! Examining Executive Functioning Deficits in Juvenile Delinquents with a History of Trauma Exposure by Melisa A. Cyperski A thesis submited to the Graduate Faculty of Auburn University in partial fulfilment of the requirements for the Degree of Master of Science Auburn, Alabama May 4, 2013 Copyright 2012 by Melisa Cyperski Approved by Steven Shapiro, Chair, Asociate Profesor of Psychology Barry Burkhart, Profesor of Psychology Jennifer Gilis Matson, Asociate Profesor of Psychology ! ! ! ii Abstract We examined whether an early childhood history of trauma explains which juvenile offenders develop Executive Functioning (EF) deficits. One hundred and eighty-eight incarcerated adolescent males were evaluated for personal trauma history and EF, from which three latent factors were formed. Despite exhibiting below average EF performance, SEM showed that childhood maltreatment and delinquent status were not mediated by EF performance. Analyses indicated that specific trauma characteristics predict juvenile offending behavior, even after controlling for EF. Salient trauma characteristics include age of first victimization, relationship to perpetrator, and combined-type victimization (i.e., physical and sexual victimization). In particular, experiences with early victimization, incestuous trauma, and combined-type abuse are related to juvenile sex offending and may be stronger predictors of prognosis than other trauma characteristics (e.g., frequency, duration). Some individuals with a history of trauma exposure and some juvenile offenders may exhibit EF deficits; but poor inhibition, cognitive flexibility, or monitoring does not appear to explain the relationship betwen trauma and delinquency. Other theories regarding the long-term efects of childhood trauma and the etiology of delinquent behavior should be explored in order to identify protective factors and inform treatment. The need for refinement in EF conceptualization and measurement also continues. ! Table of Contents Abstract.........................................................................................................................................ii List of Tables................................................................................................................................v List of Illustrations.......................................................................................................................vi Introduction...................................................................................................................................1 Executive Functions..........................................................................................................1 Neuropsychological Deficits and Delinquency................................................................8 Neuropsychological Deficits and Psychopathology among Individuals Exposed to Trauma................................................................................................................13 Trauma History among Juvenile Delinquents.................................................................18 Trauma as a Predictor of Neuropsychological Deficits and Ofender Characteristics: The Present Study.................................................................................................................19 Hypotheses......................................................................................................................21 Method ......................................................................................................................................22 Site of Study....................................................................................................................22 Participants......................................................................................................................23 Measures.........................................................................................................................24 Procedure........................................................................................................................26 Results.........................................................................................................................................27 Overview of Analyses.....................................................................................................27 Descriptive Statistics.......................................................................................................33 ! ! ! ii Evaluation of Model Fit..................................................................................................33 Evaluation of Direct and Indirect Efects.......................................................................34 Discussion...................................................................................................................................38 Age of First Traumatic Experience.................................................................................39 Duration of Traumatic Victimization..............................................................................41 Frequency of Traumatic Experiencing............................................................................41 Relationship to Perpetrator of Traumatic Experience.....................................................43 Type of Traumatic Victimization....................................................................................44 The Relative Contribution of Executive Functioning.....................................................46 Clinical Application........................................................................................................53 Limitations and Future Directions..................................................................................55 References .................................................................................................................................58 ! ! ! iii! ! ! ! ! ! List of Tables Table 1 .......................................................................................................................................69 Table 2 .1....................................................................................................................................70 Table 2 .2....................................................................................................................................71 Table 2 .3....................................................................................................................................72 Table 2 .4....................................................................................................................................73 Table 3........................................................................................................................................74 Table 4 .1....................................................................................................................................75 Table 4 .2....................................................................................................................................76 Table 4 .3....................................................................................................................................77 Table 4 .4....................................................................................................................................78 Table 4 .5....................................................................................................................................79 ! ! ! iv! ! ! ! ! ! List of Illustrations Illustration 1.1 ............................................................................................................................80 Illustration 1.2 ............................................................................................................................81 Illustration 1.3 ............................................................................................................................82 Illustration 1.4 ............................................................................................................................83 Illustration 1.5 ............................................................................................................................84 ! ! ! 1 Introduction For years, researchers have tried to identify the predictors of childhood behavioral problems and psychological disorders. In theory, if predictors can be acurately identified, psychologists and other mental health profesionals might be able to prevent problems from arising or, at the very least, to intervene in a targeted and specific way. Many predictors have been identified for juvenile delinquency?which itself is a legal clasification?including, but not limited to: decreased verbal abilities, low SES, and early childhood abuse or neglect. Some juvenile delinquents also demonstrate neurocognitive deficits in executive functioning (EF). In the past, researchers have been unable to identify the direct link betwen EF performance and delinquency. Given that decreased EF performance is often asociated with psychological disorders such as Oppositional Defiant Disorder (OD), Atention-Deficit/Hyperactivity Disorder (ADHD), and Post-Traumatic Stres Disorder (PTSD); functional dificulties in academic achievement; regulating emotions; and interpersonal problems, it is likely that indirect variables also contribute to the relationship betwen EF and juvenile offending. Researchers have speculated that the intermediary variables may be social maturity, ability to read social cues, impulsivity, or the presence of symptoms asociated with ADHD. The purpose of this study is to examine another possible explanation for the relationship betwen EF and delinquency: early trauma exposure impairs the appropriate development of EF abilities and creates a subgroup of criminal offenders with EF impairment. Executive Functions EF is a multifaceted construct conceptualized as the cognitive proceses that underlie, ! ! ! 2 organize, and execute goal-directed or problem-solving behavior. Mediated by the prefrontal cortex (PFC) and asociated interconnections, EF is instrumental in orchestrating activity across the brain that directs the identification of a goal, as wel as the means to strategize and efectively acomplish the goal. Proper EF requires the ability to use cognitive proceses such as memory, atention, reasoning, problem solving, and ultimately results in the self-regulation of emotions and behaviors. In sum, EF encompases the proceses and skils necesary for adaptive functioning in everyday life (Bergeron & Valiant, 2001; Best & Miler, 2010; Garcia-Barrera, Kamphaus, & Bandalos, 2011). Components of executive functioning. Given its broad application to adaptive functioning, the construct of EF covers a wide array of diferent components and abilities. These varying cognitive components asociated with EF are most easily understood via behavioral description. Note that researchers do not agree on a single conceptual model of EF and vary both the number and label of functional components. Thus, only a limited number of theories wil be highlighted here. Garcia-Barrera and colleagues (2011) propose that the components of EF include problem solving, updating Working Memory (WM), atentional control, behavioral control, and emotional control. To aid conceptualization, basic behavioral descriptions of each component are as follows: (a) problem solving is the planning, decision making, and organizing of information in order to achieve a goal; (b) updating WM is defined as the ability to proces and manipulate information in acordance with task demands; (c) atentional control is conceptualized as the ability to focus, sustain, and shift concentration and awarenes at wil; (d) behavioral control is the self-regulation of behavior, including inhibition or impulse control; and (e) emotional control, the ability to self-regulate afect in response to internal and external environmental cues (Garcia-Barrera et al., 2011). ! ! ! 3 Similarly, Roberts and Pennington (1996) propose that WM and inhibition are the central components of EF. Their definition of WM is similar to that which is described above. Inhibition is conceptualized as the suppresion of behaviors that are irrelevant to the task-at-hand and protect the self from interference. To these components, the theory proposed by Miyake et al. (2000) would also add the component of shifting. Cognitive flexibility/set-shifting is defined as the ability to adapt or modify current strategies acording to changing task demands. As aforementioned, models describing the components of EF commonly include problem solving, updating WM, self-regulation (i.e., atentional/behavioral/emotional control), inhibition, and cognitive flexibility/set-shifting. Other notable EF theorists, including Barkley (1997a; 1997b), also include components such as internalized speech, reconstitution, decision-making, planning, organization, performance monitoring, verbal fluency, or goal establishment (Dick & Overton, 2010; McCafrey, Lynch, & Westervelt, 2011). Components measured in popular EF batteries. When atempting to describe and measure the components of EF, many of the concepts overlap, making it dificult to pin down just how many pure components of EF exist. Given this construct impurity, researchers and psychometricians turn to factor analysis to help determine which components are being asesed in popular EF bateries. Several popular EF bateries apply a three-factor model with high intercorrelations betwen components. Behavior Rating Inventory of Executive Functioning (BRIEF; Gioia, Isquith, Guy, & Kenworthy, 2000). Parent-derived behavioral ratings of children aged 5-18 years old with various clinical diagnoses were used to analyze four competing models of the factor structure of the BRIEF. Confirmatory Factor Analysis (CFA) revealed that the best fit for the data was a three- factor model: Behavioral Regulation, Emotional Regulation, and Metacognition. Al three factors ! ! ! 