520230 research-article2014
JIVXXX10.1177/0886260513520230Journal of Interpersonal ViolenceTurchik and Hassija
Article
Female Sexual Victimization Among College Students: Assault Severity, Health Risk Behaviors, and Sexual Functioning
Journal of Interpersonal Violence 2014, Vol. 29(13) 2439–2457 © The Author(s) 2014 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0886260513520230 jiv.sagepub.com
Jessica A. Turchik, PhD1,2 and Christina M. Hassija, PhD3
Abstract The purpose of the present study was to examine the relationship between college women’s sexual victimization experiences, health risk behaviors, and sexual functioning. A sample of 309 college women at a mid-sized Midwestern university completed measures assessing sexual victimization, sexual risk taking, substance use behaviors, sexual desire, sexual functioning, prior sexual experiences, and social desirability. Severity of sexual victimization was measured using a multi-item, behaviorally specific, gender-neutral measure, which was divided into four categories based on severity (none, sexual contact, sexual coercion, rape). Within the sample, 72.8% (n = 225) of women reported at least one experience of sexual victimization since age 16. Results from MANCOVAs and a multinomial logistic regression, controlling for social desirability and prior sexual experience, revealed that sexual victimization among female students was related to increased drug use, problematic drinking behaviors, sexual risk taking, sexual dysfunction, 1National
Center for PTSD, VA Palo Alto Health Care System, Menlo Park, CA, USA University, CA, USA 3Department of Psychology, California State University San Bernardino, CA, USA 2Stanford
Corresponding Author: Jessica A. Turchik, National Center for PTSD, VA Palo Alto Health Care System, 795 Willow Rd., Menlo Park, CA 94025, USA. Email: [email protected]
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
2440
Journal of Interpersonal Violence 29(13)
and dyadic sexual desire. In addition, findings indicated that women exposed to more severe forms of sexual victimization (i.e., rape) were most likely to report these risk-taking behaviors and sexual functioning issues. Implications for sexual assault risk reduction programming and treatment are discussed. Keywords female sexual victimization, sexual assault, sexual violence, risk-taking behavior, sexual functioning, college students Sexual victimization among women is a significant public health concern in the United States. Findings from the National Intimate Partner and Sexual Violence Survey indicate that one in five women (18.3%) has experienced rape and 44.6% have experienced some other form of sexual violence at some time in their lives (Black et al., 2011). College women are at particularly high risk for sexual victimization in the United States, with estimates reporting that approximately 20% to 25% of female undergraduates will experience attempted or completed rape at some point during their college careers (Fisher, Cullen, & Turner, 2000). In addition to the psychological and physical health consequences related to sexual assault experiences (Campbell, Dworkin, & Cabral, 2009; Zinzow et al., 2011), college women who experience sexual assault may also be at increased risk for engagement in health risk behaviors (e.g., Brener, McMahon, Warren, & Douglas, 1999; Gidycz, Orchowski, King, & Rich, 2008) and compromised sexual functioning (e.g., van Berlo & Ensink, 2000). A number of studies have demonstrated positive correlational relationships between sexual victimization and health risk behaviors. Greater substance use, particularly with respect to alcohol misuse, has been routinely linked to sexual victimization in college samples (e.g., Abbey, Zawacki, Buck, Clinton, & McAuslan, 2004; Brener et al., 1999; Kilpatrick, Acierno, Resnick, Saunders, & Best, 1997; Messman-Moore, Coates, Gaffey, & Johnson, 2008; Turchik, 2012). The positive association between substance misuse and sexual victimization is of particular concern, given the already high rates of problematic substance use cited among college populations (Johnston, O’Malley, Bachman, & Schulenberg, 2007). Similarly, sexual risk-taking behaviors, such as engaging in casual sex, sex with multiple partners or while under the influence of alcohol or drugs, and inconsistent use of contraception, which are also relatively common on college campuses (e.g., Grello, Welsh, & Harper, 2006; Gullette & Lyons, 2006; Ravert et al., 2009; Turchik & Gidycz, 2012), have been associated with an increased risk of
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
Turchik and Hassija
2441
sexual victimization (Brener et al., 1999; Deliramich & Gray, 2008; Johnson & Johnson, 2013; Messman-Moore et al., 2008; Stoner et al., 2008; Turchik, 2012). A study by Gidycz et al. (2008) revealed that college women with a history of sexual victimization were 3.3 to 4.6 times more likely to have engaged in early sexual intercourse (at or prior to age 15) and 4.5 times as likely to have multiple sex partners as compared with women without a history of sexual victimization. While these relationships are correlational in nature, the positive association between sexual victimization and sexual risk taking is of great importance, as engagement in these risky behaviors can lead to additional negative consequences such as sexually transmitted infections, unintended pregnancies, and increased risk for sexual revictimization (Campbell et al., 2009). The positive association between sexual victimization and health risk behaviors has been interpreted in a few ways and findings regarding their temporal relationship have been mixed. First, some researchers have suggested that women who engage in health risk behaviors are at greater risk for sexual victimization (see Gidycz, McNamara, & Edwards, 2006). For instance, a woman’s problematic drinking may lead to impaired judgment, thereby amplifying her vulnerability for sexual violence. Recently, Mouilso, Fischer, and Calhoun (2012) found that frequent binge drinking and frequent drinking predicted a sexual assault among a sample of college freshmen women, but sexual victimization did not impact frequency of alcohol use. However, others have opined that increased engagement in health risk behaviors may be a consequence of being victimized. In a prospective investigation of college women, Gidycz et al. (2008) found that sexual victimization among college women was associated with an increased likelihood of reported alcohol, tobacco, and marijuana use. Accordingly, a sexual assault victim may begin to abuse substances as a means to cope with psychological and emotional distress emanating from the assault experience. Alternatively, the relationship between sexual victimization and health risk behaviors may also be reciprocal in nature (see Champion et al., 2004, for discussion of these relationships). In a 2-year longitudinal investigation, Kilpatrick et al. (1997) evaluated the direction of relationships between substance abuse and sexual victimization using a national probability sample of 3,006 women and found support for a reciprocal relationship. Specifically, use of drugs, but not abuse of alcohol, was associated with increased odds of a new assault in the subsequent 2 years, and following a new assault, odds of both alcohol abuse and drug use were significantly increased after a new assault, even among women without a prior history of substance use or assault. In this case, a woman who is sexually victimized may subsequently experience depression or other psychological problems and begin binge
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
2442
Journal of Interpersonal Violence 29(13)
drinking and partying in an effort to distract herself. Alternatively, a woman who abuses substances and goes home with strangers for sex may be at an increased risk of being exposed to potential assailants and less able to successfully resist an attack. The direction of relationships between substance misuse and sexual victimization remains unclear; however, this robust association is of great importance, particularly among college populations where problematic substance use and sexual assault are highly prevalent. An additional, although somewhat less studied, negative correlate of sexual victimization is sexual dysfunction. Findings from a national probability sample indicate that 40% of American women experience sexual dysfunction and that such difficulties are strongly related to prior sexual victimization for women (Laumann, Paik, & Rosen, 1999). Specifically, exposure to variants of sexual violence has been positively associated with impaired sexual functioning (Becker, Skinner, Abel, & Cichon, 1986; Sanjuan, Langenbucher, & Labouvie, 2009; Turchik et al., 2012; van Berlo & Ensink, 2000) and gynecological problems (Campbell, Lichty, Sturza, & Raja, 2006; Golding, Wilsnack, & Learman, 1998). The relationships between these variables, however, have been largely unexamined in college populations. While not specific to sexual functioning, one study of college women found that women who experienced date rape reported reduced sexual self-esteem in areas