpetrating violence at least once. These high rates of dating violence

are similar to those found in previous studies with college students.

Straus (2004) reported on physical perpetration rates among 8,666 college

students in 31 different countries around the world. Overall 29.0% of the college students

reported physical perpetration. In Gover and colleagues’ (2008) college sample, 29%

reported physical perpetration and 54% reported psychological perpetration.

Victimization studies have also reported high rates of dating violence. Gover et al.

(2008) reported that 22% of the college students were victims of physical violence and

52% were victims of psychological violence. Harned (2002) found that 39% of females

reported sexual victimization.

It is important to recognize that the current study found similar victimization and

perpetration rates and patterns compared to previous dating violence studies. This

suggests that this aspect of sample characteristics cannot explain the differences in

findings between the present study and those of previous dating violence studies. This

will be especially relevant to keep in mind for the discussion below.

Comparing Current Study to Previous Studies

There could be numerous reasons why this study found limited evidence for the

intergenerational transmission of violence or role of attitudes (e.g., acceptance) in

predicting dating violence in young adults. To explore this issue further, the present study

will be compared to similar studies on methodological issues such as sample

characteristics, measurements, research design, and statistical methods. Studies

investigating the intergenerational transmission of violence will be discussed first

followed by studies of acceptance.

Intergenerational Transmission of Violence. While researchers have found

support for the intergenerational transmission of violence, the findings from the current

study provided limited support. The discussion below will first compare the current study

to studies that have focused on physical dating violence victimization and then compare it

to studies focusing on physical dating violence perpetration.

Victimization. Several studies have found a link between childhood exposure and

dating violence victimization (Gover et al., 2008; Hendy et al., 2003; Jankowski et al.,

1999; Malik et al., 1997; O’Keefe, 1998). These studies have recruited samples that have

similarities and differences compared to the sample used in the current study. While

previous studies have focused on college students (Gover et al., 2008; Hendy et al., 2003;

Jankowski et al., 1999) or high school samples (Malik et al., 1997; O’Keefe, 1998), the

present study sampled college students. Even though the discrepancy in findings from

the high school samples could be attributed to age or developmental differences, this

explanation would not be valid for the differences between the college samples.

Moreover, most of the studies recruited larger samples than the present study (n =

2,541 for Gover et al., 2008; n = 608 for Hendy et al., 2003; n = 1,342 for Jankowski et

al., 1999; n = 719 for Malik et al., 1997) but one of them had a similar sample size (n =

232 for O’Keefe, 1998). Since the present study found small effect sizes and low power,

it is possible that sample size limitations did not allow for detecting an effect of the

exposure variables predicting victimization. However, the one study with a similar

sample size (O’Keefe, 1998) did find an effect, suggesting that the sample size used in

the present study should be adequate to produce an effect if there is one. Nevertheless,

most of the studies used larger sample sizes, which suggest that a future study would

benefit from recruiting a larger sample.

Additionally, all of the previous studies included both males and females while the

present study recruited women only. More importantly, two studies combined both males

and females in their analyses (Jankowski et al., 1999; Malik et al., 1997) but the other

studies ran separate analyses for males and females (Gover et al., 2008; Hendy et al.,

2003; O’Keefe, 1998). Even when separating analyses by gender, two of the studies had

larger samples of females than the current study (n = 1,530 for Gover et al., 2008; n = 444

for Hendy et al., 2003). O’Keefe (1998) recruited 138 females in her study, suggesting

that the sample size for the current study would be adequate. However, most studies still

included larger samples than the present study and could therefore have benefited from

recruiting more participants.

Moreover, most of the studies that have found a link between childhood exposure

and subsequent dating violence have recruited mostly Caucasian participants (Gover et

al., 2008; Hendy et al., 2003; Jankowski et al., 1999), which was also the case in this

study. Two studies focused on minorities in which one study recruited equal proportions

of African-Americans, Asian-Americans, and Latino/a (Malik et al., 1997) and the other

recruited mostly Latino/a (O’Keefe, 1998). Since these studies have found support for

the intergenerational transmission of violence with both similar and dissimilar sample

characteristics (e.g., ethnicity/race) compared to the present study, ethnicity does not

seem to impact whether there is an effect or not.

The research design could also influence whether there is a relationship between

childhood exposure to violence and later dating violence victimization. The present study

as well as the studies mentioned above all used a cross-sectional design. However, this

study used an online questionnaire administration while most of the other studies

administered the questionnaires in groups of participants (Gover et al., 2008; Malik et al.,

1997; O’Keefe, 1998). One collected most of the data in classrooms but also allowed

participants to take the surveys home to complete and later drop off in a box (Hendy et

al., 2003). One study did not specify how they collected the data (Jankowski et al.,

1999). The studies that administered the questionnaires in person have more control in

their studies compared to what was done in the current study. They are able to control for

several confounding factors such as distractions (e.g., influence of other people and

things in the environment). They are also able to answer questions if necessary. An

online questionnaire does not control for any of these potential confounding factors;

therefore it is possible that the results were impacted by any of those third variables.

