Correlates of Disclosure Cessation After Sexual Assault

Abstract

Contacts with responders after sexual assault may influence further disclosure, but this possibility has not been explored empirically. Thus, this study investigates associations between survivors’ contacts with responders and their decisions to discontinue disclosure. Fifty-four college students with a history of unwanted sexual experiences described 94 ordered contacts with responders. Results indicate that survivors’ perceptions of responsiveness were not associated with continued disclosure, but survivors were more likely to continue disclosing when they perceived more rape myth acceptance from responders and when the assault was more recent. These findings highlight survivors’ tenacity in meeting their needs, even after problematic responses.

After sexual assault, survivors must decide whether to disclose to a variety of responders¹ (Ullman, 2010). Each time survivors disclose to a responder—whether informal (e.g., friends, family) or formal (e.g., therapists, police)—they may receive information about other help, tangible aid in accessing other responders, and/or implicit or explicit encouragement or discouragement from making further disclosures. There is evidence that these contacts with responders influence whether and how survivors continue to disclose (Ahrens, 2006; Campbell, Greeson, Fehler-Cabral, & Kennedy, 2015). However, the sexual assault literature has focused primarily on correlates of contacting particular types of responders (e.g., reporting the assault to the police), first and last disclosures, and characteristics of overall help-seeking experiences (e.g., whether or not survivors disclose to anyone), rather than understanding disclosures as steps in a process that might affect later steps. To address this gap, the current study investigates the association between contacts with responders in survivors’ early disclosure experiences and their decisions to continue or discontinue disclosing.

Disclosures of sexual assault represent decisions to make the assault known to certain responders within the context of a broader help-seeking process. After sexual assault, survivors may have a variety of needs, including emotional support, legal assistance, and medical attention. It follows, then, that many disclosures are the initiation of explicit help-seeking efforts in an attempt to meet these needs (Dworkin, Pittenger, & Allen, 2016). Other survivors, particularly those without pressing needs, may make disclosures that are not explicit efforts to seek help, and instead represent efforts to inform others of recent life events or support other survivors who disclose to them. Because both types of disclosure involve survivors negotiating the aftermath of sexual assault, both are considered in the current study to be components of the post-assault help-seeking process. This process also involves other components beyond initial disclosures, such as ongoing contacts with community responders.

Typically, survivors make disclosures to informal responders (e.g., friends, family; Ahrens, Campbell, Ternier-Thames, Wasco, & Sefl, 2007; Fisher, Daigle, Cullen, & Turner, 2003; Ullman, 1996b, 1999; Ullman & Filipas, 2001). Fewer approach formal sources (Campbell, Wasco, Ahrens, Sefl, & Barnes, 2001; Fisher et al., 2003; Krebs, Lindquist, Warner, Fisher, & Martin, 2007; Paul, Walsh, et al., 2013; Ullman, 1996a; Wolitzky-Taylor et al., 2011). About a third do not tell anyone at all (Fisher et al., 2003; Krebs et al., 2007; Wolitzky-Taylor et al., 2011). When survivors decide to tell others, this process can involve multiple contacts with a variety of responders and can last for weeks to years (Ahrens, Stansell, & Jennings, 2010).

Despite this complexity, research has characterized only narrow elements of help-seeking processes. For example, research has attempted to enumerate facilitators and barriers of disclosure to informal responders and police (for a review, see Sabina & Ho, 2014). Other work has focused on understanding first (Ahrens et al., 2007) and last (Ahrens, 2006) responder contacts. Although this work is important to characterize specific elements of the help-seeking process, a recent review of this literature conducted by Sabina and Ho (2014) suggested that the focus on whether or not responders have been accessed in isolation paint a limited picture by not attending fully to the process through which survivors negotiate contacts with multiple responders. Indeed, recent theoretical frameworks have characterized disclosure and help seeking for violence against women as processes that unfold dynamically over time and are influenced by survivors’ changing needs and the reactions that they receive (Liang, Goodman, Tummala-Narra, & Weintraub, 2005; Sylaska & Edwards, 2014; Ullman, 2010).

