PSYCHOLOGICAL SCIENCE

PSYCHOLOGICAL SCIENCE Research Article Traumatic Impact Predicts Long- Term Memory for Documented Child Sexual Abuse Kristen Weede Alexander,' Jodi A. Quas,2 Gail S. GotsImams Simone Gheiti," Robin S. Edelstein,3 Allison D. ReAllich,3 Ingrid M. Cordon,3 and David P.H. Jones(' ! California State University; Sacramento; 2Univenity of California, Irvine; 'University of California, Davis; 4Research Institute on. Judicial Systems, National Research Council, Bologna, Italy; sPolicy Research Associates, Inc., Delmar, Nev York; and 6University of Oxford, Park Hospital for Children, Oxford, England ABSTRACT—Prospective studies of adults' memories of documented child sexual abuse (CSA) reveal that the majority of individuals remember their victimization. However. the accuracy of these memories has rarely been investigated scientifically. The present study examined predictors of memory accuracy and errors 12 to 21 years after abuse ended for individuals with legal experiences resulting from documented CSA. Severity of posttraumatic stress disorder (PTSD) symptomatology was positively associated with memory accuracy. However, individuals nominating CSA as their most traumatic life event exhib- ited relatively accurate memory regardless of indicators of PTSD. Predictors of memory errors were also identified (e.g., less maternal support). These results indicate that, in addition to understanding the role of traditional cog- nitive factors, understanding an event's traumatic impact is important for predicting the accuracy of long-term memory for reported CSA. Although research has focused on forgetting of child sexual abuse (CSA), virtually no studies have examined the accuracy of long-term memory for such trauma among adults who remember being victimized. We conducted a longitudinal study involving victims of documented and prosecuted GSA to investigate, prospectively, the accuracy of long-term memory of sexual abuse in childhood. Address correspondence to Gail S. Goodman. Department of Pm.. chology. One Shields Ave.. University of California. Da via. CA 95616. ggoodmanenctlavis.edu. or to Jodi (inns. Department of Psychology and Social Behavior. University of California. Irvine. CA 92697. e-mail: Mnarlitori.mlu. Volume 16—Banta I We specifically sought to examine how reactions to trauma (i.e., trauma symptoms. cog ,e appraisals) affect memory (see also Williams. 1995). Reganling trauma symptoms. much cur- rent interest focuses on memory functioning in relation to posttraumatic stress disorder (PTSD; McNally. 2003), a psy- chopathology that is defined by anxiety and memory problems (Sullivan & Cowan, 2002) and is a possible sequela of CSA (Browne & Finkelhor. 1986). Research suggests that trauma survivors, especially those with PTSD, overattend to trauma- related cues (e.g., for rape victims, the wont rape on the mod- ified Stroop test; e.g., Field et al.. 2001: Foa. Feske, Murdock, Kozak. & !McCarthy. 1991) and evince heightened memory and reduced forgetting of trauma-related words (e.g., Amir. Mc- Nally. & Wiegartz. 1996: McNally. Metzger, Lasko, Clancy. & Pitman. 1998: Paunovic, Lunclh, & Oest. 2002). One possible explanation of such findings is that traumatized individuals, especially those with PTSD. develop semantic networks ("fear networks") within which trauma-related infor- mation becomes stored (e.g.. Foa, Steketee, & Rothbaum, 1989; but see Foe & Riggs. 1993, and for review, Dalgleish. 2004). These networks may predispose traumatized individuals to be hypervigilant to trauma-related information and retain it par- ticularly well. These networks could also affect retention of the original trauma (e.g.. by fostering greater rehearsal) and influ- ence memory for similar experiences or traumatic events gen- erally (e.g., by activation of related nodes in memory). Accordingly, one might predict that abused individuals, espe- cially those who exhibit high levels of PTSD symptomatology. have robust CSA memories. despite findings that link trauma backgrounds with information processing deficits for non- trauma-related infommtion (e.g.. Ike r. Shobe. & Kililstmni. 2000: Edwank. Finsh. Anda. Felitii. & Node berg, 2001: C.0 00,0i t 2015 Anwritan No. 6,4opgal Society 33 3534-009 Page 1 of 8 EFTA_000 10253 EFTA00159901 Memory for Child Sexual Abuse Hunter & Andrews, 2002; McNally, Lasko, Macklin, & Pitman. 1995: cf. Metcalfe & Jacobs, 2000. reganling "hot" and "cold" memory systems). Independent of potential effects of PTSD symptomatology on memory of C.SA, cognitive appraisals of distressing events may also affect CSA memories. Events appraised as highly negative are remembered especially well (e.g., Bennsett, 2002), pethaps because of their salience or distinctiveness. They may activate physiological (e.g., amygrIala; LeDoux, 2000) and behavioral (e.g., rehearsal; Christianson, 1992) responses that help con- solidate or maintain memories, possibly including memories for whole categories of experiences. Moreover, for highly negative incidents, information directly related to the cause of the stress is prioritized in memory, with such information often retained better the greater the distress (Christianson, 1992); this