Latent Profiles of Social Reactions to Sexual Assault Disclosure Among Undergraduate Women

Sexual Assault and Distress

Importantly, extensive research has documented the detrimental impact that sexual violence has on the mental and physical health of survivors (Campbell et al., 2008, 2009; Chen et al., 2010; Dworkin et al., 2017). In a review of the literature on the prevalence of psychiatric disorders among survivors of sexual assault, Campbell et al. (2009) noted that up to 65% of victims develop posttraumatic stress disorder (PTSD) and up to 51% experience depression. Substance use problems are also common among victims; 13%–49% become dependent on alcohol and over half use other illicit substances. A systematic review and meta-analysis also found that victims of sexual violence have an increased risk for all forms of psychopathology assessed, including PTSD, depression, and substance abuse (Dworkin et al., 2017). Further, the observed effects did not change when broader definitions of sexual assault (i.e., including incapacitated, verbally coerced, and non-penetrative assaults) were included. This highlights the negative impact that sexual victimization has on mental health outcomes. Thus, it is imperative to understand the factors that impact post-assault recovery.

Notably, recovery from sexual assault is not only dependent on the individual survivor, but the socio-cultural context in which survivors are embedded also affects subsequent mental health outcomes (Campbell et al., 2009; Dworkin et al., 2019). In particular, extensive research has examined the social reactions that victims receive after they disclose sexual assault experiences to formal (e.g., mental or medical health providers) and informal (e.g., friends and family) sources of support and how these reactions may impact post-assault outcomes (Dworkin et al., 2019; Ullman, 2010a). Indeed, most victims will disclose to at least one other person about their experience of sexual assault (Ahrens et al., 2009; Ullman, 2010b). Up to 75% of college women will tell someone, most frequently a female friend, about their sexual assault experience. Further, among college women who experience a recent sexual victimization (over a 4-month follow-up period), over half will also disclose to someone (typically a female and/or male peer; Orchowski & Gidycz, 2015). This indicates that a sizeable portion of survivors will disclose shortly after a victimization is experienced.

Reactions to Sexual Assault Disclosure

The reactions that victims receive after they disclose experiences of sexual assault are construed broadly as positive or negative (Ullman, 2000). Ullman (2000) identified two types of positive and five types of negative social reactions. In general, positive reactions are helpful and supportive to the victim. Positive reactions include two types: (a) emotional support and (b) informational/tangible aid. Emotional support reactions include responses such as showing understanding, empathic listening, comforting, believing the survivor, expressing that she is not to blame, and telling her she is loved. Informational/tangible aid reactions include responses that help the victim access information or services such as coping resources, counseling, and medical care. In contrast, negative reactions are unsupportive or harmful. Negative reactions include victim blaming, treating the victim differently (e.g., withdrawing from the victim or suggesting she is “tainted” by the victimization), taking control (e.g., making decisions for the victim), egocentric responses that prioritize the needs of the disclosure recipient over the victim, and responses that encourage avoidance (e.g., telling the survivor not to think or talk about the event). These negative reactions fall under two broad categories based on their explicit negative valence and perceived harmfulness: (a) turning against and (b) unsupportive acknowledgment (Relyea & Ullman, 2015b). Turning against reactions are more explicitly negative and include victim blaming, treating the survivor differently or stigmatizing her, and infantilizing responses. Unsupportive acknowledgment reactions are mixed responses that acknowledge the assault but do not provide enough support to the survivor and may include egocentric responses, distracting responses, and making decisions for the survivor.

There is considerable evidence that survivors who received negative reactions to disclosure experienced worse post-assault outcomes, including increased posttraumatic stress symptoms, depression, and substance use (Dworkin et al., 2019; Littleton, 2010; Peter-Hagene & Ullman, 2014; Ullman, 2010a). Among undergraduate women, certain reactions (i.e., controlling the survivor’s decisions) were associated with increased symptoms of posttraumatic stress, depression, and anxiety after controlling for assault severity and social desirability (Orchowski et al., 2013). Notably, distinct patterns of associations emerged for the two domains of negative reactions: turning against and unsupportive acknowledgment. Relyea and Ullman (2015b) found that although both types of negative reactions predicted posttraumatic stress and maladaptive coping, only unsupportive acknowledgment was related to depression and positive coping, and only turning against was related to self-blame. Other research, however, found no association between the two types of negative reactions and depression (Relyea & Ullman, 2015a), and when examined longitudinally, only turning against reactions remained a significant predictor of posttraumatic stress (Ullman & Relyea, 2016). Thus, it appears important to make a distinction between turning against and unsupportive acknowledgment reactions, rather than collapsing all negative responses together. Further, examining how mental health outcomes relate to different combinations of turning against and unsupportive acknowledgement reactions that survivors receive may help clarify some of the mixed findings in the existing literature.