4 were significantly correlated with each other in the moderate to strong range. The authors suggest that the high degree of intercorrelation demonstrates the unitary, but also fractional nature of the EF construct (Gioia, Isquith, Retzlaf, & Epsy, 2002). Delis-Kaplan Executive Function System (D-KEFS; Delis, Kaplan, & Kramer, 2001). The D-KEFS is an EF test batery designed for children and adults, ages 8 to 89. The D-KEFS is purported to tap into constructs of cognitive-flexibility/set-shifting, inhibition, response generation, concept formation, categorization and efective use of fedback, deductive reasoning, integration of information into current memory systems, planning, and rule learning (Strauss, Sherman, & Spreen, 2006). To ases these theoretical components, in a study of normal children and adults, Latzman and Markon (2010) determined that the D-KEFS tasks are best conceptualized under a three-factor model which aligns with the Miyake et al. (2000) theory of EF: Conceptual Flexibility, Monitoring, and Inhibition. The Conceptual Flexibility and Inhibition latent variables are consistent with previous descriptions of the EF components. Monitoring is likened to Updating WM and is defined by the authors as the active proces of evaluating new information with respect to the current task and including the new information into the individual?s WM as needed (Latzman & Markon, 2010). However, the results from factor analytic studies can be misleading. Asesing and analyzing the underlying components of EF may be a misrepresentation of the construct because conducting a factor analysis suggests that there are indeed distinct, separable components when, in reality, current measures of EF are plagued with task impurity (Dick & Overton, 2010). Indeed, EF is complex. Involving the orchestration of many cognitive proceses, the very construct of EF implies interconnections within the brain?s networks. Therefore, there may be many components of EF, not al of which are addresed in asesment bateries (For more ! ! ! 5 thorough reviews se: Brocki & Bohlin, 2004; Miyake et al., 2000; Roberts & Pennington, 1996; Sergeant et al., 2002; Strauss et al., 2006). Overal, despite asesment and methodological shortcomings, EF has been the focus of many studies and has been linked to many behavioral problems and psychological disorders throughout the lifespan (Pennington & Ozonoff, 1996; Wilcutt, Doyle, Nigg, Faraone, & Pennington, 2005). Therefore, while researchers may not agree upon the components or how to best ases for EF, it is important to understand how EF develops and what happens when things go awry. Development of executive functioning. Behavioral and neuroimaging research support the notion that foundational EFs develop during the preschool years, with many children exhibiting intact inhibition, WM, and set shifting prior to the age of 5. In their developmental review of EF, Best and Miler (2010) describe that preschool children make significant gains in these foundational EFs, with modest improvements made on more advanced and complex tasks (e.g., planning) as they age. In addition to performance-based growth, experimenters have also measured brain activity. When children and adolescents complete EF tasks in fMRI studies, the results typicaly reveal activity in the PFC at a young age. Older children (ages 9 to 11) demonstrated more localized, specific activation paterns acording to task demands, which suggests that pre-tens exhibit les brain activation in regions of the PFC that are uncorrelated with requisite task performance. Therefore, after initialy acquiring executive functioning abilities during childhood, adolescents exhibit increased eficiency in their neurocognitive abilities. Similarly, Difusion Tensor Imaging (DTI) studies, which measure inter-neuronal connectivity, demonstrate increased myelination from other brain regions to the PFC with age. Increased myelination to the PFC indicate an increased eficiency of skils through adolescence ! ! ! 6 and adulthood, suggesting that the related skils are likely used more frequently and with greater ease than during childhood (Best & Miler, 2010). However, it should be noted that while neuroimaging studies reveal more specific and eficient neural activity with age, few gains in EF performance are observed. For example, when measuring inhibition via the Status task or Knock and Tap game, children show improved performance from ages 3 to 6 with no further significant improvements through age 12. In contrast, computerized tasks such as the Go-No Go task and the Continuous Performance Test (CPT) continue to show some improvement after age 8. Age related diferences are confounded with task impurity and methodological errors such that one test may not measure a specific executive component, but rather a host of interrelated abilities. Additionaly, methods of test administration may be more familiar to a certain subset of children and, thus, give them a performance advantage. For example, older children may be more familiar than young children with computer applications and would then benefit from gains in comfort and eficiency with a computerized test administration over paper-and-pencil formats (Best & Miler, 2010). For tests of WM, researchers use tasks that require maintenance and manipulation of information in order to tap into ?executive control,? but this often requires inhibition and, sometimes, set-shifting as wel. However, preschoolers demonstrate mastery over simple WM tasks including a one-back nonverbal facial recognition task. On more complex WM tasks requiring greater executive control, performance continues to improve with age until approximately 15 years. Finaly, on tasks of set-shifting, preschool children betwen the ages of 3 and 4 can succesfully shift betwen two rules. Yet, as with WM, while task demands increased, improvements were sen through adolescence and into young adulthood (Best & Miler, 2010). ! ! ! 7 Thus, children experience rapid development executive functioning abilities and other cognitive skils during preschool and early primary school years. Experts believe that betwen the ages of 3 and 7, children gain the ability to exercise mastery and control over their emotions and behavior. At this time, children begin to demonstrate advances in mental representation, mental flexibility, the ability to distinguish complex categories, and to take others? perspective (Grabel & Knight, 2009; Zelazo, Muller, Frye, & Marcovitch, 2003). However, while the emergence of EF skils appears to be during early childhood, research also demonstrates that EFs are modified or improved to some degree until they reach maturity in adolescence. The proces of EF and PFC maturation is in acordance with the way the brain develops globaly. The CNS develops through a series of progresive steps and interrelated proceses such that ?the nervous system continues to remodel and change throughout the entire period of development in response to environmental influences and geneticaly programed events? (Mendola, Selevan, Gutter, & Rice, 2002, p. 189). The interdependent and continuous nature of brain development is critical to the understanding of EF such that interruptions to the early development of EF may prohibit or delay later maturation and succesful skil acquisition. Insult or injury to the frontal cortex and crucial interconnections in early childhood may prevent the proceses required for subsequently developing increased eficiency and more advanced problem solving abilities. Furthermore, while the frontal lobe and PFC are crucial to EF, the complex nature of EF-asociated cognitive proceses lends itself to interconnections with many other areas of the brain. Considering that the interconnections of the brain develop and strengthen with age, it is reasonable to asume that areas of the brain asociated with EF may also suffer and fail to fully develop following an early insult or injury (Best & Miler, 2010). ! ! ! 8 Many diverse cognitive skils and their asociated neuroanatomical regions must function together in order to produce the higher-order abstract thinking dictated by EF, including but not limited to abilities such as perception, language, memory, planning, and concept formation (Delis et al., 2001). Thus, executive dysfunction (ED) may influence a broad range of abilities and domains across development. EF deficits can influence such abilities as negotiating interpersonal relationships (Ozonoff, 2001), succeding in academic tasks, inhibiting inappropriate thoughts or behaviors, and efectively managing one?s emotions. ED may also contribute to the etiology of certain psychological disorders, or at least, may increase an individual?s risk for developing disorders such as ADHD, Autism Spectrum Disorder (ASD), or PTSD (DePrince, Weinzierl, & Combs, 2009). Beyond the increased risk for developing impairment and/or a psychological disorder, the complexity and multiplicities of EF are important to note because dysfunction may be the result of a breakdown at any point in the system throughout the course of development, which renders mild and/or specific impairments dificult to identify (Delis et al., 2001). Researchers have already identified a number of factors that predict neurodevelopmental and EF deficits, including: low socioeconomic status (SES), malnutrition, maternal substance use, neurotoxin exposure (e.g., methylmercury), maltreatment (physical, sexual, and/or emotional), and traumatic brain injury (TBI; Mendola et al., 2002; Satler, 2008). However, because the exact proces that disrupts the appropriate formation of EF is unknown, any number of other predictors may also exist. Neuropsychological Deficits and Delinquency In addition to clinical populations such as individuals with an ASD or ADHD, violent adult offenders have also reliably demonstrated neuropsychological deficits. Preliminary studies ! ! ! 9 on neuropsychological functioning in offending populations indicated that death row inmates and homicidal offenders exhibit significant neurological deficits, along with hyperactivity, atention dificulties, and aggresive emotionality, when compared to non-offending peers (Lewis, Pincus, Bard, & Richardson, 1986; Santila & Hapasalo, 1997; Veneziano, Veneziano, LeGrand, & Richards, 2004). In addition, more neuropsychological impairment is observed among individuals who have commited violent and non-violent sex offenses, as compared to non-sex offenders. Brain imaging studies corroborate these findings and demonstrate frontal lobe dysfunction among violent offenders and temporal lobe dysfunction among sex offenders (Palone & Voelbel, 1998; Veneziano et al., 2004). Beyond pure ED, many adult and adolescent offenders also demonstrate broad cognitive impairment, with particular dificulties on measures of verbal inteligence (Kely, Richardson, Hunter, & Knapp, 2002). Experts hypothesize that pre-existing deficits in verbal inteligence, social skils, or other domains may be exacerbated by poor EF abilities. Neuropsychological deficits may further restrict social problem solving skils, inhibit efective procesing of relevant environmental stimuli, or may limit internalized self-speech such that decision making skils are impaired and the imediate reward outweighs the potential long-term goal consequences (Bergeron & Valiant, 2001; Kely et al., 2002). However, while ED sems conceptualy wel evident, and despite the cognitive and neuropsychological deficits observed among adult offenders, research regarding the neuropsychological functioning of adolescents is inconclusive and results may be dependent on the types of measures used, as wel as which intervening variables are acounted for or what subset of the delinquent population is being asesed. For instance, distinctions in EF performance can be made among lifecourse-persistent delinquents and adolescent-limited ! ! ! 