Abstract
The current study sought to examine whether symptoms of posttraumatic stress disorder (PTSD), adolescent sexual assault (ASA), and drinking motivations (e.g., drinking to regulate emotional experiences) mediate the relationship between a history of childhood sexual abuse (CSA) and subsequent alcohol-related problems among college women. Participants were 579 female students at a Midwestern university. Participants were recruited as part of a larger longitudinal study that investigated risk and resiliency factors related to sexual revictimization. Using a serial mediation model, the current study found that the proposed constructs mediated the relationship between CSA and subsequent alcohol-related problems via two separate paths. In one path, CSA was associated with PTSD, which in turn predicted drinking to regulate emotional experiences, which then was related to alcohol-related problems in adulthood. In the second path, CSA was related to ASA, which in turn predicted drinking to regulate emotional experiences, which then was related to alcohol-related problems in adulthood. These results suggest that individuals with a history of CSA are more likely to experience both revictimization in adolescence and PTSD symptoms in adulthood, which may lead to alcohol-related problems via drinking to regulate emotional experiences. These findings suggest the importance of incorporating skills training in adaptive emotion regulation strategies into treatment for individuals with a history of CSA and ASA.
Keywords
Childhood sexual abuse (CSA) is highly prevalent in the general population. Approximately 7% of men and 20% of women are estimated to experience CSA, and 20% to 27% of surveyed women reported experiencing an unwanted sexual experience prior to the age of 18 (Finkelhor, Hotaling, Lewis, & Smith, 1990; Pereda, Guilera, Forns, & Gomez-Benito, 2009). CSA can take many forms, including asking or pressuring a child to perform sexual acts, exposing oneself to a child for sexual gratification, engaging in physical sexual contact with a child, and involving a child in the production of pornography. A history of CSA increases one’s risk for a range of adverse outcomes, including depression, disordered eating, suicide attempts, and posttraumatic stress disorder (PTSD; Chen et al., 2010). In addition, CSA has been linked to sexual revictimization in adolescence (Krahe, Scheinberger-Olwig, Waizenhofer, & Koplin, 1999; Wager, 2012). For example, survivors of CSA are more likely to report unwanted sexual contact in adolescence compared with those without a history of CSA (Krahe et al., 1999; Wager, 2012).
In addition to the increased incidence of psychopathology and revictimization following CSA, extant literature has demonstrated a consistent link between CSA and increased alcohol consumption (e.g., Messman-Moore & Long, 2003; Sartor, Agrawal, McCutcheon, Duncan, & Lynskey, 2008). Specifically, survivors of CSA are more likely to begin drinking at a younger age and report more problems related to alcohol use, such as alcohol use disorders (Brems, Johnson, Neal, & Freemon, 2004; Kilpatrick et al., 2000; Zlotnick et al., 2006). Furthermore, Spak, Spak, and Allebeck (1998) found that sexual abuse before the age of 13 was a particularly robust predictor of alcohol use disorders, in addition to early age of first alcohol intoxication (i.e., before the age of 15) and early onset of psychological problems.
Because a strong relationship has been established between CSA and subsequent alcohol-related problems, researchers have attempted to develop theoretical models that help explain this phenomenon. For example, Polusny and Follette (1995) theorized that survivors of CSA drink alcohol as a means of coping with their internal psychological distress. More specifically, they suggested that internal psychological distress acts as a mediator between CSA and drinking to cope with that internal psychological distress (Polusny & Follette, 1995). Relatedly, Cooper and colleagues (Cooper, Agocha, & Sheldon, 2000; Cooper, Frone, Russell, & Mudar, 1995) theorized that individuals engage in maladaptive behaviors and activities (such as drinking alcohol) to regulate both negative and positive emotional experiences. For example, survivors of CSA may drink alcohol to mask or reduce negative emotions and/or elicit or heighten positive emotions. Similar research has demonstrated that survivors of CSA struggle with emotional management, which may include an inability to adaptively regulate both positively and negatively valenced internal experiences (Cloitre, Cohen, & Koenen, 2006). Thus, survivors of CSA may utilize alcohol consumption as a method of both positive and negative emotional management.