Research on positive reactions to disclosure is more limited and findings about positive reactions and recovery are mixed (Ahrens et al., 2007; Campbell, Ahrens et al., 2001a; Campbell, Dworkin, 2001b; DiMauro & Renshaw, 2021; Orchowski & Gidycz, 2015; Sigurvinsdottir & Ullman, 2016; Ullman & Peter-Hagene, 2014, 2016). For example, Ullman and Peter-Hagene (2014) found that positive social reactions had a direct positive relation with PTSD, but an indirect negative relation with PTSD via increased perceived control over recovery, whereas Orchowski and Gidycz (2015) found no univariate associations between positive social reactions and PTSD, depression, or anxiety. Ullman and Peter-Hagene (2016) failed to find a prospective association between positive social reactions and PTSD symptoms, whereas Sigurvinsdottir and Ullman (2016) did find that positive social reactions prospectively predicted increased depression and PTSD symptoms. In contrast, Relyea and Ullman (2015b) found that, after taking into account turning against and unsupportive acknowledgment, positive social reactions were related to decreased depression symptoms. More recently, findings from a review and meta-analysis indicated that positive reactions did not have a buffering effect against psychopathology (Dworkin et al., 2019). It is possible that positive reactions are not sufficiently protective in the presence of concurrent negative reactions. Thus, it may be important to consider the pattern of all reactions received by survivors.

Notably, most women who disclose experiences of sexual victimization receive a mix of both negative and positive social reactions to disclosure (Ahrens et al., 2009; Campbell, Ahrens, et al., 2001). However, extant research has not yet identified the complexities in such reactions, due to the variable-centered approach that assesses the degree to which positive or negative social reactions relate to a particular outcome independently, rather than in the context of one another (e.g., Orchowski & Gidycz, 2015; Ullman & Peter-Hagene, 2016). Given the pattern of mixed findings pertaining to reactions to disclosure that most women receive, person-centered analytic approaches may prove beneficial in understanding patterns or combinations of social reactions among victims. Further, researchers have advocated for the use of person-centered approaches in the study of violence against women (Nurius & Macy, 2008; Swartout & Swartout, 2012) as these approaches “can complement the existing findings and generate a more nuanced picture of the patterns of relations between our variables and participants” (Swartout et al., 2011, p. 313).

One notable exception to the variable-centered approach in the disclosure literature is Woerner et al.’s (2018) study, which conducted a latent profile analysis of social reactions to intimate partner violence (IPV) disclosure. They identified three latent profiles of subgroups of women in their community sample of female victims of partner violence with varied experiences of total positive and total negative social reactions to disclosure. In the three-profile solution, 34.7% of women (n = 60) were grouped in the low negative/high positive, 14.4% of women (n = 25) in the high negative/low positive, and 50.9% of women (n = 88) in the low negative/low positive profiles. Importantly, differences emerged in mental health among the different groups: women in the high negative/low positive reactions profile reported greater depression and PTSD symptom severity than women in the other two. There was also a significant interaction between IPV severity and latent profile membership, in that IPV severity was positively associated with depression and PTSD symptom severity for the low negative/high positive and for the low negative/low positive profiles. However, IPV severity was not associated with depression and PTSD symptom severity for the profile characterized by high negative/low positive reactions. This profile’s scores on depression and PTSD were high across all levels of IPV severity, highlighting the detrimental effects of negative reactions in particular. Such person-centered approaches can help identify how the pattern of positive and negative reactions that victims receive upon disclosure relates to outcomes (Woerner et al., 2018). Person-centered approaches have yet to be employed to understand the social reactions among victims of sexual assault and, given the prevalence of sexual assault in college campuses (Cantor et al., 2015; Krebs et al., 2009), college women are an important group to consider.

Aims of the Current Study

The overarching aim of the current study was to extend prior research on social reactions to disclosure of sexual assault by using a person-centered analytic approach, namely latent profile analysis, in order to identify subgroups of victims with varying patterns/combinations of different types of positive and negative reactions among a sample of college women. We extend Woerner et al.’s (2018) study to sexual assault victimization and by further examining college women’s experiences with two types of positive reactions, emotional support and informational/tangible aid, and two types of negative reactions, turning against and unsupportive acknowledgment. Collapsing all types of negative reactions together has been noted as a limitation in previous studies (Relyea & Ullman, 2015b). An examination of specific types of both negative and positive responses allows for a more nuanced understanding of women’s experiences with disclosure reactions and their association with distress. We hypothesized that, similar to latent profiles of reactions to disclosure of IPV (Woerner et al., 2018), we would identify at least three subgroups of membership, or latent profiles, characterized by a mix of varying degrees of emotional support, informational/tangible aid, turning against, and unsupportive acknowledgment reactions.

Sexual assault characteristics may also relate to the specific pattern of social reactions survivors receive, since certain assault characteristics are related to whether a survivor discloses sexual victimization, the number and type (informal vs. formal) of support sources told, and the reactions received (Orchowski & Gidycz, 2012; Peter-Hagene & Ullman, 2015; Starzynski et al., 2005; Ullman & Filipas, 2001a). Specifically, women were more likely to disclose assaults that involved alcohol use at the time of the assault (both victim and perpetrator use), assaults with greater severity (i.e., perpetrator use of violence, perceived life threat, and physical injury sustained), and stranger or lesser-known perpetrators (Orchowski & Gidycz, 2012; Starzynski et al., 2005). Assaults with greater severity may be related to both increased positive and negative social reactions to disclosure (Ullman et al., 2007, 2008); but victims of alcohol-involved assaults (Ullman & Filipas, 2001b) may receive greater negative reactions upon disclosure. Ullman et al. (2006) found that victims of stranger assaults receive greater negative reactions, although more recently Edwards, Ullman, et al. (2020) did not find a relation between closeness between victim and perpetrator and general negative or positive reactions. Assault characteristics are also related to mental health outcomes following sexual victimization (Ullman et al., 2006, 2007; Ullman & Najdowski, 2010). Thus, in the current study we examined differences between profiles in the following characteristics of the assault experience itself: victim intoxication, perpetrator intoxication, victim injury at the time of the event, and perpetrator type (i.e., stranger, partner, ex-partner, and acquaintance). We hypothesized that the assaults of women who experienced greater turning against and unsupportive acknowledgment reactions would be characterized by greater intoxication, physical injury, and stranger perpetrators. We did not make a priori hypotheses about profiles with greater positive reactions to disclosure given limited previous findings.

