Empirical Article ASSOCIATION FOR

Empirical Article ASSOCIATION FOR PSYCHOLOGICAL SCIENCE Long-Term Memory in Adults Exposed to Childhood Violence: Remembering Genital Contact Nearly 20 Years Later Deborah Goldfarb', Gail. S. Goodman 2, Rakel P. Larson 3, Mitchell L. Eisen', and Jianjian Qins oen:Lament of Psychology, Florida International university: 'Department of Psychology, University of California, Davis. 'Department of Psychology, Pities College. 'Department of Psychology, California State University, Las Angeles; and ^Derailment of Psychology. California State University. Sacramento Psycholopc.al Science I -I6 0 The Authors) LOIS AnKle reuse guidelines sigcpub.conliournailtpeimmans DOI- I0.1 I77/216770260130742 wenvpsychologica6cience ocp/CPS OSAGE Abstract Recent changes in statutes of limitations for crimes against children permit accusations of decades-old child sexual abuse to be considered in court. These laws challenge scientists to address the accuracy of long-term memory of genital contact. To examine theoretical, clinical, and legal concerns about long-term memory accuracy, children who in the 1990s (Time 1) were 3 to 17 years old and experienced a documented child maltreatment medical examination that included genital touch were interviewed between 2012 and 2014 (Time 2), as adults, about the medical experience. Almost half of the adults reported the childhood genital contact. Child sexual abuse and greater depression in adulthood predicted greater memory accuracy. No participant falsely reported chargeable offenses that did not occur, even when such offenses had been falsely suggested in a childhood interview. Some participants erred with regard to specific and misleading questions implying less egregious acts. Ramifications for theory and application are discussed. Keywords maltreatment, longitudinal memory Received 3/31/1R; Revision accepted 8/2/18 There are some memories that time may never erase, but questions arise as to whether genital contact expe- rienced in childhood is one of them. Recent research confirms the possibility of false memories of childhood sexual encounters, including as intensified in vulner- able individuals by debated clinical techniques (Bot- toms, Shaver, & Goodman, 1996; Lilienfeld, 2015; Loftus, 1996). Yet clinical and memory theories should also address the matter of enduring memories for genital touch actually experienced in childhood. Currently, there is a pressing need for scientific stud- ies on this topic because in "historic" child sexual abuse cases (where prosecution occurs years after the alleged assault), the accuracy of adults' memory for childhood genital contact is paramount, with concerns about inac- curacies amplified when the adults have trauma histories (Conway, 2013; Howe, 2013; Loftus, 1996; Otgaar, Muds, Howe, & Merchkelback, 2017). Society is grappling with how to respond to such cases (D. A. Connolly, Chong, Coburn, & Lutgens, 2015; Howe & Knott, 2015; Wells, Morrison, & Conway, 2014), as reflected in prosecutions of Penn State coach Jerry Sandusky, Michigan State ath- letic physician Larry Nassar, Bay Area child psychiatrist William Ayres, former Speaker of the U.S. House of Representatives Dennis Hasten, and in the U.S. Senate's Judiciary Committee hearings on the confirmation of Judge (now Justice) Brett Kavanaugh. As few, if any, published studies have analyzed the accuracy of adults' memories for a verified abuse-related childhood event that includes concurrently documented genital contact, the question of how accurately adults remember such experiences has gone largely unanswered (hut see Alexander et al., 2005; Widom & Morris, 1997; Williams, 1994). To shed light on this issue, we analyzed adults' memories for verified childhood genital contact after an 18-to-20-year delay. Corresponding Author Gail S. Goodman, Department of Psychology, University of California, One Shields Ave.. Davis. CA 95616 E-mail. pasordnianerucdaviciedu 3534-013 Page 1 of 16 EFTA_000 10280 EFTA00159928 2 Goldfarb et al. Long-Term Memory for Significant Childhood Events In general, memories fade over time for both children and adults (Hirst et al., 2015; La Rooy, Pipe, & Murray, 2007), making adults less sure of their childhood mem- ories and more subject to suggestive influences (Loftus & Pickrell, 1995). Yet memories of highly emotional (compared to neutral) events are often less susceptible to forgetting (LaBar & Cabeza, 2006; Yonelinas & Ritchey, 2016): Individuals who experienced traumatiz- ing events, such as a natural disaster, an impending airplane crash, or an injury necessitating an emergency room visit, recall the event years and sometimes even decades later (Bauer et al., 2016; Fivush, McDermott Sales, Goldberg, Bahrick, & Parker, 2004; McKinnon et al., 2015; Peterson, 2015; Van Abbema & Bauer, 2005). For example, adolescents and adults accurately remember injuries and assaults experienced 6 to 13 years prior (Goodman et al., 2003; Greenhoot, McClo- skey, & Glisky, 2005). Young children's ability to remember details of a medical event decreases over an initial 3-month period and becomes stable by 6 months (Ornstein et al., 2006). However, for children old enough to remember a stress- ful medical test, delays of months or years do not nec- essarily increase inaccuracies or suggestibility (Quas et al., 1999). Even when encoding occurred in the sec- ond year of life, during what later is typically labeled a period of -childhood amnesia," a subset of older children and adults remember salient and distinctive emotional events despite significant delays (McDermott Sales, Finish, Parker, & Bahrick, 2005; Peterson 2015; Usher & Neisser, 1993; Williams, 1994). Still, young age and long delays typically predict the waning accuracy and decreas- ing detail of long-term memory, including of child sexual abuse (Goodman et al., 2003), and predict adults' sus- ceptibility to false suggestion (Howe & Knott, 2015; Qin, Ogle, & Goodman, 2008). There are factors, however, that guard against false childhood recollections (e.g., Pezdek, Finger, & Hodge, 1997). Relatively strong memories of