of moral judgment, control, and adaptiveness as compared with those who had not been raped (Shapiro & Schwarz, 1997). Turchik (2012) examined the relationship between sexual victimization and sexual dysfunction in a sample of male college students, finding that those who experienced rape were more likely to report sexual dysfunctions than those without victimization histories. Given the high rates of sexual victimization among this population, sexual dysfunction may be a significant concern for college women and therefore warrants further investigation. The purpose of the present study was to examine the relationship between sexual victimization, health risk behaviors (i.e., alcohol, drug, tobacco use, and sexual risk-taking behaviors), and sexual functioning (i.e., sexual desire and sexual functioning) in a sample of female college students. Specifically, each of these constructs were measured without linking the questions to an assault experience, thereby mitigating the potential for participants to be influenced by their responses to sexual victimization items. Moreover, rather than comparing women with a prior history of victimization to those without, sexual victimization was evaluated and divided into four categories based on severity (i.e., none, sexual contact, sexual coercion, rape) and examined in relation to health risk behaviors and sexual dysfunction. Additionally, given the sensitive nature of the study topics, concerns regarding validity of selfreport assessment, and students’ potential need for impression management,
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
Turchik and Hassija
2443
we included a measure of social desirability to help control for level of accurate responding. We hypothesized, based on prior literature, that female college students who reported sexual victimization would report more frequent engagement in health risk behaviors, greater sexual dysfunction, and reduced sexual desire and that these relationships would be strongest among those who reported more severe sexual victimization experiences.
Method Participants Participants were 309 undergraduate women, from a medium-sized Midwestern University, who were between the ages of 18 and 22 (M = 18.9). The majority of the participants identified as heterosexual (98.7%), Caucasian (93.9%), and Christian (80.3%), and indicated that they were never married (100%) and had parents whose annual incomes were more than US$50,000 (75.8%) and were in their first or second year of college (83.4%). Approximately 42.1% of the students reported that they were not in any kind of romantic relationship, 31.4% were dating one or more people, and 26.2% were in a long-term monogamous relationship of 6 months or longer. Most of the participants (84.1%) reported having a prior consensual sexual experience (oral, anal, or vaginal sex). Of an initial 310 women, one participant had a significant amount of missing data (i.e., not completing about half of the measures) and was removed from the analyses leaving a total sample of 309. There was very little missing data (less than 3%) in the current study, and missing data were handled using multiple imputation prior to data analyses.
Measures Demographics. Participants completed several questions regarding sociodemographic factors. Sexual functioning. An item was constructed and used to assess sexual functioning problems during solitary and/or dyadic situations based on the sexual dysfunction disorders outlined in the Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM-IV-TR; American Psychiatric Association, 2000). Participants were asked the following: Do you experience any of the following sexual problems (i.e., lack of sexual desire, lack of orgasm, aversion to sexual contact, pain associated with sex, difficulty with lubrication and tightening of the vaginal muscles that interferes with sex) at least 25% of the time during sexual situations?”
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
2444
Journal of Interpersonal Violence 29(13)
A total sexual dysfunction score was derived based on a total count of endorsed problems ranging from 0 to 6. Sexual desire. The Sexual Desire Inventory (SDI; Spector, Carey, & Steinberg, 1996)—a self-report measure that assesses the strength of a person’s sex drive and their desired frequency of sexual behavior—was used to assess dyadic and solitary sexual desire. Factor analytic evidence has supported the existence of two factors, dyadic (eight items) and solitary (three items; Spector et al., 1996). Dyadic desire scores range from 0 to 62 and solitary scores range from 0 to 23, with higher scores representing a greater level of sexual desire. Two-week test–retest reliabilities were .72 and .84 for the solitary and dyadic desire scales (Turchik & Garske, 2009); and Cronbach’s alphas for the current study were .87 and .92, respectively. Substance use. Substance use was assessed using a modified version of the Drinking and Drug Habits Questionnaire (DDHQ; Collins, Parks, & Marlatt, 1985), a 31-item measure that assesses alcohol and drug use. The DDHQ contains standardized definitions of what constitutes a drink (e.g., one 4 oz. glass of wine) and asks participants to report the average number of drinks consumed weekly, with numbers ranging from “0” to “11 or more,” yielding a total weekly drinking score ranging from 0 to 77. A second subscale assessed problem drinking behaviors (e.g., getting sick after drinking, legal repercussions from drinking and driving), which was answered “yes” or “no” with a score range of 0 to 9. A third subscale assessed drug use by asking participants about their usage of 13 different drugs (e.g., marijuana, cocaine) on a 4-point “never used” to “regularly use” scale with a score range of 0 to 39. A 14th item assessed tobacco use on the same 4-point scale with a score range of 0 to 3. Two-week test–retest reliabilities range from .86 to .94 (Turchik & Garske, 2009). In the current study, the internal consistency alphas for the weekly alcohol, problem drinking behaviors, and drug use subscales were .78, .70, and .70, respectively. Sexual risk-taking behavior. The 23-item Sexual Risk Survey (SRS; Turchik & Garske, 2009) was used to assess the frequency of sexual risk behaviors in the past 6 months. The survey was designed for college students with or without sexual experience and measures a broad range of sexual behaviors. Each item is scored 0 to 4 with a possible scale total range of 0 to 92, with higher scores indicating greater risk taking. The total score of the SRS can be used or five subscale scores can be calculated; both the total score and the subscale scores were used in the current study. The SRS has evidenced convergent and discriminant validity as well as good internal consistency and 2-week
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
Turchik and Hassija
2445
test–retest reliability (Turchik & Garske, 2009). In the current study, the internal inconsistency reliability was .87 for the total SRS score and .88, .80, .75, .81, and .65 for the Sexual Risk Taking with Uncommitted Partners, Risky Sex Acts, Impulsive Sexual Behaviors, Intent to Engage in Risky Sexual Behaviors, and Risky Anal Sex Acts subscales, respectively. Social desirability. The Marlowe–Crowne Social Desirability Scale (MCSD; Crowne & Marlowe, 1960) is a 33-item self-report instrument consisting of items “defined by behaviors which are culturally sanctioned and approved, but which are improbably of occurrence” (p. 350). The measure was designed to measure deviant response sets such as “faking good” or “faking bad.” The measure is often used to assess social desirability, especially on self-administered tests. The measure has evidenced convergent validity and the 1-month test–retest reliability was found to be .89 (Crowne & Marlowe, 1960). In this study, the internal consistency alpha was .75. Sexual victimization. Sexual victimization was assessed using the Sexual Coercion Tactics Scale (SCTS), which assesses sexual coercion used by and used on both men and women since the age of 16 (Struckman-Johnson, Struckman-Johnson, & Anderson, 2003). Participants were asked to indicate how many times they have either used coercive tactics to get someone to engage in sexual behaviors or how many times they engaged in sexual behaviors because someone used these tactics on them and the gender of the other person involved. There were 18 different tactics that were listed for each of three types of sexual acts (sexual behavior, oral sex, and anal/vaginal sex) on both forms (whether they were the user or recipient of the tactics) of the survey. Only female victimization was used in the current study, both by male and female perpetrators. Victimization status was broken down into four victimization categories consistent with prior studies of female sexual victimization (e.g., Koss et al., 2007) and prior categorization of the SCTS (see Turchik, 2012): no victimization, unwanted sexual contact, sexual coercion (e.g., engaging in unwanted sex because perpetrator lied, threatened, or used their authority), and rape. In the current study, Cronbach’s alpha of the SCTS victimization scale was .86.