Further, we examined differences in the following assault-related beliefs: self-blame, perpetrator-blame, and sexual assault acknowledgment, given their association with reactions to disclosure and/or psychological outcomes following victimization (Frazier, 2003; Kennedy & Prock, 2016; Littleton et al., 2007; Sigurvinsdottir & Ullman, 2015; Wilson & Scarpa, 2017). Among victims of sexual violence, self-blame (conceptualized as a cognitive attribution that the victim makes in which she blames herself for the assault/abuse) is related to a host of negative psychological outcomes (Kennedy & Prock, 2016). Negative social reactions to sexual assault disclosure (and turning against reactions, in particular) are related to increased self-blame, whereas positive reactions are related to decreased self-blame (Relyea & Ullman, 2015b). However, positive reactions may not reduce self-blame prospectively (Ullman & Najdowski, 2011). Thus, we expected that latent profiles characterized by greater turning against reactions would be related to increased self-blame, while profiles characterized by emotional support reactions (which include responses such as telling the survivor she is not to blame) may be related to decreased self-blame in the current cross-sectional analyses. To our knowledge, perpetrator-blame has not been examined in relation to positive and negative reactions to sexual assault disclosure; however, it is likely that the different social reactions that victims receive are also related to blame assigned to the perpetrator of the assault. We expected the pattern for perpetrator-blame to be the reverse, with profiles with fewer turning against reactions and greater emotional support reactions to be associated with increased perpetrator-blame.

The types of social reactions that women receive upon sexual assault disclosure may also be related to how women make sense of and label those experiences (Littleton et al., 2007). Research on acknowledgment of rape, or victim’s labeling an experience that meets the definition of rape as “rape” (as opposed to using non-victimizing terms such as “bad sex” or “serious miscommunication” to describe their experience, Littleton et al., 2006), has established that unacknowledged rape (or sexual assault) is common among victims (Cleere & Lynn, 2013; Wilson & Miller, 2016). Littleton et al. (2007) suggested that victims may appraise the assault as either a more serious occurrence or may remain unacknowledged depending on the nature of the response to disclosure by the recipient (e.g., if others become very angry/upset vs. minimize the severity of the sexual assault). Women vary in their post-trauma outcomes based on acknowledgment status, although findings are somewhat mixed. Some research indicates that unacknowledged rape is related to greater psychological distress and hazardous alcohol use (Clements & Ogle, 2009; McMullin & White, 2006), whereas in other studies acknowledged victims experience more negative outcomes, including greater posttraumatic stress and avoidance coping (Littleton et al., 2006; Wilson & Scarpa, 2017). Thus, it is important to understand how different social reactions relate to sexual assault acknowledgment status. We hypothesized that profiles that received greater negative and positive reactions may be related to acknowledgment of assault since receiving more negative (turning against and unsupportive acknowledgement) and positive (emotional support and informational/tangible aid) reactions would communicate to the survivor that the assault was a significant event. Further, survivors who disclose the event as a sexual assault (as opposed to a less significant event or “miscommunication”) may be more likely to elicit both types of reactions from disclosure recipients. Given the dearth of previous research, we did not make specific hypotheses about the types or combinations of social reactions in relation to acknowledgment status.

Finally, we explored the associations between subgroup membership and mental health outcomes (i.e., posttraumatic stress, depression, and alcohol use). Given the significant relation between negative social reactions and psychopathology from previous variable-centered (Dworkin et al., 2019) and person-centered (Woerner et al., 2018) research, we expected that profiles characterized by greater turning against and unsupportive acknowledgment reactions would be associated with increased posttraumatic stress and depression symptoms and greater hazardous drinking. However, we did not advance any more specific hypotheses given that we did not know what particular latent profiles our analyses would identify.

Participants

The initial sample consisted of 1482 undergraduate women (part of a parent study examining rape, shame, and psychological distress) enrolled in a mid-sized public university in a Midwestern region of the United States. The research questions of the current study are unique and have not been previously published. Only women who had a history of sexual assault since age 14 and had disclosed their experiences to at least one other person were included in the current study. Sexual assault was defined as any experiences of unwanted kissing/fondling, oral, vaginal, or anal sex, in which the perpetrator used force/threats of force or alcohol/substance-related coercion tactics. Of the original sample, 27.5% (n = 407) reported an experience of sexual assault since the age of 14, and 73.7% (n = 300) of these women had disclosed their sexual assault to at least one other person. Thus, the final sample for the current analyses included 300 undergraduate women who had disclosed an experience of sexual assault.

Participants were on average 18.8 years old (SD = 1.0). Most women in the sample identified as White (94.2%), followed by African American (3.8%), Asian or Asian American (2.1%), Pacific Islander/Hawaiian (1.0%), and Native American (0.1%). Five (1.7%) participants identified as biracial, multiracial, or another race not listed previously. Percentages add up to more than 100% because some participants identified with more than one race. Participants were also asked about ethnicity in a separate question and 6.5% identified as Hispanic/Latina. Most identified as heterosexual/straight (89.4%). The remaining participants identified as bisexual (8.6%), lesbian/gay (1.7%), and asexual (0.3%). The majority of participants (80%) reported an annual household income of $75,000 or more. A significant proportion of women (41.3%) were in committed dating relationships at the time of study completion.

