Journal of Social Work Practice in the Addictions, 13:269-294, 2013 Copyright Taylor & Francis Group, LLC ISSN: 1533-256X print/1533-2578 online DOI: 10.1080/1533256X.2013.814483 Child Maltreatment and Adult Substance Abuse: The Role of Memory LAURA ELWYN, PHD Senior Research Associate/Analyst, Advocates for Human Potential, Kerhonkson, New York, USA CAROLYN SMITH, PHD Professor, School of Social Welfare, University at Albany, Albany, New York, USA Child maltreatment is a risk factor for substance abuse in adulthood. This study examines whether memory of maltreatment is a necessary link in the path leading from prospectively measured childhood maltreatment to adult substance use problems. Ofcial Child Protective Services reports and adult retrospective recall of childhood maltreatment were used to predict illegal drug use and alcohol problems in adulthood controlling for covariates. Memory was a necessary link in the path between prospective reports of maltreatment and alcohol problems, and it was an important link in the path between prospective reports and illegal drug use. Implications for prevention and treatment are discussed. KEYWORDS alcohol problems, child maltreatment, illegal drug use, memory, prospective measurement, retrospective measurement Maltreatment in childhood has long been associated with adult substance abuse. As many as two thirds of people receiving services for drug or alcohol abuse report histories of child maltreatment (Najavits, Weiss, & Shaw, 1997; Ouimette, Kimerling, Shaw, & Moos, 2000; Swan, 1998). However, many questions about the possible pathways between childhood maltreatment and subsequent substance abuse remain unanswered. One question is why some maltreated children grow up to abuse substances and others do not. Perhaps there are additional factors that tip the balance in one direction or the other. Received October 25, 2012; revised February 6, 2013; accepted March 15, 2013. Address correspondence to Laura Elwyn, Senior Research Associate/Analyst, Advocates for Human Potential, 29 Rock Haven Road, Kerhonkson, NY 12446, USA. E-mail: ljelwyn@ gmail.com 269 270 L. Elwyn and C. Smith Another question is based on differences in the association between childhood maltreatment and adult substance abuse when retrospective measures of maltreatment are used compared to prospective measures of maltreatment. These differences suggest the possibility that adults who have problems with substance use are more likely to remember maltreatment in childhood than adults without substance use problems. Using an urban community sample that provides longitudinal data from early adolescence through the early 30s, this study presents ndings about the role that memory of maltreatment plays in the relationship between prospectively measured maltreatment and adult substance use. According to the 2010 National Survey on Drug Use and Health (Substance Abuse and Mental Health Services Administration [SAMHSA], 2011), more than 22 million Americans annually abuse or are dependent on drugs or alcohol. The personal and societal cost of substance abuse is large. Economic costs to the United States were estimated at $223.5 billion for excessive alcohol consumption in 2006 (Bouchery, Harwood, Sacks, Simon, & Brewer, 2011) and $193 billion for drug abuse in 2007 (National Drug Intelligence Center, 2011). As reected in the series of objectives in Healthy People 2020 (U.S. Department of Health and Human Services, 2010) reduction of substance abuse and related problems is a major public health target. The literature on the etiology of alcohol and drug abuse, a review of which is beyond the scope of this study, provides evidence that many factors might contribute to the development of substance abuse. These factors include aspects of family environment such as parental substance use and abuse (Chassin, Curran, Hussong, & Colder, 1996; Stein, Leslie, & Nyamathi, 2002), parent monitoring and supervision, and parent-child attachment (Brook, Morojele, Pahl, & Brook, 2006; Fergusson, Boden, & Horwood, 2008). This study focuses on a particular aspect of family environment, child maltreatment, and its association with subsequent substance abuse. Child maltreatment, a common and serious individual and public health problem in American society, generally encompasses physical, sexual, and emotional abuse, as well as several different dimensions of neglect. National estimates of the extent of maltreatment are based on data from the National Child Abuse and Neglect Data System (NCANDS) and the fourth National Incidence of Child Abuse and Neglect study (NIS-4). In 2009, NCANDS recorded about 3.3 million referrals of alleged maltreatment to state Child Protective Services (CPS) agencies and identied 702,000 victims of 763,000 substantiated cases of maltreatment (U.S. Department of Health and Human Services ACF, 2011). In 2005-2006, NIS- 4, which relied on reports from a sample of professionals in a position to recognize maltreatment, estimated that approximately 1.25 million children experienced maltreatment using their harm standard (involving acts resulting in demonstrable harm), and slightly fewer than 3 million childrenor about Memory in Child Maltreatment and Adult Substance Abuse 271 1 in every 25 childrenwere maltreated using their endangerment standard, a less restrictive criteria that includes acts of endangerment as well as harm (Sedlak et al., 2010). As detailed later, many studies have found a positive association between maltreatment in childhood and later substance abuse. However, an indication that how maltreatment is remembered might be a factor in the development of substance abuse comes from differences in the association between maltreatment and substance abuse when maltreatment is measured prospectively rather than retrospectively. Prospective data on child maltreatment, measured approximate to the time the maltreatment occurs, is usually drawn from ofcial records such as CPS (e.g., Smith, Ireland, Thornberry, & Elwyn, 2008) or family courts (e.g., Widom & Shepard, 1996). One advantage of prospective data is that because they measure maltreatment earlier in time they can more readily be used to establish causal versus correlational associations with subsequent outcomes (Briere, 1992). Another advantage is that CPS and court records of maltreatment usually involve substantiated ndings by uninvolved parties (CPS workers, judges) of child abuse or neglect based on statutory criteria. Therefore the events involved are corroborated and meet reasonably clear and stringent standards of maltreatment. However, ofcial prospective data also have disadvantages. The most notable of these is that ofcial records capture only a portion of cases of child maltreatment (Hardt & Rutter, 2004; Sedlak & Broadhurst, 1996; Smith et al., 2008). Retrospective reports of adults recalling events of maltreatment that occurred in their childhood or adolescence is a more common measure of maltreatment and has been used in many outcome studies. Retrospective reports have been shown to capture events that were not documented prospectively in childhood; that is, maltreatment that never came to the attention of authorities (Della Femina, Yeager, & Lewis, 1990; Hardt & Rutter, 2004; Kendall-Tackett & Becker-Blease, 2004; Smith et al., 2008). However, retrospective reports are measuring the recall or memory of childhood maltreatment rather than the maltreatment itself. As these memories are generally measured at the same time as adult outcomes, they bear a correlational rather than potentially causal relationship to outcomes such as substance abuse (Briere, 1992). In addition, because they involve memory over long periods of time, they are subject to forgetting and other functions of memory and related cognitive processes. Hardt and Rutter (2004), in a review examining the validity of retrospective self-reports of childhood adversity, noted that on average across studies about one third of adults do not report maltreatment that was documented in their childhood or adolescence. Thus, although prospective ofcial measurement understates the prevalence of maltreatment, retrospective self-report measures also understate the prevalence of maltreatment although for different reasons. This raises the issue of the role of memory about the maltreatment experience. 