Although these theoretical models suggest that the interactions that survivors have with responders after sexual assault are likely to influence later disclosure decisions, this topic has received minimal research attention. In one of the only studies to our knowledge that has addressed this topic, a recent qualitative study examined the pathways through which adolescent sexual assault survivors access help from formal systems and identified that, in many cases, contact with informal responders was related to continued engagement with help seeking (Campbell et al., 2015). However, adolescents who have been sexually assaulted are likely to have their help-seeking decisions regarding formal responders taken out of their hands (e.g., if a parent compels the adolescent to contact police), and research is needed to establish whether contacts with responders are associated with disclosure continuation for adult survivors. Furthermore, it is important to understand how the nature of offered social reactions might be predictive of disclosure continuation. The literature on social reactions to sexual assault has generally either focused on survivors’ perceptions of response quality or the actual type of response (e.g., victim blame) offered. Survivor perceptions and response type are overlapping but distinct constructs (Dworkin, Newton, & Allen, 2016) that may be differentially related to disclosure continuation.

First, the type of response offered by responders—aside from how these responses are perceived—might be associated with disclosure continuation or cessation. Responses involving the articulation of rape myths, including excusing the perpetrator, minimizing the seriousness of the assault, or blaming the victim (McMahon & Farmer, 2011), have been characterized as insensitive from a best-practice standpoint (Ullman, 2010). Indeed, in a sample of survivors who had been “silenced”—or discouraged from further disclosure by an early disclosure—all reported being blamed for their assault (Ahrens, 2006). In contrast, victim blame from informal responders has been related to an increased likelihood of disclosure to mental health professionals (Starzynski, Ullman, Townsend, Long, & Long, 2007), suggesting that survivors who experience victim blame (and perhaps, other forms of rape myth acceptance) may need additional support from novel sources. It is also possible that more helpful responses could discourage further disclosure by negating the need for further assistance seeking, but this has not been documented in the literature to date.

Second, survivors’ perception of the quality of social responses that they receive might be related to their continued disclosure. Survivors vary in the degree to which they view behaviors consistent with rape myth acceptance in a negative light (Ahrens & Aldana, 2012; Dworkin, Newton, & Allen, 2016). A minority of survivors view victim blame and other problematic responses as helpful or consistent with their needs, particularly if they blame themselves or have a close relationship with the responder. Thus, when understanding how interactions with responders might influence continued disclosure, it is important to consider both the actual behaviors engaged in by responders and survivors’ perceptions of these behaviors. Perceptions of response quality involve evaluations of the degree to which needed services or supports are provided and the degree of harmfulness/helpfulness of responses (both immediately in terms of discomfort/comfort and ultimately in terms of longer term impact; Campbell & Raja, 1999; Dworkin, Newton, & Allen, 2016). Broadly speaking, responses that are perceived negatively appear to discourage further disclosure (Staller & Nelson-Gardell, 2005; Ullman & Najdowski, 2011) by leading survivors to question the utility of further disclosure, increasing self-blame, or prompting survivors to doubt whether their experience counted as rape (Ahrens, 2006). Experiences with responders that are perceived as being harmful, for example, may lead survivors to expect that disclosure will continue to have negative consequences (Staller & Nelson-Gardell, 2005), and they may decide to disengage from a given responder and/or from seeking support from other responders (Ahrens et al., 2007; Ahrens et al., 2010; Ullman, 1999).

Survivors’ unfolding needs over time have also been identified as theoretical correlates of their disclosure decisions (Liang et al., 2005; Sylaska & Edwards, 2014; Ullman, 2010). One way to capture these changing needs is to investigate the level of need that survivors begin with as well as the passage of time as correlates of disclosure decisions. Although there are many possible ways to characterize survivors’ level of post-assault need, assault severity may be one important factor that determines the nature and urgency of survivors’ post-assault needs (e.g., resolution of distress, medical attention, offender accountability; Sylaska & Edwards, 2014). More severe assaults are expected to be associated with continued disclosure because they would be more likely to be associated with concomitant distress (Brown, Testa, & Messman-Moore, 2009). Then, as time passes, these needs would be expected to change. Ahrens and colleagues identified distinct groups of survivors based on their disclosure timing after sexual assault, including those who disclose only in the immediate aftermath of assault and those who continue to disclose over time (Ahrens et al., 2010). As the time since assault increases, the utility of contacting medical responders and the police may reduce because of the destruction of physical evidence, so survivors who disclose late may be left with only informal responders to contact. Furthermore, as time passes after sexual assault, distress typically decreases (Breslau, 2009; Weaver & Clum, 1995), which may reduce the need for continued help seeking. However, the degree to which the timing of contact with responders is associated with continued disclosure remains unexamined.