relation between memory and distress implies that increased severity of CSA would result in more accurate memory for main features of the abuse. Thus, overall, although alternative hypotheses exist (e.g., traumatic reactions and negative appraisals could overwhelm coping abilities or activate defense mechanisms, leaving victims with reduced or distorted processing of CSA experiences; Ten, 1991: van der Kolk, 1997), there is reason to predict that individuals who appraise their CSA as a particularly traumatic experience would remember it well, es- pecially if they evince high levels of FISH symptomatology, and further that more seven• abuse would result in better CSA memory. In addition to the aforementioned trauma-specific factors, numerous factors correlated with trauma undoubtedly affect memory. Thus, the impact of trauma must be considered in the context of other potentially important variables. Increased se- verity of CSA may result in increased trauma, but severity can comprise such factors as duration and frequency, which them- selves may affect memory accuracy independently of traumatic impact. Moreover, severe abuse is often aggressively prose- cuted. Insofar as extensive legal involvement results in elab- orative rehearsal, greater stress, and a more distinct experience, children with extensive legal involvement may remember their CSA experiences particularly well. Another set of variables potentially correlated with trauma concerns interpersonal relationships. Close relationship to the perpetrator is associated with lack of maternal support when child victims disclose their abuse (Everson, Hunter, Runyon, Edelsohn, & Coulter, 1989). To the extent that unsupportive mothers talk less often with their children about the CSA than supportive mothers do or fail to legitimate their children's al- legations, long-term memory may be adversely affected (Goodman, Quas, Battentran-Faunce, Riddlesberger, & Kuhn, 1994; Haden, Mayne, & Fivush, 1997). Furthermore, individ- uals abused by an emotionally close caregiver may feel espe- cially betrayed (Freyd, 1996). Such feelings may reduce the clarity and accessibility of memories (e.g., through cognitive avoidance). However, a close relationship to the abuser (e.g., 34 intrafamilial abuse) may be associated with a detailed knowl- edge base, which may support memory. Thus, relationship variables must he considered, despite the unclear direction of their effects. Factors that are not necessarily related to trauma but affect long-term retention (e.g., age, delay) also likely influence memory of CSA. Older children provide more accurate and complete memory reports than do younger children (e.g., Quas et al., 1999), which suggests that children who were older when the abuse ended may have more accurate long-term memories than children who were younger. Additionally, delay often heralds forgetting; a shorter delay between the event and memory interview should engender better memory. Further- more, compared with females, males may be less likely to dis- close CSA (e.g., Finkelhor, Hotaling, Lewis, & Smith, 1990; Widom & Morris, 1997; but see Goodman et al., 2003) and may have less detailed autobiographical memories about emotional events (Davis, 1999). If males are hesitant to talk about CSA, their memory of it might fade over the years. In the present study, memory accuracy was considered in light of these pos- sible influences. THE PRESENT STUDY Our goal was to examine the accuracy of adolescents' and young adults' memory for CSA that ended 12 to 21 years previously. We hypothesized that individuals displaying more vrsu symptomatology and reporting CSA as their most traumatic life event would exhibit better memory for the abuse than individ- uals displaying less PTSD symptomatology and reporting an- other event as their most traumatic experience. Further, increased abuse severity, greater maternal support, more ex- tensive legal involvement, older age when the abuse ended, shorter delay, and being female were expected to make inde- pendent contributions to memory accuracy. METHOD Participants and Procedure Participants had been involved during the 1980s in a study of the short-term effects of criminal prosecutions on GSA victims (Goodman et al., 1992). In that study. detailed information was collected on two hundred seventeen 3- to 17-year-olds' back- grounds. mental health, and abuse and legal experiences.' The data included demographic information (e.g.. ethnicity. socio- economic status), ratings of the children's behavioral adjust- ment at the outset of prosecution (i.e., Achenbach's. 1991, Child Behavior Checklist. or Gael.), information about the abuse (e.g., victim-perpetrator relationship, abuse duration). and facts 'One participant from the original mudy M a 2181 isnot included because the defendant ma% out t years older. a neve-nary element of the crime for prose- cution. Volume In—Number I 3534-009 Page 2 of 8 EFTA_000 10254 EFTA00159902 LW. Alexander ri al. related to the children's legal experiences (e.g., extent of legal involvement).2 Ten to 16 years following the original study, when participants were adolescents and young adults, they were recontacted to join a study of attitudes