272 L. Elwyn and C. Smith Retrospective studies examining the relationship between child maltreatment and substance abuse generally involve clinical samples of adults or adolescents in treatment for substance abuse, trauma, or other problems, as well as a range of community samples. Studies compare both whether persons with substance abuse disorders report a higher prevalence of child maltreatment than a comparison population without substance abuse disorders, as well as whether persons reporting maltreatment display a higher prevalence of substance use problems than comparison groups with no maltreatment. A wide range of studies have found associations between adult substance use problems and retrospectively reported histories of maltreatment (A, Brownridge, Cox, & Sareen, 2006; Bartholow et al., 1994; Bensley, Spieker, Van Eenwyk, & Schoder, 1999; Burnam et al., 1988; Duncan, Saunders, Kilpatrick, Hanson, & Resnick, 1996; Fergusson, Horwood, & Lynskey, 1996; Fergusson & Lynskey, 1997; Kunitz, Levy, McCloskey, & Gabriel, 1998; Langeland & Hartgers, 1998; MalinoskyRummell & Hansen, 1993; Moncrieff & Farmer, 1998; Polusny & Follette, 1995; Saunders, Kilpatrick, Hanson, Resnick, & Walker, 1999; Scott, 1992; Simpson & Miller, 2002; Spak, Spak, & Allebeck, 1998; Wilsnack, Vogeltanz, Klassen, & Harris, 1997). There have been fewer reports of null ndings (Dinwiddie et al., 2000; Langeland & Hartgers, 1998; Simpson & Miller, 2002). There are many fewer prospective studies that examine whether maltreated children grow up to have higher rates of substance abuse than nonmaltreated children, and results are more equivocal. Although some studies have found a positive association between childhood maltreatment and adult substance abuse (Jasinski, Williams, & Siegel, 2000; Schuck & Widom, 2001), other studies have found no association between prospective measures of maltreatment and adult substance use problems (McCord, 1983) or have found an association for women but not for men (Ireland & Widom, 1994; Widom, Ireland, & Glynn, 1995; Widom & White, 1997). Perhaps of most interest, albeit rare, are studies comparing prospective and retrospective measures using the same sample. Tajima, Herrenkohl, Huang, and Whitney (2004) found that although the retrospective self-reports of physical abuse by emerging adults strongly predicted alcohol and marijuana abuse, prospective reports by their mothers more weakly predicted marijuana abuse and did not predict alcohol abuse. Widom, Weiler, and Cottler (1999) found that although ofcial prospective measures of maltreatment were not associated with adult drug use problems, retrospective self-reports of childhood maltreatment in the same sample robustly predicted increased risk for drug use problems in adults of both genders. These ndings suggest a weaker association between prospective measures of maltreatment and adult substance use problems compared to retrospective measures of maltreatment and adult substance use problems. This raises the question of whether retrospectively reported memory of maltreatment mediates the relationship between childhood maltreatment and Memory in Child Maltreatment and Adult Substance Abuse 273 adult substance use problems. In other words, is a reported memory of maltreatment an important or even essential link in the chain leading from an experience of maltreatment in childhood to problem substance use as an adult? As noted earlier, a substantial number (on average a third) of people with prospectively measured ofcial records of maltreatment do not report a memory of maltreatment in adulthood. To help us understand why this might be so, we look briey at evidence about autobiographical memory. Much of what humans experience is either not remembered, forgotten rapidly, or forgotten more slowly over time (Neisser, 1986; Thompson, Skowronski, Larsen, & Betz, 1996), and memory for the \"gist\" or central meaning of a remembered event is the most enduring (Barclay, 1986; Neisser, 1986). In addition, what humans remember are not reproductions or literal copies of past events but reconstructions (Barclay, 1986; Conway, 1996; Neisser, 1986; Singer & Salovey, 1993; Thompson et al., 1996). Thus, autobiographical memory is not an objective recounting of events, but is subject to processes that include reconstruction and selection relevant to the self. A salient feature of autobiographical memory is its strong and essential relationship to the self; in fact, some theorists believe autobiographical memory to be an intrinsic part of the self (Conway & Pleydell-Pearce, 2000; Howe & Courage, 1997; Singer & Salovey, 1993). Barclay (1986) described identity formation as \"the development of a sense of sameness or consistency about one's self\" (p. 86). This generalized schematic concept of the self is built from experience and social interaction in the world and in turn organizes current experience and is the basis of future expectations. Markus (1977) showed that self-schemata selectively inuences how new experience is processed and that this selection supports existing theories of self. Subsequent research has supported and extended these ndings; for example, by mapping the development of self-identity from early childhood through emerging adulthood (Bakermans-Kranenburg & van IJzendoorn, 1993; Barclay, 1986; Habermas & Bluck, 2000; McAdams, 2001; Mikulincer, 1998). This evidence supports the idea that events of maltreatment in childhood might be incorporated into autobiographical memory and self-identity differently by different people. Thus it could explain, at least in part, why maltreatment that was ofcially documented in childhood is not always reported in adulthood. Other factors have been shown to be associated with child maltreatment or substance abuse and might exert a spurious effect on these relationships. Parental substance use has been linked both to childhood maltreatment (Famularo, Kinscherff, & Fenton, 1992; Kirisci, Dunn, Mezzich, & Tarter, 2001; Zielinski & Bradshaw, 2006) and to substance use problems in offspring (Chassin et al., 1996; Stein et al., 2002). The relationship between child and parent has also been shown to moderate the association between maltreatment and subsequent substance use problems (Bolen & Lamb, 2002; Chandy, Blum, & Resnick, 1997; Leifer, Kilbane, & Grossman, 2001). Thus 274 L. Elwyn and C. Smith these variables need to be accounted for in examining the relationships in our hypothesized model described next. CONCEPTUAL FRAMEWORK As discussed in the literature review, there is a body of evidence that links maltreatment in childhood or adolescence to an increased risk for substance abuse problems in adulthood; however, this link appears to be stronger for retrospective self-report as compared with prospective ofcial report, suggesting that processes of memory play a role in the pathways between maltreatment and adult substance use problems. Consistent with theories of the development of autobiographical memory, we propose that the internal schema of the self, other people, and the world in general developed throughout childhood and adolescence is the lens through which events of maltreatment are interpreted, incorporatedor notas part of a selfidentity, and remembered. These theories postulate that from early childhood each person develops a set of internal constructs of the self, others, and the world that provide the basis for cognitive processes involved in interpersonal relationships, emotional well-being, and a variety of behaviors including substance use. This internal map will be inuenced by sexual abuse, physical abuse, or neglect by caregivers. However, these internal constructs in turn will inuence how events such as maltreatment are interpreted and remembered. Our theorized model results in three specic hypotheses. Hypothesis 1: Prospectively measured maltreatment in childhood or adolescence will have a positive association with substance use