Sexual Victimization Using Latent Class Analysis

Sexual Victimization Continuum and Severity

Koss, Gidycz, and Wisniewski (1987) identified four types of unwanted sex including rape (i.e., sexual intercourse due to alcohol/drugs or physical force), attempted rape (i.e., attempted but not completed sexual intercourse by alcohol/drugs or physical force), sexual coercion (i.e., sexual intercourse due to continual arguments, pressure, or abused authority), and sexual contact (i.e., sex play due to arguments, pressure, or abused authority). Studies typically assess sexual victimization experiences on a continuum of severity, which suggest that physical force resulting in intercourse (i.e., completed rape) is the most psychologically traumatic, and that greater incidents of sexual victimization are related to lower levels of psychological and behavioral health (Classen, Palesh, & Aggarwal, 2005; Ullman, Townsend, Filipas, & Starzynski, 2007).

Although forcible rape has been considered the most severe victimization experience, substance facilitated sexual assault and verbal coercion have also been associated with deleterious outcomes. Some studies have found that forcible rape and incapacitated rape show comparable levels of psychological distress (Brown, Testa, & Messman-Moore., 2009). McCauley and colleagues (2009) found among a nationally representative sample of young women that survivors of incapacitated and substance facilitated sexual assault were comparable to other sexual assault survivors on past year major depressive episode and post-traumatic stress disorder (PTSD). Verbal coercion has also related to greater psychological distress as Norwood and Murphy (2011) found among a community sample of women, in which sexual coercion was a unique predictor of PTSD symptoms, whereas forcible rape was not. The relationship of sexual victimization diminished, however, when controlling for psychological abuse. Moreover, some studies report lower self-esteem (Testa & Dermen, 1999; Tyler, Hoyt, & Whitbeck, 1998) and greater psychological distress (Zweig, Barber, & Eccles, 1997) among women who were sexually coerced than those who were raped. Although penetration is presumed to be the most severe sexual act of victimization, some research has found poor psychological sequelae among sexual assault survivors when sexual intercourse did not occur (Resnick, Kilpatrick, Dansky, Saunders, & Best, 1993).

Sexual victimization of men has also been associated with deleterious psychosocial correlates (Larimer, Lydum, Anderson, & Turner, 1999; Walker, Archer, & Davies, 2005). Unlike the growing research on differential correlates of women’s sexual victimization, however, few studies have explored the psychological correlates of men’s victimization by tactic and sexual outcome. One notable exception is Turchick’s (2012) recent study among 302 male college students who were classified by victimization as non-victims, unwanted sexual contact, sexual coercion, and rape. Results indicated that over half of the men reported some form of sexual victimization, and men who were raped showed greater sexual risk taking and substance use behaviors than their non-victimized counterparts.

Limitations in Multivariate Explorations of Sexual Victimization

Despite the negative psychosocial outcomes associated with different forms of sexual victimization, few studies have investigated sexual victimization based on all three types of common perpetrator tactics (i.e., verbal coercion, substance facilitated assault, forcible rape) within one study. One notable exception is Abbey, BeShears, Clinton-Sherrod, and McAuslan (2004) who investigated sexual victimization based on perpetrator tactic among a community sample of women, focusing on experiences of sexual intercourse by verbal coercion, intoxicated sexual assault, and physical force. Using discriminant function analysis, participants were classified by their most severe experience based on a continuum of coercion types, victim perception, and recency of assault. Results showed common and divergent classifications based on the tactic used. Physical force had unique assault characteristics from intoxicated and verbal coercion, whereas physical and intoxicated assault showed greater similarities in perceived seriousness and severity compared to verbal coercion. Despite these differences, all survivors of sexual victimization experienced some negative outcome, with the majority of women experiencing multiple sexually coercive tactics. For example, 94% of forcible rape survivors and 88% of intoxicated sexual assault survivors also experienced verbal coercion. Brown and colleagues (2009) also explored differential consequences related to forcible rape, incapacitated rape, and verbal coercion among a large sample of community and college women. Among both samples, forcible rape victims reported the greatest psychological trauma followed by incapacitated assault victims; verbal coercion victims reported the fewest traumatic symptoms. After controlling for number of items endorsed, experiences meeting legal definitions of rape (i.e., forcible and incapacitated rape) were significantly related to greater traumatization.