10! delinquents such that the former group demonstrates more EF deficits, including more impulsive temperaments and decreased cognitive abilities, especialy in the verbal domain (Donnelan, Ge, & Wenk, 2000; Kennedy, Burnet, & Edmonds, 2011). While compeling, when considered in tandem with other literature that seks to examine the relationship betwen EF and delinquency, the results are hardly conclusive for any subset of adolescent offenders. Several early studies indicated that approximately one-third of juvenile delinquents demonstrate ED when tested using a batery of measures such as the Wisconsin Card Sorting Task (WCST), Porteus Mazes, Trails B, and Controlled Oral Word Asociation Test (COWAT) (Skoff & Libon, 1987). In contrast, other early studies failed to diferentiate betwen delinquent and non-delinquent groups when analyzing EF performance on the same types of tasks (Appelof & Augustine, 1985). More recently, Bergeron and Valiant (2001) conducted a study using similar EF measures as in the aforementioned seminal studies. Analyzing the neurocognitive delays and personality characteristics of adolescent and adult offenders, the researchers found that both offenders and their non-offending peers demonstrated ED on some measures of EF, but no deficits on other EF measures. In particular, offenders exhibited significantly poorer performance on the Qualitative Score of the Porteus Maze Test, which alegedly taps into the EF components of planning and foresight. They also performed more poorly than non-offending peers on the Conceptual Level score of the Paragraph Completion Method (PCM) which infers the individual?s abstract reasoning abilities, thinking style, and social competence. However, offenders and non-offenders performed similarly on the WCST, a popular EF measure of abstract reasoning and cognitive flexibility (Bergeron & Valiant, 2001). Bergeron and Valiant (2001) purported that the unstable EF performance among ! ! ! 11! offenders was due to the primary brain region asociated with each task (i.e., orbitofrontal- ventromedial vs. dorsolateral prefrontal). Additionaly, they reported that the Porteus Maze and PCM tasks demonstrate task impurity within and outside the EF construct, and the authors suggest their batery, and the PCM in particular, may tap into measures of social judgment rather than pure neuropsychological evaluation. Thus, the inconsistencies in the EF scores for offenders are reflective of conceptual and methodological shortcomings rather than pure neurocognitive abilities, leading to few conclusions about ED among offending populations (Bergeron & Valiant, 2001). Moffit (1990) suggests that inconsistencies in EF performance among delinquents might be due to individual histories with ADHD. In support of this claim, one study demonstrated that delinquents with ADHD have more executive dysfunction than peers without a past history of ADHD symptomatology (Moffit, 1990; Moffit & Henry, 1989). Additionaly, nearly 46% of juvenile delinquents are estimated to have problems with atention (O?Brien, Langhinrichsen- Rohling, & Sheley-Tremblay, 2007), which is much higher than the national average of 6 to 8% (Weis, 2008). Conversely, in a study examining the relationship betwen aggresion, executive functioning, and ADHD, Seguin and colleagues (1999) found that boys with a history of physical aggresion demonstrated reliable ED in WM, even after controlling for an ADHD diagnosis. Thus, it is likely that the relationship betwen juvenile delinquency and EF cannot be explained completely by the presence of ADHD symptoms, especialy for perpetrators of violent and aggresive crimes. To examine whether there is something unique about individuals who commit particular types of crimes, researchers have atempted to discriminate betwen juvenile sex offenders ! ! ! 12! (JSOs) and non-sex offenders (NJSOs). Veneziano and colleagues (2004) compared the EF performance of JSOs and NJSOs, as measured by the WCST, COWAT, Tower of London, and Trail Making Parts A and B. Comparisons betwen the groups yielded mixed results, depending on the task and domain measured. Generaly, on most measures of EF in their study, there was no statisticaly significant diference betwen the offender groups. However, NJSOs did demonstrate significantly beter performance on the Part B of the Trail Making Test than their sex-offending peers, indicating more advanced cognitive set-shifting skils. The authors suggest that the improved cognitive flexibility demonstrated by the NJSOs may represent a beter developed ability to identify stimuli within their environment and to use these cues more efectively when deciding a course of action (Veneziano et al., 2004). It sems imperative to investigate this claim more systematicaly and directly. Furthermore, select individuals from both groups (i.e., JSOs and NJSOs) showed impairment on some EF tests, but not others. Both sets of boys demonstrated lower than average performance on the Tower of London task. Adolescent offenders required more Total Moves and had a lower Total Initiation Time than is typical, which suggests that offending youths take les time to think or plan before engaging in a behavior relative to their non-offending peers (Veneziano et al., 2004). Alternatively, engaging in les planning prior to task initiation may be reflective of increased levels of impulsivity or decreased motivation for succesful task completion. A lack of significant diference betwen JSOs and NJSOs on EF performance measures is contrary to the research conducted with adult offenders. However, Veneziano and colleagues (2004) reportedly used a sample of non-violent sex offenders. It is posible that decreased EF performance is asociated with violent crimes, rather than the type of offense (sex vs. non-sex). ! ! ! 13! Alternatively, it is possible that precipitant stresors or early environmental variables, such as a history of trauma exposure, contribute more to the development of adverse outcomes and executive dysfunction than does the severity or type of offense (Farris, 2007; Mendola et al., 2002; Veneziano et al., 2004). Overal, researchers are stil unsure as to whether juvenile delinquents as a whole demonstrate neuropsychological deficits. In addition to the potential explanations described above, discrepancies betwen the juvenile and adult offender literature may also be due to diferences in the type of task being asesed or age-related task invariance. Finaly, variations in the relationship betwen EF and offending among juveniles and adults may be the result of natural heterogeneity and behavioral inconsistencies during adolescence. Fluctuations in behavior and thought paterns may occur because adolescence is often characterized by increased disinhibition and experimentation with diferent social roles, al of which may contribute to the occurrence of criminal offenses. Thus, it is reasonable to conclude that adolescents? performance on any psychological measure may be inconsistent. Variability in EF performance may be particularly common among adolescent offenders but not in adults, given that the brain is stil developing and changing during this time period. Neuropsychological Deficits and Psychopathology among Individuals Exposed to Trauma Several other populations demonstrate poor cognitive and executive functioning performance, including individuals with a history of trauma exposure. Childhood exposure to trauma, such as physical, emotional, or sexual maltreatment, is asociated with a number of undesirable outcomes, including neuropsychological deficits, subsequent psychopathology (i.e., depresion, anxiety, PTSD), and decreased academic achievement (van der Kolk, 2003). In particular, children who experience traumatic maltreatment early in life are more likely than ! ! ! 14! children who did not, to develop depresion, PTSD, delinquency, aggresive tendencies, substance abuse disorders, and hypersexualized behaviors, among other maladaptive outcomes (Grabel & Knight, 2009). Children exposed to trauma may also be at risk for developing low self-estem, suicidal ideation, guilt, and be at increased vulnerability for future victimization (Kendal-Tacket, Wiliams, & Finkelhor, 1993; Walker, Carey, Mohr, Stein, & Sedat, 2004). The negative efects of trauma are also long term. Adult women who experienced either child physical or sexual maltreatment before age 17 continued to demonstrate elevations on measures of internalizing symptomatology, including depresion and anxiety, as compared to individuals with no history of abuse (Wind & Silvern, 1992). Throughout their lifetime, trauma survivors of both sexes experience increased rates of mood disturbance, anxiety, disordered personality, maladaptive eating and substance use, ADHD, and oppositional defiant behavior (Walker et al., 2004). Not surprisingly, one of the most common disorders to develop after a traumatic experience is PTSD. Research examining the neuropsychological functioning of adults with PTSD has demonstrated decreased cognitive/executive functioning such as concentration, learning, and WM. Poor EF performance among this population results in decreased reasoning and decision-making abilities, as wel as poorer impulse and emotional control (Walter, Palmieri, & Gunstad, 2010). Research surrounding the neuropsychological performance of children with PTSD is les developed, but is critical to understand given the developmental consequences of neuropsychological deficits described above. In a study of 14 children, Beers and De Belis (2002) found that children with maltreatment-related PTSD performed more poorly in multiple cognitive domains, including learning and memory, visual-spatial functioning, problem solving, and atention, when compared ! ! ! 15! to children who had not been maltreated. After Bonferroni correction, children with PTSD exhibited significant decreased performance in the areas of atention and abstract reasoning/executive functioning. Deficits in atention and EF are consistent with the theory that trauma in early childhood interrupts neuroanatomical development and procesing, with particular detriment to the frontal lobe and its asociated networks (Beers & De Belis, 2002). However, due to their methodology, Beers and De Belis (2002) were unable to compare the EF performance of children with maltreatment related PTSD to children who were exposed to trauma, but did not subsequently develop PTSD. Instead, the authors used a comparison group of non-trauma exposed children and were, thus, unable to disentangle the neuropsychological efects of trauma exposure from PTSD symptomatology. Beers and De Belis (2002) did report, though, that EF performance bared no relation to the type or severity of PTSD in children. Similarly, in a later study, Samuelson, Krueger, Burnet, and Wilson (2010) compared the EF performance of children with maltreatment-related PTSD (PTSD+) to children who had experienced a trauma, but did not show signs of PTSD (PTSD-). Al children in the sample witnesed intimate partner violence at a young age. Both groups demonstrated below average performance on tasks measuring EF (e.g., WCST, COWAT, Trail Making B), atention (e.g., Stroop Color & Word Test), and intelectual ability (e.g., selected subtests from the WISC-III). Neither PTSD+ nor PTSD- children demonstrated WM deficits. A lack of WM impairment is contradictory to the adult literature, which indicates poor WM performance and decreased hippocampal volume among adults with PTSD, but is consistent with other studies of childhood PTSD (Beers & De Belis, 2002; Samuelson et al., 2010). Overal, the relationship betwen trauma and EF in children appears to be mediated by the trauma experience itself, not PTSD symptomatology. Thus, PTSD symptomatology is not a ! ! ! 16! focus in the present study except as a means of identifying a clas of individuals with significant trauma history. Salient trauma characteristics. Regarding trauma generaly, DePrince, Weinzierl, and Combs (2009) examined EF among a community sample of trauma-exposed children. EF performance was measured as a composite variable, averaging performance on tasks of WM, inhibition, interference control, and procesing speed. Children who experienced at least one familial trauma (i.e., sexual abuse, physical abuse, witnesing domestic violence) displayed poorer EF performance overal than children exposed to non-familial trauma (i.e., motor vehicle acident, natural disaster), suggesting that al trauma exposure is likely not equal. In fact, the adverse outcomes of child abuse are mediated in part by the type, frequency, and severity of trauma (Kendal-Tacket et al., 1993; van der Kolk, 2003). As aforementioned, familial trauma, in particular, places children at an increased risk for EF-related dificulties. The heightened severity asociated with familial-type trauma is particularly important given that authorities estimate that nearly 80% of al child abuse is commited by a child?s parent, while another 10% is perpetrated by a close relative (van der Kolk, 2003). Additionaly, DePrince et al. (2002) showed that the reported number of familial-trauma incidents contributed unique variance in the prediction of EF composite scores whereas the number of non-familial trauma incidents failed to predict EF performance. The unique predictive ability of the number of reported familial trauma incidents indicates that the frequency of severe trauma experiences may be an important factor in determining the prognosis of trauma-related dificulties and may help guide treatment decisions. Trauma during sensitive periods of development. Adverse outcomes later in life are also predicted by the age at which the child was first traumatized. Grabel and Knight (2009) found ! ! ! 