problems in adulthood. Hypothesis 2: Retrospectively reported memory of maltreatment in early adulthood will mediate the association between prospectively measured child or adolescent maltreatment and adult substance use problems. Hypothesis 3: Prospectively measured maltreatment will have no direct association with adult substance use problems when retrospectively reported memory of maltreatment is included in the model (i.e., memory of maltreatment will fully mediate the association between prospectively measured maltreatment and adult substance use problems). METHODS Sample Data for this research come from the Rochester Youth Development Study (RYDS), a longitudinal study designed to increase understanding of the development of antisocial behavior and drug and alcohol use in an 275 Memory in Child Maltreatment and Adult Substance Abuse TABLE 1 Measures Time Period Adulthood Early adulthood Birth through late adolescence Late adolescence Adolescence Categorical Measure Illegal drug use Any alcohol problem Any memory of maltreatment Any substantiated maltreatment Illegal drug use Any alcohol problem Parental substance use Male African American Hispanic White Percentage (n) 33.62% 12.70% 28.84% 19.80% 26.97% 10.22% 29.95% 72.90% 68.00% 17.00% 15.00% (270) (102) (244) (198) (240) (91) (233) (729) (680) (170) (150) Continuous Measure 3.4 (0.44) 1.18-4.00 803 846 1,000 890 778 1,000 M (SD) Range Adolescent-parent relationship Total n 947 urban community sample. During the 1987-1988 academic school year, 1,000 seventh- and eighth-grade students were randomly drawn from the Rochester, NY, public school district. To overrepresent youth at high risk for antisocial behavior, boys and students from census tracts with high crime rates were oversampled. Table 1 presents characteristics of the RYDS participants. Due to the oversampling 73% of participants were male. Participants were drawn from an urban, high-minority school district; 68% of participants were African American, 17% were Hispanic, and 15% were White. There are three phases of the RYDS. During Phase 1, separate interviews with each student and his or her primary caregiver were held every 6 months over the course of 4.5 years (M age = 13-18). After a gap of 2 years, Phase 2 involved annual interviews held for 3 years with each participant, now in his or her early 20s. Following a longer gap, Phase 3 involved two interviews a few years apart held with participants now in their late 20s to early 30s. Data from all three phases are used in this study. In addition to interviews, data were collected from ofcial sources, including the Department of Social Services of the local county, which provided substantiated reports of maltreatment on RYDS participants from birth through age 17. All human subject research protections have been observed throughout the RYDS, which is monitored by the Institutional Review Board of the University at Albany. Retention of RYDS participants has been good: At the end of Phase 2, 85% of participants and 83% of primary caregivers participated, and at the end of Phase 3, 76% of participants were retained. Earlier analyses found no evidence of differential attrition for delinquency measures including drug use and only minor differences have been found on demographic measures between participants remaining and participants lost to 276 L. Elwyn and C. Smith the study (Krohn & Thornberry, 1999; Thornberry, Krohn, Lizotte, Smith, & Tobin, 2003). Analyses for this study found no substantive differential attrition related to substance use problems, maltreatment, or covariates (details available from author). Measures The two outcomes measures used in this study are any illegal drug use and any alcohol-related problems from both of the adult (Phase 3) interviews. For the measure of illegal drug use, all respondents were asked whether they had used an index of illegal drugs during the past year, including marijuana, crack cocaine, cocaine other than crack, heroin, PCP, barbiturates, amphetamines, inhalants, hallucinogens, and other nonprescription drugs such as ecstasy or steroids. The measure used in this study is any use of illegal drugs at either of the adult interviews, where no use is coded as 0 and any use is coded as 1. As shown in Table 1, 34% of RYDS participants reported illegal drug use in adulthood. Respondents who reported at least monthly use of any alcoholic beverage were asked a series of 11 questions about problem alcohol use, for example, whether participants had \"tried to cut down on drinking and found that you could not.\" Questions included problems with personal relationships, functioning at work or school, legal problems, health problems, and increased tolerance or an inability to stop or reduce alcohol use. These problem use indicators are based on common alcohol and drug screening instruments. The measure used in this study is an endorsement of one or more problems experienced in the past year at either adult interview, with no reported alcohol-use-related problems coded as 0 and one or more alcohol-use-related problems coded as 1. As shown in Table 1, 13% of the RYDS participants reported one or more alcohol-related problems in adulthood. MEMORY OF CHILD MALTREATMENT Retrospective self-reports on childhood maltreatment were collected in early adulthood (Phase 2) when participants had a mean age of 22.7 years. Participants were asked a series of questions about specic events of neglect, physical abuse, and sexual abuse that they experienced before the age of 12. These questions were adapted from the work of Widom (e.g., Widom et al., 1999) and others (Finkelhor, 1979). For neglect, respondents were asked six questions about specic types of neglect, such as \"Were there circumstances where your parents or caretaker left you home alone when you didn't feel old enough to take care of yourself or keep yourself safe?\" For physical abuse, participants were asked eight questions, such as \"Did parents/caretaker hit you with something hard like a stick or baseball bat?\" Memory in Child Maltreatment and Adult Substance Abuse 277 For sexual abuse, 11 questions were asked, such as \"Did another person touch your sex organs?\" In each category, respondents were also asked whether they remembered any other maltreatment and then were asked to describe it. For this study, memory of maltreatment was coded as 1 if participants endorsed any memory of maltreatment and 0 otherwise. As presented in Table 1, 29% of RYDS participants had a memory of one or more events of maltreatment in childhood. OFFICIAL PROSPECTIVE CHILDHOOD MALTREATMENT Substantiated child maltreatment data were collected from the CPS of Monroe County, the county of residence for all RYDS participants. For each participant a search was conducted to nd any substantiated reports of maltreatment between birth and 1992, when mean age of participants was 17. Details on every incident were coded according to the classication system developed by Cicchetti (Barnett, Manly, & Cicchetti, 1993; Smith & Thornberry, 1995) for which there is ample evidence of strong reliability and validity (Bolger, Patterson, & Kupersmidt, 1998; Cicchetti & Barnett, 1991). Interrater reliability among coders was 0.8. An extensive description of the subtypes and dimensions of ofcial substantiated maltreatment (and the selfreported maltreatment described earlier) for the RYDS sample are detailed in a separate analysis (Elwyn, 2011). In part because the different subtypes are often cooccurring (41% of those maltreated had reports of multiple subtypes) in the following analyses one or more substantiated reports of maltreatment is coded as 1 and no maltreatment is coded as 0. As shown in Table 1, 20% of the RYDS sample had at least one substantiated report of maltreatment. COVARIATES To control for potentially confounding effects several other measures are included in the multivariate analyses, including adolescent substance use problems, parental substance use, and adolescent relationship with parent. (In earlier models, we also included a measure of family poverty, but as this was not associated with any outcome or