These studies offer important advances to the sexual victimization literature by exploring differential correlates of psychological assessments related to the three most commonly reported and investigated sexual victimization tactics: verbal coercion, incapacitated assault, and forcible rape. However, the research lacks in that both studies explored psychosocial correlates by ranking participants a priori based on the most severe tactic experienced and only included unwanted penetration. This measurement approach and analysis does not allow for the investigation of multiple experiences within the individual and patterns of victimization, exploration of sexual behaviors other than penetration, or the psychosocial variables related to these patterns.

Personal and Group Level Variables Related to Sexual Victimization

Research suggests that certain groups with specific identifying variables might be at greater risk for sexual victimization than others. One consistent finding indicates that childhood sexual abuse survivors are more likely to experience later revictimization than their non-abused counterparts (Macy, Nurius, & Norris, 2007; Swartout, Swartout, & White, 2011; Webster, 2001).

Adolescents are also overrepresented in sexual victimization rates, representing 50% of sexual assault survivors (Catalano, 2005). Potential risk factors may include the environmental context related to adolescent development. For example, 53% of high school girls were found to have experienced sexual assault (Young, Grey, & Boyd, 2009), and those with older male partners were significantly more likely to experience unwanted or coerced sex (Marin, Coyle, Gomez, Carvajal, & Kirby, 2000). College environments have unique contexts of sexual coercion including increased expectations for sex and drinking (Abbey, 2002), and rape supportive attitudes among male dominated groups (Forbes, Adams-Curtis, Pakalka, & White, 2006; Humphrey & White, 2000). Rates as high as 69% have been documented for experiences of unwanted sexual contact in college environments (Fisher, Cullen, & Turner, 2000). Despite overrepresentation of sexual coercion in mid to late adolescence, few studies have compared the experiences of sexual coercion across adolescent contexts such as high school and college.

Although women are more likely to report sexual victimization than men, we know considerably less about male sexual victimization in comparison. Research suggests that men are more often coerced through the use of verbal and substance facilitated coercion strategies, more so than physical coercion tactics (Sorenson & Siegel, 1992; Struckman-Johnson & Struckman-Johnson, 1994). A small but nationally representative sample of male sexual assault victims reported that the majority of their experiences did not include substance use or penetration (Light & Monk-Turner, 2009). Exploring the victimization patterns among men is particularly imperative given the dearth of research in this area.

The Present Study

Research consistently documents the harmful effects of sexual victimization on psychological health. However, few studies have explored psychological correlates related to different perpetration tactics and sexual outcomes. Notable exceptions have found unique relations based on the tactic used and suggest that outcomes other than sexual intercourse relate to deleterious sequelae. This complicates the notion of the sexual violence continuum, yet studies typically classify survivors based on the most “severe” type of victimization. Macy (2008) challenges researchers to advance traditional assessments of victimization using binary yes/no measures and encourage multivariate analyses to assess victimization patterns. Understanding the unique patterns of sexual victimization and their influence on psychological adjustment is imperative in sexual violence prevention and counseling. Following Macy’s recommendations for sophisticated analyses in sexual violence research, this study expands on the work of Abbey et al. (2004) and Brown et al. (2009) by using LCA to explore sexual victimization patterns based on the three most commonly classified coercive tactics (i.e., verbal coercion, substance facilitated, physical force), while also including a range of sexual outcomes (i.e., kissing or fondling; attempted vaginal, oral, anal intercourse; completed vaginal, oral, anal intercourse). LCA is a person-centered approach that allows researchers to characterize participants by various indicators, in this case, characteristics of sexual victimization, and explore to these classes. Such analyses methodologically advance one-dimensional dichotomous examinations of sexual coercion, based on presumed severity, by offering multi-dimensional assessments based on various indicators.

