Women’s Self-Report of Sexual Victimization: An Experimental Examination of the Influence of Race,Mode of Inquiry,Setting,and Researcher Contact

Abstract

Female college students (N = 255) provided self-report of adult sexual assault (ASA) and child sexual abuse (CSA) in one of eight randomly assigned conditions which manipulated three methodological variables (level of researcher contact, setting, and inquiry mode). The impact of methodology on reporting was examined, with race as a moderator. Non-White participants reported significantly more CSA than White participants out of lab, but differences were not observed in lab. White participants reported significantly more ASA than non-White participants on pencil-and-paper surveys, but there were no differences on computer surveys. Interpretation of race-related differences in self-reported victimization may require consideration of methodological effects.

Keywords

methodology sexual aggression rape sexual assault child sexual abuse race experimental design

Introduction

Prevention, prosecution, and treatment efforts aimed at curbing the negative effects of sexual aggression (e.g., experiences of adult sexual assault [ASA] and child sexual abuse [CSA]) depend largely on obtaining accurate research information about the prevalence, correlates, and consequences of these experiences. Unfortunately, from a research standpoint, it has long been recognized that sexual aggression research poses a daunting methodological challenge (e.g., Smith, 1987, as cited in Cook, Gidycz, Koss, & Murphy, 2011). Instances of sexual aggression typically occur in private and, thus, are witnessed solely by the victim and the perpetrator. As such, victim or perpetrator self-report is perhaps the only realistic way to assess prevalence and characteristics of sexual aggression. There are many challenges in identifying representative samples of sexual aggression perpetrators because admitting to perpetration is socially undesirable and could even risk legal consequences; thus, victim self-report often becomes the focus of research (Strang, Peterson, Hill, & Heiman, 2013). Even when researchers rely on criminal databases or institutional (e.g., military, university, criminal justice) data, researchers are indirectly tapping into victim’s self-report (Cook et al., 2011). Thus, the current study aimed to examine methodological factors that may influence victims’ self-report of sexual victimization; understanding these factors may help researchers to collect more useful data about the prevalence and characteristics of sexual aggression.

Although reliance on self-report is all but unavoidable in sexual aggression research, the method is certainly not without its limitations. To accurately measure instances of sexual victimization, researchers must recruit samples that include victims of sexual aggression, victims must recall instances of aggression when prompted, and victims must be willing to disclose recalled events to researchers (Cook et al., 2011). Bias may be introduced at any of these steps in the self-report process and could result in skewed conclusions about the rates and characteristics of sexual aggression. Social desirability is one potential source of bias which has been proposed (e.g., Kim, Dubowitz, Hudson-Martin, & Lane, 2008). It has been suggested that individuals may consciously edit their responses about sexual victimization experiences to avoid negative social judgments. For example, commonly held myths about rape blame victims by suggesting that women who are raped were “asking for it” (Peterson & Muehlenhard, 2004); thus, women may not want to admit to sexual victimization for fear of being blamed for the experience. Although the social desirability literature related specifically to sexual victimization is limited, there is some indication that social desirability may affect disclosure of intimate partner violence (IPV) victimization (e.g., Bell & Naugle, 2007; Sugarman & Hotaling, 1997).

In recognition of the potential for bias, a substantial literature has been devoted to developing and improving self-report measures of sexual victimization (e.g., Cook et al., 2011; Fisher, 2009; Follingstad & Rogers, 2013; Koss, 1992). Much of this research has centered on evaluating the impact of question wording and ordering effects on rates of sexual victimization disclosure. Arguably, these efforts have been fruitful, in that use of behaviorally specific measures, such as the Sexual Experiences Survey (SES; Koss et al., 2007; Koss, Gidycz, & Wisniewski, 1987), has resulted in consistently higher rates of rape disclosure than non-behaviorally specific questions, which had often been used previously (i.e., simply asking participants whether they have experienced “rape” or “sexual assault”). Compared with non-behaviorally specific measures, these behaviorally specific questions also more clearly map onto the specific experiences targeted (e.g., legally defined rape) independent of social judgments or interpretation of the acceptability of the behavior (e.g., Cook et al., 2011; Fisher, 2009).

In spite of innovations in measurement of sexual victimization, there may still be means for further improving self-report methodology. It has been suggested that by expanding our consideration of factors beyond measure construction, it may be possible to identify additional methodological factors which affect participants’ decisions related to disclosure of sexual victimization (Fisher, 2009). Methodological factors such as inquiry mode (e.g., pencil-and-paper, computer, face-to-face interview), amount of contact with researcher, and setting of completion (e.g., in lab, out of lab) may differentially influence response biases, such as social desirability, and may ultimately affect participant willingness to disclose experiences of sexual victimization.

Inquiry Mode

Many studies have attempted to examine the impact of inquiry mode on participants’ self-report of a broad range of sensitive content. One of the more consistent findings related to mode of inquiry and self-report is that participants are generally less willing to disclose sensitive information in face-to-face interviews than they are through pencil-and-paper surveys or computer-administered self-interviews (CASI; e.g., Richman, Weisband, Kiesler, & Drasgow, 1999). However, the most commonly hypothesized explanation for this effect is not really related to mode of inquiry but rather to researcher contact; that is, the presence of a researcher in face-to-face interviews serves as a stronger pull for socially desirable responding than a faceless survey packet or computer screen (McCallum & Peterson, 2012). Other studies that have compared pencil-and-paper surveys with equivalent computer-based surveys have provided mixed results for mode-related differences in self-report of sensitive behavior (e.g., for a more thorough review of inquiry mode and self-reported sexual behavior, see McCallum & Peterson, 2012).