negative, conse- quential childhood experiences combined with age improvements in metacognitive abilities, such as accu- rately realizing that one does not know an answer (Koriat, Goldsmith, & Pansky, 2000; Lyons & Ghetti, 2010), may support the ability of many adults either to accurately report salient childhood events that occurred decades prior or to use a conservative response strategy (e.g., saying, "I don't remember") if one has forgotten or is unsure of what happened. There is disagreement, however, about the contribution of forgetting to recol- lection error, as well as about the role that lack of confidence in one's memory plays in resisting false suggestions (Loftus, 1996; Rubin & Wenzel, 1996; Wickens, 1998; Wixted, 2004). There are legitimate con- cerns that delay can lead to increased guessing, schema- driven commission errors, and false memories (e.g., Kleider, Pezdek, Goldinger, & Kirk, 2008). Although few studies of memory accuracy have included delays as long as 20 years (Bahrick, Bahrick, & Wittlinger, 1975), it seems likely that, as the time between an event and a memory task increases, individual differences will be evident in the adoption of a conservative versus liberal response strategy, the latter of which could increase suggestibility (Singer & Wixted, 2006). Individual Differences and Memory Individual differences in trauma history and psychopa- thology may affect the accuracy of long-term memory for stressful life events. In this regard, maltreatment history and posttraumatic stress disorder (PTSD) have been of much interest, especially to clinicians. Some researchers find that maltreatment history and/or PTSD symptomology are associated with increased accuracy of remembering abuse-related experiences (Alexander et al., 2005; Eisen, Goodman, Qin, Davis, & Crayton, 2007). For example, children with a history of sexual abuse omit fewer details regarding a forensic anogenital examination than children with no such history (Katz, Schonfeld, Carter, Leventhal, & Cicchetti, 1995). Prior childhood sexual victimization, especially when associ- ated with PTSD, may provide a knowledge structure within which to encode abuse-related acts or increase the saliency (including trauma relevance) of such expe- riences (Baker-Ward et al., 2015; Frankenhuis & Weerth, 2013). Another mental health problem of particular interest is depression, which is associated with a child maltreat- ment history (Brown, Cohen, ohnson, & Smailes, 1999), increased overgeneral memory (Williams & Broadbent, 1986), and increased rumination and recall of negative life events (e.g., S. L. Connolly & Alloy, 2018; Hertel & El-Messidi, 2006; Matt, Vazquez, & Campbell, 1992). Greater rumination of negative childhood experiences may keep such memories alive, leading to greater accuracy. It has also been proposed, however, that individuals with trauma-related psychopathology, such as PTSD or depression, are less conservative in responding and more likely to err in reporting events (Otgaar et al., 2017; Windmann & KrUger, 1998). Coupled with a trauma history, PTSD and depression may thus be pre- dictors of increased correct memory of negative life events but also of greater error (e.g., susceptibility to misleading questions). The hypothesized memory errors related to maltreatment may be driven by mental 3534-013 Page 2 of 16 EFIA_000 10281 EFTA00159929 Remembering Genital Touch 3 health symptomology, perhaps resulting from trauma, rather than by maltreatment itself (Eisen et al., 2007; Goodman et al., 2016). Furthermore, gender differences in memory for emo- tional childhood events have been documented, with males compared to females remembering fewer emo- tional childhood experiences (Davis, 1999). Such dif- ferences may he particularly likely for an emotional event that is sexual in nature, such that males may he more reluctant than females to remember or disclose sexual details (thereby increasing the extent of omis- sion errors; Ullman & Filipas, 2005; Widom & Morris, 1997). Effects of Misleading Questioning in Childhood on Adult Memory Increased memory error may occur after children are exposed to false suggestions in interviews (Ceci Bruck, 1995); yet when memory is strong, misleading questions can increase the accuracy of long-term mem- ory in children and adults (Peterson, Parsons, & Dean, 2004; Putnam, Sungkhasettee, & Roediger, 2017; Quas et al., 2007). Moreover, memory rehearsal (e.g., via repeated interviews or conversations with others regarding the event) may reinstate accurate memory but can also lead to error (Cordon, Pipe, Sayfan, Melinder, & Goodman, 2004; Ornstein et al., 2006; Peterson, 2015; Peterson, Pardy, Tizzrard-Drover, & Warren, 2005). However, the effects of prior misleading interviews in childhood on the accuracy of adults' memory after an almost 20-year delay for an event involving genital contact have not been previously published. The Present Study This project is part of a longitudinal study of memory in children exposed to violence. In 1994 (Time I), because of suspicions of child maltreatment, authorities removed participants from their homes and placed them in a forensic hospital unit for evaluation (not for illness), where participants experienced an anogenital exam by a physician as part of a 5-day child-maltreatment investigation. As the anogenital exam was part of the standardized forensic medical procedure for the hospi- tal unit, virtually all children received such an exam, and it followed a set format, including the doctor administering both visual and manual inspection and penetration of the genital and rectal areas to enable swabbing for venereal disease. Researchers were pres- ent during the anogenital exam and documented what occurred, including all genital and anal contact. Nearly 20 years later, between 2012 and 2014, 30 participants were located and interviewed as adults about their memories of the experience. On the basis of aforementioned research, we predicted that partici- pants who were older at Time 1 and female would be more likely to report genital touch, but also that some of the youngest Time 1 participants (i.e., 4 years old) would accurately remember such contact—an abuse- related analogue for a legally chargeable act that did occur. We also examined accuracy in response to spe- cific and misleading questions, including questions that could lead to memory errors with legal relevance. Given research showing that prior experience of sexual abuse in childhood may provide a framework for encoding and/or may increase the personal significance of genital touch (Katz et al., 1995), we predicted that individuals with (vs. without) a child sexual abuse history would be more likely to remember this documented event. As to the possible influence of psychopathology on mem- ory, it was predicted that higher levels of current PTSD symptoms and greater depression would he associated with more accurate memory of the anogenital exam, including the genital contact, but also to greater error in response to misleading questions. Finally, exposure to a misleading interview in childhood was expected to be related to inaccuracy of memory in adulthood. Method Participants At Time 1 (1990s), when they experienced an anogeni- tal examination as part of a forensic investigation of maltreatment allegations, the 30 participants ranged in age from 4 to 17 years (M = 8.37 years, SD = 3.61; 20 females). They ranged in age from 23 to 36 years (M = 27.80 years, SD = 3.55) when interviewed at Time 2, approximately 20 years later (M = 19.03 years, SD = .32, range = 18 to 20), about their memories of the exami- nation involving genital contact. Participants included non-Hispanic Whites (13.3%), African Americans (80%), and Latinos/as (6.7%). For analyses, ethnicity was coded as African American = 1 and non-African American = 0 (M = .80, SD = .41). As adults, participants also tended to be single (67%) and of low socioeconomic status (57% reported making less than $20,000 per year). Half of the Time 2 participants (n = 15) were interviewed (with open-ended, specific, and misleading questions) about the anogenital exam at Time 1, whereas the other half had not been interviewed about the exam at Time 1 (see Eisen, Qin, Goodman, & Davis, 2002, for details). A central hypothesis of the present study concerned memory in adults with Time 1 histories of child sexual abuse compared to those with no Time 1 history of that 3534-013 Page 3 of 16 EFIA_000 10282 EFTA00159930 4 Goldfarb et at type of maltreatment. Thus, children classified as child sexual abuse victims were those whose Under the Rain- bow (UTR) cases were determined to be "indicated" for child sexual abuse by the Department of Child and Family Services (DCFS) after extensive investigations by local law enforcement, child-welfare authorities, and UTR specialized staff (i.e., medical, mental health, and social work professionals). DCFS records were also checked for past indicated sexual abuse. For the present research, if the child had an indicated case of sexual abuse at or prior to Time 1 (even if he or she had experienced ocher forms of maltreatment, such as neglect, which was common, or physical abuse), the child was considered a sexual abuse victim = 19). If the child had no known sexual abuse case (current or past at Time 1), the child was not considered a sexual abuse victim (n = 11), although the child might have suffered founded physical or psychological abuse or neglect, or had no founded child abuse case (e.g., 4 neglect, 3 nonabused controls). As discussed below, Time 1 maltreatment status was unknown for 3 partici- pants. For the present study, child sexual abuse status was coded as child sexual abuse history = 1 and no such history = 0 (M = .70, SD = .47).' At Time 2, 5 par- ticipants reported having experienced sexual assault as an adult and 25 reported no such experiences. There was no significant difference in Time 2 report of adult sexual assault between those who had experienced child sexual abuse as of Time 1(19%) and those who had not (10%), Fisher's Exact Test, p = 0.285, ns. Time 2 participants Os = 30) were not significantly different from Time 1 participants who did not take part in the Time 2 interview (n = 183) in terms of age, gen- der, ethnicity, and memory accuracy at Time 1 for either the exam generally or the genital contact specifically, ts(15-30) < 11.871. The Time 2 sample, however, con- tained more child sexual abuse victims than the original sample, t(27) > 13.311, p = .001. The sample size was based on prior research, including effect sizes, on long- term memory for emotional events (Peterson, 2015; Talarico & Rubin, 2003). Measures and procedure The longitudinal study was approved by the university's institutional review board and carried out in accordance with the provisions of the World Medical Association Declaration of Helsinki. At Time 1, consent for follow- up had been obtained. At Time 2, researchers located the participants' current physical or postal address, email address, and/or phone number by extensively searching available databases, including Google, LexisNexis, and TLO, and social network sites, such as Facehook and MySpace. Contacts were made (via phone, email, and/or letter) to participants, inviting them to take part in the research. Once participants were reached, trained female researchers "blind" to Time 1 measures, including to memory performance and maltreatment history, con- firmed participants' identity (i.e., name, birthdate, race, gender, and city in which the participant grew up). Participants were told that the study's purpose was to interview children who had grown up in Chicago in the 1990s. After consent was obtained and confidential- ity ensured, participants answered a series of demo- graphic and background questions, which allowed for rapport building before the memory portion of the interview commenced. For the memory interview, participants were cued to the target event by our saying that we wanted to ask about the time they stayed at a hospital unit, the UTR program, in the 1990s as a child or adolescent, and that "there were a lot of other children" there. Of note, participants were never informed of the purpose of their stay at the hospital, including