Procedure Participants were recruited from undergraduate psychology courses and volunteered via an online experiment scheduling system, participating for partial class credit. The study was advertised as a research project examining social, health, and personality factors and was part of a larger study conducted in
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
2446
Journal of Interpersonal Violence 29(13)
compliance with the university’s Institutional Review Board that focused on measure validation (for more detailed information, see Turchik & Garske, 2009). Participants were administered paper-and-pencil surveys; the survey measures were given in a fixed order within small group settings of 20 participants or less in large classrooms where participants were spread out to ensure privacy. Due to the sensitive nature of the study material, a clinical psychology graduate student was available during all study sessions in case participants experienced distress; however, no participants exhibited or reported any problems or distress related to taking part in the study. Each participant received a debriefing form that included local resources for sexual education, sexually transmitted infections and pregnancy testing, and counseling services.
Data Analyses Pearson bivariate correlations were examined for sexual victimization status (coded as 0 = none, 1 = unwanted sexual contact, 2 = sexual coercion, and 3 = rape) and all of the health risk and sexual functioning variables included in the analyses described below. Two one-way MANCOVAs were utilized to test the effect of victimization status on health risk behavior engagement and sexual functioning. Past sexual experience (whether the person has engaged in consensual oral, anal, or vaginal sex; 0 = no, 1 = yes) and social desirability scores were included as covariates in each analysis. The first MANCOVA was used to examine the risk-taking variables (weekly drinking amount, problematic drinking behaviors, tobacco use, drug use, sexual risk-taking behaviors) and the second, the sexual functioning variables (dyadic sexual desire, solitary sexual desire, number of sexual dysfunctions). If the multivariate test was significant, the univariate ANCOVAs were examined, and if these were significant, Bonferroni post hoc tests were then used to further examine the pairwise comparisons. Given the effect of victimization status on sexual risk-taking behaviors, a third follow-up MANCOVA was also run to explore the effect of sexual victimization on the five SRS subscale scores. Finally, a multinomial logistic regression was conducted to examine differences in the reported specific sexual dysfunction behaviors by victimization status. Effects sizes are presented in the form of partial eta squared and can be generally interpreted as follows: .01 a small effect size, .06 medium, and .14 large (Cohen, 1977). Examination of bivariate scatterplots and screening for outliers was conducted before each analysis and revealed no violations of MANCOVA assumptions; however, significant Box’s M tests suggest possible violations of the assumption of homogeneity so the more conservative Pillai’s Trace statistic was used in place of Wilks’s Lambda (Tabachnick & Fidell, 2007).
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
Turchik and Hassija
2447
Results Frequency of Victimization and Bivariate Correlations In the current study, 72.8% (n = 225) of female participants reported at least one experience of sexual victimization since age 16. When examining sex of the assailant, 1.3% reported at least one victimization experience with a female perpetrator and 72.5% reported at least one victimization experience with a male perpetrator. When participants were placed in mutually exclusive groups based on the most severe type of victimization reported, 27.2% (n = 84) reported none, 25.9% (n = 80) unwanted sexual contact, 19.4% (n = 60) sexual coercion, and 27.5% (n = 85) completed rape. Bivariate correlations revealed that all variables except solitary sexual desire were significantly associated with sexual victimization status (see Table 1).
Sexual Victimization and Health Risk Behaviors A one-way MANCOVA was conducted to examine the effect of sexual victimization on health risk behavior engagement. The multivariate test was significant, Pillai’s trace = .18, F(15, 903) = 3.93, p < .001, η2p = .06, observed power = 1.00, indicating significant differences among the victimization categories on the combined health risk variables. The covariates, sexual experience and social desirability were also significant, Pillai’s trace = .15, F(5, 299) = 10.26, p < .001, η2p = .15, observed power = 1.00 and Pillai’s trace = .11, F(5, 299) = 7.67, p < .001, η2p = .11, observed power = 0.99, respectively. The univariate tests were then performed, and victimization category differences were significant for drug use, problematic drinking behaviors, and sexual risk taking when the alpha was set at .01. The effect sizes ranged from small to medium. A comparison of adjusted means revealed a number of differences between the victimization groups (see Table 2). Given the significance of sexual risk taking using the full SRS score, a one-way MANCOVA was conducted to examine the effect of sexual victimization status on the five SRS subscales scores. MANCOVA results indicated significant differences among the victimization categories on the combined sexual risk subscales, Pillai’s trace = .18, F(15, 903) = 3.90, p < .001, η2p = .06, observed power = 1.00. The covariates, sexual experience and social desirability, both significantly influenced the combined dependent variable, Pillai’s trace = .19, F(5, 299) = 14.34, p < .001, η2p = .19, observed power = 1.00 and Pillai’s trace = .06, F(5, 299) = 3.46, p < .01, η2p = .06, observed power = 0.91, respectively. The univariate tests were then conducted on the dependent variables, demonstrating that victimization status differences were
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
2448
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
1
*p ≤ .05. **p ≤ .01.