Although many studies have examined mode-related differences for self-reported sexual behavior broadly, relatively few have looked directly at self-report of sexual victimization. Currently, the limited research related to mode of inquiry and questions regarding non-consensual sex have yielded mixed findings. There is some indication that women are more likely to disclose sexual assault related to alcohol use through a web survey than they are in a phone interview (Parks, Pardi, & Bradizza, 2006). Also, a study conducted with pregnant women presenting to a prenatal clinic found higher disclosure rates of IPV on a computer-based screener than during an interview in an initial visit with a provider (Chang et al., 2012). Another study indicated higher rates of sexual assault disclosure by participants through pencil-and-paper inquiry than through CASIs (Testa, Livingston, & VanZile Tamsen, 2005). However, the latter study was limited by low response rates in the computer condition, which illustrates the need to consider inquiry mode-dependent participation bias, as only 61.4% of contacted participants showed up for the computer condition in comparison with the 87.6% of participants who completed and returned pencil-and-paper surveys. Yet another study of sexual victimization disclosure found no difference in disclosure rates via computer-based, face-to-face interview, or pencil-and-paper modes of inquiry, though participants did indicate a preference for computer-based reporting (DiLillo, DeGue, Kras, Di Loreto-Colgan, & Nash, 2006). Thus, there is some limited support for the idea that computer-administered surveys may result in higher rates of disclosure (Chang et al., 2012; Parks et al., 2006) and be better tolerated by participants (DiLillo et al., 2006) than interview or paper-and-pencil surveys. However, it remains unclear why the effects of mode do not appear consistent across studies. One explanation that has been proposed is that additional underlying methodological factors (e.g., amount of contact with the researcher and/or the setting of completion), which are conflated with mode, are differentially affected by study designs and, in turn, contribute to inconsistent results (McCallum & Peterson, 2012).

Researcher Contact

Broadly speaking, it has long been recognized that participants’ level of contact with a researcher can affect the degree to which participants feel comfortable disclosing sensitive information (e.g., Jourard & Friedman, 1970). This relationship is particularly relevant when considering disclosure of sexual victimization, given that fear of negative social judgment has been identified as a factor that governs victims’ willingness to disclose these experiences (e.g., Sable, Danis, Mauzy, & Gallagher, 2006). Although level of researcher contact is not typically manipulated intentionally in the context of sexual victimization research, it is often affected by the inquiry mode selected (McCallum & Peterson, 2012). As noted above, researchers have consistently observed higher rates of self-reported sexual behavior in lower contact modes such as pencil-and-paper and computer surveys, than in higher contact modes, such as telephone-based or face-to-face interviews, in which participants interact directly with researchers to complete survey questions (e.g., Reddy et al., 2006). Differences in researcher contact often have been presumed to account for these mode-related differences. We are not aware of any published studies specifically related to level of researcher contact and disclosure of sexual victimization, but it is plausible that lower rates of researcher contact might promote higher rates of disclosure.

Setting of Completion

Participants are often given the freedom to complete pencil-and-paper and Internet-based surveys in a setting of their choosing, though the impact of the setting selected has rarely been evaluated. In one recent study, researchers found that participants who were given the opportunity to complete surveys wherever they chose opted to do so in situations with a range of potential social distractors (e.g., in the presence of family or friends or in the midst of a lecture; Hardré, Crowson, & Xie, 2012). Furthermore, researchers found a mode-dependent difference, which suggested that participants completing pencil-and-paper measures were more likely to select private locations than those completing computer-based surveys. Although there are no studies of which we are aware that have looked at the impact of setting on self-reports of sexual victimization, it is possible that the aforementioned social distractors and lack of privacy impact responding much like researcher contact, with participants decreasing disclosure of victimization due to concerns about sensitive information being revealed to those in the vicinity. Thus, lab-based surveys may ensure greater privacy and thus higher rates of disclosure.

Race

The impact of race on general self-disclosure in research has long been recognized (e.g., Jourard & Lasakow, 1958). Non-white participants are often found to be more resistant to disclosure of personal information in the context of a research interview or survey questionnaire, particularly when questions relate to sensitive behaviors or experiences (e.g., Johnson & Richter, 2004). Furthermore, several studies have shown differences in the effects of mode on disclosure as a function of race (Aquilino, 1994; Aquilino & LoSciuto, 1990; Hewitt, 2002). Although the modes being compared (e.g., CASI, phone interview, face-to-face) and the behaviors being examined (e.g., drug use, number of sexual partners) varied across studies, the observation of larger mode effects among non-White participants has been consistent. However, setting of completion and researcher contact was confounded within the modes being compared in existing research; these factors were not considered independent of mode, and their unique impact cannot be evaluated.

Unfortunately, there are no published studies of which we are aware that have considered race as a moderator for the impact of inquiry mode, researcher contact, or setting on sexual victimization disclosures. However, there is conceptual justification for these potential relationships. One proposed explanation for the observed differences is a race-dependent difference in beliefs concerning the trustworthiness of others (perhaps due to different levels of contact with discrimination among White and non-White participants), which translates into greater concerns related to confidentiality among non-Whites in some inquiry modes (Aquilino, 1994). However, this hypothesis was only partially supported in an examination of confidentiality concerns and mode among African American, Hispanic White, and non-Hispanic White participants (Aquilino, 1994), suggesting that other factors also may contribute to race-related differences in disclosure across methodologies.

As previously noted, people are often resistant to disclose sensitive information to researchers regardless of race, out of concerns that their responses may not remain confidential, or that they may be judged negatively for the information that they provide. These concerns may be magnified for non-White participants who likely have had substantial direct experiences with prejudicial judgment. In addition, non-White participants may also be concerned that the information provided will further fuel existing racial stereotypes, and these concerns may lead to higher awareness of and motivation for impression management efforts (e.g., Johnson & Richter, 2004). Although not directly related to mode of inquiry or victimization disclosure, there has been some bogus pipeline research that suggests greater impression management efforts among non-White minority groups. The bogus pipeline is a social psychology paradigm that utilizes a fake lie detector device to convince participants that dishonest answers will be detected, thereby motivating more honest responding. When comparing responses in a bogus pipeline and a standard testing condition, researchers observed that African American participants were more likely than White participants to inflate self-reports of their overall self-esteem under standard testing conditions; in other words, the African American participants showed a greater discrepancy than White participants in responses in the bogus pipeline versus the standard testing conditions (Zeigler-Hill, Wallace, & Meyers, 2012). Given the limited research that is available, it seems plausible that any methodological factor that affects participants’ willingness to disclose may have an even greater effect for non-White participants as compared with effects observed for White participants.

Present Study

Our primary aim in this study was to examine the relationship between methodological factors—inquiry mode, researcher contact, and setting—and self-report of sexual victimization. By considering these three factors in parallel, we can better isolate the unique impact of each on rates of victimization disclosure. Consideration of race as a potential moderating factor will serve to further advance the understanding of the impact of methodological factors on disclosure as well.