Procedures

Study procedures were approved by our university’s Committee on the Use of Human Subjects in Research. Study participants were recruited from the undergraduate research subject pool of our psychology department and received course credit for participation. Inclusion criteria were that participants needed to be 18 years or older and identify as a woman. Research consent was obtained in person in the research laboratory. Participants received instructions to access and complete the survey online via Qualtrics after completing consent procedures.

Measures

Latent Profile Analysis: Profile Identification

Latent profile analysis is a person-centered approach and was used to identify groups in the current sample based on positive and negative reactions received when participants disclosed their sexual assault. A series of two- to six-profile models were tested via Mplus (Version 8; Munthén & Munthén, 1998–2017). Selection of the number of profiles was based on interpretative meaningfulness and statistical relevance. When examining model fit, a series of statistical indices were examined including the Akaike information criterion (AIC; Akaike, 1987), the Bayesian information criteria (BIC; Schwarz, 1978), sample size adjusted Bayesian information criterion (SSABIC; Schlove, 1987), the entropy value (Ramaswamy et al., 1993), the Lo–Mendell–Rubin adjusted likelihood ratio test (LMR; Lo et al., 2001), and the bootstrapped likelihood ratio test (BLRT; McLachlan & Peel, 2000). Lower AIC, BIC, and SSABIC values indicate better model fit, whereas entropy values closer to 1 suggest better classification (Nylund et al., 2007). When examining latent profiles in the current sample, fit statistics did not specify one profile emerging as a better fit in all categories (see Table 2). The AIC and Log Likelihood were lowest in the five-profile while the BIC and SSABIC were lowest in the four-profile solution. Notably, the BIC indicated a four-profile solution and has been demonstrated as a strongest indicator of fit (Nylund et al., 2007). Entropy for the four-profile was highest which indicated the best classification precision. The LRT indicated the four-profile, five-profile, and six-profile solutions did not fit significantly better than the previous solution. Finally, the BLRT was significant (p < .001) for all profile solutions. Based on best goodness-of-fit indices (Nylund et al., 2007), we retained the four-profile solution. Frequency information about the disclosure recipients for the full sample and for the specific latent profiles is contained in online Supplementary Table 1.

OPEN IN VIEWER

Table 2.

Fit Indices for Latent Class Models with Two to Six Profiles Based on Reactions to Disclosure.

	2 Profiles	3 Profiles	4 Profiles	5 Profiles	6 Profiles
AIC	2203.39	2103.22	2012.45	1953.95	2050.34
BIC	2251.19	2169.40	2057.02	2076.90	2081.22
Sample adjusted BIC	2209.97	2112.31	1924.08	1968.11	1954.21
Log likelihood	−1088.70	−1033.61	−983.23	−948.98	−987.25
Entropy	.870	.878	.894	.864	.840
Lo, Mendell, Rubin	2 v 1 value = 293.04 p = .00	3 v 2 value = 106.43 p = .00	4 v 3 value = 97.33 p = .32	5 v 4 value = 66.17 p = .22	6 v 5 value = 74.12 p = .10
BLRT	−1240.38 p = .001	−1088.70 p = .001	−1033.61 p = .001	−983.23 p = .001	−977.06 p = .001
n for each class	C1 = 35C2 = 257	C1 = 189C2 = 66C3 = 37	C1 = 170C2 = 43C3 = 63C4 = 16	C1 = 32C2 = 140C3 = 58C4 = 46C5 = 16	C1 = 28C2 = 37C3 = 139C4 = 58C5 = 24C6 = 6

Social Reaction Differences by Latent Profile

The four profiles were given labels to describe the frequency of turning against, unsupportive acknowledgment, emotional support, and informational/tangible aid reactions received. The profiles were labeled based on how the positive and negative social reactions differed from the overall sample means and based on how each compared to scores of the other profiles. Some profiles were categorized by overall positive (i.e., emotional support and informational/tangible aid) or negative (i.e., turning against and unsupportive acknowledgment) patterns of social reactions. Profile 1 (n = 170) was labeled Moderate Emotional Support/Low Negative and was characterized by moderate levels of emotional support, low levels of tangible/informational aid, and low levels of any negative reactions. Profile 2 (n = 43) was labeled Moderate Emotional Support/Moderate Negative and was characterized by moderate levels of both turning against and unsupportive acknowledgment reactions, moderate emotional support, and low tangible/informational aid. Profile 3 (n = 63) was labeled High Positive/Some Unsupportive Acknowledgment and was characterized by high emotional support and informational/tangible aid, some unsupportive acknowledgment, and low turning against reactions. Profile 4 (n = 16) was labeled Moderate Positive/High Negative and was characterized by high turning against and unsupportive acknowledgment, some emotional support, and moderate tangible/informational aid. We then conducted a series of one-way ANOVAs to examine profile differences on social reactions subscales. Significant group differences emerged for all four subscales, turning against, unsupportive acknowledgment, support, and informational aid. Post-hoc comparisons with Bonferroni corrections were conducted, and significant group differences are detailed in Table 3.

OPEN IN VIEWER

Table 3.

Social Reactions Questionnaire Subscale Means by Profile.