The operationalization of variables is another factor to consider. All the studies

used the CTS (e.g., either the 1979 or the 1996 version) to measure dating violence. Some

researchers made some scoring adjustments, such as changing the 7-point likert scale to a

3-point scale (1 = never, 2 = once or twice, 3 = many times; Hendy et al., 2003) and

combining both physical and sexual dating violence items into one variable (8 items for

physical victimization and 1 for sexual; O’Keefe, 1998). Hendy and colleagues did not

specify what type of violence they measured but most likely it was physical violence.

Because the present study also used the CTS to measure physical dating violence

victimization, this did not likely impact the differences in results between studies.

All of the previous studies that found support for the intergenerational

transmission of violence for victimization used the CTS to measure the two exposure

variables of witnessing interparental abuse and childhood abuse as well. In the current

study the CTS was not used to measure the exposure variables. Hendy and colleagues

(2003) continued to use their changed likert scale. One study used the CTS for

measuring witnessing interparental abuse but not for child abuse (Jankowski et al., 1999).

They also only asked one question about physical child abuse, which focused on specific

acts of violence. For the current study four items were created to measure witnessing

interparental abuse, which are based on the distinction Straus and colleagues (1996) make

between minor and severe violence. Even though the present study did not use the CTS

to measure witnessing interparental abuse, it used similar questions as the other studies

that support the intergenerational transmission of violence. Therefore, there appears to be

more similarities than differences between how witnessing interparental abuse was

operationalized across the studies, suggesting that the measurement of this variable did

not likely affect the non-significant findings.

In the current study two questions from Briere’s CMIS-SF measure (1992) were

used to measure physical child abuse. One of them asked about specific acts of child

abuse and the other one was a global question (“To the best of your knowledge, before

age 17, were you ever physically abused?”). Again, the specific question is similar to the

CTS physical abuse content while the general is not. However, the CTS includes several

question about different minor and severe violent acts while the current study only

included two questions all together. Overall, as for operationalization of witnessing

interparental violence, operationalization of child abuse appear to be more similar than

different across the studies, suggesting that this did not impact the differences in findings

between the current study and previous studies.

More importantly, O’Keefe (1998) was the only study that had a sample size

similar to that of the current study, but she focused on a subsample (i.e., 232 participants

out of 1,012) that reported more severe violence exposure. She only included students

who reported witnessing high levels of interparental abuse (i.e., at the 75th percentile of

higher) while the present study included females who have not witnessed interparental

abuse as well as those who have. O’Keefe also asked participants to report on their worst

year as children for the exposure variables. This suggests that a future study would

benefit from either recruiting a larger sample or focus on a sample with more severe

violence exposure.

Overall, in regards to operationalization of childhood exposure, there appears to

be more similarities than differences in the measurement of the childhood exposure

variables comparing the current studies and previous studies. Nevertheless, this study

could have benefited from including a broader range of questions related to childhood

exposure.

The statistical method used for the analyses as well as how the different variables

were created can also impact the findings. Except for two studies (Hendy et al., 2003;

Malik et al., 1997), the comparison studies dichotomized all variables. In Malik and

colleagues’ (1997) study the exposure variables were continuous while the dating

violence variables were dichotomous (Malik et al., 1997). Hendy and colleagues (2003)

used continuous variables. In the current study, the exposure variables were dichotomous

and the physical victimization variable was continuous. All the comparison studies used

regression analyses (either logistic or regular regression analysis). Overall, the present

study used similar statistical methods as the other studies.

Nevertheless, there were some noteworthy differences between the current study

and some of the other studies. Jankowski et al. (1999) created separate variables for

witnessing interparental abuse based on which parent was the perpetrator (i.e., same sex

parent, opposite sex parent, both parents, or none). More importantly, they controlled for

physical child abuse in their analyses while this study used physical child abuse as a

predictor for victimization.

Taken together, a future study should include a larger sample size, expand on the

childhood exposure measures, and administer the surveys in a more controlled setting.

The current study could also have benefited from inclusion of participants with exposure

to more severe violence.

Even though the findings of the current study did not find childhood exposure to

be predictive of later physical victimization when averaging across violence types, there

are some similarities to results from previous studies that have found support for the

intergenerational transmission of violence. Previous studies have found an effect both for

witnessing interparental abuse and being a victim of physical child abuse and subsequent

physical dating violence victimization. Three studies found an effect for witnessing

abuse and subsequent physical victimization (Hendy et al., 2003; Gover et al., 2008;

Jankowski et al., 1999). However, Jankowski and colleagues (1999) found that

witnessing both parents abuse each other predicted physical victimization for both female

and male college students while the other two studies only found an effect for female

college students. Gover and colleagues (2008) found that witnessing father-to-mother

inflicted violence predicted physical victimization and Hendy and colleagues (2003)

found that witnessing mother-to-father inflicted violence predicted physical victimization.

In the current study witnessing interparental abuse predicted physical dating violence

victimization in the post-hoc analyses (e.g., exposure to physical violence as a child and

acceptance of physical violence were entered as predictors for physical victimization).

Since Jankowski and colleagues (1999) combined males and females in their

analyses, it is difficult to compare their findings to the current study. Instead, one could

argue that the differences in findings suggest that victimization is not the same for

females and males. The other two studies found that witnessing interparental abuse

predicted victimization (Hendy et al., 2003; Gover et al., 2008), as in the current study.