In this study, we examined the impact of methodological variables specifically on women’s self-reported history of CSA and ASA. To assess the impact of mode, women in our study were assigned to complete either pencil-and-paper or computer-based surveys. To assess the impact of researcher presence, participants were assigned to receive high or low researcher contact. To assess the impact of setting, participants were assigned to complete the surveys in a research lab or out of the lab (in a location of their choosing). Thus, there were eight possible conditions to which participants were randomly assigned, represented in a 2 (paper vs. Internet) × 2 (high vs. low researcher contact) × 2 (in-lab vs. out-of-lab completion) research design.

Hypotheses

We predicted that inquiry mode, researcher contact, and setting would all uniquely affect participants’ self-reported history of CSA and ASA. Based on the literature described above and based on our assumption that conditions that promoted a greater sense of confidentiality and anonymity would result in more open reporting of victimization histories, we predicted that computer-based inquiry mode, low researcher contact, and out-of-lab completion would promote higher reported rates of sexual victimization than pencil-and-paper mode, high-contact, and in-lab completion, respectively.

We also hypothesized that race would moderate the degree to which each of the three methodological factors affected reporting. We anticipated that White individuals would have less motivation than other racial groups to edit responses, because non-White individuals may have concerns that they will fuel racial prejudice or stereotyping if they admit to a history of victimization. As such, we predicted that White participants would demonstrate less difference across methodological conditions than individuals who are racial minorities. Thus, our hypotheses were as follows:

Hypothesis 1: Women will report higher rates of CSA in computer conditions, low researcher contact conditions, and out-of-lab completion conditions, than they will in pencil-and-paper conditions, high-contact conditions, and in-lab conditions, respectively. These relationships will be moderated by race, such that non-White participants will show bigger differences between conditions than White participants.

Hypothesis 2: Women will report higher rates of ASA in computer conditions, low researcher contact conditions, and out-of-lab completion conditions, than they will in pencil-and-paper conditions, high-contact conditions, and in-lab conditions, respectively. These relationships will be moderated by race, such that non-White participants will show bigger differences between conditions than White participants.

Method

Participants

Participants for this study were 255 women between the ages of 18 and 62 (M = 23.26, SD = 6.44) recruited from a psychology subject pool at a Midwestern Urban Public University. Participants were provided with course credit as compensation for participation in the study. Between 39 and 51 potential participants were randomly assigned to each condition, and between 22 and 44 participants in each condition completed the study (see Table 1). Differential completion rates were observed after assignment to condition, ranging from 56% (high-contact, out-of-lab, pencil-and-paper) to 90% (low-contact, in-lab, pencil-and-paper). This project was part of a larger study, and differential rates of dropout for this study are examined in greater detail elsewhere (McCallum & Peterson, 2015).

Table 1.

Summary of Procedures for Each Experimental Condition.

Condition	Description of procedures
High-contact, in-lab, computer (n = 30; 17 White and 13 non-White)	Participants scheduled appointments online, presented to lab space, were greeted by researchers, were guided through informed consent by researchers, and completed surveys via computer.
High-contact, in-lab, pencil–paper (n = 28; 15 White and 13 Non-White)	Same as above, except participants completed surveys via pencil-and-paper format.
Low-contact, in-lab, computer (n = 28; 15 White and 13 non-White)	Participants scheduled appointments online, presented to lab space with no researcher present, read through informed consent alone in lab, and completed surveys via computer.
Low-contact, in-lab, pencil–paper (n = 42; 21 White and 21 non-White)	Same as above, except participants completed surveys via pencil-and-paper format.
High-contact, out-of-lab, computer (n = 39; 23 White and 16 non-White)	Participants scheduled appointments online, contacted researchers via phone, were guided through informed consent by researcher, and were provided a web link and completed the survey via computer. Participants also followed up with researchers via phone after completing the survey.
High-contact, out-of-lab, pencil–paper (n = 22; 13 White and nine non-White)	Participants picked up a pencil-and-paper survey from campus, scheduled appointments online, contacted researchers via phone, were guided through informed consent by researcher, and completed the pencil-paper survey. Participants also followed up with researchers via phone after completing the survey. Participants then returned the surveys to a drop box on campus.
Low-contact, out-of-lab, computer (n = 34; 22 White and 12 non-White)	Participants were emailed a web link to the computer-based survey, they read through informed consent, and completed the computer-based survey.
Low-contact, out-of-lab, pencil–paper (n = 30; 17 White and 13 non-White)	Participants picked up a pencil-and-paper survey from campus, read through the informed consent, and completed the survey at a location of their own choosing. Participants then returned the surveys to a drop box on campus.

Participants primarily identified themselves as White/European American (61.7%) or Black/African American (33.3%). Participants also identified as Native American/Alaskan Native (4.2%), Asian/Asian American (6.5%), and Other (3.4%). These categories were not exclusive, and participants were able to identify with multiple racial groups. For the purposes of analyzing differences between White and non-White participants, participants who identified any racial category other than exclusively White/European American were classified as non-White participants. Thus, participants who identified as bi- or multi-racial were considered non-White, even if White was one of the racial groups with which they identified; this was based on the assumption that a bi-racial participant would have similar concerns about negative stereotypes and social judgment as those experienced by other non-White participants. Based on these criteria, 56.7% of participants were classified as White and 42.7% were classified as non-White. White and non-White participants were compared on other demographic factors (age, income, relationship status), and no significant differences were observed between the two groups on any of the factors examined.

Measures

Demographics

Basic demographic information was collected using a 15-item Demographic Questionnaire which was developed for this project and included questions about age, year in school, religious affiliation, income, race, ethnicity, country of birth, and relationship status.