that they were at the UTR as children as part of a forensic evaluation investigating maltreatment allegations. Participants were first asked a free-recall question concerning their general experience at the UTR program ("Please tell me everything you remember about being there"). They were then prompted to provide any additional informa- tion they could remember ("Is there anything else you remember about it? Even the smallest details are of interest to us."). Participants were then asked to recall everything they could remember about the medical exam at the UTR, the one where "small white patches Jelectrode patches) and wires were placed on your chest to mea- sure your heart beat." Note that for this part of the Time 2 interview, like the initial free-recall question, no cues were given to inform participants that they had received an anogenital exam. Two free-recall questions were asked: "Please tell me everything you remember about the doctor examination in as much detail as passible" and "Is there anything else you remember about it?" One open-ended question (e.g., "What parts of your body did the doctor examiner) and 25 closed-ended questions about the examination followed. Closed- ended questions consisted of 16 specific (e.g., "Did the doctor have you bend over?") and nine misleading questions that presumed false information (e.g., "When the doctor gave you the shot/inoculation, was it in your upper arm, upper thigh, or in your buttocks?" though participants did not receive an inoculation) designed to assess memory accuracy and suggestibility, respec- tively. Inaccurate responses for nine of the specific and seven of the misleading questions were commission errors (e.g., choosing an option when asked, "Did the 3534-013 Page 4 of 16 EFIA_000 10283 EFTA00159931 Remembering Genital Touch 5 nurse wash off your whole body at the start of that medical exam or was it during it?" when in fact the children's bodies were not washed then), and inaccu- rate responses for seven of the specific and two of the misleading questions were omission errors (e.g., agree- ment to "I know it is hard to remember back all that time, but there wasn't a chair in the room, was there?" when in fact there always was a chair in the exam room). As both commission and omission errors to misleading questions index suggestibility, they were combined as incorrect responses. A subset of the closed-ended questions (n = 7) asked about forensically relevant details that might well he related to an inves- tigation of inappropriate or abusive behavior on the part of the doctor or nurse (e.g. "Did the doctor take your clothes off at the start of the exam?" when, in fact, the doctor did not). Two of these seven forensically relevant questions concerned memory of the genital contact that actually did occur during the anogenital exam ("Did the doctor examine your genitals fprivate pans) during that exami- nation?"; if participants responded "yes," they were asked, "Did the doctor examine both your genital and rectal areas or just the genital area?" as a follow-up); this permitted us to examine omission errors of acts that were potentially chargeable legally. Two of the seven abuse-related questions concerned acts that did not occur and that (as with genital touch) could also, on their own, potentially lead to legal charges ("Did the doctor or nurse hit you during that medical exam?" "At the end, did the doctor kiss you?"). This allowed us to examine potentially chargeable commission errors. At the end, participants were asked one final question ("Do you remember anything else about your doctor exam that day?"). Questions were roughly balanced for correct yes and no answers. Participants also completed a battery of psychopa- thology measures (all with strong psychometric proper- ties and appropriate for age and race/ethnicity). Of note here, at Time 2 they completed the 40-item Trauma Symptom Checklist (TSC; Elliot & Briere, 1992) and the 49-item Posttraumatic Diagnostic Scale (PDS; Foa, Cashman, Jaycox, & Perry, 1997). For our sample, rel- evant means and standard deviations on these measures were: TSC total score, M = 23.92, SD = 14.08; TSC depression, M = 6.17, SD = 3.91; PTSD avoidance, At = .70, SD = .67; and PTSD arousal, M = .82, SD = .83. To assess how frequently participants had discussed their memory of the UTR program, Time 2 participants were also asked, "How frequently have you discussed your stay at Mt. Sinai Hospital with others?" Participants responded using a 5-point scale (1 = never, 6 = very frequently), = 1.58, SD = .88.2All participants were debriefed at the end of the interview (e.g., told that it was normal not to remember everything from the UTR, asked how they were doing). As many participants did not remember portions of the event in question or the UTR at all, special attention was paid to assure the participants that some questions might not have applied to them and that we asked the same questions regardless of an individual's specific experience. At debriefing, participants were given infor- mation on support hotlines they could contact. After completion of the interviews, research assis- tants (RAs) blind to hypotheses transcribed and de- identified the interviews (removing any identifying information not relevant to the accuracy of the medical examination). Of central interest was memory for the documented genital contact (i.e., vaginal or penile touch). Interview responses across free-recall, open- ended, and closed-ended questions were coded on a checklist for report of genital contact, false denial of genital contact (omission errors), and "don't know"/"don't remember" responses. Participants' answers to the Time 2 specific questions were scored as proportion correct, commission errors, omission errors, and 'don't know"/'don't remember" replies. Responses to Time 2 misleading questions were coded as proportion correct, incorrect, and "don't know"/'don't remember" responses. To analyze the accuracy of par- ticipants' overall reporting, including monitoring of their lack of report, particularly after such a long delay, a genital report variable