These two studies found effects based on which parent inflicted the violence. The present

study did not separate witnessing by perpetrator gender.

Two of the above-mentioned studies also found a link between childhood abuse

and subsequent physical victimization (Hendy et al., 2003; Gover et al., 2008). Gover

and colleagues (2008) found that physical child abuse was a significant predictor for

physical and psychological victimization for females while Hendy and colleagues (2003)

found that mother inflicted child abuse predicted victimization for both females and

males. Other studies have found similar results. In Malik and colleagues’ (1997) high

school sample both males and females who reported being abused as children were more

likely to report victimization than those without an abuse history. O’Keefe (1998) only

found a link between physical child abuse and later victimization for females. The

current study did not find evidence for physical child abuse predicting physical dating

violence victimization. There could be several explanations for this such as the limited

number of questions about childhood abuse in the present study compared to previous

studies. It is possible that only those women who experienced more severe child abuse

answered yes to the questions used in the present study, which means that women

exposed to milder forms of child abuse were categorized with non-victims.

Since the current study did not find much evidence supporting the

intergenerational transmission of violence, it would also be important to discuss other

studies that have failed to find a link between childhood exposure and later victimization.

In Lichter and McCloskey’s (2004) longitudinal sample of adolescents, those who

reported witnessing their father abuse their mother were not more likely to experience

dating violence. In O’Keefe and Treister’s (1998) study of the background-situational

model with high school students, they did not find support for the intergenerational

transmission of violence. O’Keefe and Treister included both witnessing interparental

abuse and physical child abuse as predictors for physical victimization.

Lichter and McCloskey (2004) used a similar sample size (n = 208) as in the

current study but their sample was different in many ways. They recruited mothers and

their adolescent children to be part of a longitudinal study that focused on long-terms

effects of intimate partner violence on children. They recruited both violent and

nonviolent mothers and their adolescents from the community. Since the researchers

specifically targeted adolescents who grew up in violent homes, it is likely that their

sample was exposed to more severe violence compared to the current sample. Moreover,

they used the CTS to measure both childhood exposure and dating violence but they used

weights to reflect differences in impact between male and female perpetrated dating

violence. These weights were based on studies by Marshall (1992a, 1992b) with college

students who reported “how serious, aggressive, abusive, threatening, and violent”

different acts were based on whether the perpetrator was male or female (Lichter &

McCloskey, 2004, p. 348). Moreover, in addition to the CTS items for measuring

exposure to father-to-mother inflicted violence, Lichter and McCloskey used 7 more

violent items derived from talking to female victims and staff at domestic violence

shelters. In addition to differences based on sample characteristics and measurements

between Lichter and McCloskey’s (2004) study and the current study, they used a

longitudinal research design with in-person interviews. Overall, their study and the

current study appear more dissimilar than similar.

O’Keefe and Treister’s (1998) study was the second study that did not find

support for the intergenerational transmission of violence. They included a much larger

sample size than the current study (n = 939) and they recruited a high percentage of

minority high school students (mostly Latino/a and African-American). O’Keefe and

Treister reported findings from the same sample as O’Keefe (1998), but included the

entire sample while O’Keefe focused on those who reported witnessing more severe

violence between their parents. Interestingly, O’Keefe (1998) found support for physical

child abuse and later victimization for females, but when including the entire sample

O’Keefe and Treister did not find any effects for child abuse or witnessing. This suggests

that the sample size of the current study was adequate but could have benefited from

recruiting individuals reporting more severe violence exposure. An alternative recruiting

method would be to sample a broader range of violent exposure instead of including so

many participants with no violence exposure. Most of the women in the present study

were not exposed to violence as children or as young adults. Also, those who reported

exposure to violence as children or adults reported mostly exposure to minor acts of

violence. A future study would benefit from including a broader range of violence

exposure.

Perpetration. Even though the current study focused on physical dating violence

victimization, it is beneficial to compare it to previous studies focusing on physical dating

violence perpetration to further explore the limited support for the intergenerational

transmission of violence in the current study. Three of the previously discussed studies

that found evidence for childhood exposure and later victimization also investigated the

effect of childhood exposure and later perpetration (Hendy et al., 2003; Malik et al.,

1997; O’Keefe, 1998). Two studies found evidence for witnessing interparental abuse

and perpetration while two found support for child abuse and perpetration. Both Malik

and colleagues (1997) and Hendy and colleagues (2003) found that females were more

likely to perpetrate violence if they had witnessed their mother abuse their father. In one

study, this was true for males as well (Malik et al., 1997). In terms of childhood abuse,

O’Keefe (1998) reported that females were more likely to report being perpetrators if

they had been abused as children while Hendy and colleagues (2003) reported that males

were more likely to report being perpetrators if they had been abused by their mothers as

children.

Three additional studies that focused solely on perpetration also found support for

the intergenerational transmission of violence (Foo & Margolin, 1995; Luthra & Gidycz,

2006; Riggs & O’Leary, 1996). Riggs and O’Leary (1996) found that witnessing

interparental abuse predicted aggression against peers but not against dating partners.