	Full Sample N = 292	Moderate Emotional Support/Low Negative n = 170	Moderate Emotional Support/Moderate Negative n = 43	High Positive/Some Unsupportive Acknowledgment n = 63	Moderate Positive/High Negative n = 16
Turning against	.39 (.59)	.13 (.17)a	1.04 (.29)b	.18 (.20)a	2.22 (.49)c
Unsupportive acknowledgement	.63 (.54)	.39 (.32)a	1.17 (.50)b	.66 (.42)c	1.64 (.64)d
Emotional support	2.54 (.91)	2.44 (.92)a	2.13 (.80)ab	3.28 (.39)c	1.80 (.74)b
Info/aid	.78 (.84)	.36 (.41)a	.57 (.57)ab	2.03 (.62)c	.90 (.83)b

Latent Profile Associations with Assault Characteristics

A series of one-way ANOVA, chi-square analyses, and Fisher’s Exact Tests were conducted to examine latent profile differences in assault characteristics (see Table 4 for group means and percentages). In the ANOVA analyses, significant group differences emerged for victim injury, F(2,289) = 12.95, p < .001. Post-hoc comparisons with Bonferroni corrections revealed the Moderate Positive/High Negative group did not differ from the Moderate Emotional Support/Moderate Negative group but reported more injuries than the other two groups. The Moderate Emotional Support/Moderate Negative group did not differ from the High Positive/Moderate Unsupportive Acknowledgment group, which also did not differ from the Moderate Emotional Support/Low Negative group. No significant group differences emerged for victim intoxication, F(3,287) = .39, p = .76, or perpetrator intoxication, F(3,288) = 1.05, p = .37.

OPEN IN VIEWER

Table 4.

Mental Health Outcomes and Assault Characteristics by Social Reactions Profile.

	Full Sample N = 292	Moderate Emotional Support/Low Negative n = 170	Moderate Emotional Support/Moderate Negative n = 43	High Positive/Some Unsupportive Acknowledgement n = 63	Moderate Positive/High Negative n = 16	Difference by Profile
PTSD symptoms	15.02 (16.11)	9.50 (11.94)a	20.81 (15.25)b	20.61 (17.97)b	35.56 (20.44)c	F(3,281) = 24.05**
Depression symptoms	5.02 (4.85)	4.30 (4.49)a	7.26 (5.89)b	5.07 (4.77)ab	6.38 (4.13)ab	F(3,281) = 4.81*
Hazardous alcohol use	8.07 (5.32)	8.06 (5.20)	9.00 (6.31)	7.23 (4.80)	9.25 (5.77)	F(3,270) = 1.11
Self-blame	3.32 (1.04)	3.15 (1.00)a	3.69 (.89)b	3.40 (1.14)ab	3.86 (1.01)b	F(3,287) = 5.11*
Perpetrator blame	2.09 (.87)	1.91 (.80)a	2.26 (.95)ab	2.33 (.86)b	2.60 (1.01)b	F(3,284) = 6.54**
Victim intoxication	2.55 (1.64)	2.53 (1.58)	2.74 (1.76)	2.41 (1.65)	2.69 (1.92)	F(3,287) = .39
Perpetrator intoxication	2.09 (1.18)	2.16 (1.23)	2.05 (1.09)	2.06 (1.15)	1.63 (1.03)	F(3,288) = 1.05
Victim injury	1.54 (.87)	1.43 (.77)a	1.88 (1.07)bc	1.44 (.74)ab	2.38 (1.15)c	F(3,286) = 9.20**
SA acknowledgement	3.79 (2.25)	3.20 (2.17)a	4.12 (2.07)ab	4.67 (2.19)b	5.69 (1.58)b	F(3,287) = 12.42**
Stranger	12.7%	12.4%	7.0%	19.0%	6.3%	Fisher’s exact = .30
Partner	27.7%	31.2%	25.6%	20.6%	25.0%	Fisher’s exact = .43
Ex-partner	18.2%	19.4%	16.3%	19.0%	6.3%	Fisher’s exact = .54
Acquaintance	41.1%	37.1%	51.2%	39.7%	62.5%	χ2(3,292) = .11

Chi-square analyses and Fisher’s Exact Tests were conducted to examine group differences on dichotomous assault characteristics (see Table 4 for percentages). Per recommendation by Blalock (1999), the Fisher’s Exact Test was conducted due to unreliability when the expected cell frequencies in a chi-square test are less than five. Participants were provided with a list of categories (e.g., acquaintance, ex-partner, partner, etc.) and asked to identify their relationship with the perpetrator. No significant associations emerged between profile and relationship with the perpetrator.