Sexual victimization in childhood

Participants’ history of CSA victimization was assessed using a modified version of a Childhood Sexual Abuse Measure (CSAM) first developed by David Finkelhor (1981). Although the validity of this specific measure has not been evaluated empirically, related measures have been discussed extensively, and are widely supported as valid retrospective measures of CSA (e.g., Finkelhor, 1994). The CSA measure consists of 11 questions about sexually abusive behaviors experienced prior to a participant’s 14th birthday, including kissing, fondling, oral sex, anal sex, and penile-vaginal intercourse initiated by someone at least 5 years older (e.g., “When you were 13 years old or younger, how many times did an older person [at least 5 years older than you] fondle you in a sexual way?”) and forced or coerced sex with someone of any age. Participants were asked to indicate how many times they had experienced each act selecting from among six response options (0, 1, 2-4, 5-7, 8 or more times, no answer). Participants were grouped into two categories—those who reported one or more act of CSA and those who reported no acts of CSA.

Sexual victimization in adulthood

Sexual victimization history in adulthood was assessed using the Sexual Experiences Survey–Short Form Victimization (SES-SFV; Koss et al., 2007). The SES-SFV is the most recent version of a well-established measure of sexual victimization. The previous version of the SES has demonstrated good reliability and validity when compared with face-to-face interviews related to victimization (e.g., Koss et al., 1987). The SES-SFV presents participants with seven different non-consensual sexual experiences, including kissing/fondling, oral sex, anal sex, and penile-vaginal intercourse (e.g., “Someone had oral sex with me or made me have oral sex with them without my consent by:”). Each of these experiences is followed by a description of five types of coercion (i.e., verbal pressure, verbal manipulation, intoxication, threat of physical harm, physical force), allowing participants to indicate how they were coerced into that specific sexual experience. Participants provide an indication of the number of times each form of coercion has taken place over the past 12 months and since they were 14 years old using a 4-point scale (0, 1, 2, or 3 or more times); for the sake of this study, we only asked about victimization since the age of 14. Similar to the CSA measure, participants were classified as having experienced some type of adult sexual victimization (i.e., a response greater than 0 on any item) or as having not experienced any adult sexual victimization (i.e., 0 on all items), based on their responses to the SES-SFV.

Procedure

The study was advertised as “a research study about sexual experiences and attitudes.” Interested participants provided contact information, were randomly assigned to one of the eight conditions, and were contacted via email with instructions for participation in the study. Instructions provided in the emails reflected the condition to which participants were assigned. For lab-based conditions, participants were required to arrange appointment times through online scheduling software to avoid confounding the researcher contact variable. For out-of-lab conditions, participants were simply provided with instructions outlining how they could complete the survey.

The instructions emailed to participants also reflected researcher contact and the mode by which participants were to complete the survey. Participants in the high researcher contact condition were informed of the interactions which would take place with the researcher, and these interactions were presented as a means to “ensure that the participant feels comfortable completing the survey.” Similarly, participants in the low-contact condition were informed of the lack of interaction with the researcher, which was also presented as a means to “ensure that the participant feels comfortable completing the survey.”

Participants in high researcher contact conditions had contact with a female researcher to minimize the potential for gender-related researcher effects. All researchers were White, thus matching race of the researcher and participants was not possible. The researchers in this study were trained undergraduate research assistants who used a detailed research protocol to ensure standardized administration across all trials. The principal investigator for the study was male and, as such, his name was listed on study materials and he facilitated some of the low-contact data collection. Although the principal investigator did not have any direct contact with participants, the possibility exists that his role had an impact on participants’ responses; however, this impact should have been consistent across conditions.

Participants in high-contact, in-lab conditions were greeted face-to-face by a researcher and were verbally guided through informed consent and study procedures. Participants in the high-contact, out-of-lab conditions were greeted by a researcher via phone and verbally guided through informed consent and procedures. Participants in all low-contact conditions were provided with written instructions and did not interact verbally with researchers at any point during the study. In-lab participants completed surveys in the lab space. Out-of-lab participants completed surveys in a setting of their own choosing. In the out-of lab, pencil-and-paper condition, participants were asked to return surveys to a predetermined drop-off point on campus. See Table 1 for a summary of procedures and samples sizes for each group.

Results reported here were part of a larger study (McCallum & Peterson, 2015). Thus, in addition to the measures reported here, participants also completed a number of additional measures of consensual sexual behavior.

Results

Test of the Hypotheses

Low researcher contact, in-lab completion, and computer-based inquiry mode conditions were predicted to be associated with greater likelihood of reported sexual victimization (CSA, ASA) than high researcher contact, out-of-lab completion, and pencil-and-paper inquiry mode conditions, respectively. Furthermore, we predicted that these effects would be moderated by race, with observed effects being greater for non-White than White participants.

Hypothesis 1: Self-report of childhood victimization

The CSAM was used to create a dichotomous measure of CSA, such that participants who reported any history of CSA on the CSAM were coded as “CSA positive,” and those who did not report any experiences of CSA were coded as negative. Although the majority of female participants reported no history of CSA (66%), scores on the CSAM ranged from 0-44 (M = 3.43, SD = 3.43), with higher scores indicating reporting of more instances of CSA, suggesting that some women in our sample experienced extensive sexual abuse in childhood.

To test our hypotheses related to the methodological variables, a series of logistic regressions were run for each of our methodological variables, using the dichotomous version of the CSAM as the dependent variable. For each regression, the methodological variable of interest (mode, researcher contact, or setting) was entered along with race in Step 1, and the interaction between race and the methodological variable of interest was added in Step 2. A summary of all of the CSA analyses can be found in Table 2.

Table 2.

Impact of Race and Methodological Variables on Participant’s Disclosure of CSA.

Variable	p value	OR	95% CI
Mode
Step 1:
Race (non-White vs. White)	.005**	2.13	[1.25, 3.62]
Mode (computer vs. paper)	.054	0.59	[0.35, 1.01]
Step 2:
Race (non-White vs. White)	.043*	2.14	[1.03, 4.48]
Mode (computer vs. paper)	.176	0.60	[0.28, 1.26]
Race by Mode	.978	0.99	[0.34, 2.86]
Setting
Step 1:
Race (non-White vs. White)	.004**	2.20	[1.29, 3.75]
Setting (in-lab vs. out-of-lab)	.398	0.80	[0.47, 1.35]
Step 2 :
Race (non-White vs. White)	.001**	3.78	[1.76, 8.92]
Setting (in-lab vs. out-of-lab)	.437	1.34	[0.64, 2.82]
Race by Setting	.049*	.34	[0.12, 1.00]
Researcher contact
Step 1:
Race (non-White vs. White)	.004**	2.16	[1.27, 3.66]
Researcher contact (high vs. low)	.350	0.78	[0.46, 1.32]
Step 2:
Race (non-White vs. White)	.22	1.56	[0.77, 3.17]
Researcher contact (high vs. low)	.114	0.54	[0.25, 1.16]
Race by Researcher contact	.185	2.07	[0.71, 6.04]

Note. CSA = child sexual abuse; OR = odds ratio; CI = confidence interval.