was created to capture not only participants' rates of correct and incorrect responding but also their "don't know" responses (-1 = incorrect recall, 0 = don't know, 1 = correct recall). Results Descriptive and correlational analyses Information on key variables is presented in Tables 1 and 2. For the variable indexing discussion of the hos- pital visit, mean imputation replaced missing data for 4 people. All significant effects are reported. Report of genital contact. As can be seen in Table 1, across the entire Time 2 memory interview (collapsing across free-recall, open-ended, and closed-ended ques- tions), a slight majority of the participants (57%) failed to report the documented genital touch (e.g., said they did not remember what parts of their bodies were exam- ined), including 2 participants who denied that such touch had occurred (e.g., said the doctor examined their upper bodies but not their private areas). However, 13 (43%) of the participants correctly reported it. For the 3534-013 Page 5 of 16 EFIA_000 10284 EFTA00159932 6 Goldfarb el al. Table 1. Percent of Adults Who at Timc 2 Reported Timc 1 Genital Contact, Denied Genital Contact, or Said "Don't Know," Analyzed by Time 1 Age Group and Gender Agc gmup at Time 1 Gender AJ 3-5 years 6-10 years 11-15 years Mak Female Type of genital contact report' (n=7) = 15) (n=8) = 10) (n= 20) (N=30) Reported genital contact 28.6% 40.0% 62.5% 20.0% 55.0% 43.3% Incorrect denial of genital contact 14.3% 6.7% 0% 20.0% 0% 6.7% -Don't know" across all question types 57.1% 53.5% 37.5% 60.0% 45.0% 50.0% 'Collapsed across free-recall, open.ended, and elosed-ended questions. subset of respondents who were asked specifically about anal touch (n = 11), omission errors were more frequent for anal (55%) compared with vaginaVpenile (7%) con- tact, 1(10) = —3.46,p = .006. Lost memory. Five out of 30 (17%) participants did not remember, or at least did not disclose, being at the UTR hospital unit at all (i.e., evincing a "lost memory"). For the overall sample, having a lost memory of the UTR was not significantly correlated with age (r = —.05; p = .808) but was significantly correlated with gender, even with age partialed (r = —.44, p = .016): Males (40%) were more likely than females (5%) to express having no knowledge of ever being at the UTR program. Not recalling the UTR was also significantly related to higher total TSC scores, r = —.37, p = .048 (with age partialed; with gender par- tialed, r = .28, us). Child age. Being older at Time 1 was associated with a greater likelihood of accurately reporting genital contact at Time 2, r = .40, p = .027. No participant who was over the age of II when the genital touch occurred falsely denied it. Of those who remembered Time 1 genital contact, 1 adult was only 4 years old (53 months) at the time. Four additional participants were of this age at Time 1, 3 of whotn said "don't know' at Time 2 and 1 of whom incor- rectly denied genital touch had occurred. None of the 30 panic-4)3ms wa.s younger than 4 years old at Time 1.3 Correlations controlling for child age. Partial correla- tions, statistically controlling for Time 1 age, assessed whether other potential theorized predictors (e.g., gender, depression, PTSD symptoms) related to long-term mem- ory of the anogenital examination (Table 3). Consistent with the lost-memory findings, males (vs. females) were significantly less likely to report genital contact and more likely to omit information. Males also had lower shop-term memory (STM) scores at Time 1, but Time 1 STM (Al = 43.87, SD = 7.47) was not significantly correlated with memory performance. The Titne 1 memory-interview Table 2. Proportion of Correct, Incorrect, and "Don't Know" Responses to the Time 2 Memory Closed-Ended Questions About the Anogcnital Examination Question type Agc group at Time 1 Gender Al 3-5 years (n=7) 6-10 years (n= 15) 11-15 years (n = 8) Male (n= 10) Female (n = 20) (N=30) Specific questions Correct .17 (.20) .34 (.29) .33 (.30) .22 (.23) .33 (30) .30 (.28) Commission .05 (.08) .05 (.05) .06 (.07) .05 (.07) .06 (.06) .05 (.06) Omission .04 (.09) .03 (.08) .03 (.05) .08 (.11) .02 (.03) .04 (.07) 'Don't know' .73 (.37) .58 (.36) .58 (.39) .65 (.39) .60 (.36) .61 (.36) Misleading questions Correct .15 (.28) .17 (.19) .12 (.13) .14 (.24) .16 (.17) .16 (.20) Incorrect .23 (.22) .17 (.20) .24 (.28) .23 (.24) .19 (.22) .20 (.22) 'Don't know' .62 (.37) .65 (.34) .64 (.36) .63 (.40) .65 (.32) .64 (.34) Note: Means are accompanied by standand deviations in parentheses. misleading questions incorrect = commission + omission errors. 3534-013 Page 6 of 16 EFIA_000 10285 EFTA00159933 Table 3. Partial Correlations of Key Variables Controlling for Participant Age at Time I Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 IS 16 I" 10 I. Gender 1.00 2. Ethnicity 0.09 1.00 3. Time I STM 0.4r -cum 1.00 4. Time I Memory Interview 0.21 0.08 —0.04 1.00 5. CSA status" 0.09 0.10 0.15 —0.04 1.00 6. Time 2 Genital Contact Report 0.38' 0.05 0.001 0.05 0.48* 1.00 7. Time 2 Npo. tic Correct" 0.14 0.09 0.32 0.13 0.27 0.25 1.00 8. Time 2 Npo. tic Comm,' 0.04 0.33 0.22 0.16 0.23 0.38o 0.70" 1.00 9. Time 2 NI.. tic Omissions -O.3T 0.15 -0.15 0.19 -0.28 -0.13 0.4P 0.52" 1.0D 10. Time 2 specific DK4 —0.03 —0.15 —0.25 -0.16 -0.19 -0.23 -0.96" -0.8P• -0.61" 1.00 It, how 2 MI. Conrecri 0.06 0.15 0.10 -0.001 0.21 0.33 0.72" 0.59* 0.40* -0.73" 1.00 12. Time 2 MI. Incorrect -0.10 0.20 0.13 0.03 0.05 -0.02 0.62" 0.71" 0.46" -0.69" 0.34 1.00 13. Time 2 ML l)K 0.03 —0.22 —0.14 -0.02 -0.16 -0.17 -0.82" -0.79" -0.52" 0.87" -0.79" -.84" 1.00 14. Time 2 Total TSC 0.29 0.02 0.17 0.08 0.32 0.11 0.34 0.09 -0.11 -0.25 0.09 0.03 -0.07 1.00 15. TSC-Depression 0.17 0.10 0.07 -0.01 0.27 0.17 0.5000.22 -0.13 -0.39. 0.26 0.26 -0.32 0.79" 1.0D 16. PTSD Avoid Severity -0.04 0.09 0.08 0.09 0.6r 0.09 0.13 0.26 -0.14 -0.12 0.07 0.12 -0.12 0.39 0.31 1.00 17. PTSD Arousal Severity 0.21 -0.19 -0.03 0.09 0.62" 0.26 0.08 0.02 -0.25 -0.01 -0.08 -0.13 0.14 0.73" 0.44o 0.64" 1.00 18. Discussing Hospital VLsitf0.21 —0.09 0.15 -0.06 0.13 0.27 0.3600.25 -0.14 0.29 0.31 0.27 —0.35 0.16 0.30 —0.12 —0.09 1.00 Nitre: N.30. except child sexual abuse (CSA) status, AN'. 