Although individuals with a history of CSA may be a population at increased risk of alcohol-related problems, relatively few studies have investigated possible mediators through which CSA might be associated with subsequent alcohol problems. In one of the few mediational studies on this topic, Grayson and Nolen-Hoeksema (2005) examined whether motives for alcohol consumption (e.g., using alcohol to reduce negative affect and/or increase positive affect) mediated the relationship between CSA and alcohol-related problems in adult women. The authors proposed two separate models to help explain why women with a history of CSA exhibit alcohol-related problems in adulthood: the distress coping model and the emotion regulation model. The distress coping model suggests only one drinking motive (i.e., drinking alcohol to reduce negative affect) mediates the relationship between CSA and alcohol-related problems. The emotion regulation model suggests two drinking motives (i.e., drinking alcohol to reduce negative affect and to enhance positive affect) mediate the relationship between CSA and alcohol-related problems. Both models suggest that CSA is related to general psychological distress, which in turn is related to drinking motive(s). The authors found that both the enhancement motive and the coping motive partially mediated the relationship between CSA and alcohol-related problems. Goldstein, Flett, and Wekerle (2010) also examined the role of drinking motives in childhood maltreatment and subsequent drinking problems in both male and female college students. They found that, for female participants, drinking to cope mediated the relationship between childhood maltreatment and alcohol-related problems, and drinking to enhance did not (Goldstein et al., 2010).
Although the aforementioned studies serve as invaluable contributions to the scant literature on determining mediators between CSA and subsequent alcohol-related problems, they are limited in several ways. First, it is impossible to make causal inferences using correlational data. Second, neither Grayson and Nolen-Hoeksema (2005) nor Goldstein and colleagues (2010) included adolescent sexual assault (ASA) in their model, despite the established relationship between CSA and subsequent ASA (see Arata, 2002, for a review). Prior research has demonstrated that sexual revictimization acts as a mediator between childhood abuse and subsequent alcohol-related problems (Najdowski & Ullman, 2009). Therefore, failing to include multiple victimizations may falsely attribute results to the experience of CSA alone. Finally, very few studies have examined the role of other psychological symptoms, such as PTSD, in the relationship between CSA and alcohol-related problems, despite the well-established relationship between CSA and PTSD symptoms (e.g., Ullman, Najdowski, & Filipas, 2009). Therefore, the aim of the current study was to examine the relationship between CSA and subsequent alcohol-related problems while addressing several gaps in the literature. Expanding upon the model proposed by Grayson and Nolen-Hoeksema, the present study combined both drinking motivation variables to create the latent variable of “drinking to regulate emotional experiences.” Both drinking to reduce negative affect and drinking to enhance positive affect represent ways in which individuals regulate internal emotional experiences (Cooper, 1994).
The present study also expanded upon Grayson and Nolen-Hoeksema’s (2005) model by including PTSD symptoms (instead of general psychological distress) as a potential mediator between CSA and drinking motivations. Specifically, it was hypothesized that CSA would be related to PTSD symptoms, which in turn would predict drinking to regulate emotional experiences. Furthermore, drinking to regulate emotional experiences and ASA were included as potential mediators of the relationship between CSA and subsequent alcohol-related problems in a longitudinal model.
Method
Participants and Procedures
Participants in the current study were part of a larger longitudinal study assessing risk and resiliency factors related to sexual revictimization. Because research has shown that sexual revictimization is more likely to affect women than men (e.g., Pereda et al., 2009), only female students were recruited for the study. Participants completed three waves of a longitudinal study that received approval from the Northern Illinois University (NIU) Institutional Review Board. Participants who completed the first session (Time 1; n = 1,045) were female students in an introductory psychology class recruited from a mass testing pool. Prerequisites for Time 1 were that participants be women 18 years of age or older and fluent in English; participants were not recruited based on the presence or absence of trauma history. At Time 1, participants completed measures of lifetime exposure to potentially traumatic events (e.g., CSA and ASA), substance use, alcohol-related problems, and PTSD symptoms. Measures were completed online and took approximately 1 hr to complete. Upon completion, participants received partial course credit for their participation. Data for Time 1 were collected between September 2006 and February 2008.