toward and experiences with the legal system. The follow-up was conducted in three phases: (a) Phase 1 was conducted via telephone interview, (b) Phase 2 was conducted via mailed questionnaires, and (c) Phase 3 was conducted via in-person interview. Exceptions to interview format were made as necessary (e.g., for participants lacking telephones, Phase I was conducted via mail or in-person in- terview). For scientific and ethical reasons, participants were not told of the researchers' knowledge of the sample's past victimization, legal involvement. or participation in the original study. Interviewers were blind to individual participants' prior experiences. In each phase, information about the participants' mental health, victimization, and legal experiences was elicit- ed. Of the original 217 eligible participants, 81% (n = 175) were interviewed at least once. For the present report. CSA memory accuracy was deter- mined using primarily information collected during Phase 3 (in- person interview), which focused on participants' CSA memo- ries, disclosure, and legal experiences. CSA was defined as exhibitionism, sexual touching, rape, oral sex, or intercourse (completed or attempted acts) that occurred before age 18 and with a person more than 4 years older (see Goodman et al.. 2003). Of the 103 individuals (2.3 males, 80 females) who completed all portions of Phase 3, 7 said they were never abused (although 3 of them had disclosed abuse in Phase 1 or 2), and 2 discussed the legal case but stated the abuse was a false report. The present study concerns the 94 individuals (19 males, 75 females) who reported the former CSA and answered questions about it (see 'fable 1).3 Of these individuals. 69.1% were non- Hispanic Caucasian, 8.5% were African American, 12.8% were Hispanic. and 9.6% were "other." Alleged perpetrators were strangers (9%), acquaintances (30%), nonparental caregivers (39%), or parental caregivers (22%). Of the cases, 44.7% in- volved genital contact and 42.6% involved penetration. At entry into the original study, maternal support following disclosure was assessed on a dichotomous scale; most individuals had supportive mothers = 83 of 91). Participants' legal involve- ment was scored as I if the child never went to the courthouse (n = 25), 2 if the child went to the courthouse hut did not testify (n = 32), or 3 if the child went to the courthouse and testified (n = 37). Participants who completed Phase 3 and disclosed the CSA were comparable to the other participants from the original sample (Goodman et al., 1992) in terms of gender, socioeco- 2Sre Goodman el al. (19921 and Goodman et at (2003) for details about the measures in the original study and identification of target eases. enpectively. sfin predictors of Mime I disclosure. see Goodman et al. j2003). An identical analysis predicting disc home in the smaller Mime 3 sample revealed similar findings. ahlsough age no longer significantly predicted disclosure. Volume IS—Number I TABLE 1 Characteristics of the Sample Variable SD Range Victim's age at end of abuse' 9.88 3.38 3-16 Victim's age at police report' 10.23 3.51 3-17 Victim's age at Phase 3* 24.23 3.55 17-31 Delay' 14.34 1.41 12-21 Socioeconomic MMus' 4.88 1.69 1-7 Victim's relations/tip to perpetrator' 2.76 0.90 1-4 Abuse seventy" 4.81 1.80 2-9 Abuse frequency( 1.88 0.89 1-3 Legal involvements 2.13 0.81 1-3 Maternal support" .91 .28 0-1 Number of posurauniatie stress disorder criteria met 3.6.5 1.95 0-6 'Ages are reported in years. Itelay is the number of seam between the err- station of alone and the Phase 3 interview. lillti0001/1101100 /011111,1 high. 7 low) was 11C0011011.41 during the original study using a modified Ihillingw head index. The soak for relationship to perpetrator ranged from us-anger (I) to pc:ranted figure I1). 'The ahum-aaverity composite index (1 - low. 12 - high) included abuse duration. extent of sexual contact. amount of force. and extent of injury £10100101111(10110,1. of the above. all from the original ease documental ion. '.Wale frrqurney was scored as I (1 tinsel. 2 (24 tinsel). or 3 ('I or mor e time.). 'fhe soak for legal intolvement ranged front minimal II) 10 testified 01. hl a tental support 01111161111104 on a dichotomous *role (0 tt no. 1 seal. nomic status, abuse severity, victim-perpetrator relationship, and original CRC!. total Tscore, ts(213-215)< 1.19. However, this subsample experienced greater legal involvement, received more maternal support, and was older at the end of the abuse than were others from the original sample, ts(164-215) 12.311, ps < .01. PESO was assessed during Phase 2 using the Post-Traumatic Stress" Diagnostic Scale (PDS), which has commendable psy- chometric pmperties (e.g., internal consistencies from .78 to .84 for the criteria: Foa, Cashman, Jaycox, & Perry, 1997). This measure provides a list of traumatic events (including CSA); respondents indicate which events happened to them and, among those, which was the most traumatic. The answer to the latter question was used to score the most traumatic event (MTE) as CSA or another event. The measure then lists a series of questions regarding the effects of participants' self-identified COTE on