27, and PTSD variables, N= 25. 5Th = short-teem memory; Comm = commission; DK = 'Don't Know"; ML = misleading. = male, I = female. = non-African American, I = African AMC/ILIA. s0 = not CSA, I = CSA. 'CI inconett recall of genital contact, 0 = don', know, I = correct recall of genital contact. `Tune 2 memory variables are all proportion scores (dosed-ended queitions). (Imputed values. 71, < .05. "p < .01. EFTA00159934 8 Goldfarb et al. Table 4. Age, Gender, Time 2 Depression, and Time 1 Child Sexual Abuse (CSA) Status Predicting Time 2 Memory for the Anogenital Exam Model Genital contact correct Proportion specific correct Proportion specific omission b SE p b SE I r b SE p Model 1 Age .06 .03 .33 1.781.02 .02 .28 1.40 <.001 .004 .02 0.09 Gender .33 .24 .25 1.36 .03 .13 .04 0.21 -.08 .03 -.48 -2.60• R2 = .20 R2 = .08 R2 = .22 g2, 24) = 3.081 1(2, 24) = 1.09 FT2, 24) = 3.46• Model 2 Depression .01 .03 .08 0.40 .04 .01 .52 2.92" -.001 .004 -.07 -0.36 AR' = .006 Are = .25 AR2= .004 1(3, 23) = 2.03 1(3,23) = 3.80* ki3, 23) = 2.26 Model 3 CSA .59 .25 .47 2.38' .09 .12 .15 .72 .04 -.27 -1.27 AR! = .16 AR/ = .02 AR2 = .05 114, 22) = 3.26• 1(4, 22) = 2.92° 1(4, 22) = 2.15 Now: Depression = '1SC depression suhscale score; GSA: 0 = not CSA, t = GSA; Genital contact collect: -1 = incorrect recall of genital contact, 0 = dont know, t = collect recall of genital contact, a = 27. 1p<.10. sp < .05. "p < .01. variable (i.e., having had a Time 1 memory interview) was not significantly related to gender, ethnicity/race, or Time 2 memory performance. Regarding psychopathology, total TSC scores were not significantly correlated with the memory variables shown in Table 3. Contrary to our predictions, Time 2 depression was not significantly related to report of genital contact, but consistent with prediction, greater depression was significantly correlated with greater memory accuracy as assessed by specific questions (correct and "don't know" responses). The PTSD avoid- ance and anxiety scores were not significantly associ- ated with memory performance and were not unique predictors in preliminary regression analyses; thus, they were maintained for control purposes only, as needed. Having engaged in more discussion of the hospital visit was associated with increases in the proportion of cor- rect answers to specific questions at Time 2. Because ethnicity, STM, total TSC score, and experience of a Time 1 memory interview were not significant predic- tors of memory in correlational and preliminary regres- sion analyses, they are not considered further. Unique predictors of memory nearly 20 years later The regression models discussed below tested the unique predictors of long-term memory of the medical examination. The first set of analyses concerned mem- ory of genital contact. The second set concerned the adults' accuracy in response to closed-ended questions about the anogenital examination generally. Throughout, in each of the sets of regressions, unless indicated oth- erwise, Time 1 age and gender were tested in the first model, depression was added in the second model, and child sexual abuse status was added in the third model (Table 4).. Genital contact. In the regression analyses of genital contact memory, the first model was not significant, p = .065, although there was a trend for those who were older at Tame 1 to be more likely in adulthood to remem- ber the childhood genital contact (p = .088), as would he expected. Gender was not a significant predictor. When depression was added, the model was also not signifi- cant. However, in the final model, with child sexual abuse status included, the model was significant: Having been a child victim of sexual abuse at Time 1 was a significant predictor in adulthood of accurately reporting of child- hood genital touch experienced during the UTR medical exam. Because frequency of discussion about the hospital visit was expected to affect memory, we conducted the regression analysis above but with the hospital-visit discussion (imputed) variable added in the third model, and then with child sexual abuse status added in the last model. The model for frequency of discussion was not significant, R2 = .32, F(4, 22) = 2.58, p = .065, R=A = .11. However, the model that included child sexual abuse status was significant, R2 = .467, F(5, 21) = 3.66, p = .016, R2A = .15; child sexual abuse remained a sig- nificant predictor of adulthood memory of childhood genital contact even after controlling for frequency of discussions: child sexual abuse status, b= .56, SE= .23, 3534-013 Page 8 of 16 EFIA_000 10287 EFTA00159935 Remembering Genital Touch 9 13 = .45, 1(21) = 2.39, p = .026. In two sets of regressions, when the PTSD variables were separately entered in second models followed by entering depression in the third models, the second models were not significant, ps 2 .078, but in the third models, child sexual abuse status remained a significant predictor, bs 2 .72, SEs 2 .32, ps 2 is 2 2.24, ps 5 .038 (ns = 22). Closed-ended questions. We were also interested in unique predictors of Time 2 accuracy in answering spe- cific and misleading questions about the anogenital examination. For proportion of correct responses to spe- cific questions, Model 1 was not significant (Table 4). However, when depression was added, the model was significant, and depression predicted greater memory accuracy. When child sexual abuse status was added, the model was also significant, but child sexual abuse status was not a significant unique predictor. We also conducted the regression above but with the discussion variable added in a second model, with depression added in the third model, and child sexual abuse status added in the last model. The second model, R1 = .18, g3, 23) = 1.65, p = .205, E2A = .09, p = .118, and the frequency of discussion (imputed) variable, b = .10, SE = .06, 14 = .32, t(23) = 1.62, p = .118, were not significant. In contrast, the model that included depression was still significant, R2 = .36, F(4, 22) = 3.08, p = .037, R2E, = .18, and depression