Of the 1,045 participants from Time 1, 812 (78%) were invited via email to complete an additional battery of measures online (these individuals consented to future contact at Time 1 and were current students at the university at Time 2). Participants who completed Time 2 received US$40 compensation or were offered partial course credit (if they were still enrolled at NIU at the time). Of the 812 participants invited, 691 (85%) completed the assessment (Time 2). Data for Time 2 were collected between February 2008 and September 2008. The average time between the Time 1 and Time 2 assessments was 27 weeks. In September 2008, participants who consented to future contact from the Time 2 sample were invited via email to complete another follow-up survey online. Five hundred eighty-eight (85%) participants from the Time 2 sample completed the third session (Time 3). Participants who completed Time 3 received US$40 compensation. Data for Time 3 were collected between September 2008 and March 2009. Out of the 588 participants from Time 3, 579 participants were included in the present analyses. Nine participants were excluded from the main analyses due to missing data on the outcome variable (i.e., alcohol-related problems).
It is important to recognize that when conducting longitudinal research, subsequent traumatic events can occur. Participants who were enrolled in the current study experienced a campus shooting between Time 1 and Time 2; therefore, exposure to the mass shooting, as well as psychological distress related to the shooting (e.g., PTSD symptoms stemming from the campus shooting), was controlled for in the current study.
The average age of participants was 18.8 (SD = 1.66) at Time 1, 19.6 (SD = 2.57) at Time 2, and 20.1 (SD = 2.12) at Time 3. Among the final sample (N = 579), 416 (71.6%) participants self-identified as non-Hispanic White, 165 (28.4%) participants self-identified as any Other race/ethnicity category, and 7 (1.2%) participants preferred not to respond to the question. A majority of participants (n = 436; 74%) were freshmen at Time 1. One hundred seventy-six participants (29.9%) reported their family income as US$80,000 or more, 77 (13.1%) reported US$60,000 to US$69,000, 62 (10.5%) reported US$50,000 to US$59,000, 47 (8.0%) reported US$40,000 to US$49,000, 46 (7.8%) reported US$70,000 to US$79,000, and 119 (20.3%) reported less than US$40,000. Fifty-two participants chose not to respond to the family income question. In regard to sexual orientation, a majority of participants (n = 483, 82.1%) self-identified as 100% heterosexual (i.e., attracted to members of the opposite sex only).
Measures
Traumatic Life Events Questionnaire (TLEQ)
The TLEQ (Kubany, Haynes, et al., 2000) is a 22-item self-report measure that assesses exposure to a broad range of potentially traumatic events (including CSA and ASA). Respondents are offered seven fixed-choice options to each type-of-event question: never, once, twice, 3 times, 4 times, 5 times, and more than 5 times. The TLEQ was administered at Time 1 and was used to dichotomously measure CSA and ASA. Items on the TLEQ pertaining to instances of CSA and ASA were worded as follows: “Before your 13th birthday: Did anyone—who was at least 5 years older than you—touch or fondle your body in a sexual way or make you touch or fondle their body in a sexual way?” and “After your 13th birthday and before your 18th birthday: Did anyone touch sexual parts of your body or make you touch sexual parts of their body—against your will or without your consent?” CSA was defined in the current study as (a) sexual contact before the age of 13 by a perpetrator at least 5 years older or (b) sexual contact against one’s will or without consent before the age of 15. ASA was defined as any sexual contact against one’s will or without consent between the ages of 15 and 18. The TLEQ has demonstrated strong test–retest reliability, as well as good content validity (Kubany, Haynes, et al., 2000).
Because previous literature suggests that CSA is correlated with other forms of childhood maltreatment (such as childhood physical abuse [CPA] and witnessing family violence in childhood; Anda et al., 2006), the current study assessed prevalence rates and bivariate correlations among all forms of childhood maltreatment as measured by the TLEQ. Items on the TLEQ pertaining to instances of CPA and witnessing family violence were worded as follows: “While growing up, were you physically punished in a way that resulted in bruises, burns, cuts, or broken bones?” and “While growing up, did you see or hear family violence (e.g., your father hitting your mother, any family member beating up or inflicting bruises, burns, or cuts on another family member)?” Eighty-nine participants (12.9%) endorsed a history of CPA, while 227 (32.96%) endorsed witnessing family violence during childhood.