Likewise, being a victim of childhood abuse predicted aggression against peers but not

against dating partners (Riggs & O’Leary, 1996). Foo and Margolin (1995) found

evidence for both witnessing interparental abuse and child abuse. More specifically,

witnessing parents abuse each other predicted male perpetration while child abuse

predicted female perpetration. However, Foo and Margolin found an effect for sexual

child abuse but not physical. Lastly, Luthra and Gidycz (2006) found that females were

more likely to perpetrate violence if they had been abused by their father as children.

The three additional studies described above that found some evidence for

childhood exposure and later perpetration had fairly similar sample sizes compared to the

current study. Riggs and O’Leary (1996) included almost twice as many participants (n =

375) but Foo and Margolin (1995) included 290 students and Luthra and Gidycz (2006)

included 200 students. This suggests that the sample size in the current study would be

adequate for investigating the intergenerational transmission of violence. However, the

current study focused on victimization while the other three previous studies focused on

perpetration. When comparing the current study to previous studies of victimization, one

suggestion was to recruit a larger sample or focus on a subsample with more severe

violence exposure. Since the three studies with similar sample sizes as the current study,

which focused on perpetration instead of victimization found some support for both

witnessing and child abuse and later dating violence, it is possible that childhood

exposure to violence impacts physical dating violence perpetration and victimization

differently.

In addition to the similar sample size, the three previously discussed studies of

perpetration have more similarities than differences compared to the current study. They

were all cross-sectional studies with college students. Two of them recruited mostly

Caucasian students (Luthra & Gidycz, 2006; Riggs & O’Leary, 1996) while one included

about 55% Caucasian students and the rest were minorities (mostly Asian-Americans;

Foo & Margolin, 1995). The current study recruited only female college students while

these three studies recruited both male and female college students. Moreover, the

previous three studies used the CTS to operationalize witnessing interparental abuse,

child abuse, and dating violence perpetration, except Riggs and O’Leary (1996) who used

the Family Violence Questionnaire (FVQ) to measure child abuse and witnessing

interparental abuse. The FVQ asks more general questions as opposed to the CTS that

asks about specific acts of violence. As discussed for the previous victimization studies,

the present study could have benefited from including more questions about childhood

exposure.

In terms of statistical methods, the current study used regression analyses, which

two of the perpetration studies did as well. Again, the exception was Riggs and O’Leary

(1996) who used a structural equation modeling technique. The current study differed

from these three studies in regards to the research design. Similarly to the previous

studies focusing on victimization, the three studies focusing on perpetration used a more

controlled setting for data collection as opposed to the current study. All three studies

administered their surveys in group settings while the current study administered them

online. As previously discussed, the current study would have benefited from

administering the questionnaires in a more control study to rule out potential third

variables.

Acceptance. The current study hypothesized that childhood exposure would

predict dating violence victimization and suggested that acceptance of dating violence

would predict victimization after controlling for childhood exposure. As for the

intergenerational transmission of violence, the current study also found limited evidence

for acceptance of dating violence as a predictor for dating violence victimization. There

was no support for acceptance as a mediator between childhood exposure and later

victimization. The current study will be compared to previous studies that have

investigated childhood exposure, acceptance, and dating violence. First, the discussion

will focus on victimization and then perpetration to try to explore why the current study

found limited evidence for acceptance as a predictor of dating violence victimization.

Victimization. Few studies have looked at the relationship between acceptance of

dating violence and dating violence victimization. One study focused on predicting

perpetration but also reported that acceptance of perpetration of violence increased the

odds of being a victim of physical dating violence (Malik et al., 1997). However,

acceptance increased the odds of being a perpetrator more than being a victim. The only

previous studies that have focused on predicting victimization by including acceptance as

a predictor were done by O’Keefe and colleagues (O’Keefe, 1998; O’Keefe & Treister,

1998).

When only including participants who reported more severe exposure to

witnessing their parents abuse each other, O’Keefe (1998) found that males who were

more accepting of dating violence were also more likely to report being victims of dating

violence. This was not the case for females. When including the larger sample of high

school students, O’Keefe and Treister (1998) found that females who reported being more

accepting of male-to-female violence were more likely to report being victims of dating

violence. The findings in O’Keefe and colleagues’ studies differed by gender, which

suggests that victimization is not the same for females and males. In the current study,

acceptance of female-to-male perpetrated physical violence predicted physical

victimization for female college students. Logically it makes sense that women who are

being abused by their male partners would be more accepting of male perpetrated

violence, as the findings of O’Keefe and Treister’s (1998) study support. The current

study found the opposite results. There could be several explanations for this. Some

could be related to methodological difference, which will be discussed below. The

women in the current study were more accepting of female perpetrated violence in

general, which could be because it is more socially acceptable for females than males to

be violent against their partners. Another contributing factor might be that most of the

victims were also perpetrators and it would make sense for female perpetrators to be more

accepting of female perpetration. A related explanation is that acceptance plays a larger

role for perpetrators than victims, which again would argue that victimization and

perpetration are separate experiences.

O’Keefe and colleagues (O’Keefe, 1998; O’Keefe & Treister, 1998) used a

different measure for acceptance of dating violence than the current study. They used

Margolin and Foo’s (1992) A Justification of Violence Scale, which is an unpublished

measure. Both measures assess for acceptance of both male and female perpetrated

physical violence. However, in contrast to Price and Byers’ (1999) measure, used in the

current study, Margolin and Foo’s measure asks about acceptance of perpetrating violence

in situations that are either humiliating or when violence is used in self-defense.