The Current Study

Although help seeking related to sexual assault has been widely characterized as a complex process that often involves a sequence of many contacts with responders, correlates of disclosure have been understood more narrowly (e.g., in relation to disclosure to particular responders). Examining the association between each contact with a responder after sexual assault and the continuation or cessation of disclosure could help to clarify the role of responders in affecting help-seeking processes. Therefore, the current study investigates correlates of survivors’ decisions to continue or cease disclosing in a sample of college students. By measuring characteristics of individual contacts with responders, we aim to determine whether aspects of each contact, as well as survivors’ changing needs over time, were associated with survivors’ continued disclosure. This study examined three central questions:

• Research Question 1: Is the perceived quality of responses—including both perceived responsiveness and rape myth acceptance—associated with disclosure cessation?

• Research Question 2: Is the timing of contact with responders associated with disclosure cessation?

• Research Question 3: Is assault severity associated with disclosure cessation?

Method

Data were collected through an online survey that took about 50 min to complete. This study received Institutional Review Board approval. Participants were undergraduates above the age of 18 who were recruited through notices posted through a psychology department subject pool. The study was described in these notices as a survey about “sexual experiences, personal beliefs, mental health, and experiences with community resources” and the consent form specified that potentially upsetting, traumatic, and/or unwanted sexual experiences would be queried in particular. Participants received course credit for participation.

Survey Measures

The online survey assessed a number of aspects of participants’ sexual experiences, contact with responders, and mental health. The following measures are relevant to the current analyses.

Sexual assault history and severity

The Sexual Experiences Survey–Short Form Victimization (SES-SFV; Koss et al., 2007) assessed participants’ history of sexual assault. This is the most widely used self-report measure of sexual assault and includes seven behaviorally-based items (e.g., “A man put his penis into my vagina, or someone inserted fingers or objects without my consent by … ”) representing a range of unwanted sexual behavior from sexual contact to completed rape experienced since age 14. Each of these items has five follow-up questions that inquire about tactics used (e.g., “Threatening to physically harm me or someone close to me”). All participants who endorsed any unwanted sexual experiences were asked which experience was the most meaningful to them and then were presented with questions about contact with responders related to that experience. Consistent with the SES-SFV scoring rules (Koss et al., 2008), a sexual assault severity variable was created to indicate the type of sexual assault that participants indicated was most meaningful (about which they answered follow-up questions). The mutually exclusive categories were sexual contact, attempted sexual coercion, sexual coercion, attempted rape, and rape.

Contact with responders

Participants who endorsed a history of sexual assault were asked to identify what types of formal responders (e.g., police, 911, therapist) they contacted as a result of their experiences by choosing from a list. They were then asked to name the first, second, and third formal responder whom they contacted. Each of these first three contacts with formal responders was supplemented with a series of follow-up questions (e.g., how long after the assault they contacted that responder, what the experience was like, when they ended contact with them, whether they contacted any additional responders after this responder). Similarly, they were asked to name up to three of the first informal responders whom they told about their experience. As with the formal responder contacts, participants were asked a series of descriptive follow-up questions about their contacts with informal responders. Finally, the (up to) three informal and (up to) three formal responders were listed together, and participants were asked to number these responders in the order that they contacted them. This limited the number of potential responder contacts that survivors could describe to six; although they were able to indicate that they contacted more than three informal or three formal responders, they did not answer follow-up questions about additional responders. Participants were able to indicate that some contacts occurred at the same time (e.g., a survivor told her mother and father at the same time); these contacts were combined into a single contact for the purpose of analyses and all variables describing their perception of their experiences were averaged. From this information, the researchers created a variable for each responder to represent whether or not participants subsequently contacted an additional responder after each responder contact (0 = continued disclosure, 1 = ceased disclosure). Survivors who indicated that they contacted additional responders after the final responder contact that they described in depth were not considered to have ceased disclosure.

For each responder contacted, the survey asked, “How long after your sexual experience did you have contact with [responder identity]?” Possible response options were 1 = within 24 hr, 2 = 1–3 days later, 3 = 3–7 days later, 4 = 1–4 weeks later, 5 = 1–4 months later, 6 = 4–8 months later, 7 = 8–12 months later, and 8 = 1 or more years later.