Distressing Events Questionnaire (DEQ)
The DEQ (Kubany, Leisen, Kaplan, & Kelly, 2000) is a self-report measure used to assess PTSD according to the Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM-IV-TR; American Psychiatric Association, 2000). When combined with the TLEQ, the DEQ provides a trauma history screen as well as a measure of PTSD symptoms and severity. The DEQ was designed to assess the three symptom clusters of PTSD (i.e., reexperiencing, avoidance, hyperarousal) via 17 items. Items are rated on a scale of 0 (absent or did not occur) to 4 (present to an extreme or severe degree). According to Kubany, Leisen, et al. (2000), a total score of 18 or above on the DEQ corresponds to significant symptoms of PTSD among women. In the current study, 131 (22.6%) women reported probable PTSD at Time 1 according to the cutoff score determined by Kubany, Leisen, et al. The DEQ has shown good test–retest reliability, excellent internal consistency, and good convergent and discriminant validity (Kubany, Leisen, et al., 2000).
In the current study, the DEQ was measured at Time 1 and Time 2. A total score of PTSD symptoms was obtained by summing the responses to the number of current symptoms endorsed by participants. The Time 1 assessment of the DEQ was used to assess PTSD symptoms associated with participants’ previously reported traumatic event (e.g., CSA). If participants endorsed exposure to multiple potentially traumatic events on the TLEQ, they were asked to respond to items on the DEQ according to the traumatic event that they endorsed as the most distressing. The Time 2 assessment of the DEQ was used as a covariate to control for PTSD symptoms related to the campus shooting. 1 All DEQ cluster scores evidenced adequate internal consistency across administrations (α values ranging from .82 to .90).
Physical exposure to mass shooting (PhysEMS)
Items used to assess PhysEMS were adapted from a measure used to assess physical exposure to the mass shooting at Virginia Tech (Littleton, Grills-Taquechel, & Axsom, 2009). Participants were asked to respond to dichotomous questions assessing proximity (e.g., “Were you in Cole Hall at the time of the shootings?”) and severity (e.g., “Do you know anyone that was wounded in the shootings?”) of exposure to the campus shooting event. PhysEMS items were administered at Time 2 and a total score was calculated by summing across the 12 items with a possible range of 0 to 12. The PhysEMS was included in analyses to control for exposure to the campus shooting.
Young Adults Alcohol Problems Screening Test (YAAPST)
The YAAPST (Hurlbut & Sher, 1992) is a self-report measure that assesses college-student alcohol problems and subsequent problematic behavior. The YAAPST has shown to have good internal consistency and good concurrent validity with measures of drinking, alcohol expectations, and alcohol abuse/dependence symptoms (Hurlbut & Sher, 1992). A subset of five items from the YAAPST was utilized in the current study to assess consequences of alcohol use (i.e., “Have you gotten into trouble at work or school because of drinking?” “Has your drinking created problems between you and your boyfriend/girlfriend (spouse) or another near relative?” “Have you neglected your obligations, your family, your work, or schoolwork for two or more days in a row because of your drinking?” “Have you awakened the morning after a good bit of drinking and found that you could not remember a part of the evening before?” and “Has your drinking gotten you into sexual situations which you later regretted?”). Participants indicated if these consequences had happened not at all, one, two, or three times since their last interview. Responses were dichotomized as ever = 1 versus never = 0 and summed to create a count score of self-reported alcohol problems at both Time 1 and Time 3.
Motivations for alcohol use
The Motivations for Alcohol Use Scale (MAU; Cooper, 1994) is a 20-item self-report questionnaire aimed to assess four separate motivations for drinking alcohol. The MAU has demonstrated good construct validity and has shown to be a reliable measure of drinking motivations (Cooper, 1994). These motivations include both internal/external and positive/negative reinforcement dimensions: the need to be sociable (external/positive), the need to regulate negative affect (e.g., coping; internal/negative), enhancement (e.g., to get high; internal/positive), and conformity (external/negative). Participants rate their relative frequency of drinking for each of the indicated reasons using a 5-point Likert-type scale ranging from 1 (almost never/never) to 5 (almost always/always). Each subscale consists of 5 items. Subscale scores are computed by averaging the scores across the 5 items on each subscale (Cooper, 1994). Only the internally focused subscales (Coping, for example, “because it helps you when you feel depressed or nervous”; and Enhancement, for example, “because it’s exciting”) were measured in the larger longitudinal study on sexual revictimization, and therefore only the internal subscales could be included in the current study. As explained by Cooper (1994), both drinking to reduce negative affect and drinking to enhance positive affect represent ways in which individuals regulate internal emotional experiences. Therefore, in the current study, we combined both variables to create the latent variable of drinking to regulate emotional experiences. Drinking motivations were assessed at Time 3. The internal consistency estimate for drinking to regulate emotional experiences was alpha = .94.