— 1. Sexual victimization severity 2. Sexual experience 3. Social desirability 4. Weekly drinking 5. Problematic drinking 6. Drug use 7. Tobacco use 8. Sexual risk taking 9. Dyadic sexual desire 10. Solitary sexual desire 11. Sexual dysfunctions 12. Risk with uncommitted partners 13. Risky sex acts 14. Impulsive sexual behaviors 15. Intent to engage in sexual risk 16. Risky anal sex
Variable/Subscale
3
—
−.03 —
.29** −.16**
2
5 .24**
6 .14*
7 .42**
8 .15*
9 .08
10 .32**
11 .40**
12 .17**
13 .36**
14 .20**
15 .12*
16
—
.47** —
.34** .23** —
.17** .14* .33** —
.46**
.09 —
.27** .41**
—
.27** —
.18**
.11 —
.16**
.12*
.20** .14* .64** .13*
.20**
.28** .08 .67** .31**
—
.29** .28** .85** .23**
.16** .09 .33** .12*
.15** .21** .19** .27**
.05
—
—
.36** .02
.20**
.17**
.13*
.11 .01 .41** .17**
.06 .38**
.27**
.14*
.07
.11 .19** .46** .24**
.29** .23** .22** .13* .41** .20** .15** .28** .33** .46** .13* .10 .16** −.18** −.31** −.23** −.14* −.24** −.16** −.07 −.15* −.16** −.16** −.22** −.15** −.03 — .54** .45** .32** .34** .19** .10 .38** .28** .20** .27** .23** .06 — .47** .21** .30** .17** .10 .15** .20** .14* .31** .23** .06
.15** .26**
4
Table 1. Bivariate Correlations Among Health Risk Behaviors, Sexual Functioning Variables, and Sexual Victimization Severity.
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
2449
5.08 (0.43) 2.55 (0.38)a
0.59 (0.22)
0.39 (0.20) 31.06 (1.23)a 3.24 (0.52) .76 (0.12)a,b
20.82 (0.58) 2.97 (0.09) 6.51 (0.18) 0.82 (0.06) 15.80 (0.55) 4.14 (0.26)
5.40 (0.22) 4.37 (0.20)
1.14 (0.11)
0.76 (0.10) 33.18 (0.63) 3.33 (0.27) 1.12 (0.06)
Weekly drinking Problematic drinking Drug use Tobacco use Sexual risk taking Risk with uncommitted partners Risky sex acts Impulsive sexual behaviors Intent to engage in sexual risk Risky anal sex Dyadic sexual desire Solitary sexual desire Sexual dysfunctions 0.88 (0.21) 35.70 (1.24)a 3.26 (0.53) .87 (0.12)c,d
1.21 (0.22)
5.22 (0.43) 3.93 (0.39)b
20.63 (1.13) 3.16 (0.18) 6.09 (0.36)b 0.81 (0.12) 14.37 (1.07)b,d 3.13 (0.52)b
Unwanted Sexual Contact
0.93 (0.24) 31.62 (1.44) 3.26 (0.61) 1.39 (0.14)a,c
1.26 (0.25)
6.21 (0.50) 5.39 (0.45)a
19.36 (1.32) 2.69 (0.21)b 6.29 (0.42) 0.76 (0.14) 18.91 (1.25)b,c 5.12 (0.60)a
Sexual Coercion
0.82 (0.20) 34.36 (1.21) 3.56 (0.52) 1.43 (0.12)b,d
1.48 (0.21)
5.11 (0.42) 5.61 (0.38)a,b
22.44 (1.11) 3.47 (0.18)a,b 7.54 (0.35)a,b 1.00 (0.12) 19.18 (1.05)a,d 6.16 (0.51)a,b
Rape
.097***
.182***
.184***
Pillai’s Trace
1.39 3.18* 0.09 7.00***
3.03
1.24 12.08***
1.15 5.33*** 3.64** 1.08 12.59*** 11.39***
Univariate F
.01 .03 .00 .07
.03
.01 .11
.01 .05 .04 .01 .11 .10
2
ηp
Note. Table includes results from three separate multivariate analyses of covariance. All means and standard errors are adjusted for the covariates of sexual experience and social desirability. Means (SE) in the same row with the same superscripts differ at p < .05 based on Bonferroni post hoc comparisons. *p ≤ .05. **p ≤ .01. ***p ≤ .001.
20.86 (1.12) 2.56 (0.18)a 6.14 (0.36)a 0.71 (0.12) 10.73 (1.07)a,c 2.16 (0.52)a
Sample M (SE)
Variables
Nonvictim
Table 2. Multivariate Analyses of Covariance Examining Effects of Sexual Victimization Status on Health Risk Behaviors and Sexual Functioning.
2450
Journal of Interpersonal Violence 29(13)
significant for two of the five subscales, Sexual Risk Taking with Uncommitted Partners and Impulsive Sexual Behaviors, with the alpha set at .01. Effect sizes were medium in magnitude. A comparison of adjusted means revealed a number of significant differences (see Table 2).