mediator and did not contribute to any model, for reasons of parsimony we exclude it from this nal presentation.) Table 1 presents the prevalence or sample means of these covariates. Late adolescent illegal drug use is the same measure as in adulthood for a yearlong period when the average age of participants was 17.7. Similarly, late adolescent alcohol problems are a measure of one or more alcohol-related problems at age 17.7. As shown in Table 1, the prevalence of illegal drug use at this age was 27% and the prevalence of any alcohol-related problem was 10%. Parental substance use is derived from the information collected from primary caregivers and is coded as 1 if the primary caregiver endorsed any 278 L. Elwyn and C. Smith one of the following over the course of Phase 1: often had three or more alcoholic drinks, used marijuana, or used other illegal drugs. It is coded as 0 if the primary caregiver did not report substance use at any interview. It should be noted that because the majority of primary caregivers were womeneither mothers or mother surrogates such as foster mothers or grandmothersthis is primarily a measure of maternal substance use rather than paternal substance use. As Table 1 presents, the prevalence of parental substance use was 30%. A measure of adolescent-parent relationship is derived from an 11-item index adapted from Hudson's Index of Child's Attitude Toward Mother Scale. The scale measures agreement on a 4-point response scale with items on the affective dimension of family life, including parent-adolescent warmth, liking, lack of hostility, and sense of parental approval. The scale can range from 1 (little attachment with the primary caregiver) to 4 (very strong attachment to the primary caregiver). The measure used in this study is from midadolescence when the mean age of participants was 14.5. The reliability for this measure is 0.87. As shown in Table 1, the mean score on adolescent-parent relationship was 3.4 (SD = 0.44) with a range of 1.18 to 4. In general, core RYDS measures, including outcomes such as illegal drug use, predictors such as self-reported maltreatment, and covariates such as adolescent-parent relationship, have been developed over the span of the project and their validity reported in many publications (e.g., Krohn, Stern, Thornberry, & Jang, 1992; Thornberry & Krohn, 2000, 2003; Thornberry et al., 2003). RESULTS The major question in this study is whether memory of maltreatment is a mediator in the association between ofcial maltreatment and substance use problems in adulthood. The concept of mediation is that the association of one variable with another variable is transmitted, either partly or completely, through a third variable; this is also referred to as an indirect effect. A useful method for assessing mediation is path analysis, which can assess the magnitude, direction, and statistical signicance of path coefcients for direct, indirect, and total effects, with indirect effects tested by the joint signicance of the involved paths (MacKinnon, 2008; MacKinnon, Lockwood, Hoffman, West, & Shets, 2002). We therefore used path analysis for multivariate analyses. Data Analyses Bivariate analyses were conducted using SAS Version 9.2. Path analysis was conducted using Mplus Version 4.21. Due to the need to model categorical Memory in Child Maltreatment and Adult Substance Abuse 279 mediators in the path analyses, probit regression models were conducted using the weighted least squares mean and variance adjusted estimator developed by Muthn (Kaplan, 2009; Muthn & Muthn, 2007). Probit regression uses the cumulative distribution function of the standard normal distribution to link the probability of a categorical outcome to a linear combination of predictors. The sign and statistical signicance of estimated probit coefcients are interpreted similarly to ordinary least squares regression. For example, a negative, statistically signicant coefcient would indicate that an increase in the predictor would result in a decrease in the probability of the outcome, controlling for other variables in the equation (Liao, 1994; Long, 1997). Chi-square difference testing using the option developed in Mplus for mean and variance-adjusted estimators (Asparouhov & Muthn, 2006) was used to examine the effect of constraining (i.e., removing) specic paths. Three global t indexes were used to assess overall t of the model to the data, Good t is indicated by a model chi-square that is not statistically signicant, a root mean square error of approximation (RMSEA) that is less than 0.05, and a comparative t index (CFI) that is greater than 0.90 (Kaplan, 2009). Modication indexes were used to assess local t of the model to the data. Although as described earlier we did not nd evidence of differential attrition on the measures used in this study, some data were missing due to attrition or nonresponse on particular items. To address missing data, we used a full information approach for weighted least squares estimation developed by Asparouhov and Muthn (2010) that maximizes available information and increases the efciency of estimates. Bivariate Associations Table 2 presents the simple bivariate associations between the measures. For categorical measures, a chi-square test was used to assess differences in prevalence of the outcome; for the one continuous measure (adolescent- parent relationship) a t test was used to assess mean differences between groups dened by the other measures. As shown in the rst two rows of Table 2, a memory of maltreatment in early adulthood increases the likelihood of both illegal drug use and alcohol problems in adulthood at a statistically signicant level. As shown in the third and fourth rows of Table 2, any substantiated CPS report in childhood or adolescence was weakly associated with illegal drug use in adulthood and not associated with any alcohol problem in adulthood. As would be expected, ofcial maltreatment was associated with a memory of maltreatment. Note, however, that less than half (44%) of those with an ofcial substantiated nding of maltreatment reported a memory of maltreatment in early adulthood. 280 L. Elwyn and C. Smith TABLE 2 Bivariate Associations Between Adult Substance Use Problems, Maltreatment, and Covariates % Adult Illegal Drug Use % Memory of Maltreatment 20.09 9.89 39.74+ 32.21+ Memory of maltreatment Yes No Substantiated maltreatment Yes No Late adolescent drug use Yes No Late adolescent alcohol problems Yes No Parental substance use Yes No Gender Male Female Race/ethnicity African American Hispanic White % Adult Alcohol Problems 46.88 28.27 Categorical Measures ( 2 Test) 15.23 12.12 56.82 24.83 43.98 25.15 35.81 26.39 40.48 9.50 34.48 28.32 41.23 30.23 17.06 10.56 31.19 27.01 38.81 20.78 15.73 5.19 28.26 30.29 33.58 31.78 35.77 10.89 13.95 19.51 29.17 24.29 32.31 Continuous Measures (t Test) Adolescent-Parent Relationship Adult Drug Use Yes Adult Alcohol Problem No Yes Memory of Maltreatment No Yes No M SD M SD M SD M SD M SD M SD 3.30 .48 3.43 .42 3.30 .46 3.40 .44 3.26 .50 3.44 .41 + p < .1. p < .05. p < .01. p < .001. Memory of Maltreatment as a Proxy In a set of analyses with details published separately (Elwyn, 2011), we also evaluated whether memory of maltreatment was simply a proxy for some dimension of ofcial maltreatment, for example, the severity of abuse. The Cicchetti classication system codes for maltreatment subtype and severity and provides the basis for coding other dimensions such as duration and number of incidents of abuse. These analyses, which extensively describe these dimensions, showed no relationship between the number of subtypes, number of incidents, duration, or severity of ofcial substantiated maltreatment with memory of maltreatment. For example, participants who Memory in Child Maltreatment and Adult Substance Abuse 281 reported a memory of maltreatment in early adulthood experienced on average the same level of ofcial maltreatment severity as participants who did not report a memory of maltreatment in early adulthood. These analyses conrmed that reported memories of maltreatment are not simply markers for the subtypes, extent, or