Analytic Strategy
Data screening was performed using IBM SPSS Statistics 20.1 statistical software. Data were inspected for any out-of-range values or univariate outliers. No out-of-range values were found; however, nine cases contained influential age outliers and therefore those values were truncated to the next highest value plus one unit. No multivariate outliers were detected. All means and standard deviations were assessed and found to be plausible. As stated previously, the current study created the latent variable of drinking to regulate emotional experiences. To create a latent factor that was not locally underidentified, item parceling (i.e., an atheoretical technique in which a measure is broken up into a smaller number of groups [parcels] by randomly combining two or more individual items; Bandalos, 2008) was used to create two indicators from the drinking to cope subscale and two indicators from the drinking to enhance subscale of the MAU, for a total of four indicators.
Main analyses were performed in MPlus 7.11 (Muthén & Muthén, 1998-2012) using maximum likelihood parameter estimation. Because the primary outcome variable was a count variable (YAAPST scores), a zero-inflated Poisson model with a robust maximum likelihood estimator and Monte Carlo numerical integration was used, as outlined by Muthén and Muthén (1998-2012). First, the direct effects of CSA, ASA, Time 1 PTSD symptoms, drinking to regulate emotional experiences, and alcohol-related problems were tested. Next, the indirect effect of CSA on Time 3 alcohol-related problems (i.e., YAAPST scores) was modeled using a serial mediational model analysis (i.e., multiple-step multiple mediation; see Hayes, Preacher, & Myers, 2010), which assumes that mediators causally influence one another.
Pearson correlations were used to assess bivariate associations between potential covariates, CSA, ASA, Time 1 PTSD symptoms, drinking to regulate emotional experiences, and Time 3 alcohol-related problems (see Table 1). Witnessing family violence, CPA, CSA, ASA, and race/ethnicity were represented as dichotomous variables, PTSD symptoms and drinking to regulate emotional experiences were represented as continuous, and Time 1 and Time 3 alcohol-related problems were represented as count variables. All abuse variables were coded as 0 = no abuse and 1 = history of abuse, while race/ethnicity was coded as 1 = non-Hispanic White and 0 = all Other races/ethnicities. The associations between covariates and the variables of interest were examined before conducting the mediation analysis. The covariates that were significantly associated with the main variables of interest were entered as control variables in the subsequent analysis. In addition, to account for the campus shooting that occurred during the longitudinal data collection period, level of exposure to the shooting and shooting-related PTSD symptoms were explored as potential predictors of all outcome variables in the present analyses. More specifically, the time elapsed between Time 1 and the shooting ranged from a few hours to 74 weeks, with an average of 27 weeks (SD = 21). The Time 2 survey was launched 17 days postshooting, and the average time between the shooting and completion of the Time 2 survey was 27 days (SD = 12). Level of shooting exposure and shooting-related PTSD symptoms were not significantly correlated with the main outcome variables in the model and thus were not included as covariates in the mediation analysis.
Descriptive Statistics and Bivariate Correlations Among All Study Variables.
Note. Race was coded as 1 = non-Hispanic White and 0 = all Other race/ethnicity categories; CPA, witnessing family violence, CSA and ASA were coded as 1 = history of abuse and 0 = no abuse history; Drinking to Regulate = mean score of the drinking to cope and drinking to enhance variables. T1 = Time 1. CPA = childhood physical abuse. CSA = childhood sexual abuse. ASA = adolescent sexual assault. PTSD = posttraumatic stress disorder symptoms. T3 = Time 3.
p < .05. **p < .01.
Results
To assess differences due to attrition between Time 1 and Time 2, those responding at Time 2 (n = 691) were compared with eligible nonresponders (n = 121) on demographics and variables measured at Time 1. Non-Hispanic White participants were more likely to complete Time 2 than participants from all Other race/ethnicity categories, χ2(1, N = 793) = 5.14, p < .05. To assess differences due to attrition between Time 2 and Time 3, those responding at Time 3 (n = 588) were compared with those who did not complete Time 3 (n = 103) on demographics and variables measured at Time 2. Participants who self-identified as non-Hispanic White were more likely to complete Time 3 than participants from all Other race/ethnicity categories, χ2(1, N = 674) = 12.4, p < .01. No other significant differences emerged.