Significant at <.05. **Significant at <.01.

For the mode-related analysis, there was a significant main effect for race in Step 1, such that the regression revealed that non-White (43%) participants were more likely than White participants (27%) to report CSA history. There was no significant effect for mode. After the addition of race-related interaction in Step 2, race remained significant, and the interaction was not significant.

For the setting-related analysis, there was again a significant main effect for race in Step 1, with no main effects observed for setting. In Step 2, race remained significant, but there also was a significant race by setting interaction. Follow-up analyses suggested that in out-of-lab conditions, non-White participants were more likely to disclose CSA (54%) than White participants (24%), χ²(1, N = 126) = 12.07, p < .001 (see Figure 1). No significant differences were observed between White and non-White participants in lab, χ²(1, N = 128) = 0.46, p = .49.

Figure 1.

CSA logistic regression, setting by race interaction effect.

There also was a significant main effect for race in Step 1 of the researcher contact-related analysis, though this dropped out of significance after the race by researcher contact interaction term was entered in Step 2. There was no main effect for researcher contact, and the interaction between race and researcher contact was not significant either.

Hypothesis 2: Self-report of adult victimization

Scores on the SES-SFV were totaled and frequency scores ranged from 0-3 across 35 different types of sexual victimization. Higher scores reflected more instances of sexual victimization, and scores ranged from 0-71 (M = 6.35, SD = 11.82), with 42.9% reporting no history of sexual victimization. SES-SFV scores were also converted into a dichotomous variable, which distinguished between participants reporting any history of adult sexual victimization and those who denied any adult sexual victimization history.

Again, a series of logistic regressions were run for each of the methodological variables, with the dichotomous SES-SFV variable serving as the dependent variable. For each regression, the methodological variable of interest (researcher contact, setting, or mode) was entered along with race in Step 1, with the interaction between race and the methodological variable being added in Step 2. See Table 3 for a summary of regression analyses for ASA disclosure.

Table 3.

Impact of Race and Methodological Variables on Participant’s Disclosure of ASA.

Variable	p value	OR	95% CI
Mode
Step 1:
Race (non-White vs. White)	.038*	0.59	[0.35, 0.97]
Mode (computer vs. Paper)	.97	0.99	[0.60, 1.63]
Step 2:
Race (non-White vs. White)	.004**	0.34	[0.16, 0.70]
Mode (computer vs. paper)	.155	0.61	[0.31, 1.21]
Race by Mode	.037*	2.95	[1.07, 8.17]
Setting
Step 1:
Race (non-White vs. White)	.037*	0.59	[0.35, 0.97]
Setting (in-lab vs. out-of-lab)	.820	1.06	[0.64, 1.75]
Step 2:
Race (non-White vs. White)	.563	0.81	[0.39, 1.66]
Setting (in-lab vs. out-of-lab)	.322	1.41	[0.72, 2.77]
Race by Setting	.218	0.53	[0.19, 1.46]
Researcher contact
Step 1:
Race (non-White vs. White)	.038*	0.59	[0.35, 0.97]
Researcher contact (high vs. low)	.125	1.48	[0.90, 2.46]
Step 2:
Race (non-White vs. White)	.263	0.68	[0.34, 1.34]
Setting (high vs. low)	.122	1.71	[0.87, 3.39]
Race by Researcher contact	.537	0.73	[0.26, 2.01]

Note. ASA = adult sexual assault; OR = odds ratio; CI = confidence interval.

Significant at < .05. **Significant at < .01.

No significant main effects were observed for any of the methodological variables entered in Step 1 across any of the models. However, there was a main effect for race in each model, suggesting that White participants (63%) were more likely than non-White participants (50%) to report a sexual victimization history. After the addition of race-related interactions in Step 2, the main effect for race dropped out of significance within the setting and researcher contact models. There were no significant interaction terms for setting or researcher contact.

Race remained significant within Step 2 of the mode-related analysis, along with a race by mode interaction, with White participants (69%) being significantly more likely than non-White participants (43%) to report sexual victimization history in pencil-and-paper conditions, χ²(1, N = 124) = 8.65, p = .003 (see Figure 2). There were no significant differences between the racial groups for computer conditions.

Figure 2.

ASA logistic regression, mode by race interaction effect.

Summary

Although race was consistently predictive of CSA and ASA disclosure, no support was found across either the CSA or ASA measure for our hypotheses related to the main effects of our methodological variables. For CSA disclosure, there was a race by setting interaction with non-White participants being significantly more likely than White participants to disclose CSA outside of the lab, but with no difference in lab. In terms of sexual victimization in adulthood, there was a significant interaction observed between race and mode that suggested that White participants were more likely than non-White participants to report sexual victimization on pencil-and-paper surveys, but there were no racial differences for computer-based surveys.

Discussion

This is the first study of which we are aware that evaluates self-reported sexual victimization in the context of the three methodological factors considered in this study. Given that reporting of victimization has important implications for prevention and intervention efforts, we see this as an area that is particularly deserving of attention. Reporting of victimization helps researchers to understand the scope of the problem and identify consequences and correlates of sexual victimization. Methodological factors are also particularly relevant for measurement of sexual victimization experiences because the experiences are relatively rare and infrequent (although certainly not as rare or infrequent as one would hope). As such, failing to identify even a small number of victimization experiences may amount to missing a substantial proportion of the experiences that have taken place. Furthermore, in applied settings, disclosure often serves as a necessary first step toward identification of appropriate resources, and even incrementally improving clinicians’ ability to connect with survivors of sexual aggression is a worthwhile endeavor.