Among the follow-up questions asked regarding each contact with formal and informal responders, the survey asked a number of questions regarding survivors’ perceptions of the quality of responses. Based on literature suggesting that perceptions of responsiveness involve the degree to which desired responses are provided, how sensitive or insensitive responders are perceived to be, and how helpful those responses are seen to be in terms of short- and long-term impact (Campbell & Raja, 1999; Dworkin, Newton, & Allen, 2016), we created a four-item scale to assess responsiveness. First, to assess the degree to which responders provided expected services/supports, the survey presented the responder identity along with the following response options: “did not provide me any of the services/supports that I hoped for,” “provided me with some of the services that I hoped for,” “provided me with all of the services that I hoped for,” “provided me with more services than I hoped for,” and “I did not have any hopes about the amount of services that I would receive.” This final response option was treated as a “not applicable” response and contacts with these responders (n = 14) were excluded listwise, thereby reducing the responder-level sample size to 94. Second, to assess helpfulness, the survey presented the prompt, “Looking back, [responder identity] was …” with the following response options: very unhelpful, somewhat unhelpful, neither helpful nor unhelpful, somewhat helpful, and very helpful. Third, to assess responder sensitivity, the survey presented the prompt, “Looking back, [responder identity] treated me …” with the following response options: very insensitively and uncompassionately, somewhat insensitively and uncompassionately, neutrally, somewhat sensitively and compassionately, and very sensitively and compassionately. Fourth, to assess how comfortable or uncomfortable the response felt, the survey presented the prompt, “At the time, my experience with [responder identity] felt …” with the following response options: very upsetting and/or uncomfortable, somewhat upsetting and/or uncomfortable, comfortable, and very comfortable. These four items were averaged to create a perceived responsiveness score for each responder. Internal consistency reliability was α = .89.

To assess rape myth acceptance of responders separate from survivors’ perceptions of whether responders treated them sensitively, we asked four questions that corresponded to the four domains of rape myth acceptance from the Illinois Rape Myth Acceptance Scale–Revised (i.e., “she asked for it,” “he didn’t mean to,” “it wasn’t really rape,” and “she lied”; McMahon & Farmer, 2011). Although these domains represent unique factors related to rape myth acceptance, they have been found to load strongly on a higher-order factor, and the internal consistency reliability of the overall scale is high, suggesting that items related to these distinct factors represent a single higher order construct (McMahon & Farmer, 2011). Specifically, the survey used the prompt, “I got the sense that [responder identity] …” with the following items: “felt that I was to blame for my sexual experience,” “felt that the person with whom I had the sexual experience was not to blame for what happened,” “thought that my sexual experience was not sexual assault,” and “did not believe me.” Participants rated each item on a three-point Likert-type scale (response options: 1 = not at all, 2 = somewhat, 3 = very much). These items were averaged to create a rape myth acceptance score for each responder. Internal consistency reliability was α = .62.

Data Reduction

From 790 participants, we selected the 206 participants who endorsed an unwanted sexual experience, ranging from unwanted sexual contact to completed rape, that had occurred in adolescence or adulthood (i.e., since age 14).

We investigated how and whether disclosure cessation could be deduced from survey responses. Presumably, at least some survivors of sexual assault in our sample would continue to disclose after they completed the survey. Complicating this issue, many participants described sexual assaults that had occurred quite proximally in time to their survey participation. Saying that these participants’ last described contact with a responder in the survey is truly the last disclosure would likely be invalid. Thus, we explored the time period in which most disclosures were made—in effect, the “critical period” of frequent disclosure—and limited the analytic sample to those who were assaulted longer than this amount of time ago to minimize the likelihood of further contacts after this point in time. By doing so, we could then have reasonable confidence that the last described contact in the survey was the last in the critical period of disclosure. Most contacts (65.9%) were made within a year of the assault, and the likelihood of contact with responders decreased over the year post-assault, although many contacts were made more than a year after the assault. Based on this information, analyses were conducted only on those participants whose most meaningful assault (about which follow-up questions regarding responder contact were asked) occurred at least a year prior to taking the survey, with the knowledge that some might have continued to disclose. This involved the exclusion of 106 participants whose most meaningful assault occurred less than a year before taking the survey. Rather than interpret the dependent variable as representing a true or permanent termination of disclosure, this was interpreted as the termination of this process within the critical period of disclosure. Participants assaulted in the past year did not differ from participants assaulted more than a year prior to survey completion in assault severity, t(185) = 1.10, p = .27, although those with more recent assaults tended to be significantly younger than those with less recent assaults, t(130.90) = 5.16, p < .001, perhaps because older participants had a larger range of time from which a meaningful assault could have occurred (e.g., 18-year-old participants could have experienced an assault in the past 4 years whereas 22-year-old participants could have experienced an assault in the past 8 years). When considering survivors’ perceptions of responders, responders contacted by survivors assaulted more than a year prior did not differ on responsiveness, t(254) = −0.89, p = .38, but were seen as endorsing rape myths significantly more strongly than responders contacted by survivors assaulted within the year before taking the survey, t(251.92) = −2.35, p = .02.