remained a signifi- cant predictor of memory accuracy in answering spe- cific questions, b = .03, SE = .01, p = .47, 1(22) = 2.50, p = .021. Similarly, when the PTSD variables were entered in second models followed by entering depres- sion in the third models, the models with the PTSD vari- ables were not significant, p > .576, but in the third model, depression remained a significant predictor, bs 2 .04, SEs 2 .015, Bs 2 .54, ts 2 2.64, ps 5 .017, (ns = 25). For proportion of specific omission errors, age and gender were again entered in the first model, which was significant. Although age was not a significant pre- dictor, gender significantly predicted more omission errors. The other models and predictors for proportion of specific omission errors were not significant. There were no other significant models for specific questions (proportion of commission and "don't know" responses). Furthermore, there were no significant models for pro- portion of misleading questions (proportion of correct, incorrect, and "don't know" responses). Discussion Controversies about memory accuracy in historic child sexual abuse cases challenge clinical and cognitive researchers to examine, and theorists to explain, the accuracy or inaccuracy of reports of childhood genital contact that actually occurred (Skeem, Douglas, & Lilienfeld, 2009). There is a pressing clinical, societal, and scientific need to know whether individuals who have experienced childhood trauma can accurately remember genital contact decades later and to identify factors that promote accurate reporting of childhood events (e.g., Goodman, Goldfarb, Quas, & Lyon, 2017). A main goal of this study was to examine the accu- racy of adults' long-term memory, after a nearly 20-year delay, for genital contact and related events experi- enced in childhood. Almost half of the adults (43%) correctly disclosed genital touch that occurred during a childhood medical exam. Most participants (93%) either accurately reported the touch or stated that they did not know if the touch occurred; only 2 participants incorrectly denied the genital touch (7%). Whereas most participants utilized a conservative response strategy (e.g., saying "don't know") in answering closed-ended questions regarding the exam, some participants revealed suggestibility, mainly as a tendency to falsely report schema-consistent information. Although genital touch during medical exams is not at issue in most child sexual abuse cases, it is the alleged crime in some his- toric prosecutions (e.g., People of the State of Michigan v. Lawrence Gerard Nassar, 2018; State of California v. William Ayers, 2013). Such touch also serves as an ana- logue to child sexual abuse that can he scientifically studied to examine memory. Predictors of long-term memory of significant childhood events Child age. The 30 individuals tested ranged widely in age (4 to 17 years) at Time 1. Across this age range, one would expect older compared to younger children to better remember the forensic experience (Peterson, 2015), and for this age pattern to carry over into adult- hood. In the present study, correlational analyses indi- cated the expected age effect, with adults who were older at Time 1 being more likely than adults who were younger at Time 1 to remember the genital touch, although the finding was not quite significant in regres- sion analyses when gender was also considered. In this regard, it is relevant that even 1 of the youngest partici- pants (53 months old at the time of the exam) recalled the genital contact. As noted by Peterson (2015), this long-term accurate memory of information encoded around or shortly after the time of infantile amnesia is particularly of note became even though many adults cannot remember events from this early period of devel- opment, some individuals who were quite young at the time of encoding are able to remember an emotional event decades later (McDermott Sales et al., 2005; Usher & Neisser, 1993). For some adults, accurate recollections 3534-013 Page 9 of 16 EFIA_000 10288 EFTA00159936 10 Goldfarb et al. of events experienced at an early age can include mem- ory of genital contact. Child sexual abuse. Consistent with Eisen et al.'s (2007) results, child sexual abuse status at Time 1 pre- dicted accurate reporting of genital touch. It is passible that the anogenital part of the exam was particularly rel- evant or salient for these children or that they were sen- sitized to such contact (Goodman, Rudy, Bottoms, & Aman, 1990). For example, for participants with a child sexual abuse history, the forensic medical exam (i.e., as part of a maltreatment investigation) likely had additional importance in determining if they would he placed in faster care or returned home. Moreover, if at Time 1 the children falsely denied that sexual assault occurred when it actually had taken place, they might have feared that the medical exam would reveal the truth, making the exam stand out in memory (Lyon, 1995). Furthermore, compared to the other adults interviewed, those with child sexual abuse histories, even in childhood may have had a richer semantic knowledge base regarding genital contact within which to encode the exam (Howe, 2011), and this may have aided accurate reporting or inference (rather than stating "don't know") 20 years later. Like Eisen et al. (2007), Katz et al. (1995) found that children in maltreatment evaluations show particularly robust memories of genital touch if the investigations were for sexual abuse. Of importance, particularly to historic child sexual abuse cases, the present study adds that such findings hold even after delays of 20 years, regardless of whether or not an earlier memory inter- view was administered. Psychopathology. One goal of this study was to deter- mine whether individual differences in trauma-related psychopathology predicted memory accuracy or error. Although PTSD symptoms were not significantly related to memory performance, participants who were more depressed at Time 2 more accurately answered specific questions about the anogenital exam. Because depressed individuals ruminate on