their lives. Depending on participants' responses to these questions, they could meet up to six PTSD criteria: ex- posure to trauma, reexperiencing symptoms, avoidance symp- toms, arousal symptoms, trauma-related symptoms lasting 3 months or more, and significant distress or impairment in daily functioning. A PTSD diagnosis is made when all six criteria are met. However, because another potentially important index is number of PTSD criteria satisfied, which may reflect more precisely the magnitude of an individual's psychopathology related to a particular trauma, we used slumber of PTSD criteria met (nVISD) as the operational index of traumatization. Of participants who completed the PDS Or = 86), 77 provided detailed information about the nature of their most traumatic 35 3534-009 Page 3 of 8 EFTA_000 10255 EFTA00159903 Memory for Child Sexual Abuse TABLE 2 Correlations Among Key Variables Variable 2 3 4 5 6 7 8 9 10 II 12 I. Victim's age at end of abuse 2. Victim's gender .05 3. Abuse severity .06 .11 4. Legal involvement .17 —.09 .06 5. Victim's relationship to perpetrator —.03 .28** .38*** -.19 - 6. Maternal support -.04 .04 -.04 .11 .09 7. Delay -.09 -.07 -.08 .25a .04 8. Abuse frequency .02 .09 -.11 .04 9. Number of posttraumatic re disorder criteria met .25* .01 .04 .20 .08 .03 -.16 10. Most traumatic event -.22' -.03 .13 -.20 .22' -.02 .30" .10 .00 11. P1011elti011 correct .07 -.01 .16 .19 .01 .29" .09 .26* - 12. P1011elti011 Commission errors .09 .13 -..30" -.24* .13 -.08 -.61' - 13. Proportion omission errors -.16 -.21' .15 -.29" .09 .24' -.26' -.20 -.43a* -.16 Note. No ranged from 73 to 94. Sr. Table I for explanations of rating scales: in addition. for wittiest's gender. male - I and female 2. and most traumatir trent was scored 0 if it was not child sexual abuse and I if it was child sexual abuse. op < .05. cep < .01. "'p < .00I (twodaikil tests). experience, which was CSA for 48 of these participants and another event for 29! Memory Coiling Memory interviews were coded for accuracy according to facts recorded in the original study. The coding scheme was based on 11 points of information (i.e., perpetrator's name, sexual acts, victim's age at onset and offset, perpetrator's age, child-perpe- trator relationship, frequency and duration of abuse, child- perpetrator living arrangement, and force and coercion in- volved). Participants' Phase 3 responses to relevant questions (e.g., "How old were you when the CSA first happened?" "What was the defendant's name?") were examined for this informa- tion. When necessary for clarification, Phase 1 interviews and Phase 2 questionnaires were consulted. Then, responses were compared with the corresponding information from the original study (Goodman et al., 1992). Proportion correct was computed by dividing the number of correct responses by the number of the 11 points of information reported by the participant. For 7 of the 11 points (e.g., whether force was involved), errors could be scored as errors of commission (exaggeration or addition of information relative to the original record: e.g., the victim stating that he or she lived with the perpetrator when our records indicated that was not the case) or omission (exclusion or downplaying of information: e.g., indicating no force was in- volved when our records indicated force was used).' Tor most caws. it was possible to determine that the CSA reported as the MTE was the CSA rase we had studied previously of the 48 participants indicating CSA as their NITE. 36 provided enough information to verify that they were referring to the documented CSA). '`Questions not scored for errors were those such at. -How old were you when the abuse ended?" If. for example. the participant was 10rears old but indicated 9 year,. it was unclear whether this should he scored as a eommissicm or omission error. The denominator for proportion emu" was 7. 36 To establish reliability, four raters independently coded 12% of the interviews. For correct responses, commission errors, and omission errors, proportion agreement ranged from .83 to .99 (Al = .93) for each pair of raters. The raters divided the re- maining interviews for coding. RESULTS Our goal was to identify predictors of memory accuracy for CSA. Dependent measures included proportion correct (M = .72, SD = .18, range = .29-1.00), proportion commission errors (Al = .14, SD = .17, range = .00—.71), and pmportion omission errors Of = .14, SD= .18, range =.00—.80). Correlations among key variables are shown in Table 2. To examine the relative contribution of each variable in predicting memory of CSA, we conducted three linear regres- sions.