Of the 98 participants whose assault occurred at least a year prior to participation, we excluded 40 participants who did not indicate that they contacted at least one responder related to the sexual assault. Four additional participants had missing data on at least one focal variable and were excluded from analyses, leaving a final sample size of N = 54.

Analyses

The clustered structure of the data required analytic methods that accounted for the correlation within a given participant’s responders to avoid violating assumptions of independence of observations (Raudenbush & Bryk, 2002). Marginal models (e.g., as estimated using generalized estimating equations) and conditional models (e.g., as estimated using mixed effects logistic regression) both can predict dichotomous outcomes (i.e., dropout or no dropout) using correlated, clustered observations (Agresti, 2007). Like multilevel models, generalized estimating equations account for clustering in data. Unlike multilevel models, which model random effects, generalized estimating equations average over random effects and account for clustering through corrections made to the standard errors (McNeish, 2014). Generalized estimating equations are preferred over multilevel models when participants have few observations (i.e., when the data are “sparse”) because they are less likely to overestimate between-group variance, and when making cluster-specific inferences is not a goal of the study (McNeish, 2014). In the current study, survivors had between one and three observations, and making survivor-specific inferences was not a goal, which justified the use of a marginal model that took repeated observations into account. Thus, we used PROC GENMOD in SAS® 9.4 to model a generalized estimating equation that accounted for repeated observations.

Results

There were 94 documented contacts with responders corresponding to 54 participants (one to three ordered responder contacts per person) who were assaulted a year or more prior to the survey and had sufficient data to be included in at least basic analyses. All results describe this subsample.

Participants were mostly women (88.89%) and heterosexual (94.44%), with an average age of 20.39 (SD = 1.56 years, range = 18–26 years). Most participants indicated that they were White (66.67%); 7.41% identified as Black/African American, 12.96% identified as Latino/a/Hispanic, and 16.67% identified as Asian or Asian American. As participants could select as many racial groups as they wished, these figures are not mutually exclusive; indeed, 9.26% of the sample identified as multiracial. Most (64.71%) of those who described a first contact described a second contact and 30.30% of those who described a second contact described a third contact. Although none described a fourth contact, all (n = 22) those who described a third contact endorsed telling more informal responders than three, so they were assumed to have persisted in disclosure after the third contact. All but four contacts were with informal responders, limiting our ability to test how the type of responder might be associated with continued disclosure.

The central focus of the current study (Research Question 1) was to examine how experiences with each responder were associated with disclosure cessation. See Table 1 for descriptive statistics regarding survivors’ perceptions of their experiences with responders. Survivors tended to see responders as highly responsive and responders tended to act in a manner inconsistent with rape myth acceptance. As expected, perceived responsiveness was significantly negatively correlated with rape myth acceptance, r = −.47, p < .01, such that responders who were seen as more responsive were less likely to behave in a manner consistent with rape myth acceptance. In bivariate analyses, only perceptions of higher rape myth acceptance and time between assault and responder contact were significantly associated with disclosure continuation, such that those who continued to disclose after contact with a responder tended to rate that responder as having higher rape myth acceptance, and responder contacts made later after the assault were less likely to be followed by additional contacts with responders.

Table 1.

Descriptive Statistics.