Latent Profile Associations with Mental Health Symptoms and Blame

One-way ANOVAs were conducted to examine latent profile differences on PTSD symptoms, depression, hazardous alcohol use, self-blame, and perpetrator blame (see Table 4 for group means). Significant group differences emerged for PTSD, F(3,284) = 24.05, p < .001, depression symptoms, F(3,281) = 4.81, p < .01, self-blame, F(3,287) = 5.11, p < .01, perpetrator blame, F(3,284) = 6.54, p < .001, and SA acknowledgement, F(2,290) = 17.33, p < .001. Post-hoc comparisons with Bonferroni corrections were conducted to examine profile differences for significant ANOVAs. For PTSD symptoms, the Moderate Positive/High Negative group reported more PTSD symptoms than the other three groups. The High Positive/Moderate Unsupportive Acknowledgment and Moderate Emotional Support/Moderate Negative groups did not differ but reported more PTSD symptoms than the Moderate Emotional Support/Low Negative group. When examining depression symptoms, the Moderate Emotional Support/Moderate Negative group reported more depression symptoms than the Moderate Emotional Support/Low Negative group. The other two groups did not differ from any other group. For self-blame, the Moderate Positive/High Negative group did not differ from the Moderate Emotional Support/Moderate Negative group though both reported higher depression symptoms than the Moderate Emotional Support/Low Negative group. The High Positive/Moderate Unsupportive Acknowledgment group did not differ from any other group. When examining perpetrator blame, the Moderate Positive/High Negative group did not differ from High Positive/Moderate Unsupportive Acknowledgment group, but both groups reported more perpetrator blame than the Moderate Emotional Support/Low Negative group. The Moderate Emotional Support/Moderate Negative group did not differ from any other group on perpetrator blame. For SA acknowledgement, the Moderate Positive/High Negative group did not differ from the High Positive/Moderate Unsupportive Acknowledgement group but both groups reported higher SA acknowledgment than the Moderate Emotional Support/Low Negative group. The Moderate Emotional Support/Moderate Negative group did not differ from any other group. Finally, no significant group differences emerged for hazardous alcohol use, F(3, 270) = 1.11, p = .35.

Profile and Sexual Assault Characteristics

Results identified four profiles of combinations of negative and positive social reactions undergraduate women received to disclosure of sexual assault experiences. On a hopeful note, the majority of women in our sample were in the first profile (moderate emotional support/low negative) that received emotional support with few negative responses. However, women in the remaining three profiles received at least some unsupportive acknowledgment reactions, which although not as overtly or explicitly negative towards survivors, are still associated with negative mental health outcomes (Dworkin et al., 2019). We also identified a small group of women who received highly negative responses with the least emotional support. Notably, this group’s PTSD score was above the cutoff on the PCL-5 for probable PTSD, suggesting that these women may be particularly vulnerable to experience negative posttraumatic outcomes and may need additional support services, such as counseling, for their symptoms.

We compared the characteristics of the most distressing sexual assault experience of college women based on latent profile membership. Consistent with our hypotheses, we found that women in profiles with greater negative reactions reported greater physical injury, suggesting the assaults may have involved greater severity/force. This is consistent with previous research that greater sexual assault severity is related to receiving greater negative and fewer positive reactions (Ullman & Filipas, 2001a; Ullman et al., 2007, 2008). Assaults involving greater victim injury may be more distressing for disclosure recipients to learn about, prompting unhelpful responses such as egocentric (e.g., becoming so upset that the victim feels compelled to reassure them) or distracting (e.g., efforts to stop the victim from talking about the assault) reactions. Further, sustaining more injuries may have contributed to disclosure to more recipients, including more formal sources if the injuries required further attention. Assaults that involved greater physical injury also may be more likely to be detected. Thus, victims with greater injuries may be more likely to be approached by others for the disclosure. Notably, disclosures elicited by others are related to greater negative reactions and fewer positive reactions (Edwards, Ullman, et al., 2020), as well as greater PTSD symptoms (Carson et al., 2021).

Contrary to hypotheses, we did not identify any differences in victim or perpetrator intoxication at the time of the assault. This contradicts some prior research findings that victims of alcohol-involved assaults, where the perpetrator and/or the victim were drinking, received greater negative social reactions to disclosure (Ullman & Filipas, 2001b; Ullman & Najdowski, 2010). This difference may be because our sample consisted of college students, given the drinking culture on college campuses and some evidence that college survivors may be less likely to receive negative reactions compared to community women for alcohol-involved assaults (Littleton et al., 2009). In the current study, we only assessed general negative and positive reactions to sexual assault disclosure and not reactions specific to the disclosure of alcohol-related assaults (Relyea & Ullman, 2015a). We may have found differences in alcohol-specific positive and negative reactions specific to the victim’s drinking prior to the assault. Edwards, Ullman, et al. (2020) found some distinct links between disclosure recipient and victim factors and general versus alcohol-specific reactions to disclosure. Given the high rates of drinking and alcohol-involved assaults among college students (Mohler-Kuo et al., 2004), future research should also consider how alcohol-specific reactions relate to assault characteristics, pattern of general social reactions, and mental health outcomes.

The latent profiles did not differ in perpetrator type, contrary to hypotheses that victims may receive more reactions that are negative when they disclose stranger assaults (Ullman & Filipas, 2001a; Ullman et al., 2006). Stranger assaults are more likely to be disclosed to formal support sources who are more likely to provide negative social reactions (Starzynski et al., 2005; Ullman & Filipas, 2001b). In the current study, however, most of the assault experiences occurred with an acquaintance or partner and disclosure to formal support sources was low, which may explain why we did not find a difference between the profiles of different social reactions in the perpetrator relationship/familiarity. Our findings are consistent with a more recent study with college students (Edwards, Ullman, et al., 2020) that also failed to find an association between general negative and positive reactions and victim-perpetrator closeness. Interestingly, this study found that perpetrator closeness related to fewer positive alcohol-specific reactions that victims received, which also suggests that it may be important to examine alcohol-specific reactions to disclosure in studies of college students.