17! that the occurrence of sexual abuse during specific developmental epochs predicted maladaptive, impulsive behaviors such as hypersexuality. However, the only epoch that served as a significant independent predictor was abuse onset betwen 3 to 7 years of age. The lack of significant findings for other developmental periods is particularly interesting given that sexual abuse occurring after age 11 was rated the highest in both frequency and intensity but stil failed to predict sexual fantasy and psychopathology. Grabel and Knight (2009) suggest that interruptions in the development of the brain and executive proceses during ages 3 to 7 may explain the poor inhibitory and impulsive nature of many juvenile sex offenders (JSOs), ultimately causing them to act more readily on their sexual fantasies. One theory as to why the age of traumatic onset is so salient is because trauma may interrupt and delay the development of crucial cognitive and neuropsychological skils. Van der Kolk (2003) explains that trauma is believed to impact: (a) the development and full maturation of specific brain structures at particular ages, (b) the physiologic and neuroendocrine responses, and (c) the capacity to coordinate and regulate cognition, emotion, and behavior. Furthermore, as described earlier, diferent areas and skils of the brain develop at diferent rates. Given the advancements in frontal cortex functioning that occur during ages 3 to 7, children may be particularly sensitive to trauma and abuse during this period (Grabel & Knight, 2009). Additionaly, early theories on stage-specific victimization take a more Piagetian or Freudian approach, implying that trauma during a given developmental period may lead to fixation or an inability to master stage-specific tasks (Finkelhor, 1995). For example, self- regulation skils?including the ability to manage atention, afect, and arousal, such as in the behavioral and emotional regulation components of EF?have been identified as a critical task of infancy and childhood. Children exposed to maltreatment at an early age have demonstrated ! ! ! 18! disruptions in self-regulatory behaviors including aggresive or disruptive tendencies, and emotional disturbance such as fear. Acording to teacher report, maltreated children are also les proficient in peer interactions than non-maltreated peers, a factor that may be important to future development of delinquent or other non-prosocial behaviors (Alesandri, 1991; Farris, 2007). Ultimately, understanding the developmental neurobiology of trauma and abuse may help to explain why and how trauma exposure leads to adverse outcomes. More importantly, understanding the neurobiology of trauma may help identify protective and intervening variables that help to decrease the negative impact of trauma. Intervention wil be especialy efective if provided quickly and early in order to help negate any long-term neurobiological changes and reroute the brain?s developmental course. Trauma History among Juvenile Delinquents In addition to the adverse outcomes of childhood trauma listed above, many juvenile delinquents report a history of childhood maltreatment. Experts estimate that approximately 26% of children who are maltreated before age 11 are later arrested as juveniles and approximately 29% of maltreated children are arrested as adults. Comparatively, individuals who were not maltreated as children were arrested at rates of 16% and 21%, respectively (Burkhart & Cook, 2010). Furthermore, in one study of 83 detained boys aged 12 to 17, as many as 95% of delinquent boys had experienced a prior trauma, with 20% meting criteria for full or partial PTSD (Becker & Kerig, 2011). When considering sex offenders in particular, children who are sexualy abused are 4.7 times more likely to be arrested for a sex crime as adults. Additionaly, JSOs self-report alarming rates of child maltreatment, including histories of physical and sexual abuse. Estimates of prior sexual victimization among JSOs reach as high as 79.4% (Burkhart & Cook, 2010). ! ! ! 19! Thus, with histories full of maltreatment, there is no ignoring the possibility that for some individuals, early traumatic experiences contribute to later delinquent behavior and thus should be evaluated when asesing and treating juvenile offending. Unfortunately, in light of the complex etiology of offending, the nature of the relationship betwen trauma and delinquency remains unclear. Trauma as a Predictor of Neuropsychological Deficits and Offender Characteristics: The Present Study Given the overlapping relationship betwen EF and both trauma and delinquency, neuropsychological deficits may play a mediating role and predict which subset of individuals who experience an early trauma later commit offenses consistent with juvenile delinquency status. It is possible that trauma during sensitive periods interrupts the development of the brain by impacting changes in neuroanatomical structures, neuroendocrine functioning, or neuropsychological control. Such changes may contribute to ED including impoverished abilities to regulate behavior and emotions. In turn, EDs may lead to behavioral outcomes such as poor academic performance, social skils deficits, impulsivity, hypersexuality, and delinquency. Thus, the primary question of this study is: could EF performance intersect the relationship betwen trauma and delinquency? In particular, a history of traumatic experience(s) during a sensitive period of development may explain which juvenile offenders develop ED and which do not. Past studies that have evaluated the relationship betwen trauma and delinquency have not included measures of neuropsychological functioning that directly evaluate the executive and inhibitory control among JO/JSOs (Grabel & Knight, 2009). Using the data derived from the D- KEFS (Delis et al., 2001) and key variables from the ABSOP/DYS Mt. Meigs database, we can identify whether trauma experiences during early childhood are asociated with true EF deficits. ! ! ! 20! We can also explore whether or not EF deficits are found across al juvenile delinquents, or whether there is a unique patern of neuropsychological and cognitive skils among juvenile sex offenders. Admitedly, adolescent self-report of trauma with no external corroboration of abuse raises methodological concerns, but interesting findings may stil arise and lead us to a beter understanding of the complex etiology of executive functioning deficits, psychopathology, and juvenile delinquency. Findings may also guide future endeavors and indicate the use of more rigorous research methodology. Ultimately, it is important for researchers and clinicians to beter understand the etiology and maintenance of offending such that we may improve the eficacy and eficiency of treatment strategies to ensure positive outcomes and decrease recidivism. In particular, identifying which offenders have neuropsychological deficits is critical in that treatment of inatention, afective dysregulation, problem solving, or other EDs may serve as a crucial first line of defense that asists the individual throughout treatment. To examine the indirect relationship betwen trauma and juvenile delinquency, we wil use structural equation modeling (SEM) to estimate the paths betwen trauma, EF variables, and delinquency. We predict that neuropsychological deficits wil provide a causal explanation for the relationship betwen trauma and delinquency, serving as a significant partial mediator. EF wil be examined using the juveniles? scores on the D-KEFS, which wil be organized into three latent variables (Conceptual Flexibility, Inhibition, and Monitoring) in acordance with the CFA conducted by Latzman and Markon (2010). The experience of trauma wil be delineated to mark the type, onset, and frequency of trauma as wel as to indicate whether the perpetrator was related to the individual or not. We believe the most salient predictive indicator wil be the age of traumatic onset, such that ! ! ! 21! adolescents who first experienced trauma during infancy and preschool/early childhood wil demonstrate poorer outcomes than those who first experienced trauma later in life. Analyses wil be conducted on several models that highlight specific trauma characteristics in order to determine which aspects of traumatic experiences map most closely onto EF deficits and subsequent delinquency. Trauma characteristics to be evaluated include age of first traumatic experience, duration of trauma, frequency of trauma, type of trauma, and relationship of the perpetrator to the victim. Each characteristic wil be examined as its own unique model in order to fully consider the significant direct or indirect pathways to delinquency. Models wil be evaluated using the following fit indices: Chi-square, covariance residuals, Root Mean Square Error Approximation (RMSEA), Comparative Fit Index (CFI), and Tucker-Lewis Index (TLI). Iterative modifications to the models wil be made explicit if indicated and theoreticaly appropriate. The model evaluated as the best wil represent the characteristics of trauma that map onto EF deficits and wil also indicate the most salient etiological pathway. Additionaly, exploratory analyses wil also be conducted on the relationship betwen additional trauma variables and more specific D-KEFS variables. Characteristics of delinquency wil also be measured explicitly. We plan to examine the relationship betwen EF and diferent offense types and severities. We predict that both sexual and non-sexual offenders wil demonstrate executive functioning deficits given their previous trauma histories. Additionaly, those individuals who were sentenced for more serious offenses (i.e., violent crimes, repeat offenders) wil demonstrate more deficits than those who commited les intense crimes. Hypotheses The models under investigation include a complement of trauma characteristics, behavioral measures of executive functioning, and delinquency characteristics. The overal ! ! ! 22! models were derived in acordance with a chronological and theoretical basis. For subcomponents of the model and indirect pathways of particular interest, se model depictions in Figures 1.1 through 1.5. While numerous relationships are likely among the psychosocial variables within the model, in this study the hypotheses of interest include: 1) The adolescents? previous trauma histories were expected to influence delinquency status by way of executive functioning deficits, with Inhibition as a particularly salient predictor of delinquent characteristics. 2) Adolescents who commited sexual offenses wil demonstrate more EF deficits relative to their non-sexualy offending delinquent peers. 3) The age of traumatic onset is predicted to map more closely onto EF deficits and thus wil represent the best fiting model when compared to the duration and frequency of trauma. In particular, adolescents who first experienced trauma during preschool/early childhood wil demonstrate the most impairment. Extended frequency or duration of trauma wil also result in more relative impairment than individuals who experienced a limited exposure. 4) Adolescents who experienced sexual trauma in childhood wil demonstrate the most impairment compared to adolescents who experienced physical abuse, neglect, or other trauma. Method Site of Study The current study was part of a larger research program conducted at the Mt. Meigs Complex, a residential facility for adjudicated juveniles operated by the Alabama Department of Youth Services (DYS). In order to be placed at Mt. Meigs, adolescents either pled guilty or were ! ! ! 23! found to be guilty by a juvenile court judge. Mt. Meigs is also the DYS site designated for the treatment of al adolescents in the state of Alabama adjudicated for a sexual offense. In collaboration wil several local universities, the Acountability Based Juvenile Sexual Ofender Program (ABSOP) was developed to provide asesment and treatment services to the youths. Incarcerated juveniles range in age from 12 to 21 years, with the majority of boys ranging from age 14 to 18. Participants Participants in this study were 188 adjudicated delinquent male juveniles, aged 10 years, 8 months to 19 years, 2 months, who were convicted of various offenses. Adolescents included in this study were admited into the facility during the period of time ranging from February 2005 to January 2008. Per self-report, juveniles were 51.1% Caucasian, 45.7% African American, 0.5% Hispanic, 2.1% Biracial, and 0.5% other. Participants were categorized into two groups, based on their offense type: 127 boys (68%) were adjudicated on a sexual offense and mandated to participate in sex offender specific treatment; sixty-one boys (32%) were adjudicated on non-sexual offenses and receiving treatment for anger management, substance abuse, impulse control training, and/or other psychological isues. For 49.8% of the juveniles, the incarceration at time of interview was their first commitment, and for 35.9% of those incarcerated, this was their first arrest. Additionaly, 42.3% of the boys received special education services. Estimates of Full Scale IQ were measured by the Wechsler Abbreviated Scale of Inteligence (WASI; Wechsler, 1999). The juveniles performed within a range of 53 to 125, with a mean score of 87.13 (SD = 13.6) and 50% of the boys scoring betwen 77.5 and 98. ! ! ! 