Responder-level variables (N = 94)	Minimum-maximum	M (SD)	Association with continued disclosurea	Correlations
				1	2
1. Perceptions of responsiveness	1 (very unresponsive)-4.75 (very responsive)	3.92 (0.89)	t(92) = 0.06
2. Rape myth acceptance of responder	1 (low rape myth acceptance)-2.75 (high rape myth acceptance)	1.22 (0.31)	t(91.66) = −2.18**	−.47**
3. Time between assault and contact with responder	1 (within 24 hr)-8 (1 or more years later)	3.54 (2.77)	t(58.02) = 2.83**	−.20	−.19
Survivor-level variables (N = 54)	Minimum-maximum	M (SD)		Correlations
				1	2
1. No. of responder contacts described	1–3	1.80 (0.71)
2. Severity of focal assault	2 (sexual contact)-6 (completed rape)	4.11 (1.64)		.00

^a0 = continued disclosure and 1 = ceased disclosure.

^**p < .01.

The final model is displayed in Table 2. With regard to Research Question 1, greater rape myth acceptance was a statistically significant correlate of continued disclosure (odds ratio [OR] = 0.16, p = .04), whereas perceived responsiveness was not a significant correlate of disclosure continuation. Addressing Research Question 2, a one-unit increase in how long after the assault the contact occurred was associated with a 22% increase in odds of cessation (OR = 1.22, p = .01), indicating that disclosures made longer after the assault were associated with a decreased likelihood of disclosure continuation. Finally, relevant to Research Question 3, assault severity was not significantly related to disclosure continuation.

Table 2.

Model Predicting Disclosure Cessation.

		95% confidence interval
Variable	OR (SE)	LL	UL	p
Intercept	5.83 (8.80)	0.08	413.85	.42
Rape myth acceptance of responder	0.16 (2.47)	0.03	0.95	.04
Perceived responsiveness	0.89 (1.36)	0.49	1.62	.71
Time between assault and contact with responder	1.22 (1.08)	1.05	1.42	.01
Severity of most meaningful type of assault	0.88 (1.15)	0.67	1.17	.38

Note. The dependent variable is coded as 0 = continued disclosure and 1 = ceased disclosure. Thus, ORs > 1 represent increased odds of dropout, whereas ORs < 1 represent increased odds of continuation.

LL = lower limit; UL = upper limit; OR = odds ratio.

Discussion

Help seeking after sexual assault is a highly complex process that includes multiple interdependent disclosures to a variety of responders. In the first in-depth analysis of sequences of disclosures after sexual assault, our findings provide new insight into the factors that might promote or discourage continued engagement with help-seeking processes. Whereas past research has focused primarily on single help-seeking decision points (e.g., first disclosures, last disclosures, contacting the police), this study takes a broader perspective to understand the characteristics of survivors and responders that are associated with the continuation of disclosure after adolescent or adult assault. Importantly, findings suggest that disclosure cessation is associated with survivors’ perceptions of responders’ behaviors, and disclosure appears to decrease over time.

Survivors’ interactions with responders after sexual assault appear to be related to their continued help seeking (Research Question 1). Survivors who contacted a responder who were perceived to have acted in a manner consistent with rape myth acceptance were more likely to continue making disclosures than those who disclosed to a responder who expressed validation, belief, and perpetrator blame. Responses that are supportive of rape myths (e.g., victim blame, denying that the experience was sexual assault) may be distressing, thus creating a need for more or other services/supports or leading survivors to continue looking for a helpful response. For example, survivors who are told that what happened to them was not truly sexual assault may seek validation of their experiences elsewhere. In this sense, the “silencing” documented by past research (Ahrens, 2006)—or initial disclosures that result in disclosure cessation—may not always represent a negative process. Instead, it might indicate that survivors’ needs have been adequately addressed, perhaps through the validation of their experiences, and no further help seeking is needed. Although it is somewhat surprising that survivors’ perceptions of the responsiveness of the people with whom they interacted—in terms of their sensitivity, the comfort of responses, and the degree to which responders provided the services/supports that they hoped for—were not a significant predictor of continued disclosure, it is possible that some responses were effective at resolving survivors’ needs, leading to disclosure cessation, whereas others led survivors to feel optimistic that future responders would respond positively, leading to disclosure continuation. It is also possible that the sample size of this study was not sufficient to detect an effect for this relationship. Taken as a whole, though, these results highlight the tenacity of survivors meeting their own needs even in the face of problematic responses.