past negative incidents (Hertel & El-Messidi, 2006), their memory of distressing childhood experiences may he better preserved than that of indi- viduals who are less depressed. Although this study did not directly address rumination or the "chicken-and-egg" question of whether memory for negative events drives depression or, alternatively, depressed individuals focus on negative life experiences (Everaert, Bronstein, Cannon, & Joormann, 2018), the present research indicates that depression is associated with comparatively accurate memory of negative childhood occurrences endured almost 20 years earlier. Gender. When interviewed about emotional childhood events, adult males tend to remember less than do adult females (Davis, 1999), including about child sexual abuse (Widom & Morris, 1997). In the present study, compared with females, males were more likely to exhibit "last memory" for being at the UTR, were less likely to report genital contact (correlational analyses), and were more likely to make omission errors in answering specific questions (regressions). Furthermore, the 2 individuals who denied that they experienced genital touch were both males. The anogenital exam differed slightly for males and females, but both genders experienced swab- bing of their genital and anal areas, as well as visual and manual inspection. In adulthood, males may find it emo- tionally difficult to discuss experiences relevant to child- hood victimization or may have avoided thinking about emotional childhood memories generally. Time 1 memory interview and IITR discussion In this study, the Time 1 interview, which included specific and misleading questions, did not taint memory reports after an almost 20-year delay. The questions asked at Time 1 varied considerably as to the format of the questions, and they covered a wide range of infor- mation about the anogenital exam, concerning fairly innocuous information (e.g., "Was there a sink in the room?" which there was) to misleading but not abuse- related questions (e.g., "When you went to the doctor's room, there was a little boy from the playroom with you, wasn't there?" when in fact there was not), to highly inappropriate legally chargeable acts (e.g., "Did the doctor or nurse hit you?" "How many times did the doctor kiss your "Did the doctor take her/his clothes off?" none of which the doctor or nurse did). Yet accu- racy did not significantly differ between those partici- pants who had a Time 1 interview and those who did not, and the correlations with Time 2 inaccuracy (e.g., commission errors) were low. Overall, as far as we could detect with a relatively small sample, there was no apparent memory-malleability effect in relation to the content of the Time 1 questions intruding into the adults' long-term memory reports. The results provide insight into whether prior inter- views in childhood predict long-term memory of emo- tional experiences (Ornstein et al., 2006; Peterson, 2015; Peterson et al., 2005). In line with Ornstein et al.'s (2006) and Peterson's (2015) results, we found that having a prior interview did not adversely or positively affect the overall accuracy of later memory. Similarly, discussion of the UTR experience with others, which could be a source of rehearsal of accurate or inaccurate 3534-013 Page 10 of 16 EFIA_000 10289 EFTA00159937 Remembering Genital Touch 11 information, generally did not significantly predict memory accuracy, as indicated with regression, although having engaged in more discussion of the hospital visit was associated with increases in the proportion of cor- rect answers at Time 2 to specific questions, as revealed in correlational analyses. It should be noted, however, that discussion of the UTR experience was self-reported as occurring infrequently, perhaps because the child maltreatment investigation was often considered a shameful or unpleasant part of childhood. Memory and suggestibility We examined memory and suggestibility for true infor- mation that was documented during the medical exam and for information that was false. The robustness of memory for what actually occurred may vary depending on the consequentiality, salience, and taboo nature of the act (Goodman et al., 1990). For example, to the extent that anal touch is more taboo than genital touch, it is of interest that participants who recalled experienc- ing genital contact underreported anal contact when asked about it directly. Although differences in disclo- sure of overall genital compared to anal touch may have arisen from children's failure to encode the anal touch at Time 1, an alternative possibility is that individuals may be more reluctant to disclose a rectal than a genital exam because of societal and socioemotional factors (e.g., embarrassment; Saywitz, Goodman, Nichols, & Moan, 1991). For abuse-related events that did not occur, no par- ticipant falsely reported that either the doctor or nurse committed a highly inappropriate abuse-related act (i.e., hitting or kissing the participant). At Time 2, most participants (57%) correctly denied having been hit or kissed by the doctor or nurse, and no participant made a commission error to these two questions. Instead, the remaining 43% of participants stated that they did not know whether these events had occurred. The individu- als who indicated they did not know if they had been hit or kissed by the doctor or nurse may have been able to use metacognitive strategies to monitor the absence of their memories (Koriat et al., 2000). Note that no commission errors were made to these two questions, even though the acts had been falsely suggested to half of the adults when they were children at Time 1. Thus, as far as we could detect, there was no Time 1 carryover effect on Time 2 memory for the highly inappropriate acts over the almost 20-year delay. Despite participants correctly denying the criminally chargeable acts of being hit or kissed, some participants falsely affirmed other abuse-related information at Time 2, including in response to both specific and misleading questions. For instance, in response to the Time 2 specific question, '