severity of substantiated maltreatment. Table 2 presents the bivariate associations between the covariates and the adult outcomes as well as with memory of maltreatment. As would be expected, illegal drug use in late adolescence increased the likelihood of illegal drug use in adulthood at a statistically signicant level. Any alcohol problem reported in late adolescence was a strong predictor of any alcohol problem reported in adulthood. Late adolescent drug use was also associated with memory of maltreatment, but any late adolescent alcohol problem was not. Parental substance use was associated with an increased likelihood of illegal drug use and any alcohol problem in adulthood. The lower part of Table 2 presents the results of t tests on the adolescent-parent relationship scale for groups dened by adult substance use problems, and memory of maltreatment. The respondents who reported illegal drug use or any alcohol problem in adulthood had, on average, a more negative perception of their relationship with their parent in adolescence. Similarly, respondents who reported a memory of maltreatment in early adulthood had a more negative perception of their relationship with their parent as adolescents than those who did not report a memory of maltreatment. Perhaps surprisingly, there was no statistically signicant difference in adolescent-parent relationship between the group with substantiated maltreatment reports compared to those without such reports, nor were there differences for the group whose primary caregiver reported substance use compared to those who did not (results not shown). Male gender was associated with an increased likelihood of illegal drug use and alcohol problems in adulthood, but not with memory of maltreatment. Finally, White participants were more likely to report an alcohol problem in adulthood. Path Analyses To look at these relationships in a multivariate context and to assess the mediated (indirect) effects we turn now to path analyses. Figure 1 presents the probit regression path model for the outcome of adult illegal drug use. The t indexes for this model were good: The model chi-square statistic was not statistically signicant (p = 0.9), the CFI was 1.0, and the RMSEA was 0.0, all indicating a good global t with the data. In addition, there were no modication indexes above 3.84, indicating good local t of the model with the data. We rst address Hypothesis 2 (H2) as to whether a memory of child maltreatment in early adulthood mediated the association between ofcial substantiated maltreatment and adult drug use; we found that it did. Figure 1 282 L. Elwyn and C. Smith Official Substantiated Maltreatment (Birth Through 17) 0.48*** Memory of Maltreatment 0.25* 0.26*** -0.54*** Adolescent-Parent Relationship -0.49*** Any Illegal Drug Use 0.46*** Any Illegal Drug Use 0.35*** Parental Substance Use 0.62*** Male ns Hispanic ns White Adolescence Late Adolescence Early Adulthood Adulthood FIGURE 1 Path model of illegal drug use. Notes. N = 864. All exogenous variables are allowed to correlate. 2 (10) = 5.43, p = .86, comparative t index = 1.00, root mean squared error of approximation = 0.00. p < .05. p < .01. p < .001. shows the two portions of the path between ofcial maltreatment and adult drug use via memory of maltreatment. Consistent with bivariate results, respondents with a history of substantiated maltreatment reports were more likely to have a memory of maltreatment in early adulthood (probit coefcient [pc] = 0.48, p < .001), and respondents with a memory of maltreatment were more likely to report use of illegal drugs in adulthood (pc = 0.26, p < .001). The product of these two paths, the indirect effect, is presented in Table 3 (pc = 0.13, p < .01) and was also statistically signicant. However, a history of illegal drug use in late adolescence also mediated the association between ofcial substantiated maltreatment and adult drug use. Ofcial maltreatment increased the likelihood that respondents engaged in illegal drug use in late adolescence (pc = 0.25, p < .05), and illegal drug use in late adolescence increased the likelihood of illegal drug use in adulthood (pc = 0.46, p < .001). The indirect effect of ofcial maltreatment on adult drug use via late adolescent drug use was 0.11 (p < .05). Also presented in Table 3 is the total effect of substantiated maltreatment on adult drug use, which is the sum of the two indirect effects. This was also statistically signicant (pc = 0.24, p < .001). This nding addressed H1 and indicated an overall association between ofcial substantiated maltreatment and adult drug use. Notably, however, we did not nd a direct effect of ofcial maltreatment on adult drug use (H3). In earlier models (not shown), we included a direct path from ofcial maltreatment to adult drug use; the coefcient for this path was 0.002 (p = .99). We also conducted a formal Memory in Child Maltreatment and Adult Substance Abuse 283 TABLE 3 Direct, Indirect, and Total Effects on Outcomes From Probit Regression Path Models Probit Coefcients (95% CI) Mediators Effect Type Memory of maltreatment Total/direct Late adolescent drug use Total/direct Late adolescent alcohol problems Total/direct Exogenous Ofcial maltreatment Indirect (via memory of maltreatment) Indirect (via late adolescent drug use or alcohol problems) Total Parental substance use Adolescent-parent relationship Male Total/indirect (via late adolescent drug use or alcohol problems) Indirect (via memory of maltreatment) Indirect (via late adolescent drug use or alcohol problems) Total Direct Indirect (via late adolescent drug use or alcohol problems) Total Hispanic Total/indirect (via late adolescent drug use or alcohol problems) Total White p < .05. p < .01. Adult Illegal Drug Use Adult Alcohol Problems 0.26 [0.15, 0.38] 0.46 [0.36, 0.56] 0.27 [0.13, 0.41] 0.13 [0.05, 0.21] 0.11 [0.01, 0.21] 0.24 [0.11, 0.37] 0.16 [0.06, 0.26] 0.14 [0.22, 0.06] 0.22 [0.33, 0.12] 0.37 [0.49, 0.24] 0.62 [0.40, 0.84] 0.13 [0.04, 0.22] 0.62 [0.40, 0.84] 0.01 [0.11, 0.13] 0.04 [0.16, 0.09] 0.53 [0.39, 0.66] 0.13 [0.04, 0.22] 0.25 [0.10, 0.39] 0.14 [0.23, 0.05] 0.16 [0.30, 0.02] 0.30 [0.46, 0.14] 0.46 [0.09, 0.83] 0.23 [0.03, 0.42] 0.68 [0.34, 1.03] 0.10 [0.08, 0.27] 0.19 [0.02, 0.35] p < .001. test to determine whether constraining this path to zero had any impact on the model; the result of this chi-square difference test was 2 (1) = 0.00, p = .99, conrming that there was no direct effect of ofcial substantiated maltreatment on illegal drug use in adulthood. These ndings indicate that participants with a history of ofcial substantiated CPS reports in childhood or adolescence have an increased likelihood of engaging in illegal drug use in adulthood. However, this is the case only because these participants have an increased likelihood of reporting a memory of maltreatment or an increased likelihood of engaging in late adolescent drug use. It is worth noting here that less than half of those respondents with a substantiated CPS report of maltreatment reported a memory of 284 L. Elwyn and C. Smith maltreatment in early adulthood. Those respondents who had a substantiated CPS report but who did not report a memory of maltreatment or engage in late adolescent drug use had no more likelihood of engaging in adult illegal drug use than those respondents with no history of ofcial maltreatment. The implications of this for intervention and treatment are discussed further in the last section. We briey turn our attention to the role of other factors in the path analysis model of illegal drug use. As shown in Figure 1, neither adolescent perception of relationship with parent nor parental substance use directly predicted illegal drug use in adulthood. Earlier models showed that the direct paths were not statistically signicant and chi-square difference tests for both showed that constraining these paths to zero was not detrimental to the model. Both predictors, however, did have a statistically signicant association with adult drug use through indirect paths. Table 3 presents the probit coefcients and 95% condence intervals for the indirect and total