Blame and Acknowledgment Characteristics

There were differences in the level of self- and perpetrator-blame reported by women between the different profiles. Providing partial support for our hypotheses, profiles characterized by moderate positive/high negative and moderate negative/moderate positive reactions had increased self-blame compared to the profile characterized by moderate emotional support/low negative reactions. This is consistent with prior research on the relation between negative reactions and self-blame among victims (Ullman et al., 2007). However, we also found that the profile characterized by the greatest emotional support did not differ in self-blame compared to profiles characterized by more negative reactions. A limitation of the cross-sectional nature of the current study is that we are not able to track the bidirectional associations of negative and positive reactions and self-blame over time. It is possible that victims who report greater self-blame may initially elicit more emotional support and informational/tangible aid reactions. However, these positive reactions may not reduce self-blame over time and self-blame may decrease the number of positive reactions prospectively (Ullman & Najdowski, 2011), whereas a more cyclical relation may exist between negative reactions and self-blame and general victim blame (Orchowski & Gidycz, 2015; Ullman & Najdowski, 2011).

We found only partial support for our hypotheses for perpetrator-blame. The profile characterized by high emotional support and informational/tangible aid had greater perpetrator-blame compared to the profile with only moderate emotional support. However, contrary to our expectations, we found that the profile characterized by high turning against reactions, such as victim blame, had greater perpetrator-blame. This latter profile was also characterized by high unsupportive acknowledgment reactions and some researchers have suggested that certain reactions considered negative (i.e., egocentric reactions of wanting to seek revenge against the perpetrator) may increase blame of perpetrators and society (Orchowski & Gidycz, 2015). Further, women in the latter profile also reported greater physical injury, which may have also contributed to perpetrator-blame based on the severity of the sexual assault.

As expected and suggested by Littleton et al. (2007), we found differences in women’s acknowledgment of the incident as sexual assault. The profiles characterized by high negative reactions and high positive reactions reported greater acknowledgment compared to the profile characterized by moderate emotional support/low negative reactions, whereas the profile characterized by moderate positive/moderate negative reactions did not differ from either. It appears that receiving greater negative or positive reactions may relate to victims acknowledging their experience as sexual assault since these reactions (regardless of valence) may indicate that the experience was a significant event. Whereas receiving fewer reactions of any kind may suggest that the experience was not a serious incident that may contribute to victims labeling the experience not as a sexual assault but as a less significant event (e.g., “bad sex”). Conversely, disclosing the experience as “sexual assault” or “rape” may elicit stronger reactions from disclosure recipients (negative or positive) compared to unacknowledged victims who may describe the event in ways that do not elicit strong reactions of any kind in disclosure recipients. In addition, acknowledgement of the assault may relate to the frequency of disclosures such that women who acknowledge their assault may be more likely to disclose and receive various reactions. Longitudinal research is needed to clarify the directionality of the current findings.

Mental Health Outcomes

As hypothesized, differences emerged in posttraumatic stress and depression symptoms between profiles of membership. Consistent with prior findings (Dworkin et al., 2019; Woerner et al., 2018), the profile characterized by high negative reactions reported greater PTSD symptoms compared to the other three profiles. Interestingly, the high positive/some unsupportive acknowledgment profile did not differ in PTSD compared to the moderate emotional support/moderate negative profile, both of which reported greater symptoms of PTSD than the moderate emotional support/low negative reactions profile. Thus, it appears that the presence of any negative reaction (in this case, unsupportive acknowledgment) is associated with distress even when disclosure recipients also provide high levels of positive reactions. This finding is consistent with accumulating evidence that while negative reactions to sexual assault disclosure are detrimental for recovery, positive social reactions do not protect or buffer against psychopathology among victims of sexual assault in variable-centered (Dworkin et al., 2019) or person-centered research (Woerner et al., 2018). Further, some unsupportive acknowledgement reactions such as telling the survivor to stop thinking about the event, although well intentioned, may encourage distraction as a way to cope (Dworkin et al., 2019). However, avoidance coping in the long term can contribute to maintenance of PTSD symptoms (Ullman et al., 2007).

The current study focused on psychopathology as outcomes and did not include other variables, such as perceived control over recovery or posttraumatic growth, which may be facilitated by positive reactions to disclosure. For example, Ullman and Peter-Hagene (2014) found that positive social reactions related to increased sense of control over recovery among survivors, which in turn related to decreases in PTSD. Thus, the benefits of emotional support and informational/tangible aid in recovery should continue to be assessed in future research. Future research may also benefit from comparing differences in outcomes when survivors received positive social reactions versus when they chose not to disclose the assault at all. ¹

We found fewer differences in depression between the profiles. Somewhat surprisingly, the only significant difference was that the moderate support/moderate negative profile reported greater depression than the moderate support/low negative, whereas the profiles characterized by high negative and high positive reactions did not differ from the rest. The relation between unsupportive acknowledgment and turning against reactions and depression has not been consistently found in prior research and may also depend on the type of assault experienced (alcohol-related vs. not; Relyea & Ullman, 2015a; 2015b). After taking into account negative reactions, positive reactions may predict decreased depression symptoms (Relyea & Ullman, 2015b), which is consistent with our results that the profile with moderate support/low negative reactions reported lower depression. Assessment of PTSD symptoms in the current study was tied to the sexual assault experience, whereas depression was assessed with a general symptom measure. Thus, depression symptoms may have been influenced by various factors beyond social reactions to disclosure not included in the current study.