° Predictors included victim's age at the end of the abuse, victim's gentler, abuse severity, ntatemal support. victim-per- petrator relationship, extent of legal involvement. delay be- tween end of abuse and current memory interview, wrsu, AllE, and the nPTSD x MTE interaction. The regression for proportion correct was significant, F(10, 64) = 3.34, p < .01 (fable 3): Maternal support, nPTSD, MTE, and the MTE x nPfSD interaction were significant predictors. Individuals who received maternal support following CSA discovery evinced more accurate long-term memory than those who did not receive such support. More PTSD criteria met and nominating CSA as the MTE were also associated with more accurate responses. Figure 1 shows the interaction between °Because of missing data. n in the regressions was 73 to 73. The patient of results was virtually identical texcepi as noted in the test) when duration or frequency, which could independently affirm memory. replaced severity. and when only rmmborated caws were considered. Volume 16—Number 3534-009 Page 4 of 8 EFTA_000 10256 EFTA00159904 K.W. Alexander ri al. TABLE 3 Regression Analysis for Variables Predicting Accuracy of Long- Term Memory for Child Sexual Abuse Variable I p Victim's age at end of abuse -.07 -0.62 .54 Victim's Fender .16 1.47 .15 Abuse severity .17 1.47 .15 Legal involvement .03 0.28 .78 Victim's relationship to perpetrator -.10 -0.82 .42 Maternal support .33 3.11 .003 Delay -0.51 .61 MTE 2.54 .01 nPTSD .23 2.07 .04 MTE x nPTSD -3.10 .0O3 Note. n 75. IIITE newt traumatic event (not child sexual abuse —0. child sexual abuse — I). nIPTSD **number of posttraumatie stress disorder criteria met. YrITE and nPTSD were standanfixed to create the interaction term. re - .14. The addition of the interaction term led to le change of .10. p <.01. See Tables I and 2 for explanations of rating scales. KITE and WISE'. For individuals designating CSA as their most significant life trauma, there was virtually no relation between nPTSD and memory for the abuse. r = —.02: memory was rel- Proportion Correct Nominated as Most Traumatic Event Childhood Sexual Abuse -O- Other 1.0 0.9 0.8 0.7 0.6 0.5 Few Many Posttraumatic Stress Disorder Criteria Met Fig. 1. Relation between postiraumatic stress disorder mite. ria satisfied and proportion correct on the measure of memory for the target childhood sexual abuse for participants indicating their most traumatic life event was child sexual abuse versus another event. Lines are plotted according to the regression coefficients. Volume 1O—Nimilei 1 naively accurate regardless of the number of PTSI) criteria satisfied. However, for individuals nominating another life event (e.g., car accident, loved one's death) as their most sig- nificant trauma, the relation between nPISI) and memory was positive. r = A4, p < .03. For these individuals. a greater number of PTSD criteria met was associated with better memory of CSA. It is possible that PTSD symptomatology is related to how often the almse was discussed with a counselor or parent, which could affect memory. To investigate this possibility, we added therapy (whether the participant reported receiving counseling during or immediately following the pmsecutiott) and frequency of GSA discussion with the caregiver (immediately following the CSA and during the delay), as indicated during Phase 2, indi- vidually to the regressions.7 We similarly included self-reported number of other CSA experiences (31 = 1.71 for participants whose MTE was GSA and At = 2.07 for participants whose MTE was not CSA) and number of non-CSA traumas = 7.02 for participants whose MTE was CSA and At = 8.90 for partici- pants whose MTE was not CSA), assessed in Phase 1. No sig- nificant it2 changes resulted" Our accuracy measure might be perceived as simply an index of consistency between the original and current reports. Although some facts used to create the accuracy variable relied solely on children's original statements, others were ob- jectively verifiable (e.g., perpetrator's name and age). When a composite variable including only proportion correct for these verifiable questions was entered as the dependent variable, the regression results remained virtually identical, 1(10, 64) = 2.57, p = .01. lite regression analysis for omission errors was also signifi- cant, 1(10, 62) = 2.17, p < .05, R2 = .26. Maternal support was the ottly significant predictor, 6 = —.30, p = .01 (greater maternal support predicted fewer omissions) 9 although gentler, 6 = —.21,p <.10, appmached significance (females produced somewhat fewer omission errors than males). When we con- ducted the analysis replacing severity with frequency (the two variables were highly correlated), greater frequency of CSA significantly predicted increased omission errors, S = .40, p = .01, despite a positive relation between frequency and 1We did not have infommi ion on recent therapy. "Other mental health indices (composite WOrt}j collected during the initial study C.BC.I.1 or present study (i.e.. Dissociative Experiences Seale. Bernstein & Putnam. PM& Youth Self Report. Achenbach. 1907: Behavior Symptom Index. Dricoraiis. 1983: Beek's Depression Inventory. Beck & Bramesderfer. 1071: Trauma Symptom Inventory. Briery. Elliott. Ilanis. & Colman. 1995) were not significantly related to memory accuracy or errors. m C 1.1911)f the 86 inditilitlaIS whose data allowed flea formal PTSD diagnosis to he made. 14 (15.9,A) reached the clinical diagnosis level (meeting the six criteria). Hawing a PTSD diagnosis was not :elated to proportion comet or commission or omission errom reganling the documented CSA. n < 1(181. largely because participants who met four in fit. PTA) criteria generally evinced accurate memory for the docomenied CSA. ''Participants who reported that their mother avoided talking about the abuse• or told them not to talk about it ma& more omission errors than other partici. pants. = —.20 and —.25. OF 78. ps <.05. 