Furthermore, Price and Byers’ (1999) measure includes acceptance of physical, sexual,

and psychological violence and the measure asks more about specific acts of violence and

less so about specific context. The contexts included in Price and Byers’ (1999) measure

involve perpetrating violence because the partner cheated or is intoxicated or because of

jealousy or love. Price and Byers’ (1999) measure also have more items (39 for male

perpetrated violence and 37 for female perpetrated violence) compared to Margolin and

Foo’s measure (8 for each perpetrator type). The number of items is more similar when

only focusing on acceptance of physical violence as Price and Byers’ (1999) measure

included 12 items for each perpetrator type for the acceptance of physical violence scale.

Foo and Margolin (1995) refer to their measure as The Attitudes About Dating Index and

report that the original version has 24 items for each perpetrator (i.e., males and females).

Overall, the many differences between the acceptance measures make it difficult to

compared findings across the studies.

Another difference between the current study and O’Keefe and Treister’s (1998)

study was related to the statistical methods. O’Keefe and Treister used a multiple

regression analysis in which they entered all predictors simultaneously while the current

study used a hierarchical regression analysis in which the exposure variables were entered

in the first step and acceptance of dating violence in the second step. In other words, the

current study controlled for childhood exposure to determine if acceptance added any

additional predictability of victimization. Moreover, O’Keefe and Treister included

fifteen predictors in their study but the current study focused on 3 predictors. Even

though O’Keefe and Treister included the same predictors as the current study

(witnessing interparental abuse, child abuse, and acceptance) they did not control for the

effects of childhood exposure and included many more predictors.

Overall, there are many differences between the current study and the two

previous studies that have looked at childhood exposure to violence, acceptance, and

victimization. Therefore, it is difficult to determine what might have impacted the

differences in outcomes.

Perpetration. The current study found limited support for acceptance of dating

violence as a predictor for dating violence victimization and as discussed above, few

studies have looked at acceptance and victimization. Therefore, studies investigating

acceptance and perpetration will also be discussed to further compare and contrast the

current study with previous studies. Even though more studies have investigated the

relationship between acceptance and perpetration than acceptance and victimization, the

research is still limited. Two studies reported correlations and odds ratios. In Price and

Byers’s (1999) correlational study, they found that high school students were more

accepting of the type of violence they had perpetrated. Students who reported

perpetrating physical violence were also more accepting of physical perpetration. Malik

and colleagues (1997) found that acceptance of violence increased the odds of reporting

being a perpetrator of physical violence. Moreover, as for victimization, O’Keefe (1998)

found that acceptance of dating violence predicted dating violence perpetration for males

but not for females.

Five additional studies have investigated the relationship between acceptance of

dating violence and perpetration. Three of them were previously compared to this study

in terms of the intergenerational transmission of violence (Foo & Margolin, 1995; Lichter

& McCloskey, 2004; Riggs & O’Leary, 1996). In two of the previous studies acceptance

of dating violence predicted physical perpetration for both male and female college

students (Foo & Margolin, 1995; Riggs & O’Leary, 1996). In a longitudinal study by

Foshee et al. (2001), they found that high school students were more likely to perpetrate

violence if they reported being more accepting of dating violence. At baseline, this was

true for predicting female perpetration but at the 1-year follow-up acceptance predicted

male perpetration. Moreover, Lichter and McCloskey (2004) found that acceptance of

male-to-female perpetrated violence predicted dating violence perpetration while

Josephson and Proulx (2008) found an effect for acceptance of female-to-male violence

and perpetration. Both studies reported the same results for males and females.

The current study measured acceptance of dating violence using Price and Byers’

(1999) measure. Two other studies that examined dating violence perpetration used the

same measure. Josephson and Proulx (2008) created a shorter version of the original

scale, which had 6 items for each type of violence (i.e., 18 items for male perpetration

and 18 for female perpetration). The researchers did not specify which items they

included in their study. Interestingly both the current study and Josephson and Proulx’s

(2008) study found that acceptance of female-to-male perpetrated physical violence

predicted physical dating violence even though this study focused on victimization while

Josephson and Proulx focused on perpetration. Josephson and Proulx also found an

indirect effect of acceptance of male-to-female violence with greater acceptance

predicting psychological perpetration, which predicted physical perpetration for both

males and females. Josephson and Proulx used a similar sample size to the current study

(n = 290) but they recruited high school males and females while the current study

focused on female college students. Their study was also cross-sectional but they used a

SEM while the current study used a hierarchical regression analysis. Also, Josephson and

Proulx did not include any childhood exposure variables in their model instead they

included variables about knowledge about abuse, self-efficacy, and the use of non-violent

conflict strategies. Taken together, the numerous differences between the present study

and Josephson and Proulx’s study make it difficult to compare the two.