It is notable that, consistent with previous research (Ullman, 2010), perceptions of responders were generally quite positive; very few participants reported having negative contacts with responders. Given the potential harm that can be caused by negative responses (Ullman, 2010), this is a heartening finding. Also consistent with past research, most responders in this sample were informal responders (e.g., friends, family), whereas few were formal responders (e.g., police, health care). Although the low frequency of contact with formal responders was expected based on past research (Campbell et al., 2001; Fisher et al., 2003; Paul, Zinzow, McCauley, Kilpatrick, & Resnick, 2013; Ullman, 1996a; Wolitzky-Taylor et al., 2011), it precluded testing the impact of contact with formal versus informal responders.

The timing of help seeking also appears to be related to disclosure cessation (Research Question 2). Consistent with our expectations, disclosures made later after the assault were less likely to be followed by further disclosures, which likely reflects survivors’ changing needs as time passes. This is consistent with past findings that a subset of survivors discloses only in the immediate aftermath of assault (Ahrens et al., 2010). Because these later contacts were made more proximally to survey completion, it is possible that they do not represent a true dropout from disclosure—indeed, in the only study to assess patterns in timing of disclosure after assault, some survivors continued to disclose over long periods of time (Ahrens et al., 2010)—but it is still likely that contacts with others about an assault decrease in frequency over time as the salience of the assault and/or need for help decrease.

Surprisingly, this study did not provide support for the idea that assault severity relates to disclosure continuation (Research Question 3). In the current study, assault severity was considered a proxy for survivors’ level of need at the onset of disclosure. Although the extant literature suggests that more severe assaults may be associated with greater need because of increased distress and ability to identify that what happened was assault (Brown et al., 2009; Dworkin, Menon, Bystrynski, & Allen, 2016; Kilpatrick, Resnick, Ruggiero, Conoscenti, & McCauley, 2007), more severe assaults might also be associated with increased stigma, which could reduce the likelihood of disclosure. Furthermore, survivors with less severe assaults may not have fewer needs than those with more severe assaults, but instead may have different needs (e.g., understanding what happened, participating in discussions of sexual experiences) that may still promote contact with responders. It is possible that assessing other areas of need (e.g., physical injury, desire for a criminal justice response) could better capture the variety of survivors’ needs as they relate to disclosure continuation. Survivors’ level of need might also be associated with dropout later in help-seeking processes (i.e., beyond the first three contacts). Because we focused only on early contacts with responders, it was not possible to detect these effects for the 41% of survivors in the current study who indicated that they continued to disclose after their third contact with a responder. It is also possible that our sample size was not large enough to detect these effects.

Limitations

The cross-sectional, retrospective nature of our study design limited the conclusions that can be drawn from this work for several reasons. First, although the subsample analyzed only included those who were assaulted more than a year prior, it is possible that those who “dropped out” of help-seeking processes ultimately made more disclosures after taking the survey. This choice also reduced our sample size, which may have limited the power necessary to detect some effects. Even with a reduced sample size—in which the chance of Type II error is greater —we were able to detect effects, strengthening our confidence in the conclusions of this study despite its small sample size. However, additional research is needed to confirm the null effects. Second, the perceptual variables were limited by the retrospective nature of our data collection; it is possible that evaluations of the responses are affected by the ultimate outcome of the assault. For example, a survivor who experienced significant traumatic stress might see contacts with responders as more ineffective at resolving her needs than one who did not experience this degree of traumatic stress. It is also possible that memory problems could interfere with the accuracy of reporting. Longitudinal investigations could resolve these limitations but are also subject to methodological limitations (e.g., observer effects, recruitment challenges). Third, because consent procedures specified that the study measures included an assessment of sexual assault history, it is possible that survivors might have been more likely to participate in this study— which is, itself, an additional disclosure—depending on the past responses that they received. Although responses received were generally viewed positively, this is consistent with past research on survivors’ perceptions of social responses, which suggests that we sampled a representative range of past disclosure experiences (Ullman, 2010). Fourth, we were necessarily limited in the detail with which we could measure survivor perceptions of responders by the length of the survey. Because we were asking participants to report on (up to) six responder contacts in the context of a long survey, we were conscious of limiting the number of questions asked about each responder to reduce participant burden. Thus, we were unable to assess survivor perceptions of responsiveness and rape myth acceptance in great detail, which precluded investigations of (for example) how different domains of rape myth acceptance might have differentially related to disclosure continuation. Future research could investigate this possibility. Fifth, this study used a convenience sample of undergraduates, who—although they are particularly vulnerable to sexual assault (Krebs et al., 2007)—may not be representative of broader populations of sexual assault survivors. However, our sample was relatively racially diverse and had experienced a range of assault types. Furthermore, improving campus responses to sexual assault has been a focus of recent efforts to reform policy and practice, so the results from this study are well-positioned to inform such efforts.