effects. Parental substance use increased the likelihood of late adolescent drug use (pc = 0.35, p < .001). The total effect on adult drug use via late adolescent drug use was also statistically signicant (pc = 0.16, p < .01). A more negative adolescent-parent relationship increased the likelihood of both a memory of maltreatment (pc = -0.54, p < .001) and of late adolescent drug use (pc = -0.49, p < .001). Both indirect paths were statistically signicant as was the total effect of parent-adolescent relationship on adult drug use (pc = -0.37, p < .001). Consistent with bivariate results, adult males were more likely to use illegal drugs than adult females, but this difference was not true in late adolescence when the likelihood of drug use for each gender was similar. In this path model we found no statistically signicant differences in late adolescent or adult drug use for groups dened by race and ethnicity. Adult Alcohol Problems Figure 2 presents the t indexes for the model of adult alcohol problems: The model chi-square statistic was not signicant (p = .95), the CFI was 1.0, and the RMSEA was 0.0, all indicating good global model t. There were no modication indexes greater than 3.84, suggesting good local t of the model to the data. As shown in Figure 2, the probit coefcient for the path between ofcial maltreatment and memory of maltreatment was 0.49 (p < .001) and the probit coefcient for the path between memory of maltreatment and any alcohol problem in adulthood was 0.27 (p < .001). As presented in the second column of Table 3, the probit coefcient for the indirect effect was 0.13 (p < .01). These ndings conrm H2, that memory of maltreatment mediated the association between ofcial substantiated maltreatment and any alcohol problems in adulthood. 285 Memory in Child Maltreatment and Adult Substance Abuse Official Substantiated Maltreatment (Birth Through 17) 0.49*** Memory of Maltreatment 0.27*** -0.52*** Adolescent-Parent Relationship -0.30* Any Alcohol Problems 0.53*** Any Alcohol Problems 0.47*** Parental Substance Use 0.43* 0.46* Male ns Hispanic 0.35* White Adolescence Late Adolescence Early Adulthood Adulthood FIGURE 2 Path model of alcohol problems. Notes. N = 864. All exogenous variables are allowed to correlate. 2 (10) = 3.83, p = .95, comparative t index = 1.00, root mean squared error of approximation = 0.00. p < .05. p < .01. p < .001. However (unlike the model for illegal drug use), we did not nd a path between ofcial substantiated maltreatment and late adolescent alcohol problems. We included this path in earlier models and found a probit coefcient of 0.20 that was not statistically signicant (p = .18). A chi-square difference test, 2 (1) = 1.82, p = .18, indicated that constraining this path to zero had no detrimental effect on the model. Thus the only mediator between ofcial maltreatment and adult alcohol problems was memory of maltreatment. We also evaluated the direct path between ofcial maltreatment and adult alcohol problems in earlier models and found a probit coefcient of 0.06 (p = .70). A chi-square difference test for constraining this path to zero indicated no effect on the model, 2 (1) = 0.15, p = .69. This result conrmed H3, that there was no direct path between ofcial maltreatment and adult alcohol problems. Therefore the indirect effect via memory of maltreatment was also the total effect (pc = 0.13, p < .01). This nding conrmed H1 that there was a positive association between ofcial substantiated maltreatment and any alcohol problems in adulthood. However, this association was completely mediated by memory of maltreatment. As in the model of illegal drug use, parental substance use increased the likelihood of adult alcohol problems through the path mediated by late adolescent alcohol problems. The probit coefcient for this mediated total effect was 0.25 (p < .001). A more negative adolescent-parent relationship 286 L. Elwyn and C. Smith increased the likelihood of both memory of maltreatment and late adolescent alcohol problems, both of which in turn increased the likelihood of adult alcohol problems. The total effect of adolescent-parent relationship on adult alcohol problems had a probit coefcient of -0.3 (p < .001). Consistent with bivariate results, male gender increased the likelihood of adult alcohol problems both directly and indirectly through late adolescent alcohol problems. White participants were also more likely to report late adolescent alcohol problems resulting indirectly in an increased likelihood of adult alcohol problems. In summary, our ndings conrmed all three hypotheses for the outcome of any adult alcohol problem. Hypotheses 1 and 2 were conrmed for the outcome of illegal drug use; however, the association between prospective substantiated maltreatment and adult illegal drug use was mediated not only by memory of maltreatment, but also by late adolescent drug use. Ofcial substantiated maltreatment had an association with both adult substance use problem outcomes. However, for adult illegal drug use this association was fully mediated by illegal drug use in late adolescence and a memory of maltreatment in early adulthood. Those respondents who had an ofcial maltreatment report but did not use illegal drugs in adolescence and did not report a memory of maltreatment in early adulthood were no more likely to use illegal drugs in adulthood than those respondents with no ofcial maltreatment report. The association between ofcial substantiated maltreatment and adult alcohol problems was fully mediated by memory of maltreatment alone. Respondents who had a substantiated nding of maltreatment by CPS as a child or adolescent only had an increased risk of alcohol use problems in adulthood if they also reported a memory of maltreatment in early adulthood. DISCUSSION The idea that the subjective construction of maltreatment as part of a life story and self-identity is what links these experiences to substance use problems in adulthood is consistent with theory about autobiographical memory (Conway & Pleydell-Pearce, 2000), which proposes that experiences are interpreted and selectively incorporated into an internal self-schema that is inuenced by relationships and experience, but also provides the framework for assigning meaning and selective importance to experience. For example, an individual whose internal self-schema was adapted to a harsh, punitive early environment might continue to conceptualize the self, others, and the world in negative ways that result in problems such as substance abuse. The ndings from this study are consistent with this conceptualization. Maltreatment in childhood or adolescence might be measured more or less objectively by substantiated CPS reports where the events of abuse or Memory in Child Maltreatment and Adult Substance Abuse 287 neglect must meet standard ofcial criteria. However, the ofcial criteria for abuse or neglect do not determine how these events were subjectively experienced and interpreted by the child or adolescent or whether and to what extent they were incorporated into the internal schema. This interpretation and incorporation of experience is also not based simply on the events of maltreatment themselves, but also on the child or adolescent's internal schema at the time. The conception of self and others that dictates how events are interpreted and incorporated has itself developed from a variety of prior experiences and relationships, particularly primary caregiver relationships. Thus events of maltreatment that are seen as the same when viewed through the lens of ofcial criteria could be interpreted and remembered as maltreatment by one child or adolescent but not by another. In adulthood, whether events are remembered and reported as maltreatment is thus based on how the experiences were initially interpreted and incorporated into the internal schema, as well as any subsequent revision of this internal schema in relation to the experience of maltreatment. It is this subjective memory of maltreatment, rather than the objective events based on ofcial criteria themselves, that increases the likelihood of substance use problems. Existing