We examined two measures of victimization to capture self-reported sexual victimization experiences in childhood and adulthood. Independent of specific methodological variance, the overall rates of CSA disclosure (34%) and ASA disclosure (57%) are generally consistent with rates that have been observed elsewhere with similar samples and measurement (e.g., Koss et al., 1987; Walsh, Fortier, & DiLillo, 2010). Contrary to our expectations, there were no main effects for any of the methodological variables in reporting of either CSA or ASA; however, there were significant interactions between participant race and methodological variables. In terms of CSA disclosure, there was a race by setting interaction, which suggested that non-White participants were significantly more likely to report a history of CSA when completing surveys in out-of-lab conditions than White participants were. There were no significant differences observed between White and non-White participants on surveys completed in lab. Although not exactly the interaction that we had predicted, the pattern of results (see Figure 1) seems consistent with the conceptual rationale behind our hypotheses. Specifically, it seems possible that non-White participants may have been underreporting in the in-lab condition (due to feeling less anonymous and due to fear of racial stereotyping), thus obscuring real differences between the White and non-White participants in that condition. In the out-of-lab condition, the non-White participants may have felt more anonymous, and thus were willing to admit to more CSA, revealing a difference between the White and non-White groups. Consistent with the observed difference, there has previously been some indications that African American children may be at a higher risk for childhood maltreatment (including CSA), though it is important to note that observed differences may have been confounded by socioeconomic factors (e.g., Lee et al., 2012).

For ASA, a race by mode interaction was identified, which suggested that White participants reported more ASA than non-White participants in the paper-and-pencil condition, but not in the computer condition. Again, this is not the pattern that we had expected, and it is difficult to interpret these results. Based on the overall pattern of results (see Figure 2), it seems that this interaction was driven by the fact that the White participants reported slightly more ASA in the paper-and-pencil conditions than in the computer conditions (although the difference did not reach the level of significance), and the non-White participants reported slightly less ASA in the paper-and-pencil conditions than in the computer conditions (although again, this did not reach the level of significance). Thus, these results seem to suggest that different modes may be needed to maximize the reporting of ASA among White versus non-White participants. It is not clear why this might be the case. It is possible that the groups differ in their level of trust or sense of anonymity on the computers. More research is needed to explore the mechanism behind this race by mode interaction.

It is difficult to draw any universal conclusions from our findings related to the interaction between race and methodological factors. We expected that methodological variables would be especially important for non-Whites because they might be particularly concerned with impression management given the potential for stereotyping and prejudice (especially given that the researchers were White); thus, they might be particularly influenced by the differing perceptions of anonymity and confidentiality across methodological conditions. This expectation was potentially supported for CSA reporting (although the results were not entirely clear), but was not well supported for ASA reporting. Notably, McCallum (2015) found evidence that non-White participants were more strongly influenced by methodological differences than White participants when they were asked to disclose consensual sexual behavior. It is possible that, for non-White participants, disclosure of sexual victimization is less influenced by methodology than disclosure of other kinds of sexual behaviors, perhaps because sexual victimization is not the person’s choice, whereas consensual sexual activity involves a personal choice to engage in potentially socially unacceptable behaviors. Clearly, more research is needed to shed light on these findings.

The racial divergence observed related to the effects of mode and setting is consequential for a number of reasons. First and foremost, it is important to recognize the possibility that methodological decisions may affect different racial groups differentially. Multiple studies have used self-report measures to draw conclusions about differences between White and non-White women in terms of risk for sexual victimization (e.g., Starzynski, Ullman, Townsend, Long, & Long, 2007; Tillman, Bryant-Davis, Smith, & Marks, 2010). It is possible that observed differences may be due, in part, to methodological factors. For example, if our observed results held across other studies, they would suggest that in-lab settings and computer measures may underestimate the true difference in risk between White and non-White participants (or alternatively, out-of-lab and paper-and-pencil measures may overestimate the difference).

More broadly speaking, there is no literature of which we are aware that has considered the relationship between race and inquiry mode for self-reported victimization experiences of any kind. Further attention to the differential impact of methodological factors on various racial groups is needed because of the potential for skewed or misleading conclusions about racial differences or similarities.

Limitations

This study has several important limitations to consider. Because our study only examined the impact of methodological factors on women’s disclosure of sexual assault and CSA, these results cannot be generalized to men. Because stereotypes suggest that only women are victims of sexual abuse and assault (e.g., Chapleau, Oswald, & Russell, 2008), men’s self-reports of sexual victimization may be even more affected by social desirability, and thus by methodological factors, than women’s. Future research would benefit from directly comparing the impact of methodological variables on men’s and women’s rates of disclosure.

Our sample for this study was relatively small. As a result, we did not have adequate power to enter all methodological variables (contact, setting, and mode) and their interactions into a single logistic regression model. Although not part of our hypotheses for this project, it is possible that the different methodological variables interact with each other to affect reporting. Our sample was also a convenience sample of college students, and as such, our results may not generalize to other populations. However, it is worth noting that this study was conducted at a racially and economically diverse urban state university with many adult learners, resulting in more variability than most college samples.

Another limitation relates to our assumptions about the meaning of observed differences in self-report. Based on the available literature related to stigma surrounding sexual violence, we assume that higher rates of self-reported CSA and sexual assault are reflective of a reduction in underreporting (e.g., Cook et al., 2011). However, though there are no conceptual reasons to expect over-reporting in the context of this study, we cannot decisively eliminate the possibility that observed differences reflect increased over-reporting of unwanted sexual experiences in some conditions rather than reduced underreporting in other conditions.

It is also important to consider several limitations related to the experimental conditions in this study. Although phone-based contact was included in high-contact out-of-lab conditions to approximate the level of contact experienced in lab, it is possible that talking on the phone with a researcher is qualitatively different than speaking with a researcher in person. Furthermore, the very nature of out-of-lab survey completion decreases our ability to control factors such as privacy in the settings in which participants elect to complete surveys. Although this is a limitation in that it leads to more variability within the out-of-lab than the in-lab conditions, it also reflects the reality of much of the survey research being conducted in which researchers similarly have little control over these factors. In that sense, this may also be considered as a strength, as our findings are more applicable to studies in which those factors are not being directly controlled.