Sexual Victimization and Sexual Functioning A one-way MANCOVA was conducted to determine the effect of sexual victimization status on sexual functioning. The multivariate test was significant, Pillai’s trace = .10, F(9, 909) = 3.37, p < .001, η2p = .03, observed power = 0.99, indicating significant differences among the victimization categories on the sexual functioning variables. The covariates, sexual experience and social desirability, were both significant, Pillai’s trace = .07, F(3, 301) = 7.78, p < 2 .001, ηp = .07, observed power = 0.99 and Pillai’s trace = .03, F(3, 301) = 3.09, p < .01, η2p = .03, observed power = 0.72, respectively. The univariate tests were then performed, and victimization category differences were significant for dyadic sexual desire and number of sexual dysfunctions with an alpha set at .02 (see Table 2), both accounting for a modest amount of the variance. A comparison of adjusted means revealed a number of significant differences. A multinomial logistic regression was then used to examine the association between level of victimization and report of each of the six types of sexual dysfunction behaviors. The overall model was significant, χ2(24, N = 309) = 88.33, p < .001, and the Nagelkerke R2 was .27 for the total model (see Table 3). Both the covariates, sexual experience, χ2(3, N = 309) = 20.55, p < .001, and social desirability, χ2(3, N = 309) = 9.59, p
JIVXXX10.1177/0886260513520230Journal of Interpersonal ViolenceTurchik and Hassija
Article
Female Sexual Victimization Among College Students: Assault Severity, Health Risk Behaviors, and Sexual Functioning
Journal of Interpersonal Violence 2014, Vol. 29(13) 2439–2457 © The Author(s) 2014 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0886260513520230 jiv.sagepub.com
Jessica A. Turchik, PhD1,2 and Christina M. Hassija, PhD3
Abstract The purpose of the present study was to examine the relationship between college women’s sexual victimization experiences, health risk behaviors, and sexual functioning. A sample of 309 college women at a mid-sized Midwestern university completed measures assessing sexual victimization, sexual risk taking, substance use behaviors, sexual desire, sexual functioning, prior sexual experiences, and social desirability. Severity of sexual victimization was measured using a multi-item, behaviorally specific, gender-neutral measure, which was divided into four categories based on severity (none, sexual contact, sexual coercion, rape). Within the sample, 72.8% (n = 225) of women reported at least one experience of sexual victimization since age 16. Results from MANCOVAs and a multinomial logistic regression, controlling for social desirability and prior sexual experience, revealed that sexual victimization among female students was related to increased drug use, problematic drinking behaviors, sexual risk taking, sexual dysfunction, 1National
Center for PTSD, VA Palo Alto Health Care System, Menlo Park, CA, USA University, CA, USA 3Department of Psychology, California State University San Bernardino, CA, USA 2Stanford
Corresponding Author: Jessica A. Turchik, National Center for PTSD, VA Palo Alto Health Care System, 795 Willow Rd., Menlo Park, CA 94025, USA. Email: [email protected]
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
2440
Journal of Interpersonal Violence 29(13)
and dyadic sexual desire. In addition, findings indicated that women exposed to more severe forms of sexual victimization (i.e., rape) were most likely to report these risk-taking behaviors and sexual functioning issues. Implications for sexual assault risk reduction programming and treatment are discussed. Keywords female sexual victimization, sexual assault, sexual violence, risk-taking behavior, sexual functioning, college students Sexual victimization among women is a significant public health concern in the United States. Findings from the National Intimate Partner and Sexual Violence Survey indicate that one in five women (18.3%) has experienced rape and 44.6% have experienced some other form of sexual violence at some time in their lives (Black et al., 2011). College women are at particularly high risk for sexual victimization in the United States, with estimates reporting that approximately 20% to 25% of female undergraduates will experience attempted or completed rape at some point during their college careers (Fisher, Cullen, & Turner, 2000). In addition to the psychological and physical health consequences related to sexual assault experiences (Campbell, Dworkin, & Cabral, 2009; Zinzow et al., 2011), college women who experience sexual assault may also be at increased risk for engagement in health risk behaviors (e.g., Brener, McMahon, Warren, & Douglas, 1999; Gidycz, Orchowski, King, & Rich, 2008) and compromised sexual functioning (e.g., van Berlo & Ensink, 2000). A number of studies have demonstrated positive correlational relationships between sexual victimization and health risk behaviors. Greater substance use, particularly with respect to alcohol misuse, has been routinely linked to sexual victimization in college samples (e.g., Abbey, Zawacki, Buck, Clinton, & McAuslan, 2004; Brener et al., 1999; Kilpatrick, Acierno, Resnick, Saunders, & Best, 1997; Messman-Moore, Coates, Gaffey, & Johnson, 2008; Turchik, 2012). The positive association between substance misuse and sexual victimization is of particular concern, given the already high rates of problematic substance use cited among college populations (Johnston, O’Malley, Bachman, & Schulenberg, 2007). Similarly, sexual risk-taking behaviors, such as engaging in casual sex, sex with multiple partners or while under the influence of alcohol or drugs, and inconsistent use of contraception, which are also relatively common on college campuses (e.g., Grello, Welsh, & Harper, 2006; Gullette & Lyons, 2006; Ravert et al., 2009; Turchik & Gidycz, 2012), have been associated with an increased risk of
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
Turchik and Hassija
2441
sexual victimization (Brener et al., 1999; Deliramich & Gray, 2008; Johnson & Johnson, 2013; Messman-Moore et al., 2008; Stoner et al., 2008; Turchik, 2012). A study by Gidycz et al. (2008) revealed that college women with a history of sexual victimization were 3.3 to 4.6 times more likely to have engaged in early sexual intercourse (at or prior to age 15) and 4.5 times as likely to have multiple sex partners as compared with women without a history of sexual victimization. While these relationships are correlational in nature, the positive association between sexual victimization and sexual risk taking is of great importance, as engagement in these risky behaviors can lead to additional negative consequences such as sexually transmitted infections, unintended pregnancies, and increased risk for sexual revictimization (Campbell et al., 2009). The positive association between sexual victimization and health risk behaviors has been interpreted in a few ways and findings regarding their temporal relationship have been mixed. First, some researchers have suggested that women who engage in health risk behaviors are at greater risk for sexual victimization (see Gidycz, McNamara, & Edwards, 2006). For instance, a woman’s problematic drinking may lead to impaired judgment, thereby amplifying her vulnerability for sexual violence. Recently, Mouilso, Fischer, and Calhoun (2012) found that frequent binge drinking and frequent drinking predicted a sexual assault among a sample of college freshmen women, but sexual victimization did not impact frequency of alcohol use. However, others have opined that increased engagement in health risk behaviors may be a consequence of being victimized. In a prospective investigation of college women, Gidycz et al. (2008) found that sexual victimization among college women was associated with an increased likelihood of reported alcohol, tobacco, and marijuana use. Accordingly, a sexual assault victim may begin to abuse substances as a means to cope with psychological and emotional distress emanating from the assault experience. Alternatively, the relationship between sexual victimization and health risk behaviors may also be reciprocal in nature (see Champion et al., 2004, for discussion of these relationships). In a 2-year longitudinal investigation, Kilpatrick et al. (1997) evaluated the direction of relationships between substance abuse and sexual victimization using a national probability sample of 3,006 women and found support for a reciprocal relationship. Specifically, use of drugs, but not abuse of alcohol, was associated with increased odds of a new assault in the subsequent 2 years, and following a new assault, odds of both alcohol abuse and drug use were significantly increased after a new assault, even among women without a prior history of substance use or assault. In this case, a woman who is sexually victimized may subsequently experience depression or other psychological problems and begin binge
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
2442
Journal of Interpersonal Violence 29(13)