24! Many participants also reported a history of abuse: 35.4% reported being the victim of sexual abuse, 26.5% reported being the victim of physical abuse, and 21.7% reported being the victim of neglect. Besides victimization, 12.2% of the boys reported witnesing a violent crime, 4.8% reported living through a natural disaster, and 4.2% reported surviving a serious acident (e.g., auto, fire). Experiencing such traumatic events may have been in addition direct sexual abuse, physical abuse, or neglect. Measures Clinical interview. Prior to entry into the treatment program, each adjudicated delinquent completed a semi-structured pre-treatment clinical interview, which takes approximately 2 to 3 hours to complete. The interview was created for on-going research at the site and was designed in acordance with the empirical literature regarding juvenile sex offender asesment and treatment. Information asesed in the interview results in 200 coded variables. Information gathered during the interview includes, but is not limited to historical data regarding the adolescents: demographics, early development, family, physical/mental health, relationships/social functioning, history of abuse/trauma, and sexual history. For a list of relevant variables utilized in analyses for this study, se Appendix. Delis-Kaplan Executive Function System (D-KEFS). The D-KEFS (Delis et al., 2001) is a neuropsychological batery consisting of nine tests that cover a spectrum of verbal and nonverbal tasks and are appropriate for use with individuals aged eight to 89 years. Designed to measure higher level cognitive functioning and components of executive functioning, the D- KEFS taps into various domains such as inhibition, problem solving, cognitive flexibility, planning, impulse control, and abstract thinking. Tests are designed to ases skils in a game- like fashion by using a cognitive proces approach such that both fundamental cognitive skils ! ! ! 25! and higher-level cognitive functions are represented in each task. Additionaly, there is no single score to represent overal EF because EF is multifarious in nature and multiple cognitive abilities (i.e., both fundamental and higher-level) are necesary for succesful performance (Delis et al., 2001). Each subtest score is normed to a mean of 10 and standard deviation of 3. In this study, the following subtests were administered: Trail Making Test, Verbal Fluency Test, Color-Word Interference Test, Sorting Test, Word Context Test, and Tower Test. However, not al subtests were used in analyses. Subtests were selected for inclusion in a manner consistent with the factor structure model suggested by Latzman and Markon (2010) and appropriate for use among 8 to 19 year olds. Refer to Table 1 for a list of which subtests? primary measure scores were included for data analysis. Test-retest reliability coeficients vary across tests, ranging from low to high (se Table 1), but suggest that the skils asesed by most D-KEFS tasks are consistent over time. Additionaly, convergent and discriminant validity also vary appropriately across subtests (Delis et al., 2001). The Trail Making Test (TMT) is a visual cancelation task in which examinees complete a series of increasingly complex connect-the-circles tasks. TMT requires underlying component skils such as visual scanning, number and leter sequencing, and motor speed. In addition, the subtest included in the current analyses aseses for the higher-order EF task of flexibility of thinking (set shifting) with the Number-Leter Switching condition. The Color Word Interference Test (CWI) aseses an individual?s ability to inhibit an automatic, over-learned verbal response in order to generate a novel, conflicting response. Four conditions are presented: name colors on a page (Color Naming); read the printed name of a color (Word Reading); name the color of ink of the printed word, which is disonant with the writen color word (Inhibition); and switch back and forth from responding in a manner ! ! ! 26! consistent with the ink name or the writen word depending on the present governing rule (Inhibition/Switching). The later two tasks were used in the current analyses to emphasize the primary executive functions of inhibition and cognitive flexibility measured by the Stroop procedure. The Verbal Fluency (VF) subtest is a timed subtest that examines the ability to generate verbal responses in acordance with specified rules. Three conditions are presented: Leter Fluency and Category Fluency both ases for vocabulary, atention, semantic organization, initiation, and procesing speed; Switching ases the individual?s ability to contact the semantic network and rapidly retrieve information from memory, as wel as to demonstrate cognitive flexibility while acurately alternating betwen rule-sets. Al three conditions were used in the current analyses to tap into lexical and semantic fluency, and the ability to simultaneously shift betwen over-learned concepts. Finaly, the Sorting Test is akin to the WCST and aseses an individual?s problem- solving abilities as they sek to identify novel groupings of stimuli. Stimuli can be organized acording to verbal information or visual-spatial features. This subtest includes two conditions that require the participant to generate the groups and then to identify the categorical feature depicted in groups created by the examiner. Examinees are asked to describe the sorts, providing the examiner with an understanding into their conceptual-reasoning skils. Acording to Delis et al. (2001), the Sorting Test taps into the individual?s problem-solving, abstract reasoning, and initiation skils. Additionaly, in order to generate novel categories as wel as to simultaneously ases verbal and visual-spatial paterns, cognitive flexibility is required. Procedure Each juvenile provided consent prior to participation and was provided with an ! ! ! 27! explanation of the procedures in place to maintain confidentiality. Following this, participants were encouraged to respond openly and honestly to al interview questions. In instances where inconsistencies were detected, researchers tried to clarify with the adolescent directly, as wel as to consult any records when available. Participants were administered the clinical interview, diagnostic interview, several rating scales and self-report measures, and the D-KEFS. In total, the asesment protocol required approximately 10 to 14 hours to complete. The protocol was administered by a combination of advanced clinical psychology graduate students and undergraduate students, al of whom received extensive training and supervision specific to working with incarcerated juveniles. Furthermore, several training sesions were conducted by the supervising licensed psychologist to ensure a standardized administration of the D-KEFS, which was administered exclusively by the graduate students. In addition, each participant?s protocol was reviewed to ensure scoring acuracy. Computer scoring software was used when available. For those items in the protocol that required manual scoring, undergraduate students were trained on proper scoring procedures and graduate students checked for acuracy. Similar procedures were used for entering and coding information in the database. Results Overview of Analyses From the total sample of 188 juvenile delinquents, a subsample of 92 (21% NJSO, 79% JSO) adolescents was identified and used strategicaly in specific analyses. The 92 boys were selected given their pertinent abuse history as each boy in the subsample reported at least one previous experience with physical or sexual abuse. Additionaly, during the intake interview, ! ! ! 28! only those boys who endorsed a history of sexual or physical victimization were asked follow-up questions about the details and specific nature of their abuse, including the age of first experiencing. Therefore, the amount of data available for use in the analysis of several questions and hypotheses was limited. For those boys who had not experienced abuse, we elected not to include them as dummy controls (coded as zeros) given that inclusion would skew the data and might over- or misrepresent the low end of traumatic experiencing. Furthermore, utilizing the trauma-only sample alowed us to examine the specific efects and unique characteristics of experiencing abuse more closely and more acurately. Ultimately, the reduced sample was used to analyze models pertaining to age of traumatic experiencing, duration of abuse, and relationship to abuse perpetrator. For data analysis, raw data were submited to SPS 19.0 (IBM Corp., 2010) and Mplus 6.12 (Muth?n & Muth?n, 2011). Given the presence of a binary categorical variable (i.e., Group Membership) and the smal sample size, the Mean-and-Variance Adjusted Weighted Least Squares (WLSMV) estimator was used (Flora & Curran, 2004; Yu, 2002). Outliers and normality of data. Ful sample. Data for al variables were normaly distributed, with aceptable skew (-0.76 to 2.25) and kurtosis (-1.59 to 3.09) values. For skew and kurtosis, values with an absolute value greater than 3 and 10, respectively, would be considered extreme (Curran, West, & Finch, 1997; Kline, 2011). Additionaly, each variable within the dataset was combed for outliers beyond the range identified by the median ? two times the interquartile range. Several univariate and bivariate outliers were identified within the dataset, primarily for variables regarding the participants? age and D-KEFS scores. However, D-KEFS related outliers were on the upper end and thus likely represented true performance, and not a lack of motivation or efort. Additionaly, ! ! ! 29! given constraints of the data set (e.g., dummy coding, smal sample size), the individuals with identified extreme scores are believed to be representative of the population and thus outliers were maintained for analysis. Similarly, several individuals were identified as a potential multivariate outlier with a Mahalanobis distance > ! 2 (18) = 42.31, p < .001. However, the existence of multivariate outliers is likely reflective of the aforementioned univariate and bivariate outliers. Therefore, al outliers and extreme scores were noted but ultimately retained in the dataset. Subsample. Data for al variables were normaly distributed, with aceptable skew (-1.47 to 2.66) and kurtosis (-1.16 to 8.25) values. Using a procedure similar to that outlined for the full sample, several univariate and bivariate outliers were identified within the dataset for variables regarding the participants? group membership, age, and duration of abuse. However, as aforementioned, given constraints of the data set (e.g., dummy coding, smal sample size), the individuals with identified extreme scores are believed to be representative of the population and thus outliers were maintained for analysis. Similarly, one individual was identified as a potential multivariate outlier with a Mahalanobis distance > ! 2 (13) = 34.53, p < .001. However, the existence of this multivariate outlier is likely reflective of his scatered performance on the D- KEFS. It is impossible to determine whether his varied performance was reflective of a true deficit, lack of motivation, or other contributing factor. Therefore, his scores were noted but ultimately retained in the dataset. Mising data. Ful sample. Within the sample, 181 individuals had complete data profiles. Among the seven individuals with incomplete data, there were five mising data paterns with les than 5% mising for any given variable. Litle?s MCAR test [! 2 (47) = 45.55, p = .53] was non- ! ! ! 30! significant, indicating that the data are mising completely at random. Subsequently, to handle mising information, data were evaluated using both pairwise deletion and multiple imputation. While the suggested method of handling mising data may be to use multiple imputation (Enders, 2010), the diferences betwen results obtained using multiple imputation and pairwise deletion were very smal. Furthermore, given that multiple imputation was not designed for use with non-normal or categorical data and significance testing cannot be conducted, pairwise deletion was utilized throughout analyses. Furthermore, pairwise deletion has been found to result in unbiased parameter estimates for MCAR data, such as is found within the full and subsamples used for this study (Brown, 2006). Subsample. Within the subsample, 87 individuals had complete data profiles. Among the five individuals with incomplete data, there were four mising data paterns with les than 5% mising on any given variable. Litle?s MCAR test [! 2 (42) = 36.42, p = .84] was non-significant, indicating that the data are mising completely at random. Pairwise deletion was also used to handle mising data in analysis of the subsample. Model fit criteria. The following fit estimates were considered: Chi-square (! 2 ), Root Mean Square Error of Approximation (RMSEA), Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), and Weighted Root Mean Residual (WRMR). As an estimate of exact-fit, ! 2 examines the discrepancy betwen the observed and implied matrices. Good fit was evaluated using the following criteria: ! 2 with p > .05, RMSEA ! .05, CFI > .95, TLI > .95, WRMR < .90 (Brown, 2006; Hu & Bentler, 1999; Yu, 2002). Overall model specification, identification, and modifications. In examining the defined structural model, causal pathways were determined in acordance with the aforementioned literature such that early traumatic experiencing is related to juvenile ! ! ! 