The other perpetration study using Price and Byers’ (1999) measure was Luthra

and Gidycz’s (2006) cross-sectional study with male and female college students. They

did not find support for acceptance as a predictor of perpetration. In contrast to the

current study, Luthra and Gidycz included dating violence victimization as a predictor for

perpetration and they reported that victimization was a strong predictor for perpetration

for both males and females. The effect was the strongest for females. Since the current

study found a strong correlation between victimization and perpetration, it is likely that

perpetration would be a strong predictor for victimization and that acceptance of

femaleto-male violence would become a non-significant predictor if perpetration were

added as a predictor.

The four perpetration studies that also included acceptance of dating violence as a

predictor operationalized acceptance differently compared to the current study. Two of

them used Margolin and Foo’s The Attitudes About Dating Index (Foo & Margolin, 1995;

Lichter & McCloskey, 2004). Both studies only used 12 out of the 24 items but they used

different items. Foo and Margolin focused on situation relating to self-defense and

humiliation. Lichter and McCloskey did not clarify which items they used. Moreover, the

remaining two studies used different measures. Riggs and O’Leary (1996) used their own

measure (AVQ), which consists of three items for each perpetrator (e.g., acceptance of

pushing, slapping, punching). Lastly, Foshee and colleagues (2001) 8 items that they

named Prescribed Dating Violence Norms to measure acceptance. Six out of the 8 items

referred to male perpetration and only 2 referred to female perpetration. Nevertheless,

these items were actually similar to the measure used in the current study even though

Foshee and colleagues only included acceptance of physical violence.

Overall, with the exception of Foshee and colleagues’ (2001) study, operationalization of

acceptance differed between the current study and the other perpetration studies.

Since Foshee and colleagues’ (2001) study measured acceptance similarly as the

current study, the two studies will be compared based on research methodology. Foshee

and colleagues recruited a larger sample (n = 1,013) and included both male and female

high school students. They used a longitudinal study design and logistic regression

analyses. Acceptance predicted female perpetration at baseline and male perpetration at

the 1-year follow up. The researchers did not separate acceptance of male and female

perpetrated violence. Overall, Foshee and colleagues’ study appear to be more dissimilar

than similar to the current study.

As for the victimization studies that included acceptance as a predictor, the

perpetration studies were more dissimilar than similar to the current study, which makes it

difficult to compare the findings. The findings from the studies that investigated the

intergenerational transmission of violence suggest that future studies should include more

items for childhood exposure and administer the surveys in a more controlled setting. A

future study should also either include a larger sample or focus on more severe violence

exposure. Nevertheless, some of the perpetration studies that used similar sample sizes as

the current study found support for the intergenerational transmission of violence (Foo &

Margolin, 1995; Luthra & Gidycz, 2006), which suggest that exposure to violence in

childhood might influence dating violence victimization and perpetration differently.

Victimization might be very different from perpetration in general.

The intergenerational transmission of violence theory suggests that being exposed

to violence as a child increases the risk for experiencing dating violence. In other words,

instrumental learning can explain how children exposed to interpersonal violence learn to

be violent themselves. In the current study it was proposed that attitudes such as

acceptance of dating violence might explain that relationship as children who been

exposed to violence have normalized the occurrence of violence in intimate relationships.

The acceptance variable was conceptualized as a cognitive variable that might

differentiate children who were exposed to violence and later experiences dating violence

compared to those who did not. It is possible that cognitive factors such as acceptance

play a greater role in predicting perpetration than victimization.

Other researchers have discussed the normalization of violence in people who

grow up in violent families. Straus and colleagues (Straus, Gelles, & Steinmetz, 1980;

Straus, & Yodanis, 1996) have discussed the normalization of violence in the context of

corporal punishment and intimate partner abuse. The researchers argue that when parents

use physical punishment to correct their children’s behaviors then the children learn that

violence is a good and acceptable strategy to deal with conflict. Straus and Yodanis state

that “normative approval of violence is based on the assumption that corporal punishment

teaches children that when someone misbehaves and won’t listen to reason, it is

appropriate to hit them” (1996, p. 827). According to Straus and Yodanis (1996), children

who are physically punished as children are more likely to be violent with other children

as well as their partners when they grow up because they have learned to use violence to

solve conflict. Furthermore, physically punished children have limited exposure to

nonviolent strategies for dealing with conflict and are therefore limited in these skills.

Because they are limited in their conflict management skills, they are more likely to use

violence in their relationships, which reinforces what they learned as children (e.g.,

violence is a good way to end a conflict). Straus and Savage (2005) made similar

arguments for people who have been neglected as children. They found that college

students that were neglected as children were more likely to perpetrate dating violence.

Straus and colleagues’ (Straus et al., 1980; Straus, & Yodanis, 1996) hypothesis that

interpersonal violence is normalized for children growing up in violent homes seems to

explain perpetration fairly well. However, it is unclear if the dynamics of perpetration are

the same as for victimization. Victims do not necessarily learn that violence works or that

violence is a good conflict strategy. It is possible that some victims may believe that

violence is a normal part of an intimate relationship. They might also have limited

conflict management skills. However, in the current study childhood exposure was not a

strong predictor of dating violence victimization. One explanation could be that some of

the women who were exposed to violence as children developed a negative reaction to

interpersonal violence (e.g., less acceptable) and are therefore less likely to being with a

dating partner who is violent. If they are in a violent relationship, they might end it

because they have not normalized the violence they were exposed to as children. They

are actually not accepting violence as a normal part of an intimate relationship. Another

issue is that a substantially greater number of women reported being victims and

perpetrators of dating violence than those exposed to violence as children. In other

words, a proportion of the women were not exposed to childhood violence but they were

victimized. This finding is consistent with other studies as researchers have reported high

proportions of dating violence in young samples (e.g., high school and college samples)

while the proportion that were exposed as children is smaller. This might explain why

childhood exposure is not a strong predictor of victimization.