One hundred fifty-one (21.9%) participants reported a history of CSA, while 94 (13.6%) participants reported a history of ASA. A chi-square analysis was computed to examine the overlap between CSA and ASA to establish the presence of revictimization. The association between CSA and ASA was significant, χ2(1, N = 650) = 268.34, p < .001, with 56.08% of CSA victims also reporting ASA.
Correlations among study variables are listed in Table 1. Of note, other forms of childhood maltreatment (CPA and witnessing family violence) were positively correlated with CSA and ASA. Furthermore, CPA was positively correlated with Time 1 PTSD, as well as Time 1 and Time 3 alcohol-related problems, whereas witnessing family violence was positively correlated with Time 1 PTSD and Time 1 alcohol-related problems. Because of these significant correlations, CPA and witnessing family violence in childhood were included as possible covariates in the mediation model. In addition, race/ethnicity was positively associated with drinking to regulate emotional experiences and Time 1 and Time 3 alcohol-related problems; therefore, race/ethnicity was also included as a covariate in the analyses.
Results also indicated that CSA was positively associated with ASA and Time 1 PTSD symptoms, such that individuals who endorsed CSA were especially likely to report ASA and endorse greater PTSD symptoms. ASA was positively related to several variables, including Time 1 PTSD symptoms, drinking to regulate emotional experiences, and Time 1 and Time 3 alcohol-related problems. Time 1 PTSD symptoms were also positively related to drinking to regulate emotional experiences and alcohol-related problems at Time 1 and Time 3.
Next, tests of significance for individual paths in the hypothesized model were tested. More specifically, a saturated path analysis was conducted in which CSA, ASA, Time 1 PTSD symptoms, and Time 1 and Time 3 alcohol-related problems were represented as observed variables in the model, and drinking to regulate emotional experiences was represented as a latent variable. Race/ethnicity, witnessing family violence in childhood, and CPA were entered as covariates (see Figure 1 for a diagram of the saturated model). The full model was compared with a more parsimonious model trimmed at p < .10, and the difference between the full and trimmed models was evaluated using the Bayesian information criterion (BIC; Raftery, 1995). Under Monte Carlo integration, there is no single referent covariance matrix, which precludes calculation of an overall model chi-square and chi-square-based fit indices. BIC provides a quantitative index of the extent to which each model maximizes correspondence between the observed and predicted variances and covariances, while minimizing the number of parameters. Based on Raftery’s (1995) guidelines, the difference between the full and trimmed models (i.e., BIC difference > 10) was in the direction of favoring the more parsimonious model. Accordingly, significant paths and correlations associated with the trimmed model (with corresponding unstandardized coefficients and critical ratio values) are depicted in Figure 2. With categorical outcomes and maximum likelihood estimation, means, variances, and covariances are not sufficient statistics for model estimation. Because of this, chi-square and related statistics are not available.
Saturated path model.
Trimmed path model of the indirect effects of ASA, Time 1 PTSD symptoms, and Time 3 drinking to regulate (drinking to regulate emotional experiences) on the relationship between CSA and subsequent alcohol-related problems (BIC difference between full and trimmed models > 10; N = 579).
CSA was directly associated with Time 1 PTSD symptoms, which in turn predicted drinking to regulate emotional experiences. Time 1 PTSD symptoms also predicted Time 1 alcohol-related problems, which in turn predicted Time 3 alcohol-related problems. Furthermore, Time 1 PTSD symptoms directly predicted Time 3 alcohol-related problems. CSA was directly related to ASA; however, CSA was negatively associated with drinking to regulate emotional experiences. On the contrary, ASA was directly associated with Time 1 PTSD symptoms and drinking to regulate emotional experiences, as well as Time 1 and Time 3 alcohol-related problems. Drinking to regulate emotional experiences was directly related to subsequent alcohol-related problems at Time 3. Of note, neither CSA nor ASA directly predicted Time 3 alcohol-related problems.