The purpose of the present study was to explore sexual victimization patterns through LCA among young men and women and the related psycho-behavioral correlates of these classes. In accordance with expert conceptualizations of adolescent development spanning into one’s early 20s (Arnett, 2000) and the disproportionate rate of sexual coercion among adolescents (Catalano, 2005), this study focused on mid and late adolescents and includes a sample of high school and college students. The following research questions and related hypotheses guided our study: (1) What patterns of sexual victimization emerge based on tactic used and sexual behavior? We hypothesized that the best fitting model would consist of multiple latent classes of sexual coercion based on the type of tactic(s) used and sexual behavior experienced, including at least one class with no victimization and one class with multiple victimization tactics and sexual behavior outcomes. (2) To distinguish classes based on prior abuse and investigate potential gender and developmental differences in victimization patterns, we asked the following: Are classes characterized by childhood sexual abuse, gender, and adolescent development? Based on the extant literature, we hypothesized that the class with multiple forms of victimization would show the greatest likelihood for childhood sexual abuse history, that college students would show greater representation in classes with substance use given increased alcohol and drug use among college populations, and that men would be less represented in classes meeting criteria for forcible rape (i.e., physical force resulting in intercourse). Finally, to explore the potential influence of various victimization patterns on psychological and behavioral outcomes, the following question was asked: (3) How do classes predict psychological and behavioral outcomes? We focus specifically on the psycho-behavioral indicators of self-esteem, psychological distress, and sexual risk taking, given that research consistently demonstrates the correlation of these variables with sexual victimization. Previous findings support evidence of differential relations in self-esteem and psychological distress based on the tactic used by the perpetrator; and exploring the relation to sexual risk taking is important in assessing adolescent risk behaviors and of particular relevance, given the overrepresentation of adolescents in sexually transmitted infections (Centers for Disease Control and Prevention, National Center for HIV/AIDS Prevention, Divisions of HIV/AIDS Prevention, 2011; Weinstock, Berman, & Cates, 2004).

Methods

Results

Research Question 1: Patterns of Sexual Victimization

LCA was conducted to determine the optimal number of classes based on tactic used (i.e., verbal, substance, physical) and sexual behavior (i.e., kissing/fondling, attempted intercourse, completed intercourse) of sexual victimization experiences. LCA fit indices determined a four-class solution as the best fitting model (see Table 1). Figure 1 summarizes the characteristics and prevalence of the four identified latent classes. Approximately half of the sample experienced some form of sexual victimization, with 54% classified as non-victims, consistent with our hypothesis of at least one non-victimized class. Of those who were victimized, the most common classification was the manipulated and forced fondle/intercourse class (27%). This class was characterized as having a high probability of verbal coercion (1.0), moderate physical coercion (0.64), and resulting most often in completed intercourse (.48), followed by kissing or fondling (.34). Equal proportions of the sample were classified in the forced fondling (9.5%) class and the poly-victimization class (9.5%). Forced fondling was characterized by a high probability of experiencing physical tactics (1.0) that overwhelmingly resulted in kissing or fondling (0.67). Consistent with our hypothesis of a class with multiple victimization experiences, the final class, poly-victimization, was characterized by a high probability of experiencing all tactics: verbal coercion (0.75), substance facilitated (1.0), and physical force (0.73) and that overwhelmingly resulted in completed sexual intercourse (0.68).

OPEN IN VIEWER

Table 1.

Model Fit indices for 1 to 5 Class Solutions of Types and Severity of Sexual Coercion Experience.

	AIC	BIC	Adjusted. BIC	LRB LRT p-value	Entropy
1-class solution	3,599.24	3,626.15	3,607.10	—	—
2-class solution	2,543.14	2,601.44	2,560.16	0.000	1.000
3-class solution	2,503.38	2,593.07	2,529.57	0.000	0.958
4-class solution	2,482.65	2,603.74	2,518.01	0.000	1.000
5-class solution	2,475.69	2,628.16	2,520.21	0.016	0.938
6-class solution	2,488.15	2,672.02	2,541.84	0.627	0.928

Note. Bold indicates best fit: The 4-class solution had the lowest adjusted BIC of eligible classes and the Bootstrap LRT indicated the 4-class solution provided a better fit than the 5-class solution. All entropy ratings indicate an acceptable fit. AIC = Akaike information criterion; BIC = Bayesian information criteria; LRP LRT = Lo-Mendell-Rubin Adjusted Likelihood Ratio Test.

OPEN IN VIEWER

Figure 1.

Defining characteristics and prevalence of each class.

Research Question 2: Class Characteristics

Latent class regression analyses were used to estimate the predictors for class membership in relation to participants’ childhood sexual abuse history, adolescent development, and gender (see Table 2). Consistent with our hypothesis, participants in the poly-victimization class showed significantly higher rates of childhood sexual abuse compared to all other classes. A small number of participants reported at least once incident of childhood sexual abuse: 20% of women and 10% of men. Manipulated and forced fondling/intercourse reported significantly greater childhood sexual abuse than non-victims. The poly-victimization class also showed significantly greater representation of women compared to all other classes, which were equivalently distributed between men and women. This supports our hypothesis that men would be underrepresented in classes with forced intercourse, yet also contradicts it as the manipulated and forced intercourse class showed equivalent proportions of men and women. Also challenging our hypothesis, there was little distinction of the classes by adolescent development. Educational context did not distinguish any of the classes; however, participants in the poly-victimization class were significantly older than non-victims.