drinking and partying in an effort to distract herself. Alternatively, a woman who abuses substances and goes home with strangers for sex may be at an increased risk of being exposed to potential assailants and less able to successfully resist an attack. The direction of relationships between substance misuse and sexual victimization remains unclear; however, this robust association is of great importance, particularly among college populations where problematic substance use and sexual assault are highly prevalent. An additional, although somewhat less studied, negative correlate of sexual victimization is sexual dysfunction. Findings from a national probability sample indicate that 40% of American women experience sexual dysfunction and that such difficulties are strongly related to prior sexual victimization for women (Laumann, Paik, & Rosen, 1999). Specifically, exposure to variants of sexual violence has been positively associated with impaired sexual functioning (Becker, Skinner, Abel, & Cichon, 1986; Sanjuan, Langenbucher, & Labouvie, 2009; Turchik et al., 2012; van Berlo & Ensink, 2000) and gynecological problems (Campbell, Lichty, Sturza, & Raja, 2006; Golding, Wilsnack, & Learman, 1998). The relationships between these variables, however, have been largely unexamined in college populations. While not specific to sexual functioning, one study of college women found that women who experienced date rape reported reduced sexual self-esteem in areas of moral judgment, control, and adaptiveness as compared with those who had not been raped (Shapiro & Schwarz, 1997). Turchik (2012) examined the relationship between sexual victimization and sexual dysfunction in a sample of male college students, finding that those who experienced rape were more likely to report sexual dysfunctions than those without victimization histories. Given the high rates of sexual victimization among this population, sexual dysfunction may be a significant concern for college women and therefore warrants further investigation. The purpose of the present study was to examine the relationship between sexual victimization, health risk behaviors (i.e., alcohol, drug, tobacco use, and sexual risk-taking behaviors), and sexual functioning (i.e., sexual desire and sexual functioning) in a sample of female college students. Specifically, each of these constructs were measured without linking the questions to an assault experience, thereby mitigating the potential for participants to be influenced by their responses to sexual victimization items. Moreover, rather than comparing women with a prior history of victimization to those without, sexual victimization was evaluated and divided into four categories based on severity (i.e., none, sexual contact, sexual coercion, rape) and examined in relation to health risk behaviors and sexual dysfunction. Additionally, given the sensitive nature of the study topics, concerns regarding validity of selfreport assessment, and students’ potential need for impression management,
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
Turchik and Hassija
2443
we included a measure of social desirability to help control for level of accurate responding. We hypothesized, based on prior literature, that female college students who reported sexual victimization would report more frequent engagement in health risk behaviors, greater sexual dysfunction, and reduced sexual desire and that these relationships would be strongest among those who reported more severe sexual victimization experiences.
Method Participants Participants were 309 undergraduate women, from a medium-sized Midwestern University, who were between the ages of 18 and 22 (M = 18.9). The majority of the participants identified as heterosexual (98.7%), Caucasian (93.9%), and Christian (80.3%), and indicated that they were never married (100%) and had parents whose annual incomes were more than US$50,000 (75.8%) and were in their first or second year of college (83.4%). Approximately 42.1% of the students reported that they were not in any kind of romantic relationship, 31.4% were dating one or more people, and 26.2% were in a long-term monogamous relationship of 6 months or longer. Most of the participants (84.1%) reported having a prior consensual sexual experience (oral, anal, or vaginal sex). Of an initial 310 women, one participant had a significant amount of missing data (i.e., not completing about half of the measures) and was removed from the analyses leaving a total sample of 309. There was very little missing data (less than 3%) in the current study, and missing data were handled using multiple imputation prior to data analyses.
Measures Demographics. Participants completed several questions regarding sociodemographic factors. Sexual functioning. An item was constructed and used to assess sexual functioning problems during solitary and/or dyadic situations based on the sexual dysfunction disorders outlined in the Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM-IV-TR; American Psychiatric Association, 2000). Participants were asked the following: Do you experience any of the following sexual problems (i.e., lack of sexual desire, lack of orgasm, aversion to sexual contact, pain associated with sex, difficulty with lubrication and tightening of the vaginal muscles that interferes with sex) at least 25% of the time during sexual situations?”
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
2444
Journal of Interpersonal Violence 29(13)
A total sexual dysfunction score was derived based on a total count of endorsed problems ranging from 0 to 6. Sexual desire. The Sexual Desire Inventory (SDI; Spector, Carey, & Steinberg, 1996)—a self-report measure that assesses the strength of a person’s sex drive and their desired frequency of sexual behavior—was used to assess dyadic and solitary sexual desire. Factor analytic evidence has supported the existence of two factors, dyadic (eight items) and solitary (three items; Spector et al., 1996). Dyadic desire scores range from 0 to 62 and solitary scores range from 0 to 23, with higher scores representing a greater level of sexual desire. Two-week test–retest reliabilities were .72 and .84 for the solitary and dyadic desire scales (Turchik & Garske, 2009); and Cronbach’s alphas for the current study were .87 and .92, respectively. Substance use. Substance use was assessed using a modified version of the Drinking and Drug Habits Questionnaire (DDHQ; Collins, Parks, & Marlatt, 1985), a 31-item measure that assesses alcohol and drug use. The DDHQ contains standardized definitions of what constitutes a drink (e.g., one 4 oz. glass of wine) and asks participants to report the average number of drinks consumed weekly, with numbers ranging from “0” to “11 or more,” yielding a total weekly drinking score ranging from 0 to 77. A second subscale assessed problem drinking behaviors (e.g., getting sick after drinking, legal repercussions from drinking and driving), which was answered “yes” or “no” with a score range of 0 to 9. A third subscale assessed drug use by asking participants about their usage of 13 different drugs (e.g., marijuana, cocaine) on a 4-point “never used” to “regularly use” scale with a score range of 0 to 39. A 14th item assessed tobacco use on the same 4-point scale with a score range of 0 to 3. Two-week test–retest reliabilities range from .86 to .94 (Turchik & Garske, 2009). In the current study, the internal consistency alphas for the weekly alcohol, problem drinking behaviors, and drug use subscales were .78, .70, and .70, respectively. Sexual risk-taking behavior. The 23-item Sexual Risk Survey (SRS; Turchik & Garske, 2009) was used to assess the frequency of sexual risk behaviors in the past 6 months. The survey was designed for college students with or without sexual experience and measures a broad range of sexual behaviors. Each item is scored 0 to 4 with a possible scale total range of 0 to 92, with higher scores indicating greater risk taking. The total score of the SRS can be used or five subscale scores can be calculated; both the total score and the subscale scores were used in the current study. The SRS has evidenced convergent and discriminant validity as well as good internal consistency and 2-week
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
Turchik and Hassija
2445
test–retest reliability (Turchik & Garske, 2009). In the current study, the internal inconsistency reliability was .87 for the total SRS score and .88, .80, .75, .81, and .65 for the Sexual Risk Taking with Uncommitted Partners, Risky Sex Acts, Impulsive Sexual Behaviors, Intent to Engage in Risky Sexual Behaviors, and Risky Anal Sex Acts subscales, respectively. Social desirability. The Marlowe–Crowne Social Desirability Scale (MCSD; Crowne & Marlowe, 1960) is a 33-item self-report instrument consisting of items “defined by behaviors which are culturally sanctioned and approved, but which are improbably of occurrence” (p. 350). The measure was designed to measure deviant response sets such as “faking good” or “faking bad.” The measure is often used to assess social desirability, especially on self-administered tests. The measure has evidenced convergent validity and the 1-month test–retest reliability was found to be .89 (Crowne & Marlowe, 1960). In this study, the internal consistency alpha was .75. Sexual victimization. Sexual victimization was assessed using the Sexual Coercion Tactics Scale (SCTS), which assesses sexual coercion used by and used on both men and women since the age of 16 (Struckman-Johnson, Struckman-Johnson, & Anderson, 2003). Participants were asked to indicate how many times they have either used coercive tactics to get someone to engage in sexual behaviors or how many times they engaged in sexual behaviors because someone used these tactics on them and the gender of the other person involved. There were 18 different tactics that were listed for each of three types of sexual acts (sexual behavior, oral sex, and anal/vaginal sex) on both forms (whether they were the user or recipient of the tactics) of the survey. Only female victimization was used in the current study, both by male and female perpetrators. Victimization status was broken down into four victimization categories consistent with prior studies of female sexual victimization (e.g., Koss et al., 2007) and prior categorization of the SCTS (see Turchik, 2012): no victimization, unwanted sexual contact, sexual coercion (e.g., engaging in unwanted sex because perpetrator lied, threatened, or used their authority), and rape. In the current study, Cronbach’s alpha of the SCTS victimization scale was .86.