31! delinquency and the mechanism for this relationship is executive dysfunction. In particular, traumatic events may disrupt proper neurophysiological development, which can later manifest as behavioral deficits in executive functioning (e.g., poor inhibition) which, in turn, increases the likelihood of delinquent behaviors and subsequent adjudication. Additionaly, JSOs are typicaly younger than NJSOs and, thus, to control for the efects of current age, the age of the adolescent at intake was included as a covariate. Al models were recursive and over-identified (Kline, 2011). However, when initialy estimating and evaluating measurement model fit, the latent variable covariance matrix was not positive definite and several modifications were indicated. Specificaly, the indicator representing Verbal Fluency: Free Sorting Description (SORT1B) was correlated with the Conceptual Flexibility latent factor at a value greater than one. Acording to Brown (2006), a non-positive definite model matrix is often likely within smal samples and within models that contain a limited number of indicators (e.g., two or three) per each latent factor. Furthermore, non-positive definite models and out-of-bound correlations, in particular, can be caused by pairwise deletion, but, as previously mentioned, pairwise deletion was the mising data method of necesity because of the limitations of the WLSMV estimator. Ultimately, SORT1B was deleted from al models. Deletion of this variable was justified on acount of its poor to adequate internal consistency ratings, marginal test-retest ratings (se Table 1), and its multicollinearity with other Conceptual Flexibility/Sorting Test indicators (for correlations, se Table 2.1). Additionaly, modification indices suggested that ! 2 would be improved by alowing correlated residual errors for variables representing Verbal Fluency: Category Switching Total Correct Responses (VF_CATSCO) and Verbal Fluency: Category Switching Total Acuracy (VF_CATSWIACC). Such a modification is conceptualy indicated and also substantialy ! ! ! 32! improved model fit. Following the deletion of SORT1B and the correlation of VF_CATSCO and VF_CATSWIACC, the re-specified measurement model stil satisfied al identification rules and statuses (Kline, 2011). Qualifying for identification, analyses proceded in acordance with two-step Modeling proces. The measurement model was first specified as a Confirmatory Factor Analysis (CFA) prior to being analyzed as SEM. As a CFA with al possible factor correlations, al re-specified models had moderate to relatively good fit and authorized proceding to the SEM pathway analyses. Statistical power. Ful sample. To determine statistical power for evaluating the overal model, we analyzed our data acording to the test of not-close fit (null hypothesis: fit is not excelent). Therefore, power is estimated as the probability of rejecting the null (i.e., rejecting poor model fit and acepting moderate to good fit) within our sample. For the full sample of 188 adolescents, power estimates for rejecting the not-close fit hypothesis were adequate to poor with les than .60 probability of correctly identifying a true efect (MacCalum, Browne, & Sugawara, 1996). Therefore, given the limited power but relatively good fit indices as described below, we can place confidence in our fit statistics and findings for models utilizing the full sample. Subsample. With les than 100 adolescents, our power estimate for rejecting the not- close fit hypothesis was very poor with les than 0.22 probability of correctly identifying a true efect (MacCalum et al., 1996). Therefore, given the limited power but relatively good fit indices as described below, we can also place confidence in our fit statistics and findings for models utilizing the subsample. ! ! ! 33! Descriptive Statistics Correlations betwen observed variables are presented in Tables 2.1 through 2.4. The means (M) and standard deviations (SD) for JSOs and NJSOs are denoted in Table 3. While follow-up analyses are needed to determine statistical significance, preliminary results generaly reveal that (1) JSOs are younger at intake/incarceration than the NJSOs, (2) JSOs first experience trauma at a younger age than NJSOs, and (3) the delinquents? EF performance, as determined by their D-KEFS Scaled Scores, varies betwen the groups but is somewhat lower than expected for their peer group overal. More specificaly, performance on the D-KEFS was below average for both JSOs and NJSOs (M scaled score JSOs = 8.01, NJSOs = 7.74). However, there was no statisticaly significant diference betwen groups [full sample: t(179) = -.91, p = .37]. Additionaly, within our adolescent sample, approximately one-third of all boys reported being the victim of at least one count of sexual abuse, one-fourth reported being the victim of physical abuse, and one-fifth reported being the victim of neglect. Evaluation of Model Fit Age of first traumatic experience. For the model which analyzed the subsample of boys who experienced traumatic victimization, the discrepancy betwen the observed and implied model matrices was non-significant, ! 2 (44) = 37.23, p = .76. Therefore, we retain the exact-fit null hypothesis. Additional support for good model fit included the following estimates: RMSEA = .00 (90% CI: .00, .05), CFI = 1.00, TLI = 1.07, and WRMR = 0.47 (Brown, 2006; Kline, 2011; Yu, 2002). Duration of traumatic victimization. For the model which analyzed the subsample of boys who experienced traumatic victimization, ! 2 (44) = 38.77, p = .70. Therefore, we retain the ! ! ! 34! exact-fit null hypothesis. Additional support for good model fit included the following estimates: RMSEA = .00 (90% CI: .00, .06), CFI = 1.00, TLI = 1.05, and WRMR = 0.45. Frequency of traumatic experiencing. For the model analyzing the full sample of delinquent adolescents, ! 2 (50) = 56.07, p = .26. Therefore, we retain the exact-fit null hypothesis. Additional support for good model fit included the following estimates: RMSEA = .03 (90% CI: .00, .06), CFI = 1.00, TLI = 1.07, and WRMR = 0.60. Relationship to perpetrator of traumatic experience. For the model analyzing the full sample of adjudicated boys, ! 2 (50) = 51.50, p = .42. Therefore, we retain the exact-fit null hypothesis. Additional support for good model fit included the following estimates: RMSEA = .013 (90% CI: .00, .05), CFI = .99, TLI = .99, and WRMR = 0.56. Type of traumatic victimization. For the model analyzing the full adolescent sample, ! 2 (56) = 64.37, p = .21. Therefore, we retain the exact-fit null hypothesis. Additional support for moderate to good model fit included the following estimates: RMSEA = .04 (90% CI: .00, .08), CFI = .93, TLI = .89, and WRMR = 0.59. Evaluation of Direct and Indirect Effects Age of first traumatic experience. Individual unstandardized and standardized factor loadings and path estimates are presented in Table 4.1; Figure 1.1 depicts the model with standardized loadings. Al indicators loaded significantly onto the D-KEFS latent factors. However, a degree of caution is noted as the D-KEFS latent factors are al significantly correlated at p ! .01. For the direct pathway betwen age of traumatic onset and delinquent group membership (JSOs = 1, NJSOs = 0) a significant relationship exists: StdYX = -.31, p < .01. Furthermore, the relationship betwen age of first physical or sexual victimization and delinquent group ! ! ! 35! membership was significant after controlling for the adolescents? age at intake. Previously, the mechanism for the relationship betwen age of first traumatic experience and juvenile delinquency was unclear. In the present study, the latent executive functions of Inhibition, Conceptual Flexibility, and Monitoring were predicted to partialy mediate the relationship betwen age of traumatic onset and delinquent group membership. However, the sum of indirect efects was not significant (p = .89), nor was any specific indirect efect (Inhibition: PRODCLIN 95% CI unstandardized estimate = -0.05, 0.03; Conceptual Flexibility: -0.02, 0.02; and Monitoring: -0.03, 0.06; MacKinnon, Fritz, Wiliams, & Lockwood, 2007). Duration of traumatic victimization. Individual unstandardized and standardized factor loadings and path estimates are presented in Table 4.2; Figure 1.2 depicts the model with standardized loadings. Al indicators loaded significantly onto the D-KEFS latent factors. However, a degree of caution is noted as the D-KEFS latent factors are al significantly correlated at p ! .01, indicating non-independent or not wholly separate constructs. For the direct pathway betwen duration of victimization and delinquent group membership (JSOs = 1, NJSOs = 0), there was no significant relationship. It was believed that extended trauma exposure would increase the severity of impairment, manifesting as executive functioning deficits and delinquent behavior. However, neither the direct efect nor the sum of indirect efects was significant (p = .64). Additionaly, there was no significant specific indirect efect (Inhibition: PRODCLIN 95% CI unstandardized estimate = -0.01, < 0.01; Conceptual Flexibility: < -0.01, < 0.01; and Monitoring: < -0.01, 0.01). Frequency of traumatic experiencing. Individual unstandardized and standardized factor loadings and path estimates are presented in Table 4.3; Figure 1.3 depicts the model with standardized loadings. Al indicators loaded significantly onto the D-KEFS latent factors. ! ! ! 36! However, a degree of caution is noted as the D-KEFS latent factors are al significantly correlated at p ! .01. A significant relationship betwen frequency of victimization (once, twice or more) and delinquent group membership (JSOs = 1, NJSOs = 0) exists for al boys who experienced trauma when compared to those boys who were not the victims of childhood physical or sexual violence. Adolescents convicted of sexual misconduct were more likely to have experienced trauma as compared to non-sexualy offending delinquents (one traumatic experience: StdYX = .29, p = .003; two or more experiences: StdYX = .25, p = .01). Furthermore, the relationship betwen the frequency of physical or sexual victimization and delinquent group membership is significant even after controlling for the adolescents? age at intake. However, it should be noted that the relationship betwen frequency of traumatic experiencing and juvenile delinquency was not partialy mediated by executive functioning. The sum of indirect efects was not significant (once: p = .82; twice or more: p = 1.00), nor was any specific indirect efect for one victimization experience (Inhibition: PRODCLIN 95% CI unstandardized estimate = -0.09, 0.10; Conceptual Flexibility: -0.17, 0.07; and Monitoring: -0.19, 0.22) or for two or more experiences (Inhibition: -0.18, 0.13; Conceptual Flexibility: -0.13, 0.19; and Monitoring: -0.18, 0.15). Relationship to perpetrator of traumatic experience. Individual unstandardized and standardized factor loadings and path estimates are presented in Table 4.4; Figure 1.4 depicts the model with standardized loadings. Al indicators loaded significantly onto the D-KEFS latent factors. However, a degree of caution is noted as the D-KEFS latent factors are al significantly correlated at p ! .01. The pathway betwen non-incestuous victimization and delinquent status was not significant (StdYX = .12, p = .17), suggesting that boys who were victims of non-incestuous ! ! ! 37! abuse were equaly likely to sexualy offend as were boys who were not the victims of physical or sexual abuse. However, the pathway betwen the relationship of the perpetrator and the boys? delinquent group membership (JSOs = 1, NJSOs = 0) was significant for those adolescents who were abused by an imediate or distant relative. In particular, incestuous trauma was related to sexual misconduct (StdYX = .37, p < .001). Furthermore, the relationship betwen incestuous victimization and group status was significant even after controlling for the adolescents? age at intake and was not significantly mediated by executive functioning. The sum of indirect efects was not significant (incest: p = .97; non-incest: p = .82). Similarly, there were no specific indirect efects for non-incestuous victimization (Inhibition: PRODCLIN 95% CI unstandardized estimate = -0.17, 0.13; Conceptual Flexibility: -0.16, 0.12; and Monitoring: -0.15, 0.17) or for incestuous abuse (Inhibition: -0.10, 0.10; Conceptual Flexibility: -0.12, 0.12; and Monitoring: -0.11, 0.13). Type of traumatic victimization. Individual unstandardized and standardized factor loadings and path estimates are presented in Table 4.5; Figure 1.5 depicts the model with standardized loadings. Al indicators loaded significantly onto the D-KEFS latent factors. However, a degree of caution is noted as the D-KEFS latent factors are al significantly correlated at p ! .01. For the direct pathway betwen type of trauma (physical, sexual, or combined) and delinquent group membership (JSOs = 1, NJSOs = 0), there are no significant direct relationships (physical: StdYX = .18, p = .11; sexual: StdYX = .22, p = .08; combined: StdYX = .15, p = .26). A priori correlations indicated that combined physical and sexual trauma was significantly related to JSO status. Therefore, the non-significant direct pathway suggests that executive functioning may partialy mediate the relationship betwen type of trauma and delinquent offending. ! ! ! 