Contrary to our hypotheses and prior work (Dworkin et al., 2019; Peter-Hagene & Ullman, 2014), we did not find any differences between the latent profiles in hazardous alcohol use. This null finding may be due to the type of sample we selected for the purposes of the current study. Alcohol use (Wechsler & Nelson, 2008) and excessive drinking (Substance Abuse and Mental Health Services Administration, 2014) are common among college students. Thus, it is possible that undergraduate women in general engage in greater alcohol use behaviors regardless of the types of reactions to sexual assault disclosure they receive. Women in our sample had experienced a range of sexual victimization experiences; however, some research indicates that victims of alcohol-involved assaults, where either the perpetrator only or both perpetrator and victim were drinking, may be more likely to engage in hazardous drinking (Messman-Moore et al., 2013). It is possible that we did not identify any differences in alcohol use since we did not consider the type of assault experienced. Our findings should be replicated in future studies with college students and community samples. Examining assault type would also help clarify the link between social reactions to disclosure and hazardous alcohol use.

Limitations and Future Directions

Results should be considered in the context of several limitations. Our sample was comprised predominantly of undergraduate women who were White, identified as heterosexual, and reported an annual household income in middle and upper-middle class ranges; thus, our findings may not generalize to other populations. The focus on cisgender, heterosexual women has been identified as a limitation of the wider social reactions literature (Dworkin et al., 2019), and the need to include racial and ethnic minority populations in trauma research has been recognized (Bryant-Davis, 2019). Survivors’ and disclosure recipients’ identities play a role in the types of social reactions that are provided (Edwards, Ullman, et al., 2020), reactions that are received (Sigurvinsdottir & Ullman, 2016; Ullman & Filipas, 2001a, 2001b), and the types of support sources disclosed (Starzynski et al., 2007; Ullman & Filipas, 2001b). Consideration of racial/ethnic and cultural identity may be particularly pertinent to research on the effects of social reactions to assault disclosure as some evidence suggests a stronger ethnic identity may be protective in the presence of negative reactions (Nikulina et al., 2019). Importantly, women’s social location plays a significant role in seeking, accessing, and receiving effective supports and help and, ultimately, mental health outcomes following victimization (Kennedy et al., 2012). Thus, future person-centered research should examine associations among sexual assault disclosure, social reactions, and mental health outcomes in more diverse samples.

The cross-sectional and retrospective design of the current study is another limitation that deserves attention. The nature of our data limits the ability to examine the directionality of observed associations. The types of reactions the different profiles were characterized by may influence how victims make sense of their sexual assault experience and assign blame for the assault and whether they acknowledge the experience as a sexual assault. However, it is also likely that self-blame, perpetrator-blame, and sexual assault acknowledgment influence when, how, and to whom survivors disclose, and reveal details about their sexual assault experience in ways that elicit different reactions from disclosure recipients. Bidirectional relations likely exist between social reactions to disclosure and victim’s assignment of blame and acknowledgment of assault. In addition, assault characteristics (e.g., injuries acquired during the assault) may reflect more severe victimization experiences and may influence post-assault mental health outcomes. Future research that combines person-centered analytic approaches with longitudinal study designs may clarify how these relations unfold over time. Use of latent class/profile membership to predict outcome variables with analyses such as ANOVA does not take into account the uncertainty in class/profile membership and may lead to biased estimations (Asparuhov & Muthén, 2014; Cheng, 2012). Thus, future research would benefit from using other approaches to compare mental health outcomes between profiles of disclosure reactions.

It should also be acknowledged that victims do not always appraise the reactions they receive to sexual assault disclosure in the same ways that researchers categorize them as “negative” or “positive” (Ahrens et al., 2009; Dworkin et al., 2018). Our labeling of reactions as negative or positive may not fully reflect how victims perceive these reactions in terms of being helpful or harmful. Thus, how victims themselves appraise these reactions continues to be an important consideration in relation to post-trauma outcomes (Dworkin et al., 2019). Additional research utilizing person-centered statistical methods may help identify whether there are certain assault characteristics or other factors (e.g., relationship to disclosure recipient) that may determine perceptions of social reactions as positive or negative among victims.

Practice Implications

Findings from the current study are consistent with previous observations of the association between negative reactions to disclosure of sexual assault and psychological distress among victims (Dworkin et al., 2019). Given that sexual assault remains a highly stigmatized experience (Kennedy & Prock, 2016), victims may be more sensitive to negative versus positive reactions. Our results highlight the importance for both formal (e.g., mental health and medical professionals) and informal (e.g., friends or partners) disclosure recipients to understand and be cognizant of the significant impact of negative reactions and to take care to avoid responding in such ways (regardless of their intentions). There is a need for education and intervention efforts on college campuses targeted at possible (formal and informal) disclosure recipients that raise awareness of the types of reactions that are harmful to victims and aim to reduce the negative social reactions and increase supportive responding. Notably, almost all of the women in our sample had disclosed their experiences of sexual assault to a friend, highlighting the importance of such interventions with college students. Preliminary findings show promising results, suggesting intervention may be effective in reducing general negative social reactions and using alcohol to cope (Edwards, Waterman, et al., 2020; Edwards & Ullman, 2018). College campuses may consider including similar interventions in their violence prevention programming that specifically address reactions to sexual assault victims. Clinicians in college counseling centers working with sexual assault survivors must take great care to avoid reacting negatively to disclosures. Further, clinicians can help victims decide whether to share their experience with others in their lives and help identify individuals who are likely to respond in supportive ways. A broader societal shift is also imperative to eliminate stigma and stigmatization of victims and instead hold perpetrators accountable for the crimes of sexual violence.