To test the total indirect effect of CSA on Time 3 alcohol-related problems via ASA, Time 1 PTSD symptoms, and drinking to regulate emotional experiences, a serial mediation model was used (i.e., multiple-step multiple mediation). Specifically, we sought to determine the relative magnitude of the specific indirect effects associated with each putative mediator (i.e., ASA, Time 1 PTSD, and drinking to regulate emotional experiences) in the hypothesized causal chain from CSA to Time 3 alcohol-related problems, while controlling for CPA, witnessing family violence, race/ethnicity, and Time 1 alcohol-related problems. Results indicated that the proposed constructs mediated the relationship between CSA and subsequent alcohol-related problems via two different paths. In the first path, Time 1 PTSD symptoms and drinking to regulate emotional experiences mediated the relationship between CSA and Time 3 alcohol-related problems (B = 0.07, z = 2.14, p < .05). In the second path, ASA and drinking to regulate emotional experiences mediated the relationship between CSA and Time 3 alcohol-related problems (B = 1.77, z = 3.03, p < .01). 2
Discussion
The current study sought to examine whether ASA, Time 1 PTSD symptoms, and motivations for alcohol use (i.e., drinking to regulate emotional experiences) mediated the relationship between CSA and subsequent alcohol-related problems among college women. These results are consistent with the argument that drinking to regulate emotional experiences has detrimental consequences. More specifically, among women with a history of CSA, using alcohol as a means of emotional management may lead to subsequent alcohol-related problems, such as missing work/school and interpersonal problems with family and friends. Previous research has also indicated that women with a history of CSA who drink alcohol to manage their emotions are especially like to have an addiction to alcohol, as well as problems with work and social relationships (Grayson & Nolen-Hoeksema, 2005).
Results from the current study indicate that the included constructs partially mediated the relationship between CSA and Time 3 alcohol-related problems via two different paths. In one path, CSA was associated with Time 1 PTSD, which in turn predicted drinking to regulate emotional experiences, which was then related to Time 3 alcohol-related problems in adulthood. The second path suggested that CSA was related to ASA, which in turn predicted drinking to regulate emotional experiences, which was then associated with Time 3 alcohol-related problems. These results remained significant after controlling for other forms of childhood maltreatment (e.g., CPA and witnessing family violence during childhood), race/ethnicity, and Time 1 alcohol-related problems. Consistent with prior research (e.g., Ullman & Najdowski, 2009), this suggests that individuals who have a history of CSA are more likely to be sexually revictimized in adolescence. These individuals may then use alcohol to regulate their internal emotional experiences, which in turn might lead to subsequent alcohol-related problems in adulthood. Because ASA was a positive predictor of drinking to regulate emotional experiences (unlike CSA), it appears that our results may reflect the cumulative effect of multiple victimizations rather than just CSA alone. Previous research has also suggested that drinking to cope with distress, negative affect, and traumatic events has an effect on problematic drinking in revictimized women (i.e., women who experienced both CSA and ASA), but not in women who have only experienced a single victimization (i.e., CSA alone; Ullman & Najdowski, 2009). More specifically, Ullman and Najdowski (2009) found that psychological symptoms only influence prospective problem drinking in women who have experienced both CSA and ASA.
While CSA was associated with ASA and Time 1 PTSD symptoms, ASA was associated with Time 1 PTSD symptoms and drinking to regulate emotional experiences. Unlike results reported by Grayson and Nolen-Hoeksema (2005), CSA was negatively associated with the drinking motivations variable in the current study. In other words, individuals without a history of CSA were more likely to use drinking to regulate their emotional experiences than those with a CSA history. We recognize that previous studies have found a significant positive relationship between CSA and drinking motivations (e.g., Grayson & Nolen-Hoeksema, 2005), and that it is surprising that the current study found a significant negative relationship between these two constructs. It is important to note, however, that Grayson and Nolen-Hoeksema defined CSA as the experience of being pressured or forced to have sexual contact before the age of 18. Therefore, their measure of CSA most likely included instances of sexual assault during adolescence as well as during childhood. In addition, previous research has suggested that proximal assaults (i.e., ASA) are more important for understanding women’s vulnerability for psychological consequences than distal assaults (i.e., Himelein, 1995; Najdowski & Ullman, 2009). That is, more recent assaults may have a more direct impact on psychological functioning than those that occurred long ago. Accordingly, this finding may partially explain why ASA, and not CSA, positively predicted drinking to regulate emotional experiences in the current study.