recommendations for treatment of persons with substance use problems are consistent with this understanding. For example, SAMHSA's (2008) Treatment Improvement Protocol on Substance Abuse Treatment for Persons with Child Abuse and Neglect Issues states: Clinicians must, therefore, understand how clients interpret their experiences. Not all abuse meets the legal or commonly held criteria for abuse; nor will all clients perceive as abusive those experiences which t the legal denition. For example, a client might report being spanked every day as a small child and might feel that he deserved the spankings because he disobeyed his mother. He might also explain that his mother loved him and that the spankings occurred within a context of caring. Such a client would deny that he had been abused as a child and would not be well served by therapists who insisted otherwise. In contrast, another client may have accepted chronic belittling and criticism while growing up and may not understand its relationship to career failures and repeated relapses. (pp. 13-14) Thus, the primary nding of this study that memory mediates the association between maltreatment and substance use problems does not engender new or revised treatment recommendations. Instead this nding provides additional empirical support for a clinical approach based on an individual's internal conceptualization of maltreatment and its impact. The nding that a more positive early adolescent relationship with the primary caregiver decreases the likelihood of memories of maltreatment and associated substance use problems in adulthood, even controlling for a history of substantiated maltreatment, has implications for prevention efforts, 288 L. Elwyn and C. Smith particularly with maltreated adolescents and their families. Interventions that are focused on strengthening family relationships between adolescents and primary caregivers might promote the development of positive self-identity during this important developmental phase, which in turn could decrease the likelihood of substance use problems later on. The results of this study thus support efforts to promote positive relationships between adolescents with a history of maltreatment and their family members. STUDY LIMITATIONS This study was affected by some limitations. In part because the study involved analysis of longitudinal secondary data initially collected over two decades ago, the choice of measures, including of substance use problem outcomes, was constrained. In addition, although we controlled for the effects of gender, race, and ethnicity in this study, there was a lack of power to examine differences between these groups in the pathways or degree of association that we examined. The characteristics of the RYDS sample, which was overselected for male participants and is from an innercity, high-minority school district, should be considered when applying these ndings. Future research in this area would further illuminate the applicability of these relationships for other groups. This study also did not specically examine differences in subtypes or other dimensions of maltreatment; further research that examined differences in the effects of maltreatment subtypes such as sexual abuse or neglect would also contribute to our understanding of the pathways between maltreatment and adult substance use problems. Examination of the relationships of ofcial maltreatment and memory of maltreatment on other adult outcomes such as depression would also increase our understanding. Finally, there might be other mediators in the pathways between childhood maltreatment and adult substance abusedepression, for examplethat will be important to identify. In general, the ndings of this study should be replicated with different samples. It should be noted, however, that the RYDS has a number of features that were invaluable in addressing the hypotheses of this study. These include the collection of data on both prospective substantiated CPS reports and retrospective self-reports on events of maltreatment for the full sample. The longitudinal nature of the RYDS was essential to the ability to relate constructs from early and late adolescence to constructs from early adulthood and adulthood. In addition, the long-term span of the RYDS appears to have resulted in a high level of trust by participants in the study and thus their willingness to accurately report on sensitive topics such as maltreatment and substance use. These strengths suggest that the ndings of this study can be looked on with some condence. Memory in Child Maltreatment and Adult Substance Abuse 289 CONCLUSION The primary nding of this study is that memory of maltreatment mediates the association between prospectively measured ofcial maltreatment and adult substance use problems. This result supports the idea that how maltreatment is interpreted and incorporated into the subjective conception of self and others is an important consideration in understanding its effect on adult substance use problems. Hopefully, this contribution improves our understanding of the link between child maltreatment and adult substance use problems, and as a result can help to inform more effective interventions. REFERENCES A, T. O., Brownridge, D. A., Cox, B. J., & Sareen, J. (2006). Physical punishment, childhood abuse and psychiatric disorders. Child Abuse and Neglect, 30, 1093-1103. Asparouhov, T., & Muthn, B. O. (2006). Robust chi square difference testing with mean and variance adjusted test statistics (Technical report). Retrieved from http:\\\\www.statmodel.com Asparouhov, T., & Muthn, B. O. (2010). Weighted least squares estimation with missing data (Technical report). Retrieved from http:\\\\www.statmodel.com Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H. (1993). A psychometric study of the adult attachment interview: Reliability and discriminant validity. Developmental Psychology, 29, 870-879. Barclay, C. R. (1986). Schematization of autobiographical memory. In D. C. Rubin (Ed.), Autobiographical memory (pp. 82-99). Cambridge, UK: Cambridge University Press. Barnett, D., Manly, J. T., & Cicchetti, D. (1993). Dening child maltreatment: The interface between policy and research. In D. Cicchetti & S. L. Toth (Eds.), Child abuse, child development, and social policy (pp. 7-74). Norwood, NJ: Ablex. Bartholow, B. N., Doll, L. S., Joy, D., Douglas, J. M., Jr., Bolan, G., Harrison, J. S., . . . McKirnan, D. (1994). Emotional, behavioral, and HIV risks associated with sexual abuse among adult homosexual and bisexual men. Child Abuse and Neglect, 18, 747-761. Bensley, L. S., Spieker, S. J., Van Eenwyk, J., & Schoder, J. (1999). Self-reported abuse history and adolescent problem behaviors. II. Alcohol and drug use. Journal of Adolescent Health, 24, 173-180. Bolen, R. M., & Lamb, J. L. (2002). Guardian support of sexually abused children: A study of its predictors. Child Maltreatment, 7, 265-276. Bolger, C. E., Patterson, C. J., & Kupersmidt, J. B. (1998). Peer relationships and self-esteem among children who have been maltreated. Child Development, 69, 1171-1197. Bouchery, E. E., Harwood, H. J., Sacks, J. J., Simon, C. J., & Brewer, R. D. (2011). Economic costs of excessive alcohol consumption in the U.S., 2006. American Journal of Preventive Medicine, 41, 516-524. 