38! However, it should be noted that there were no significant indirect efects when analyzing the relationship betwen type of traumatic victimization, juvenile delinquency, and executive functioning. The sum of indirect efects was not significant (physical: p = .64; sexual: p = .48; combined: p = .68). Likewise, there were no specific indirect efect for physical abuse (Inhibition: PRODCLIN 95% CI unstandardized estimate = -0.70, 0.34; Conceptual Flexibility: -.017, 0.29; and Monitoring: -0.36, 0.89) sexual abuse (Inhibition: -0.47, 0.24; Conceptual Flexibility: -0.20, 0.11; and Monitoring: -0.54, 0.22), or combined abuse (Inhibition: < -0.01, < 0.01; Conceptual Flexibility: < -0.01, < 0.01; and Monitoring: < -0.01, < 0.01). Discusion The current study sought to improve understanding of the etiology of juvenile offending behavior by analyzing two salient themes within the delinquency and clinical literature: traumatic experiencing and executive dysfunction (ED). Specificaly, we hypothesized that executive functioning would serve as a partial mediator of trauma and delinquency?that adolescents? previous trauma histories would influence delinquency status by way of executive functioning deficits. We also predicted that the age of traumatic onset and severity of the type of trauma (i.e., sexual, combined) would be the most salient trauma characteristics, and as such, would map more closely onto EF deficits and delinquency than other characteristics of victimization (e.g., frequency, duration). In order to examine our hypotheses more specificaly and systematicaly, we evaluated five multiple-mediational models representing various trauma characteristics, including: the age of first experience, duration of victimization, frequency of experiencing, relationship to perpetrator, and type of victimization. Al models were determined to be of moderate to good fit, but there were no significant indirect pathways. That is, even though EF performance contributed ! ! ! 39! to the overal fit of the model, we found litle to no evidence of EF mediation, and instead, verified the potency of childhood trauma. The findings for each trauma characteristic are discussed in turn below, followed by remaining implications pertinent to EF. Overal, al models represented the data with moderate to good fit, indicating that we have acurately captured real clinical phenomenon. In particular, there sems to be an interesting relationship betwen specific characteristics of childhood sexual or physical trauma and delinquent offending. For example, the age of traumatic onset appears to be a predictor of sexual offending such that many JSOs first experience physical or sexual abuse prior to age 7. Additionaly, boys who were the victims of a familial/incestuous abuse are also more likely to commit later sexual misconduct. Finaly, children who experienced both physical and sexual victimization may be more likely to sexualy offend. Therefore, it appears as though some specific characteristics of trauma offer significant predictive validity and help to explain the etiology of juvenile offending. Age of First Traumatic Experience In support of our hypothesis, age of traumatic onset was a salient predictor of the type of delinquent offense behavior. In particular, there was a negative relationship betwen age of experiencing and offense group status, suggesting that JSOs experienced trauma at a younger age than NJSOs. Follow-up analyses indicate that the modal age for onset of physical or sexual victimization among JSOs was five years of age (median = 7), whereas NJSOs most frequently reported experiencing trauma in late childhood or early adolescence (after age 10.5). The potency of early traumatic experiencing is wel-documented in the literature. For example, Grabel and Knight (2009) discovered that sexual abuse during crucial developmental epochs (i.e., ages 3 to 7) predicted maladaptive, impulsive behaviors. Additionaly, Keiley and ! ! ! 40! colleagues (2001) discovered that early childhood victimization was related to more negative outcomes across a number of domains than was victimization later in life. In particular, physical victimization prior to age five was asociated with higher levels of both internalizing and externalizing symptomatology as perceived by parents and teachers, but later-victimization was asociated with elevated levels of externalizing behavior problems only. Findings of both internalizing and externalizing problems among children abused earlier in life sem to relate to JSO, at least anecdotaly. At DYS Mt. Meigs/ABSOP, JSOs are much diferent in temperament and presentation than NJSOs who were convicted of other delinquent crimes such as arson and drug possesion and among whom higher levels of externalizing dificulties sem more characteristic. However, despite potential externalizing disorders and contrary to our expectations, age of experiencing was not related to EF performance in either group of delinquents. There were no significant pathways betwen age of onset and Inhibition, Conceptual Flexibility, or Monitoring. Furthermore, EF factor scores did not impact the significant relationship betwen age of onset and delinquent behavior. Results indicate, then, that earlier traumatic experiencing may lead to juvenile offending, but offense behaviors are not directly related to EF deficits like self- regulation, atention, or otherwise maintaining goal directed behavior. Rather than exhibiting ED and its asociated behavioral or emotional control dificulties, impairments from early child abuse may manifest instead as internalized impairment such as hypersexuality, anxiety, depresion, or dificulty navigating interpersonal relationships (Farris, 2007; Grabel & Knight, 2009). Many children who experience trauma early in life show imediate and delayed interpersonal skils deficits including poor atachments, poor perspective- taking abilities, and withdrawal from social situations (Alesandri, 1991; Fonagy & Target, ! ! ! 41! 1997). Such interpersonal deficits may relate to delinquency or aggresive interpersonal violence, in particular, and should be examined more systematicaly. Duration of Traumatic Victimization The model representing the relationship betwen duration of trauma, EF, and delinquency was determined to be of good fit, but had no significant individual pathways. Therefore, it is likely that these three constructs are related, but only tangentialy or partialy. In particular, duration of trauma may be a pertinent indicator of outcome and prognosis, but it is not the most salient trauma predictor. The limited predictive power of the duration of traumatic experiencing is consistent with our hypothesis and with previous literature, which indicates that age of onset would be a more salient predictor than duration or frequency (Grabel & Knight, 2009). Despite preliminary empirical support, this lack of significant findings is somewhat contradictory to lay expectations and alternative theories which propose that longer traumatic experiencing would have more negative implications, possibly resulting from interrupted stage- specific task acquisition or typical neuropsychological development (Finkelhor, 1995; van der Kolk, 2003). Similarly, finding limited impact for the duration of traumatic experiencing is somewhat contrary to previous findings which demonstrate that longer total duration of traumatic victimization results in more adverse consequences, particularly increased internalizing psychopathology and suicidality (Farris, 2007). However, one possible explanation is because simply experiencing any trauma at al may be a potent enough event to invoke adverse consequences. Increased duration or frequency may add litle predictive validity above and beyond the initial experiential component. Frequency of Traumatic Experiencing Results indicate that the specific frequency of traumatic experiencing is not critical to ! ! ! 42! understanding juvenile delinquency or executive functioning. Instead, similar to the model representing the duration of traumatic victimization, just one traumatic experience is more impairing than none and an increased number or lapse of experiences provide litle additional predictive utility. These findings are corroborated by other studies in which an array of traumatic experiencing, including physical or sexual asault and even traumatic bereavement, were asociated with poor outcomes later in life such as subsequent psychopathology and social impairment when compared to individuals who experienced no trauma at al (Krupnick et al., 2004). Indeed, childhood exposure to abuse is linked to many deficits later in life including psychopathology, suicidal ideation, decreased academic performance, delinquency, aggresion, and substance abuse, among other (Grabel & Knight, 2009; Kendal-Tacket et al., 1993; van der Kolk, 2003; Walker et al., 2004). However, the adverse consequences of trauma may supersede specific characteristics of the traumatic episode, such that children who experience direct victimization, who witnes violence exposure, or who are subjected to other non-victimized traumas may al experience trauma-related impairment. For instance, Howard and colleagues (2012) suggest that children who have simply witnesed violence are more inclined to perpetuate abuse against others than children who were direct victims of abuse, indicating that the adverse consequences of trauma supersede specific characteristics of the episode. Similarly, Barroso and colleagues (2008) observed a number of negative outcomes such as drug use and gang afiliation among boys who were exposed to high levels of community violence. It is likely, then, that individuals who have either witnesed or experienced trauma early in life are subjected to interruptions in their typical developmental trajectory and are to later impairment. Along these lines, our results suggest that only one ! ! ! 43! traumatic event appears to be necesary to interrupt the course of development in some manner and to lead to adverse outcomes. Relationship to Perpetrator of Traumatic Experience There was a significant pathway betwen the relationship of the perpetrator to the victim and juvenile delinquency status. In particular, incestuous trauma was related to sexual misconduct. Boys who were physicaly or sexualy victimized by non-related (non-incestuous) individuals were equaly likely to sexualy offend as they were to commit other delinquent acts. As with other results from this study, no direct or indirect relationship with EF was indicated. Previous research supports these findings and suggests the particularly adverse nature of incestuous trauma. In particular, experiencing incestuous trauma is related to a trajectory of interpersonal dificulties, general conflict, and internalizing symptomatology like depresion. There is limited evidence of externalizing disorders among individuals who were victimized by a family member, and for those individuals who do exhibit problems with externalizing behaviors, the acting out tends to be sexual in nature and does not reflect general misconduct or a lack of self-regulation skils (Alexander & Anderson, 1997; Farris, 2007). Additionaly, Ulman (2007) reported that adults who experienced incestuous sexual victimization in childhood incurred more PTSD symptoms and disclosed the details of their abuse later than victims of non-relative sexual abuse, thus delaying the receipt of intervention services. Conflict about disclosing incestuous trauma is consistent with the betrayal theory of trauma in which children may forget or deny abuse in order to continue to have their emotional and survival needs met by caretakers or significant adults (Freyd, 1996). If a child fels betrayed by his caretaker or is otherwise not engaged in a succesful and satisfying relationship with the parent figure, he may atempt to have these needs met elsewhere and may do so in socialy ! ! ! 44! inappropriate ways, such as through precocious or forced sexual activity (Kerig & Becker, 2010; Yates & Prescott, 2011). Similarly, children who experience familial-perpetrated victimization may develop incongruous atachments or inappropriate models of social learning. Early atachment relationships influence an individual?s internalized conceptualization of interpersonal relationships, providing a basis for perceptions and expectations of the self and others (Cicheti & Howes, 1991). In other words, early atachment relationships lay the track for later interpersonal style and may determine whether an individual approaches relationships with trust and confidence or with an expectation to be rejected and hurt (Kerig & Becker, 2010). Furthermore, by traumatic exposure, children may also come to emulate the relational style and behaviors of their caretakers?directly recapitulating the caretaker?s aggresive or abusive tendencies and transforming the child from victim to victimizer. Previous researchers have demonstrated correlations betwen sexualy offending and early severe sexual abuse conducted by a close relative (Burton, Miler, & Shil, 2002). Findings from our study corroborate the relationship betwen familial-induced traumatic victimization and sexual offending, but do not necesarily support the social learning and victim- to-victimizer hypothesis, given that JSOs and NJSOs both experienced traumatic physical and sexual victimization. Therefore, the specific reason why some victims of childhood maltreatment later victimize others is stil to be determined and likely cannot be explained fully by atachment style, social learning theory, or ED. Research on the predictive ability of specific trauma characteristics should continue (Ryan, 2002). Type of Traumatic Victimization In this model, there are no statisticaly significant direct or indirect relationships. ! ! ! 