37 3534-009 Page 5 of 8 EFTA_000 10257 EFTA00159905 Memory for Child Sexual Abuse initially receiving counseling related to the case, r = .30, p .05. The regression model for commission errors was not sig- nificant, FPO, 62) = 1.35, le= .18. For the three memory measures, the findings remained vir- tually identical when we excluded the 12 people who indicated another CSA case (not our target case) was their MTE; for ex- ample, the regression for the proportion correct was again sig- nificant, 1(10, 52) = 4.79,p C .001. Also, the effect of gender became significant for proportion correct, B = .23, p C .05; females were more accurate than males. DISCUSSION The present study examined the accuracy of long-term memory for CSA using a prospective design. Individuals who designated CSA as their MTE and individuals with more vrsn symptom- atologr were particularly accurate in their memory of the doc- umented CSA case, which suggests an important influence of trauma symptoms and cognitive appraisals on long-term CSA memory. Evidence demonstrates that memory for emotional events often endures and that trauma-related information is generally retained well, especially among individuals with PTSD (Paunovic et al., 2002). Our results are largely consistent with this evidence. Regarding the significant interaction involving MD crite- ria, it is possible that when CSA (our documented case or an- other incident) was indicated as the MTE, individuals encoded and rehearsed central details of the GSA enough to maintain accurate memory over the long term, regardless of PTSD symptoms. For them, as they reported, CSA was an emotional event, and the emotion or trauma-related semantic network that resulted may have facilitated memory retention (Foa et al., 1989; LeDoux, 2000). Among victims who indicated a non-CSA event as their most traumatic, those with fewer MD symptoms may not have focused on or may have actively tried to forget the CSA, leading to poorer long-term memory. In contrast. those who reported greater vrsu symptomatology (for a non-CSA trauma) may have been preoccupied with trauma or possessed fear networks that generalized to traumatic events overall, supporting relatively accurate long-term memory of GSA. It is also possible that this latter group had discussed the abuse more (e.g., in recent psychotherapy), although apparently not with a therapist right after the target CSA or with their nonoffending caregiver up to the time of the present interview. Maternal support following disclosure of the abuse signifi- cantly predicted a higher proportion correct and fewer omission errors. Previous research has revealed benefits of maternal support on children's memory for distressing experiences (e.g., Goodman et al., 1994). In our study, maternal supportiveness may have facilitated formation of a coherent representation of the abuse. Also, children with supportive mothers may have felt comfortable discussing the. CSA with other people. Both 38 greater coherence of an initial event representation and in- creased discussion could lead to maintenance of accurate memory. When the entire sample was included in the analyses, men's and women's memory accuracy did not differ significantly. However, when the sample was limited to participants speci- fying the target ease as their MTE, results demonstrated that females provided more correct information than did males. Former research has suggested gentler differences in what males and females define as abusive and in their memory for emotional events (e.g., Davis, 1999; Widom & Morris, 1997). To the extent that males may be less willing to discuss GSA. perhaps particularly when the CSA is appraised as a highly traumatic personal event, they may have fewer rehearsal opportunities from the time the abuse ends, even in individuals willing to disclose their victimization. The resulting lack of rehearsal may affect accuracy of long-term memory. Frequency of abuse was associated with a greater proportion of omission errors. There are several possible interpretations of this finding. For instance, frequent events are associated with schematization in memory, which could lead to omission errors. Victims may also downplay incest or abuse in which they were frequently (and perhaps actively) involved (Bidmse & Good- man. 2000). Although the regression predicting commission errors was nonsignificant, two potentially important factors were identified in correlational analyses, warranting further research. Abuse severity predicted lack of commission errors, a result consis- tent with studies indicating that stress is often (although not always) associated with accuracy (e.g., Christianson, 1992). Also, greater legal involvement predicted fewer commission errors. Legal involvement might he associated with elaborative rehearsal, which facilitates long-term retention. The additional stress of possibly or actually testifying may have further solidified memory, thus reducing the likelihood of memory errors. In the regression analyses, age