As proposed by Grayson and Nolen-Hoeksema (2005), the distress coping model assumes that internal psychological distress mediates the relationship between CSA and drinking to reduce negative affect. While previous research (e.g., Goldstein et al., 2010; Grayson & Nolen-Hoeksema, 2005) has tested whether a history of CSA is associated with anxiety and/or depression symptoms, which in turn predicts drinking to cope, results from the current study suggested that a history of CSA is also associated with PTSD symptoms, which in turn predicts drinking to regulate emotional experiences. These findings imply that some survivors of CSA may use alcohol to regulate emotions that are related to symptoms of PTSD. In fact, Najdowski and Ullman (2009) also found that CSA was associated with greater symptoms of PTSD and subsequent problem drinking.
Although this study contributes to the scant literature investigating mediators between CSA and alcohol-related problems, it is not without limitations. First, the sample consisted of undergraduate women from a Midwestern university, and therefore results may not be generalizable to the rest of the population. Future research is needed that assesses sexual assault, PTSD symptoms, drinking motivations, and alcohol-related problems in both women and men of diverse ages. This research is particularly warranted given previous research findings that alcohol serves different regulatory functions for men and women (e.g., Goldstein et al., 2010). Second, some of the measures utilized in the current study (e.g., TLEQ) required retrospective reporting, which could be subject to recall bias. Third, because all childhood abuse variables were measured dichotomously, the current study was unable to consider within-group variations based on severity/frequency of abuse. Future studies may wish to measure these abuse variables continuously rather than dichotomously to assess the possible impact of within-group variation on outcomes. Fourth, ASA, PTSD symptoms, and drinking to regulate emotional experiences only partially mediated the relationship between CSA and subsequent alcohol-related problems. Future research should assess other emotion regulation mediators of this relationship, such as experiential avoidance. Fifth, the current study did not assess childhood or adolescent alcohol use. Previous research has found that troubled children may start drinking alcohol at a young age (Dube et al., 2006); therefore, it is possible that childhood alcohol use may have preceded and/or contributed to sexual assault in adolescence (rather than alcohol-related problems only occurring after ASA). Accordingly, it is possible that CSA led to early alcohol consumption, which in turn led to increased risk of ASA and alcohol-related problems in college. Finally, it is important to acknowledge the possible effects of the campus shooting that occurred between Time 1 and Time 2 of data collection. Although none of the shooting-related variables were significantly associated with any of the variables utilized in the mediation analysis, participants’ experience of the campus shooting may have affected results obtained in the current study. For example, proximity to the shooting and PTSD symptoms associated with the shooting may not have adequately captured the impact of the shooting. Thus, these variables may not have been able to completely control for the impact of the shooting on participants’ reported alcohol-related problems.
The present study suggests that individuals with a history of CSA are more likely to experience revictimization in adolescence and PTSD symptoms in adulthood, which may lead to alcohol-related problems via drinking to regulate emotional experiences. These drinking motivations may contribute to alcohol-related problems in adulthood. Although it is not entirely surprising that individuals may use alcohol to cope with negative internal experiences, it is somewhat surprising that some individuals use alcohol to enhance positive experiences. Results from the current study suggest that some survivors of CSA and/or ASA (who also reported PTSD symptoms) use alcohol to enhance positive internal experiences, and that they may have difficulty experiencing positive emotions without the use of alcohol. Symptom criteria for PTSD in the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; APA, 2013) were recently changed to include “persistent inability to experience positive emotions.” It is perhaps the case that survivors of CSA and/or ASA are motivated to consume alcohol because they experience low positive affect due to PTSD symptoms. This is an important consideration that should be addressed clinically. For example, though a history of CSA may warrant intervention and teaching of adaptive emotion regulation strategies, these needs may be particularly great in the context of subsequent assault and/or posttraumatic stress symptomatology. In addition, findings suggest that it is necessary to teach individuals effective emotion regulation strategies for coping with distress as well as strategies for experiencing and increasing positive affect, such as engaging in positive visual imagery.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by grants to the second author from the Joyce Foundation, the National Institute for Child and Human Development (1R15HD049907-01A1), and the National Institute of Mental Health (5R21MH085436-02).