Research Implications

Although this study provides evidence for the importance of experiences with responders in survivors’ help-seeking processes, more work is needed. Generally, research using larger samples would be useful to capture variation in responder quality and types. In particular, research including a larger subsample of survivors who had disclosed to formal responders would be useful. This research could also more directly assess survivors’ needs and intent in disclosing the assault to various responders, and then assess the degree to which responders met these specific needs, to assess the construct of helpfulness in a more precise manner. It could also determine whether similar factors are associated with help-seeking decisions after adult and adolescent assault, which we were not able to explore given our limited sample size and relatively low prevalence of disclosure to formal providers (because disclosure to formal providers would be more likely to be influenced by outside forces among adolescents; Campbell et al., 2015). In addition, given that this research suggests that some survivors might stop disclosing when their needs are effectively met, exploring the circumstances under which this might occur using qualitative methods could be useful. Finally, research could explore the timing and frequency of disclosure longitudinally, given the evidence from this work that disclosure decreases over time, and attempt to ascertain the point at which survivors stop telling additional people with more certainty. This research could explicitly ask survivors whether they intend to tell additional responders at the end of the data collection period, although whether participants could accurately report on their future behavior is uncertain.

Practice Implications

In most cases, disclosure of sexual assault is likely to be a necessary precondition for meeting the dual goals of promoting survivor well-being and offender accountability/community safety. For example, psychological treatment for trauma related to sexual assault experiences requires the disclosure of these experiences, and offenders typically cannot be identified and arrested without survivors’ disclosures to police. Thus, it is important to use the findings of the current study regarding the unfolding nature of the disclosure process to understand how to keep survivors engaged in these processes if they so choose. However, it is also important to acknowledge survivors’ agency in making decisions about their own healing processes and the presence of circumstances under which disclosures might have diminishing returns (e.g., a high likelihood of future invalidation). Knowing that disclosures decrease over time, it is also important to direct efforts to maximize engagement in help seeking in the early aftermath of an assault. Given that these findings suggest that rape myth endorsement plays an important role in decisions about whether to continue disclosing, practitioners working with survivors of sexual assault are encouraged to attend to the types of implicit rape myths that they may be communicating through their actions. In addition, efforts to improve peer responses to sexual assault should continue to encourage potential responders to avoid, for example, questioning the assault in a way that could be perceived as denying its seriousness.

In conclusion, the current study provides new evidence for the complexity of survivors’ help-seeking processes after sexual assault. Although the study design precludes causal conclusions, results suggest some potential mechanisms by which each contact with a responder might influence disclosure continuation, including the degree to which responders are perceived to be responsive or supportive of rape myths.

Author Biographies

Emily R. Dworkin, PhD, is a postdoctoral fellow at the University of Washington School of Medicine. She received a PhD in clinical/community psychology from the University of Illinois at Urbana–Champaign and completed her clinical internship at the University of Mississippi Medical Center/VA Medical Center. Her research investigates the recovery experiences (e.g., help-seeking decisions, interactions with community responders and social networks) of survivors of sexual assault and other traumas as they relate to well-being, with a focus on the development of PTSD and substance use disorders.

Nicole Allen, PhD, is a professor of psychology at the University of Illinois at Urbana– Champaign. She received her PhD in ecological/community psychology from Michigan State University. Her research program has four major thrusts: (a) to examine systems change processes in the response to social issues and the role of collaborative settings in facilitating such change, (b) to explore the experiences of individuals as they navigate complex systems, (c) to investigate the effectiveness of social interventions that aim to alter the contexts of individuals’ lives to promote health and well-being, and (d) to examine the gendered nature and context of intimate partner violence.