at end of abuse was not sig- nificantly associated with long-term memory. In a previous analysis based on a larger sample (from Phase 1, n = 174; Goodman et al., 2003), we found that age predicted whether participants disclosed CSA. Nondisclosing participants were not included in the present study. Moreover, the present sample did not contain individuals whose abuse ended before they were 3 years old. Age at end of ahuse may he related to childhood amnesia or "lost memory" but not to accuracy for core, salient features among individuals whose abuse ended at age 3 or after and who remember GSA. Finally, victim-perpetrator relationship and delay were sta- tistically unrelated to memory of CSA. Relationship between the victim and perpetrator is likely more complicated than our classification captured. Delay may have lost its predictive power given the long time interval, and subsequent forgetting, between the abuse and memory interview. Volume la—Number 3534-009 Page 6 of 8 EFTA_000 10258 EFTA00159906 K.W. Alexander to A. Some limitations of our study should he noted. The general- izability of our findings to nonreported CSA cases is unknown, and most CSA is never reported to authorities. Because we tested memory in a prosecution sample, disclosure and dis- cussion of CSA had typically taken place during childhood. Our results do not necessarily reflect memory accuracy for cases in which the first disclosure occurs after lengthy delays (i.e., in adulthood). Moreover, if the initial allegations were false or distorted, our findings would be affected. lack of statistical power may have reduced our ability to detect certain effects, such as those of age and delay. Furthermore, future studies should focus on additional memory characteristics (e.g., com- pleteness, emotional content). Nevertheless, by examining trauma-related factors, the present study sheds light on theoretical and applied issues regarding memory for traumatic experiences. Although repli- cation is needed, our results indicate that in addition to tradi- tional cognitive factors, an event's traumatic impact predicts accuracy of long-lemur memory for GSA. Acknowledgments—This article is based on work supported by the National Science Foundation (NSF) under Grant 0001369. Any opinions, findings, conclusions, or recommendations expressed in this article are those of the authors and do not necessarily reflect the views of NSF. We thank Lois Buckman, Deena Day, Patricia England, Gina Gallo, Jenny Garfein, John KihIstrom, Jennifer Noll, Linda Port, Lydia Prato, Juliana Ras- kauskas, Leslie Rudy, Jennifer Schaaf, Daniel Stroski. Eliza- beth Taub, and Penelope Trickett. We also thank members of the Adams, Arapahoe, and Denver County District Attorney Offi- ces, especially District Attorneys James Peters, Robert Grant, and William Hitter, and Deputy District Attorneys John Jordan, Jill Straus, and Norman Brisson. REFERENCES Aehenbach. T.M. (1991). Manual for the Child Behavior thecklist14-18 and 1991 Profile. Burlington: University of Vernon. Department of Psychiatry. Achenbaeh. T.M. (1997). Manual for the Thung Adult Self-Report and Young Adult Debark), Checklist. Burlington: University of Ver- mont. Department of Psychiatry. Amu.. N.. McNally. R., & Wiegartz. P.S. (1996). Implicit memory bias for threat in posttraunratie stress disorder. Cognitive Therapy and Research. 20.62.5-6.35. Beck. A.T..& Beamesderfer. A. (1974). Assessment of depression: The Depression Inventory. Psychological Measurements in Psycho- pharmacology. 7. 151-169. Bernstein. F..M.. & Putnam. F.W. (1986). Development. reliability. and validity of a dissociation scale. Journal of Nervous and Menial Disease. 174.727-735. Berntsen. D. (2(X)2). Iiinnel memories for autobiographical events: Central details are remembered more frequently front shocking than from happy experiences. Memory & Cognition. .30. 1010- 1020. Volume 16—Number Bidrose. S.. & Goodman. C.S. (2000). Testimony and evidence: A scientific case study of memory for child sexual abuse. Applied Cognitive Psychology, 14. 197-214. Bremner. J.D.. Shobe. K.K.. & KihIstrom. J.F. (20(X)). False memories in women with self-reported childhood sexual abuse: An empir- ical study. Psychological Science. 11. 33.3-337. Brien, J.. Elliott. D.M.. Hanis. K.. & Colnran. A. (1995). Trauma Symptom Inventory: Psychometrics and association with child- hood and adult victimization in clinical samples. Journal of In- terpersonal Violence. 10. 387-401. Browne. A.. & Finkelhor. D. (1986). Impact of child sexual abuse: A review of the research. Psychological Bulletin. 99.66—i7. Christianson. S.-A. (1992). Emotional stress and eyewitness memory: A critical review. Psychological Bulletin. 112. 284-309. Dalgleish. T. (2004). Cognitive approaches to posttraumatic stress disorder: The evolution of multheptesentational theorizing. Psy- chological Bulletin. 130 228-260. Davis. P. (1999). Gender differences in autobiographical memory for childhood emotional experiences. Journal of Personality and Social Psychology. 