290 L. Elwyn and C. Smith Briere, J. (1992). Methodological issues in the study of sexual abuse effects. Journal of Consulting and Clinical Psychology, 60, 196-203. Brook, J. S., Morojele, N. K., Pahl, K., & Brook, D. W. (2006). Predictors of drug use among South African adolescents. Journal of Adolescent Health, 38, 26-34. Burnam, M. A., Stein, J. A., Golding, J. M., Siegel, J. M., Sorenson, S. B., Forsythe, A. B., & Telles, C. A. (1988). Sexual assault and mental disorders in a community population. Journal of Consulting and Clinical Psychology, 56, 843-850. Chandy, J. M., Blum, R. W., & Resnick, M. D. (1997). Sexually abused male adolescents: How vulnerable are they? Journal of Child Sexual Abuse, 6(2), 1-16. Chassin, L., Curran, P. J., Hussong, A. M., & Colder, C. R. (1996). The relation of parent alcoholism to adolescent substance use: A longitudinal follow-up study. Journal of Abnormal Psychology, 105(1), 70-80. Cicchetti, D., & Barnett, D. (1991). Toward the development of a scientic nosology of child maltreatment. In W. Gove & D. Cicchetti (Eds.), Thinking clearly about psychology: Essays in honor of Paul E. Meehl: Vol. 2. Personality and psychopathology (pp. 346-377). Minneapolis, MN: University of Minnesota Press. Conway, M. A. (1996). Autobiographical memory. In E. L. Bjork & R. A. Bjork (Eds.), Memory (2nd ed., pp. 165-194). San Diego, CA: Academic Press. Conway, M. A., & Pleydell-Pearce, C. W. (2000). The construction of autobiographical memories in the self-memory system. Psychological Review, 107, 261-288. Della Femina, D., Yeager, C. A., & Lewis, D. O. (1990). Child abuse: Adolescent records vs. adult recall. Child Abuse and Neglect, 14, 227-231. Dinwiddie, S., Heath, A. C., Dunne, M. P., Bucholz, K. K., Madden, P. A. F., Slutske, W. S., . . . Martin, N. G. (2000). Early sexual abuse and lifetime psychopathology: A co-twin control study. Psychological Medicine, 30, 41-52. Duncan, R. D., Saunders, B. E., Kilpatrick, D. G., Hanson, R. F., & Resnick, H. S. (1996). Childhood physical assault as a risk factor for PTSD, depression, and substance abuse: Findings from a national survey. American Journal of Orthopsychiatry, 66, 437-448. Elwyn, L. J. (2011). Adult memory and appraisal of maltreatment in childhood: Implications for problem substance use. Dissertation Abstracts International Section A: Humanities and Social Sciences, 72(6-A), 2156. Famularo, R., Kinscherff, R., & Fenton, T. (1992). Parental substance abuse and the nature of child maltreatment. Child Abuse and Neglect, 16, 475-483. Fergusson, D. M., Boden, J. M., & Horwood, L. J. (2008). The developmental antecedents of illicit drug use: Evidence from a 25-year longitudinal study. Drug and Alcohol Dependence, 3, 1-13. Fergusson, D. M., Horwood, L. J., & Lynskey, M. T. (1996). Childhood sexual abuse and psychiatric disorder in young adulthood: II. Psychiatric outcomes of childhood sexual abuse. Journal of the American Academy of Child and Adolescent Psychiatry, 34, 1365-1374. Fergusson, D. M., & Lynskey, M. T. (1997). Physical punishment/maltreatment during childhood and adjustment in young adulthood. Child Abuse and Neglect, 21, 617-630. Finkelhor, D. (1979). Sexually victimized children. New York, NY: Free Press. Habermas, T., & Bluck, S. (2000). Getting a life: The emergence of the life story in adolescence. Psychogical Bulletin, 126, 748-769. Memory in Child Maltreatment and Adult Substance Abuse 291 Hardt, J., & Rutter, M. (2004). Validity of adult retrospective reports of adverse childhood experiences: Review of the evidence. Journal of Child Psychology and Psychiatry, 45, 260-273. Howe, M. L., & Courage, M. L. (1997). The emergence and early development of autobiographical memory. Psychological Review, 104, 499-523. Ireland, T., & Widom, C. S. (1994). Childhood victimization and risk for alcohol and drug arrests. The International Journal of the Addictions, 29, 235-274. Jasinski, J. L., Williams, L. M., & Siegel, J. (2000). Childhood physical and sexual abuse as risk factors for heavy drinking among African-American women: A prospective study. Child Abuse and Neglect, 24, 1061-1071. Kaplan, D. (2009). Structural equation modeling: Foundations and extensions. Thousand Oaks, CA: Sage. Kendall-Tackett, K., & Becker-Blease, K. (2004). The importance of retrospective ndings in child maltreatment research. Child Abuse and Neglect, 28, 723-727. Kirisci, L., Dunn, M. G., Mezzich, A. C., & Tarter, R. E. (2001). Impact of parental substance use disorder and child neglect severity on substance use involvement in male offspring. Prevention Science, 2, 241-255. Krohn, M. D., Stern, S. B., Thornberry, T. P., & Jang, S. J. (1992). The measurement of family process variables: The effect of adolescent and parent perceptions of family life on delinquent behavior. Journal of Quantitative Criminology, 8, 287-315. Krohn, M. D., & Thornberry, T. P. (1999). Retention of minority populations in panel studies of drug use. Drugs and Society, 14, 185-207. Kunitz, S. J., Levy, J. E., McCloskey, J., & Gabriel, K. R. (1998). Alcohol dependence and domestic violence as sequelae of abuse and conduct disorder in childhood. Child Abuse and Neglect, 22, 1079-1091. Langeland, W., & Hartgers, C. (1998). Child sexual and physical abuse and alcoholism: A review. Journal of Studies on Alcohol, 59, 336-348. Leifer, M., Kilbane, T., & Grossman, G. (2001). A three-generational study comparing the families of supportive and unsupportive mothers of sexually abused children. Child Maltreatment, 6, 353-364. Liao, T. F. (1994). Interpreting probability models: Logit, probit, and other generalized linear models. Thousand Oaks, CA: Sage. Long, J. S. (1997). Regression models for categorical and limited dependent variables. Thousand Oaks, CA: Sage. MacKinnon, D. P. (2008). Introduction to statistical mediation analysis. New York, NY: Taylor & Francis. MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Shets, V. (2002). A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7(1), 83-104. Malinosky-Rummell, R., & Hansen, D. J. (1993). Long-term consequences of childhood physical abuse. Psychological Bulletin, 114, 68-79. Markus, H. (1977). Self-schemata and processing information about the self. Journal of Personality and Social Psychology, 35(2), 63-78. McAdams, D. P. (2001). The psychology of life stories. Review of General Psychology, 5, 100-122. McCord, J. (1983). A forty year perspective on effects of child abuse and neglect. Child Abuse and Neglect, 7, 265-270. 292 L. Elwyn and C. Smith Mikulincer, M. (1998). Adult attachment style and individual differences in functional versus dysfunctional experiences of anger. Journal of Personality and Social Psychology, 74, 513-524. Moncrieff, J., & Farmer, R. (1998). Sexual abuse and the subsequent development of alcohol problems. Alcohol and Alcoholism, 33, 592-601. Muthn, L. K., & Muthn, B. O. (2007). Mplus user's guide (4th ed.). Los Angeles, CA: Muthn & Muthn. Najavits, L. M., Weiss, R. D., & Shaw, S. R. (1997). The link between substance abuse and posttraumatic stress disorder in women: A research review. The American Journal on Addictions, 6, 273-283. National Drug Intelligence Center. (2011). National Drug Intelligence Center national threat assessment: The economic impact of illicit drug use on American society. Washington, DC: Department of Justice. Neisser, U. (1986). Nested structure in autobiographical memory. In D. C. Rubin (Ed.), Autobiographical memory (pp. 71-81). Cambridge, UK: Cambridge University Press. Ouimette, P. C., Kimerling, R., Shaw, J., & Moos, R. H. (2000). Physical and sexual abuse among women and men with substance use disorders. Alcoholism Treatment Quarterly, 18(3), 7-17. Polusny, M. A., & Follette, V. M. (1995). Long-term correlates of child sexual abuse: Theory and review of the empirical literature. Applied and Preventive Psychology, 4, 143-166. Saunders, B. E., Kilpatrick, D. G., Hanson, R. F., Resnick, H. S., & Walker, M. E. (1999). Prevalence, case characteristics, and long-term psychological correlates of child rape among women: A national survey. Child Maltreatment, 4, 187-200. Schuck, A. M., & Widom, C. S. (2001). Childhood victimization and alcohol symptoms in females: Causal inferences and hypothesized mediators. Child Abuse and Neglect, 25, 1069-1092. Scott, K. D. (1992). Childhood sexual abuse: Impact on a community's mental health status. Child Abuse and Neglect, 16, 285-295. Sedlak, A. J., & Broadhurst, D. D. (1996). Executive summary of the third National Incidence Study of child abuse and neglect. Washington, DC: U.S. Department of Health and Human Services. Sedlak, A. J., Mettenburg, J., Basena, M., Petta, I., McPherson, K., Greene, A., & Li, S. (2010). Fourth National Incidence Study of child abuse and neglect (NIS-4): Report to Congress, executive summary. Washington, DC: U.S. Department of Health and Human Services, Administration for Children and Families. Simpson, T. L., & Miller, W. R. (2002). Comcomitance between childh