It is also worth noting again that all of the researchers in this study were White. Although we did not observe any race-related effects for in-lab conditions—the only conditions in which participants met the researcher face-to-face—it is possible that the researchers’ race was inferred in other conditions (based on, for example, voice or name). Thus, it is possible that the racial differences observed in some conditions were related to the racial mismatch between the non-White participants and the White researchers as opposed to the racial minority status of the non-White participants. In other words, perhaps the interactions with race would have looked different if our researchers had been non-White. This deserves future research attention. A final limitation to consider is related to the classification of non-White participants in this study. All non-White participants were grouped together based on the assumption that minority status was the most important factor influencing rates of disclosure across mode. Although minority status likely did play a role in the observed effects, it is also possible that unique factors related to participants’ specific racial, ethnic, or cultural background also may have contributed to the effect as well. Given that the majority of non-White participants in our sample identified as African Americans (71.3%), it is likely that the race by mode interactions observed are largely the result of differences in responding between participants identifying as White and those who identify as African American. As such, it is unclear whether the differences observed are more reflective of minority versus majority status or whether they also reflect cultural differences between White and Black participants. Addressing this limitation would be a particularly useful contribution to existing literature related to differences in reporting among Black and White women in America (e.g., Starzynski et al., 2007; Tillman et al., 2010). Future studies should consider taking a more nuanced approach toward examination of race-related methodological effects.

Notwithstanding these concerns, this study serves as a well-controlled exploration of methodological influences on self-report of ASA and CSA. The random assignment of participants to conditions provides reasonable assurance that the observed results are directly related to the specific methodological variables. Furthermore, though more complete examination of race would be valuable, this study marks the first consideration of this seemingly important demographic variable as a possible moderator of methodologically dependent effects. We hope that future studies will build on our findings to better clarify interactions between participant characteristics (e.g., race, gender, sexual orientation) and methodological variables.

Research Implications and Future Directions

Researchers continue to work toward refining methodology to better capture the elusive experience of sexual victimization. The results of this study suggest that consideration of a broader spectrum of methodological factors beyond measure construction may be warranted to maximize the ability and willingness of participants to self-report sexual victimization. Although our results do not lend themselves to a “magic formula” for increasing reports of sexual victimization, they do support the value of further research related to the impact of researcher contact and mode of inquiry on participants’ reports of sexual victimization. Furthermore, these results suggest that it may be important for all future studies related to race and sexual aggression to be more conscious of the potential impact of methodological factors on self-report across groups.

More research is needed to replicate our findings and to identify the mechanism through which interactions between race and methodological variables arise. Currently there is a body of literature on sexual victimization consisting of data collected through both paper-and-pencil and computer-based modes; understanding the potential impact of these different modes is necessary if researchers hope to reconcile and generalize findings across modes of data collection. Ultimately, it may be necessary to consider the interaction between multiple methodological variables in parallel, as none of the variables examined in this study operates independently.

It may be useful for future studies to consider other modes as well (e.g., phone-based interviews, face-to-face interviews). Similarly, it may be useful to consider the context in which participants elect to complete self-report measures. Approximately 35% of our out-of-lab participants elected to complete their surveys in a setting that was not completely private. It also may be useful to consider the impact of methodological variables on the self-report of other forms of maltreatment and physical violence. It is unclear if the impact of mode and researcher contact is unique to sexual victimization experiences or reflects a broader tendency related to self-report of other sensitive experiences.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Biographies

Ethan B. McCallum is a psychologist at the Minneapolis VA Medical Center in Minneapolis, Minnesota. He obtained his PhD in clinical psychology from the University of Missouri–St. Louis in St. Louis, Missouri, and completed a post-doctoral fellowship at the Minneapolis VA Medical Center in 2014. The data presented in this article were gathered as a part of his dissertation project. His current clinical and research efforts focus primarily on effective treatment of post-traumatic stress disorder in the presence of other co-occurring disorders.

Zoë D. Peterson is an associate professor of psychological sciences and director of the Sexual Assault Research and Education Initiative at the University of Missouri–St. Louis. She earned a doctorate in psychology with a clinical emphasis from the University of Kansas in 2005, and she completed a post-doctoral fellowship at the Kinsey Institute for Research in Sex, Gender, and Reproduction at Indiana University in 2007. She currently remains a Research Fellow at the Kinsey Institute and is serving as an associate editor of the Journal of Sex Research. Her research focuses on unwanted sex, sexual coercion, and sexual assault.

References

Aquilino

W. S.

(1994). Interview mode effects in surveys of drug and alcohol use: A field experiment. Public Opinion Quarterly, 58, 210-240. doi:10.1086%2F269419

Aquilino

W. S.

LoSciuto

L. A.

(1990). Effects of interview mode on self reported drug use. Public Opinion Quarterly, 54, 362-395. doi:10.1086%2F269212

Bell

K. M.

Naugle

A. E.

(2007). Effects of social desirability on students’ self-reporting of partner abuse perpetration and victimization. Violence and Victims, 22, 243-256. doi:10.1891/088667007780477348

Chang

J. C.

Dado

Schussler

Hawker

Holland

C. L.

Burke

J. G.

Cluss

P. A.

(2012). In person versus computer screening for intimate partner violence among pregnant patients. Patient Education & Counseling, 88, 443-448.

Chapleau

K. M.

Oswald

D. L.

Russell

B. L.

(2008). Male rape myths: The role of gender, violence, and sexism. Journal of Interpersonal Violence, 23, 600-615. doi:10.1177/0886260507313529

Cook

S. L.

Gidycz

C. A.

Koss

M. P.

Murphy

(2011). Emerging issues in the measurement of rape victimization. Violence Against Women, 17, 201-218. doi:10.1177/1077801210397741

DiLillo

DeGue

Kras

Di Loreto-Colgan

Nash

(2006). Participant responses to retrospective surveys of child maltreatment: Does mode of assessment matter? Violence and Victims, 21, 410-424. doi:10.1891/vivi.21.4.410

Finkelhor

(1981). Sexually victimized children. New York: Free Press.

Finkelhor

(1994). Current information on the scope and nature of child sexual abuse. Future of Children, 4, 31-53. doi:10.2307/1602522

10.

Fisher

B. S.

(2009). The effects of survey question wording on rape estimates: Evidence from a quasi-experimental design. Violence Against Women, 15, 133-147. doi:10.1177/1077801208329391

11.