OPEN IN VIEWER

Table 2.

Class Characteristics: Proportion of Childhood Sexual Abuse, Gender, and Adolescent Development.

	Poly-Victimization	Force Fondling	Manipulated and Forced	Non-Victim
CSA	0.29a,b,c	0.08a	0.15b	0.08c,b
Gender	0.73a,b,c	0.42a	0.49b	0.54c
Education level	0.80	0.68	0.97	0.76
Mean age	18.82a	18.65	18.52	18.38a

Note. Values are significantly different from values with the same subscript in the same row at p < .05. CSA = childhood sexual abuse.

Research Question 3: Predicting Psycho-behavioral Correlates

Results of the equality tests of means indicated that there were significant differences in psycho-behavioral correlates related to class membership (Table 3). Consistent with our hypothesis, participants in the poly-victimization class showed the greatest level of psychological-behavioral consequences. This class reported significantly lower levels of self-esteem, greater psychological distress, and higher sexual risk taking, than the forced fondling, manipulated and forced fondling/intercourse, and non-victim classes. The manipulated and forced fondling/intercourse class reported significantly lower levels of self-esteem than the non-victims class, χ²(1, 657) = 8.53, p < .01. The forced fondling, χ²(1, 657) = 10.89, p < .01, and manipulated and forced fondling/intercourse χ²(1, 657) = 72.30, p < .001, classes engaged in significantly higher sexual risk taking than the non-victim class.

OPEN IN VIEWER

Table 3.

Tests of Mean Differences on Class Membership and Psycho-behavioral Outcome.

	Poly-Victimization		Forced Fondling			Manipulated and Forced			Non-Victim
	M	None χ2 a	M	None χ2	Poly χ2 b	M	None χ2 a	Poly χ2 b	M
Self-esteem	30.72	11.73**	34.01	1.95	14.16***	34.44	8.53**	26.73***	33.05
Psychological distress	15.50	20.60***	13.43	1.56	5.95*	12.37	0.43	22.07***	12.62
Sexual risk taking	18.30	59.40***	12.29	10.89**	8.00**	14.37	72.30***	6.00*	6.84

a

Values are based on parameterization of non-victim class as the reference group.

b

Values are based on parameterization of poly-victimization class as the reference group; tests of mean differences were not significant between forced fondling and manipulated and forced classes.

*

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

Discussion

Research in sexual violence typically classifies participants based on the most severe assault experienced, without taking into consideration that experiences considered less severe show significant relations to psychological consequences. The aim of this study was to fill a gap in the literature by identifying sexual victimization patterns based on type of tactic used and outcome of sexual behavior, and explore related psycho-behavioral correlates. Through using sophisticated analyses to statistically classify participants based on complex victimization patterns, we identified the presence of four distinct classes of sexual victimization. Approximately half of the sample experienced some form of sexual victimization, with the most frequent pattern emerged being the manipulated and forced intercourse class. This finding suggests that the most common victimization for the young men and women in this sample occurred through a combination of verbal coercion strategies and physically forceful tactics, which resulted in sexual intercourse about half of the time. Participants were equally classified in the forced fondling class and the poly-victimization class; thus, roughly a quarter of the sample experienced kissing and fondling due to physical force, or were victims of multiple tactics to have unwanted sexual intercourse. These findings offer a more complete picture of sexual victimization patterns by acknowledging intersecting tactics and sexual outcomes without prioritizing physical violence and intercourse in sexual violence research. Results challenge the traditional assessment of categorizing victimization based on the most “severe” type, as this study suggests that survivors typically experience more than one tactic with an array of sexual outcomes.

Consistent with our hypothesis, participants in the poly-victimization class demonstrated significantly poorer psychological and behavioral outcomes compared to all other classes, with the non-victimized class demonstrating the least distress. This supports the previous literature, which has found that greater frequency of sexual victimization (Classen et al., 2005) and multiple types of intimate partner violence tactics (Cavanaugh et al., 2012) relate to higher levels of psychological distress. Poly-victims were also most likely to experience childhood sexual abuse and had higher representation of women than men. Classen and colleagues’ review notes that women who experience childhood sexual abuse are two to three times as likely to experience adult sexual assault.