Limitations

There are several limitations with the current study. The effect sizes were small

and the power was low, which increases the risk for Type II errors. In other words, the

current study might have missed an effect that was present because of the increased

chance of false negatives. A future study should recruit a larger sample or focus on a

sample with more severe violence exposure to increase power and hopefully effect sizes

as well.

The exclusion of males could be considered both a limitation and an advantage of

the present study. Only female college students were recruited because few studies have

looked at victimization and previous studies have found that females report more

victimization compared to males (Follingstad et al, 1991; Gover et al., 2008; Lichter &

McCloskey, 2004; Whitaker et al., 2007). Females have also reported more negative

outcomes related to dating violence compared to males (Follingstad et al., 1991; Kaura &

Lohman, 2007; Molidor & Tolman, 1998; O’Keefe & Treister, 1998). Focusing on

females could be an advantage because so few studies have explored risk factors for

victimization. However, the exclusion of males could also be considered a limitation

because the findings cannot be generalized to males.

Moreover, the current sample was mostly Caucasian and all participants were

college students. Even though previous studies have found high proportions of dating

violence in college students (Straus, 2004), the results from the current study cannot

necessarily be generalized to other ethnicities and non-college students. College students

are in general considered a highly functional and healthy sample.

Another limitation with the current study was the operationalization of the

exposure variables. A future study should expand on the items measuring witnessing

interparental abuse and child abuse. Moreover, witnessing interparental abuse was only

related to witnessing physical violence with the assumptions of a general effect of being

exposed to any kind of violence as a child. However, the current study found a more

specific effect, which suggests that future studies should ask about several types of

violence (e.g., physical, sexual, and psychological). Moreover, the present study only

focused on dating violence victimization in the past year. It is possible that women coded

as non-victims were exposed to dating violence prior to that time frame.

The current study used self-reports and retrospective recall of violence

experiences, which further limits the findings. Also, the use of online questionnaires

administration limited internal validity and therefore future studies should administer the

questionnaires in a more controlled setting. This might also limit the proportion of

missing data, which was another limitation with the current study. Several of the

variables had more than 5% of missing data (e.g., the acceptance and dating violence

variables) and even though the results from the regression analyses were similar with and

without missing data, a future study would benefit from limiting the amount of missing

data.

Another limitation was the violations of assumptions. Several of the variables

were not normally distributed as there were more participants who reported no violence

or few violent experiences, which impacted the assumptions of linearity of residuals and

homoscedasticity. These assumptions were violated even after transforming the

nonnormally distributed variables and therefore the non-transformed variables were used

in the study. However, a future study might benefit from recruiting a broader range of

violent experiences.

Future Studies

Future studies examining the variables of the present study would benefit from

administering the surveys in a more controlled setting. They should also recruit a broader

range of violence exposure experiences or focus on more severe violence exposure.

Since few studies have investigated childhood exposure as a predictor for dating violence

victimizations, future studies should continue focusing on females or males separately to

better understand the risk factors and potential mediators and moderators. Moreover,

several studies have found different results for males and females, which suggests that

future studies should not assume that violence exposure and dating violence are the same

for males and females.

Future studies could benefit from several measurement considerations. More

questions should be asked about childhood exposure to violence. Witnessing

interparental abuse should include physical, sexual, and psychological violence. Also,

additional questions should be asked to investigate if childhood exposure to violence

normalizes violence in interpersonal relationships. Future studies should ask participants

how normal or common they perceive different violent acts to be in an intimate

relationship. Participants could also be asked about the perceived impact of being

exposed to violence as a child as well as in dating relationships. It is possible that

acceptance of dating violence is not a significant predictor because they do not think the

violence is impacting them negatively (e.g., mild violence at low frequency).

Another measurement consideration for future studies relates to how acceptance is

measured. The present study measured childhood exposure and dating violence

victimization by asking women about their personal experiences (e.g., acts of violence

that they were victims to). However, acceptance of dating violence was measured by

asking general questions about how acceptable certain acts of violence are perceived to

be. Future studies might benefit from having victims rate how acceptable they are of

their own victimization (e.g., the perpetration acts they have experienced). Future studies

would also benefit from defining the construct of acceptance. To date, no studies have

clearly defined what acceptance is and what it is not.

Moreover, the current study asked about dating violence experiences but did not

ask if they were still in that relationship. It is possible that some of the women who

experiences abuse ended the relationship because they did not accept the violence. A

future study should ask about contextual factors such as if they are still dating the

perpetrator.

The current study found higher correlations between perpetration and acceptance

than between victimization and perpetration (see Table 4) suggesting that acceptance of

dating violence is a better predictor for perpetration than victimization. A related issue is

the high correlation between victimization and perpetration, which makes it difficult to

investigate risk factors for victimization without discussing the fact that most victims are

also perpetrators. At least for college students, it seems like victimization rarely occurs

independent of perpetration. A future study would benefit from asking about the context

of the violent acts to better understand situational factors that might impact victimization

versus perpetration.