Follingstad

D. R.

Rogers

J. M.

(2013). Validity concerns in the measurement of women’s and men’s report of intimate partner violence. Sex Roles, 69, 149-167. doi:10.1007/s11199-013-0264-5

12.

Hardré

P. L.

Crowson

H. M.

Xie

(2012). Examining contexts-of use for web-based and paper-based questionnaires. Educational and Psychological Measurement, 72, 1015-1038. doi:10.1177/0013164412451977

13.

Hewitt

M. K.

(2002). Attitudes towards interview mode and comparability of reporting sexual behavior by personal interview and audio computer-assisted self-interviewing: Analyses of the 1995 National Survey of Family Growth. Sociological Methods & Research, 31, 3-26. doi:10.1177/0049124102031001001

14.

Johnson

P. B.

Richter

(2004). Research note: What if we’re wrong? Some possible implications of systematic distortions in adolescents’ self-reports of sensitive behaviors. Journal of Drug Issues, 34, 951-970. doi:10.1177/002204260403400412

15.

Jourard

S. M.

Friedman

(1970). Experimenter-subject “distance” and self-disclosure. Journal of Personality and Social Psychology, 15, 278-282.

16.

Jourard

S. M.

Lasakow

(1958). Some factors in self-disclosure. Journal of Abnormal Psychology, 56, 91-98. doi:10.1037/h0043357

17.

Kim

Dubowitz

Hudson-Martin

Lane

(2008). Comparison of 3 data collection methods for gathering sensitive and less sensitive information. Ambulatory Pediatrics, 8, 255-260.

18.

Koss

M. P.

(1992). The underdetection of rape: Methodological choices influence incidence estimates. Journal of Social Issues, 48, 61-75.

19.

Koss

M. P.

Abbey

Campbell

Cook

Norris

Testa

. (2007). Revising the SES: A collaborative process to improve assessment of sexual aggression and victimization. Psychology of Women Quarterly, 31, 357-370. doi:10.1111/j.1471-6402.2007.00385.x

20.

Koss

M. P.

Gidycz

C. A.

Wisniewski

(1987). The scope of rape: Incidence and prevalence of sexual aggression and victimization in a national sample of higher education students. Journal of Consulting and Clinical Psychology, 2, 162-170. doi:10.1037//0022-006X.55.2.162

21.

Lee

Cronley

White

H. R.

Mun

Stouthamer-Loeber

Loeber

(2012). Racial differences in the consequences of childhood maltreatment for adolescent and young adult depression, heavy drinking, and violence. Journal of Adolescent Health, 50, 443-449. doi:10.1016/j.jadohealth.2011.09.014

22.

McCallum

E. B.

Peterson

Z. D.

(2012). Investigating the impact of inquiry mode on self-reported sexual behavior: Theoretical considerations and review of the literature. Journal of Sex Research, 49, 212-226. doi:10.1080/00224499.2012.658923

23.

McCallum

E. B.

Peterson

Z. D.

(2015). Effects of experimenter contact, setting, inquiry mode, and race on women’s self-report of sexual attitudes and behaviors: An experimental study. Archives of Sexual Behavior, 44, 2287-2297. doi:10.1007/s10508-015-0590-5

24.

Parks

Pardi

Bradizza

(2006). Collecting data on alcohol use and alcohol-related victimization: A comparison of telephone and Web-based survey methods. Journal of Studies on Alcohol and Drugs, 67, 318-324.

25.

Peterson

Z. D.

Muehlenhard

C. L.

(2004). Was it rape? The function of women’s rape myth acceptance and definitions of sex in labeling their own experiences. Sex Roles, 51, 129-144. doi:10.1023/B:SERS.0000037758.95376.00

26.

Reddy

M. K.

Fleming

M. T.

Howells

N. L.

Rabenhorst

M. M.

Casselman

Rosenbaum

(2006). Effects of method on participants and disclosure rates in research on sensitive topics. Violence and Victims, 21, 499-506. doi:10.1891/vivi.21.4.499

27.

Richman

Weisband

Kiesler

Drasgow

(1999). A meta-analytic study of social desirability response distortion in computer-administered and traditional questionnaires and interviews. Journal of Applied Psychology, 84, 754-775. doi:10.1037/0021-9010.84.5.754

28.

Sable

M. R.

Danis

Mauzy

D. L.

Gallagher

S. K.

(2006). Barriers to reporting sexual assault for women and men: Perspectives of college students. Journal of American College Health, 55, 157-162.

29.

Starzynski

L. L.

Ullman

S. E.

Townsend

S. M.

Long

L. M.

Long

S. M.

(2007). What factors predict women’s disclosure of sexual assault to mental health professionals? Journal of Community Psychology, 35, 619-638. doi:10.1002/jcop.20168.

30.

Strang

E. T.

Peterson

Z. D.

Hill

Y. N.

Heiman

J. R.

(2013). Discrepant responding across self-report measures of men’s coercive and aggressive sexual strategies. Journal of Sex Research, 50, 458-469. doi:10.1080/00224499.2011.646393

31.

Sugarman

D. B.

Hotaling

G. T.

(1997). Intimate violence and social desirability: A meta-analytic review. Journal of Interpersonal Violence, 12, 275-290. doi:10.1177/088626097012002008

32.

Testa

Livingston

VanZile Tamsen

(2005). The impact of questionnaire administration mode on response rate and reporting of consensual and nonconsensual sexual behavior. Psychology of Women Quarterly, 29, 345-352. doi:10.1111/j.1471-6402.2005.00234.x

33.

Tillman

Bryant-Davis

Smith

Marks

(2010). Shattering silence: Exploring barriers to disclosure for African American sexual assault survivors. Trauma, Violence, & Abuse, 11, 59-70. doi:10.1177/1524838010363717

34.

Walsh

Fortier

M. A.

DiLillo

(2010). Adult coping with childhood sexual abuse: A theoretical and empirical review. Aggression and Violent Behavior, 15, 1-13. doi:10.1016/j.avb.2009.06.009

35.

Zeigler-Hill

Wallace

M. T.

Meyers

E. M.

(2012). Racial differences in self-esteem revisited: The role of impression management in the Black self-esteem advantage. Personality and Individual Differences, 53, 785-789. doi:10.1016%2Fj.paid.2012.06.007