There were fewer psycho-behavioral consequences related to the other victimization classes, forced fondling and manipulated and forced intercourse. One interesting finding that is consistent with the literature was the significantly lower self-esteem rates for manipulated and forced intercourse compared to non-victims. Given the high probability of experiencing verbal coercion tactics within this class, these results partially support previous findings related to verbal coercion and self-esteem (Tyler et al., 1998) and thus may be less likely to ward off unwanted advances, or are easier targets for manipulation. This class did not significantly differ from the forced fondling class, however, and showed higher self-esteem than the poly-victimization class, contrary to the research suggesting that verbal coercion victims have lower self-esteem than forcible rape survivors (Testa & Dermen, 1999). The experience of multiple types of victimization may have a powerful impact on self-worth. A potential mediating effect may partially explain these findings through the role of attributions of blame. Brown and colleagues (2009) found that participants experiencing verbal coercion showed significantly higher levels of self-blame than survivors of incapacitated assault or forcible rape; they also found that a greater number of victimization experiences related to cognitive distortions. This relationship may in turn impact self-esteem. Our interpretation is speculative, however, as we did not explore cognitive distortions; thus, we encourage future research to investigate mediating relationships. All victimization classes reported greater sexual risk taking than the non-victimized class. The relation between unwanted sexual experiences and risky sexual behaviors has been well documented (Biglan et al., 1995; Zweig, Sayer, Crockett, & Vicary, 2002) and recent longitudinal research provides initial evidence that sexual victimization leads to later sexual risk taking (Lang et al., 2011). Future research should expand this literature by investigating longitudinal outcomes as they relate to victimization patterns and characteristics.

Some unexpected yet interesting findings arose as well. There were no gender differences between the forced fondling, manipulated and forced intercourse, and non-victim classes. These results partially support previous research, which found that men more often experience verbal coercion strategies, and are less likely to experience penetration with regard to the forced fondling class (Light & Monk-Turner, 2009). However, the manipulated and forced intercourse class challenges assumptions that men do not experience victimization through physical means or that victimization does not result in completed intercourse, as the probability of sexual intercourse was about 50% for this class. These results offer important additions to the literature on the sexual victimization of men, and suggest similarities between young men’s and women’s victimization histories. Our hypothesis regarding adolescent development was weakly supported, as classes showed no significant difference based on educational level or age with the exception of poly-victimization being significantly older than non-victims. Given the small sample of high school students, results should be interpreted with caution and future research would benefit from continued exploration on developmental differences in sexual victimization.

Limitations

With the strengths of this study, results should be interpreted in the context of this study’s limitations. Although our operationalization of sexual victimization included the tactic used and sexual behavior outcome, we did not assess for the number of incidents within the individual nor types of experiences within an incident. Thus, the relationship between lifetime and incident specific victimization is unknown. Given the focus of this study on sexual victimization and psycho-behavioral outcomes, we did not explore other risk factors related to behavioral and psychological health variables and thus could not control for these in the analyses. Limitations common in retrospective and self-report assessments arise as some participants may not have accurately recalled their experiences or may have withheld disclosing private information. Finally, additional concerns include the cross-sectional design and limited geographic representation, preventing cause-and-effect interpretations and limiting generalizability.

Conclusion

Despite these limitations, this study offers important contributions in understanding patterns of sexual victimization among adolescents and related psycho-behavioral outcomes. The large sample size and statistical sophistication provide advances to the sexual violence literature. Findings reveal four distinct patterns of sexual victimization that relate uniquely to psychological and behavioral health indicators. Our findings offer important advances in the assessment of sexual victimization by exploring classifications based on intersecting experiences, and highlight some of the weaknesses and potential fallacies in classifying survivors based on presumed severity. Relations between victimization, health risk behavior, and self-esteem highlight the importance of unwanted sexual experiences across the spectrum and can better inform prevention efforts by acknowledging aspects such as subtle manipulation tactics and unwanted kissing or fondling in interventions. Assessing for patterns of sexual victimization experiences allows clinicians to form a more complete understanding of adolescent victimization while also raising awareness of potential psychological and behavioral corollaries.