Procedure Participants were recruited from undergraduate psychology courses and volunteered via an online experiment scheduling system, participating for partial class credit. The study was advertised as a research project examining social, health, and personality factors and was part of a larger study conducted in
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
2446
Journal of Interpersonal Violence 29(13)
compliance with the university’s Institutional Review Board that focused on measure validation (for more detailed information, see Turchik & Garske, 2009). Participants were administered paper-and-pencil surveys; the survey measures were given in a fixed order within small group settings of 20 participants or less in large classrooms where participants were spread out to ensure privacy. Due to the sensitive nature of the study material, a clinical psychology graduate student was available during all study sessions in case participants experienced distress; however, no participants exhibited or reported any problems or distress related to taking part in the study. Each participant received a debriefing form that included local resources for sexual education, sexually transmitted infections and pregnancy testing, and counseling services.
Data Analyses Pearson bivariate correlations were examined for sexual victimization status (coded as 0 = none, 1 = unwanted sexual contact, 2 = sexual coercion, and 3 = rape) and all of the health risk and sexual functioning variables included in the analyses described below. Two one-way MANCOVAs were utilized to test the effect of victimization status on health risk behavior engagement and sexual functioning. Past sexual experience (whether the person has engaged in consensual oral, anal, or vaginal sex; 0 = no, 1 = yes) and social desirability scores were included as covariates in each analysis. The first MANCOVA was used to examine the risk-taking variables (weekly drinking amount, problematic drinking behaviors, tobacco use, drug use, sexual risk-taking behaviors) and the second, the sexual functioning variables (dyadic sexual desire, solitary sexual desire, number of sexual dysfunctions). If the multivariate test was significant, the univariate ANCOVAs were examined, and if these were significant, Bonferroni post hoc tests were then used to further examine the pairwise comparisons. Given the effect of victimization status on sexual risk-taking behaviors, a third follow-up MANCOVA was also run to explore the effect of sexual victimization on the five SRS subscale scores. Finally, a multinomial logistic regression was conducted to examine differences in the reported specific sexual dysfunction behaviors by victimization status. Effects sizes are presented in the form of partial eta squared and can be generally interpreted as follows: .01 a small effect size, .06 medium, and .14 large (Cohen, 1977). Examination of bivariate scatterplots and screening for outliers was conducted before each analysis and revealed no violations of MANCOVA assumptions; however, significant Box’s M tests suggest possible violations of the assumption of homogeneity so the more conservative Pillai’s Trace statistic was used in place of Wilks’s Lambda (Tabachnick & Fidell, 2007).
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
Turchik and Hassija
2447
Results Frequency of Victimization and Bivariate Correlations In the current study, 72.8% (n = 225) of female participants reported at least one experience of sexual victimization since age 16. When examining sex of the assailant, 1.3% reported at least one victimization experience with a female perpetrator and 72.5% reported at least one victimization experience with a male perpetrator. When participants were placed in mutually exclusive groups based on the most severe type of victimization reported, 27.2% (n = 84) reported none, 25.9% (n = 80) unwanted sexual contact, 19.4% (n = 60) sexual coercion, and 27.5% (n = 85) completed rape. Bivariate correlations revealed that all variables except solitary sexual desire were significantly associated with sexual victimization status (see Table 1).
Sexual Victimization and Health Risk Behaviors A one-way MANCOVA was conducted to examine the effect of sexual victimization on health risk behavior engagement. The multivariate test was significant, Pillai’s trace = .18, F(15, 903) = 3.93, p < .001, η2p = .06, observed power = 1.00, indicating significant differences among the victimization categories on the combined health risk variables. The covariates, sexual experience and social desirability were also significant, Pillai’s trace = .15, F(5, 299) = 10.26, p < .001, η2p = .15, observed power = 1.00 and Pillai’s trace = .11, F(5, 299) = 7.67, p < .001, η2p = .11, observed power = 0.99, respectively. The univariate tests were then performed, and victimization category differences were significant for drug use, problematic drinking behaviors, and sexual risk taking when the alpha was set at .01. The effect sizes ranged from small to medium. A comparison of adjusted means revealed a number of differences between the victimization groups (see Table 2). Given the significance of sexual risk taking using the full SRS score, a one-way MANCOVA was conducted to examine the effect of sexual victimization status on the five SRS subscales scores. MANCOVA results indicated significant differences among the victimization categories on the combined sexual risk subscales, Pillai’s trace = .18, F(15, 903) = 3.90, p < .001, η2p = .06, observed power = 1.00. The covariates, sexual experience and social desirability, both significantly influenced the combined dependent variable, Pillai’s trace = .19, F(5, 299) = 14.34, p < .001, η2p = .19, observed power = 1.00 and Pillai’s trace = .06, F(5, 299) = 3.46, p < .01, η2p = .06, observed power = 0.91, respectively. The univariate tests were then conducted on the dependent variables, demonstrating that victimization status differences were
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
2448
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
1
*p ≤ .05. **p ≤ .01.