Furthermore, in the current study it was assumed that acceptance of dating

violence occurred after being exposed as a child but before experiencing dating violence.

It is possible that people become more or less accepting based on their dating experiences

and not because of childhood exposure.

Conclusions

The present study aimed to understand how childhood exposure to violence might

lead to becoming a victim of physical dating violence. Previous studies have found

evidence for childhood exposure increasing the risk for later dating violence exposure

(Gover et al., 2008; Hendy et al., 2003; Jankowski et al., 1999; O’Keefe, 1998; Riggs &

O’Leary, 1996) but most of them have focused on predicting perpetration. Children who

have been abused or have witnessed their parents abuse each other learn that conflict can

be solved by using violence (e.g., instrumental learning). The assumption behind the

current study was that the research on perpetration would translate into victimization. In

other words, it was predicted that women who were exposed to violence as children

would be more likely to become victims of physical dating violence.

Additionally, the current study predicted that cognitive factors such as acceptance

of violence would predict victimization above and beyond childhood exposure.

Acceptance of dating violence was also hypothesized to mediate the relation between

childhood exposure and subsequent dating violence. Initially, the present study predicted

a general or global impact of violence exposure such that being exposed to any kind of

violence as a child would increase the risk of becoming a victim of physical violence. The

data did not support this hypothesis. The findings from the current study showed a more

specific impact of being exposed to physical violence as a child predicting being a victim

of physical violence as an adult. Similar results were found for sexual and psychological

violence. When specifying types of violence, acceptance of dating violence was a

significant predictor but it did not mediate the relation between childhood exposure and

subsequent dating violence victimization.

There are several important contributions of the present study to the research on

intimate partner violence among young adults. One is the focus on victimization, as most

of the previous studies have focused on perpetration. A common assumption is that the

dynamics of perpetration is the same for victimization. However, previous studies report

different results for perpetrators and victims, suggesting that they are qualitatively

different. The findings from the current study further suggest that victimization is not the

same as perpetration. The intergenerational transmission of violence theory, which builds

on Bandura’s (1977) social learning theory, propose that children who are exposed to

interpersonal violence learn to be violent towards others. In other words, experiences

with violence teach children to become perpetrators because they do what they saw

worked as a conflict management strategy. Violence has the potential to end the conflict

and give perpetrators the outcome they want. Children who see these are arguably more

likely to use violence themselves to solve conflict. It is not necessarily the same for

victims. The current study found limited support for the intergenerational transmission

for victimization, suggesting that the relation between childhood exposure to violence and

dating violence victimization is more complex than for perpetration.

Related to the contribution of focusing on victimization is the focus on women.

The present study sampled women only because previous studies have shown higher rates

of victimization (Follingstad et al, 1991; Gover et al., 2008; Lichter & McCloskey, 2004;

Whitaker et al., 2007) and more negative outcomes related to female victimization than

male victimization (Follingstad et al., 1991; Molidor & Tolman, 1998; O’Keefe &

Treister, 1998). Previous dating violence studies have also found different results for

females and males (Foo & Margolin, 1995; Hendy et al., 2003; Gover et al., 2008; Luthra

& Gidycz, 2006; O’Keefe, 1998; O’Keefe & Treister, 1998), suggesting that they

experience dating violence differently. The present study highlights the need of at least

not assuming that dating violence is the same for females and males. With the limited

research on victimization and the mixed findings for males and females, it is important to

study them separately. Moreover, the adult domestic violence research has shown that

women are more likely to be victims than men (Dobash et al., 1992). If victimization

among younger samples of women could be better understood then that could increase

the understanding of domestic violence victimization.

Another contribution of the present study was the evidence for a more specific

effect as opposed to a global effect of childhood exposure. It was predicted that

childhood exposure to any type of interpersonal violence would increase the chance of

being a victim of physical violence. The findings did not support that hypothesis but

instead there was evidence for a more specific effect such that exposure to physical

violence increased the risk of being physically victimized.

Additionally, the present study predicted that attitudes such as acceptance of

violence would increase the likelihood of physical victimization. Even though the

support for acceptance as a predictor was limited, it is important to recognize that

cognitive factors can play a role predicting behaviors such as dating violence. The

intergenerational transmission of violence theory shows evidence for instrumental

learning in being a perpetrator (with some limited evidence for becoming a victim) but

the findings from the current study suggest that cognitive factors such as acceptance can

influence that process. This relates to Straus and colleagues’ (Straus et al., 1980; Straus,

& Yodanis, 1996) research on how children growing up in violent homes normalize or

become more tolerant of violence.

The findings from the present study are informative for prevention program of

dating violence. Prevention programs should incorporate information on perpetration and

victimization as well as the bidirectionality of dating violence. Adolescents and young

adults could also benefit from learning about healthy relationships since those exposed to

violence as children are likely to lack non-violent conflict management skills. They

could benefit from learning alternative ways to solve conflict.

Introduction Predicting Dating Violence Victimization among College Women

Students also viewed