76.498-510. Derogatis. LR. (1983). SCL-90-R: Administration. scoring, and pro- cedures manual. Townsend. MO: Clinical Psychometric Research. Edwards. YJ.. Fivush. R.. Anda. R.F.. Felitti. Vi.. & Nordenberg. D.F. (2/01). Autobiographical memory disturbances in childhood abuse sunivota. Journal of Aggression. Maltreatment and Dau- m°. 4. 247-263. Everson. M.D.. Hunter. W.M., Runyon. D.K.. F.delsohn. C.A.- & Coulter. FILL (1989). Maternal support following disclosure of incest. American Journal of Orthopsychiatry. 59. 197-207. Field. N.P.. Classen. C.. Butler. L.D., Koopman. C.. Zarmne. J.. & Spiegel. D. (2001). Rev ictimization and information processing in women survivors of childhood sexual abuse. Journal of Anxiety Disorders. 15.459-169. Finkelhor. D.. /totaling. G.. Lewis. I.A.. & Smith. C. (1990). Sexual abuse in a national survey of adult men and women: Prevalence. characteristics. and risk factors. Child Abuse and Neglect. 14. 19-28. Foa. E.B.. Cashman. I... Jaycox. L. & Peny. K. (1997). The validation of a self-report measure of posuraumaiie stress disorder. The Postuatunatic Diagnostic Scale. Psychological Assessment. 9. 445 .51. Foa. E.B.. Feske. U.. Murdock. T.. Kozak. MJ.. & McCarthy. PR. (1991). Processing of threat-related information in rape victims. Journal of Abnormal Psychology. 100 156-16.2. Foa. E.B..& Riggs. D.S. (1993). Post-traumatic stress disorder in rape victims. In J. Oldham. M.B. Rata. & A. Tasman (Eds.). Reline of psyrkiatey (Vol. 12. pp. 27.3-303). Washington. DC: American Psychiatric Association. Foa. F.R.. Stekeleix. C.. & Rothbaum. B.O. (1989). Behavioral/cogni- tive conceptualization of post-traumatic stress disorder. Behavior Therapy. 20. 155-176. Freyd.J.J. (1996). Betrayal trauma: The logic of forgetting childhood abuse. Cambridge. MA: Harvard University Press. Goodman. GS.. Chetti. S.. Quas. J.A., Edelstein. R.S.. Alexander. K.W. Redlich. A.D.. Cordon. I.M.. & Jones. D.P.H. (2003). A prospective study of memory for child sexual abuse: New Endings relevant to the repressed-memory controversy. Psychological Science. 14. 11.3-118. Goodman. GS.. Quas, J.A.. Batterman-Faunce. J.M.. Riddlesberger. M.M.. & Kuhn. J. (1994). Predictors of accurate and inaccurate memories of traumatic events experienced in childhood. Con- sciousness and Cognition..3. 269-294. 39 3534-009 Page 7 of 8 EFTA_000 10259 EFTA00159907 Memory for Child Sexual Abuse Goodman. C.S.. Taub. E.P.. Jones, D.P.H.. England. P.. Port. LK.. Rudy. L. & Prado. L. (1992). Testifying in criminal court. Mono- graphs of the Society for Research in Child Development. 57(5. Serial No. 229). Haden. C.A.. Rayne. R.A.. & Fivush. R. (1997). Des eloping narrative structure in parent-child reminiscing arras the premthool yews. Development& Psychology. .33. 295-307. Hunter. F..C.M.. & Andrews. R. (2002). Memory for autobiographical facts and events: A comparison of women reporting childhood sexual abuse and non-abused controls. Applied avatar Psy- chology, 16. 575-588. LeD0ux. J.E. (2000). Emotion circuits in the brain. Annual Review of Neuroscience. 23. 155-184. McNally. R.J. (200.3). Progress and controversy in the study of post- traumatic stress disonler. Annual Review of Psychology. 54.229- 252. McNally. RJ.. Lasko. N.B.. Macklin. M.L. & Pitman. R.K. (1995). Autobiographical memory disturbance in combat-related post- traumatic stress disorder. Behaviour Research and Therapy. 33. 619-630. McNally. RJ.. Metzger. L.J.. Lasko. N.D.. Clancy. S.A.. & Pitman. R.K. (1998). Directed forgetting of trauma cues in adult survivors of childhood sexual abuse with and without posttraumatie stress disonler. Journal of Abnormal Psychology. 107. 596-601. Metcalfe. J.. & Jacobs. W.J. (2000). 'Hot* emotions in human recol- lection: Toward a model of traumatic memory. In F.. Tulving (Ed.). Memory. consciousness, and the brain: The Tallinn Conference (pp. 228-242). Philadelphia: Psychology' Pits. 40 Paunovic. N.. Lindh. & Oea. 1.: C. (2002). Attentional and memory bias for emotional information in crime victims with acute posttraumatie stress disorder. Journal of Anxiety Disonlent. 16. 675-692. Quas.J.A.. Goodman. C.S.. Blames\ S.. Pipe. Craw. S.. & Minn. D. (1999). Emotion and memory: Children's long-term remem- bering. forgetting, and suggestibility. Journal of Experimental Child Psychology; 72. 135-270. Sullivan. C.M., & Gorman. J.M. (2002). Finding a home for post- traumatic stress disorder in biological psychiatry: Is it a disorder of anxiety. mood, stress, or memory? Psychiaitic Clinics of North America. 25. 463-3/4 8. Tern LC. (1991). Childhood traumas: An outline and overview. American Journal of Psychiatry. 1411. 10-20. van der Kolk. B.A. (1997). The psychobiology of prottraumatie stress disorder. Journal of Clinical Psychiatry. 58. 16-24. Widow. C.S.. & Morris. S. (1997). Accuracy of adult recollections of childhood victimization. Part 2: Childhood sexual abuse. Psy- chological Assessment. 9.3416. Williams. L.M. (1994). Recall of childhood trauma: A prospective study of women's memories of child sexual abuse. Journal of Consulting mud Clinical Psychology. 62. 1167-1176. Williams, I-M. (1995). Recovered memories of abuse in women with documented child sexual victimization histories. Journal of hatunatic Stress. 8.649-6673. (RECEIVED 9/15103: ACCEPITh 10/24/03) Volume 116—Number I 3534-009 Page 8 of 8 EFTA_000 10260 EFTA00159908