— 1. Sexual victimization severity 2. Sexual experience 3. Social desirability 4. Weekly drinking 5. Problematic drinking 6. Drug use 7. Tobacco use 8. Sexual risk taking 9. Dyadic sexual desire 10. Solitary sexual desire 11. Sexual dysfunctions 12. Risk with uncommitted partners 13. Risky sex acts 14. Impulsive sexual behaviors 15. Intent to engage in sexual risk 16. Risky anal sex
Variable/Subscale
3
—
−.03 —
.29** −.16**
2
5 .24**
6 .14*
7 .42**
8 .15*
9 .08
10 .32**
11 .40**
12 .17**
13 .36**
14 .20**
15 .12*
16
—
.47** —
.34** .23** —
.17** .14* .33** —
.46**
.09 —
.27** .41**
—
.27** —
.18**
.11 —
.16**
.12*
.20** .14* .64** .13*
.20**
.28** .08 .67** .31**
—
.29** .28** .85** .23**
.16** .09 .33** .12*
.15** .21** .19** .27**
.05
—
—
.36** .02
.20**
.17**
.13*
.11 .01 .41** .17**
.06 .38**
.27**
.14*
.07
.11 .19** .46** .24**
.29** .23** .22** .13* .41** .20** .15** .28** .33** .46** .13* .10 .16** −.18** −.31** −.23** −.14* −.24** −.16** −.07 −.15* −.16** −.16** −.22** −.15** −.03 — .54** .45** .32** .34** .19** .10 .38** .28** .20** .27** .23** .06 — .47** .21** .30** .17** .10 .15** .20** .14* .31** .23** .06
.15** .26**
4
Table 1. Bivariate Correlations Among Health Risk Behaviors, Sexual Functioning Variables, and Sexual Victimization Severity.
Downloaded from jiv.sagepub.com at Bobst Library, New York University on May 29, 2015
2449
5.08 (0.43) 2.55 (0.38)a
0.59 (0.22)
0.39 (0.20) 31.06 (1.23)a 3.24 (0.52) .76 (0.12)a,b
20.82 (0.58) 2.97 (0.09) 6.51 (0.18) 0.82 (0.06) 15.80 (0.55) 4.14 (0.26)
5.40 (0.22) 4.37 (0.20)
1.14 (0.11)
0.76 (0.10) 33.18 (0.63) 3.33 (0.27) 1.12 (0.06)
Weekly drinking Problematic drinking Drug use Tobacco use Sexual risk taking Risk with uncommitted partners Risky sex acts Impulsive sexual behaviors Intent to engage in sexual risk Risky anal sex Dyadic sexual desire Solitary sexual desire Sexual dysfunctions 0.88 (0.21) 35.70 (1.24)a 3.26 (0.53) .87 (0.12)c,d
1.21 (0.22)
5.22 (0.43) 3.93 (0.39)b
20.63 (1.13) 3.16 (0.18) 6.09 (0.36)b 0.81 (0.12) 14.37 (1.07)b,d 3.13 (0.52)b
Unwanted Sexual Contact
0.93 (0.24) 31.62 (1.44) 3.26 (0.61) 1.39 (0.14)a,c
1.26 (0.25)
6.21 (0.50) 5.39 (0.45)a
19.36 (1.32) 2.69 (0.21)b 6.29 (0.42) 0.76 (0.14) 18.91 (1.25)b,c 5.12 (0.60)a
Sexual Coercion
0.82 (0.20) 34.36 (1.21) 3.56 (0.52) 1.43 (0.12)b,d
1.48 (0.21)
5.11 (0.42) 5.61 (0.38)a,b
22.44 (1.11) 3.47 (0.18)a,b 7.54 (0.35)a,b 1.00 (0.12) 19.18 (1.05)a,d 6.16 (0.51)a,b
Rape
.097***
.182***
.184***
Pillai’s Trace
1.39 3.18* 0.09 7.00***
3.03
1.24 12.08***
1.15 5.33*** 3.64** 1.08 12.59*** 11.39***
Univariate F
.01 .03 .00 .07
.03
.01 .11
.01 .05 .04 .01 .11 .10
2
ηp
Note. Table includes results from three separate multivariate analyses of covariance. All means and standard errors are adjusted for the covariates of sexual experience and social desirability. Means (SE) in the same row with the same superscripts differ at p < .05 based on Bonferroni post hoc comparisons. *p ≤ .05. **p ≤ .01. ***p ≤ .001.
20.86 (1.12) 2.56 (0.18)a 6.14 (0.36)a 0.71 (0.12) 10.73 (1.07)a,c 2.16 (0.52)a
Sample M (SE)
Variables
Nonvictim
Table 2. Multivariate Analyses of Covariance Examining Effects of Sexual Victimization Status on Health Risk Behaviors and Sexual Functioning.
2450
Journal of Interpersonal Violence 29(13)
significant for two of the five subscales, Sexual Risk Taking with Uncommitted Partners and Impulsive Sexual Behaviors, with the alpha set at .01. Effect sizes were medium in magnitude. A comparison of adjusted means revealed a number of significant differences (see Table 2).
Sexual Victimization and Sexual Functioning A one-way MANCOVA was conducted to determine the effect of sexual victimization status on sexual functioning. The multivariate test was significant, Pillai’s trace = .10, F(9, 909) = 3.37, p < .001, η2p = .03, observed power = 0.99, indicating significant differences among the victimization categories on the sexual functioning variables. The covariates, sexual experience and social desirability, were both significant, Pillai’s trace = .07, F(3, 301) = 7.78, p < 2 .001, ηp = .07, observed power = 0.99 and Pillai’s trace = .03, F(3, 301) = 3.09, p < .01, η2p = .03, observed power = 0.72, respectively. The univariate tests were then performed, and victimization category differences were significant for dyadic sexual desire and number of sexual dysfunctions with an alpha set at .02 (see Table 2), both accounting for a modest amount of the variance. A comparison of adjusted means revealed a number of significant differences. A multinomial logistic regression was then used to examine the association between level of victimization and report of each of the six types of sexual dysfunction behaviors. The overall model was significant, χ2(24, N = 309) = 88.33, p < .001, and the Nagelkerke R2 was .27 for the total model (see Table 3). Both the covariates, sexual experience, χ2(3, N = 309) = 20.55, p < .001, and social desirability, χ2(3, N = 309) = 9.59, p