45! However, there was an a priori correlation betwen combined-type trauma (i.e., physical and sexual victimization) and delinquent status that was no longer significant in the model. It was hypothesized that more intense trauma would result in more EF impairment and a more severe offense. The results provide preliminary support for this hypothesis, given that the a priori correlation betwen combined-type trauma and delinquency is no longer significant when controlling for ED. Therefore, it is possible that EF plays a mediating role in the relationship betwen type of trauma and type of delinquent offense. However, given the lack of a significant direct relationship with EF and no statisticaly significant indirect efects, there is only minimal evidence of mediation (Baron & Kenny, 1986). Similarly, no specific relationship with any one EF factor was identified, suggesting that analyzing ED overal may be relevant, rather than looking at specific facets of EF. Previous literature demonstrated that combined physical and sexual child abuse predicted the highest incidence of PTSD symptomatology in adult women. No diferences were found for the impairment and psychopathology of women who experienced only physical or only sexual abuse (Schaaf & McCanne, 1998). Similarly, Krupnick and colleagues (2004) identified a cumulative efect of trauma such that individuals who experienced multiple counts of sexual and physical trauma were more likely to experience PTSD, MD, and substance abuse problems than individuals with no trauma history or with only a single instance of either physical or sexual asault. While our study did not examine PTSD symptomatology directly, our preliminary analyses indicated an a priori correlation betwen combined abuse type and delinquent group membership. However, after completing the SEM analyses, there was no longer a significant relationship. Therefore, it is possible that EF mediated the relationship betwen combined abuse ! ! ! 46! and delinquency, especialy in light of the fact that individuals who experience combined-type abuse often experience more severe symptoms of PTSD and individuals with PTSD often exhibit EF deficits (Krupnick, 2004; Schaaf & McCanne, 1998; Walter et al., 2010). Therefore, future studies may wish to look at the relationship betwen trauma, EF, and various types of impairment both within and outside a delinquent population. The Relative Contribution of Executive Functioning Notably, though, while the models? results overal indicate an acurate fit to the data and an interesting efect, the depicted models may not be parsimonious. In particular, the relative contributions of EF are limited. It does not appear as though EF mediates the relationship betwen trauma and delinquency given that there were no significant indirect efects in any model tested. Thus, in light of the fact that the EF factors relate only loosely and non- significantly to al trauma characteristics and delinquency, the models may stil be acurate if we pare or eliminate many of the EF pathways and retain only the trauma variables. The relationship betwen childhood trauma and delinquent behavior is strong enough to be evaluated independently. Stil, al models tested were determined to be of good fit with the inclusion of the EF factors and item loadings. Further investigation of the relationship betwen EF and trauma or EF and delinquency may be warranted, but would need to bare in mind several important considerations about EF conceptualization and measurement. Regarding EF deficits, we are left with the question as to why there were no significant findings as previous studies have demonstrated ED among delinquent individuals and trauma survivors alike (DePrince et al., 2009; Veneziano et al., 2004). ! ! ! 47! When first considering the intersection betwen trauma and EF, there may be natural variation in EF performance within our sample of victimized youths such that individuals who experience child maltreatment may or may not exhibit ED. As indicated in previous literature, trauma is believed to impact: (a) the development and full maturation of specific brain structures at particular ages, (b) the physiologic and neuroendocrine responses, and (c) the capacity to coordinate and regulate cognition, emotion, and behavior (van der Kolk, 2003). However, these anatomical or neuropsychological may not have specific efects on EF. Instead, interruptions in typical brain development may impact any number of interconnecting neural systems in which case the impairment would need to be severe in order to impact al coordinated systems and manifest as a pure EF deficit (Dick & Overton, 2010). Therefore, trauma may impact functioning via other les physiological developmental factors such as social learning, atachment styles, acquisition of stage-specific tasks, or any number of other social-behavioral systems (Finkelhor, 1995; Schaaf & McCanne, 1998). Alternatively, the relationship betwen EF and trauma may intersect only at the crossroads of psychopathology. While individuals diagnosed with various Axis I disorders (e.g., ADHD, PTSD) often exhibit EF impairment (Grabel & Knight, 2009; Kendal-Tacket et al., 1993; Walker et al., 2004; Wind & Silvern, 1992), it is possible that EF deficits only manifest in cases of severe psychopathological and symptomatic distres. Within our sample of juvenile delinquents, many boys exhibit symptoms of ADHD and/or PTSD but may or may not carry a diagnosis. While children and adults with clinical levels of symptomatic distres-related PTSD have demonstrated EF impairment (Beers & De Belis, 2002; Walter et al., 2010), there is some evidence to suggest that children with sub-clinical levels of PTSD do not show diminished EF performance and complete tasks as acurately as children without symptoms of post-traumatic ! ! ! 48! stres (Carrion, Garret, Menon, Wems, & Reis, 2007). However, other studies suggest that al children with and without PTSD who witnesed domestic violence exhibited below average performance on tests of EF, atention, and cognitive ability (Samuleson et al., 2010). Thus, future comparative analyses of trauma and EF should sek to parse out the efect of pathological impairment, or at least should diferentiate betwen clinical and sub- or non-clinical individuals. The relationship betwen EF and trauma may also have been limited in this study by which EF factors were selected for inclusion. Factors of Inhibition, Conceptual Flexibility, and Monitoring are primarily cognitive and behavioral in nature. However, afective self-regulation skils are also considered by some theorists to fal under the umbrela of EF (Garcia-Barrera et al., 2011). Previous research indicates that individuals who experience childhood abuse often develop anxiety and mood disorders or otherwise exhibit high rates of internalizing symptomatology (Farris, 2007; Keiley et al., 2001; Wind & Silvern, 1992). The scope of this study may have been too narrow when defining EF and focused too heavily on factors related to externalizing disorders like ADHD, rather than analyzing factors which are more closely related to self-regulation and thus may be more closely related to internalizing disorders. When considering the relationship betwen EF and delinquency, previous research within this population has demonstrated inconclusive and inconsistent findings that are highly dependent on study methodology (Bergeron & Valiant, 2001; Veneziano et al., 2004). Within our adolescent sample, many boys demonstrated ED as measured by the D-KEFS. Mean EF performance was below average for both JSOs and NJSOs, but despite this, there was no statisticaly significant diference betwen delinquent groups. Of note, the adolescents? below average EF performance may also have been impacted by their below average IQ scores and future studies may wish to control for the relative contribution of intelectual ability. ! ! ! 49! Nonetheles, similarities in overal EF performance across JSOs and NJSOs are consistent with the previous literature, which describes an inconsistent main efect for delinquent group (Veneziano et al., 2004). Therefore, ED may be exhibited among al adjudicated adolescents, suggesting that al juvenile delinquents?regardles of offense type?demonstrate relative skils deficits in domains like inhibition, cognitive flexibility, monitoring, problem solving, and emotional control. EF performance may not be a function of delinquent offense type. Instead, EF performance may difer when comparing juvenile delinquents to a normative sample, when comparing aggresive to non-aggresive offenders, or when comparing first-time offenders to repeat offenders (Veneziano et al., 2004). Future researchers may wish to analyze other specific criminal profiles in order to glean more information about which delinquents exhibit EF deficits and which do not. Most likely, there is a range of EF performance within the delinquent population with some offenders demonstrating deficits in inhibition, planning, and self- regulation while other offenders have intact EF. Similarly, EF performance may be related to functional impairment and specific behavioral deficits, not necesarily to the characteristics of the adjudicated offense. For example, EF performance may be related to PTSD symptomatology and aggresion, which may influence?but not cause?risk-taking or delinquent behavior (Grabel & Knight, 2009). Considerations of the true domains of impairment afected by ED are reflective of isues with ecological validity. Many EF tasks are plagued by poor ecological validity such that there is incongruence betwen the task-measured performance and the construct that the task is intended to represent. For example, the Trail Making Test (Condition 4) is designed to ases for the higher-order EF task of cognitive flexibility/set-shifting (i.e., set-shifting). However, the task of ! ! ! 50! drawing a line betwen numbers and leters is not one that the typical individual incurs in the real world and is, instead, a novel task contained within the asesment context. A more ecologicaly valid test of cognitive flexibility/set-shifting would require an individual to alter betwen various real-world rule sets, such as in a school seting when children must know when it is appropriate to talk aloud and when it is not. Rating scales such as the BRIEF (Gioia et al., 2000) or the new Delis Rating of Executive Function (D-REF; Delis, 2012) can also provide additional ecological validity by having multiple raters evaluate real-time behavior in various contexts. Other isues in the measurement and conceptualization of EF also may have contributed to the lack of identifiable discrepancies in performance (Bernstein & Waber, 2007). For example, the tasks measured by the D-KEFS may not be appropriate for identifying specific EF deficits in an adolescent sample. While EF appears to emerge in early childhood, executive skils are stil being refined and improved throughout adolescence (Best & Miler, 2010). In particular, adolescents have demonstrated linear improvement with age on tasks measuring advanced cognitive proceses such as selective atention and problem solving, but demonstrated stable performance on planning tasks like the Tower of London (Blakemore & Choudhury, 2006). To elaborate, EF is often conceptualized and measured as a cognitive proces construct in which both fundamental cognitive skils and higher-order cognitive functions are represented (Delis et al., 2001). Some theorists purport that the basic cognitive components and EF skils (e.g., WM) must be intact before an individual can develop and utilize more advanced neurocognitive skils (e.g., set-shifting; Best & Miler, 2010). Thus, by measuring and combining scores on tasks that require both specific and advanced cognitive proceses, the D-KEFS and other EF tests might be insensitive to the unique developmental paterns of EF and have limitations in specificity or other age-based invariance ! ! ! 51! (Goldberg et al., 2005; Latzman & Markon, 2010). More specificaly, age related diferences might be confounded with task impurity. One test may measure a host of interrelated abilities, not just one specific executive component even when tasks?such as those comprising the D- KEFS?were designed with the intention of parsing out fundamental and higher-order proceses (Dick & Overton, 2010; Delis et al., 2001; Strauss et al., 2006). Using D-KEFS contrast scores may be indicated to beter acount for more basic skil deficits; however, contrast scores are acompanied by additional psychometric shortcomings. Therefore, given limitations in task impurity, there is no guarante that D-KEFS performance captured each adolescent?s true EF abilities. Performance on tasks of EF requires the coordination of many cognitive proceses, not just executive skils (Dick & Overton, 2010). For example, Strauss and colleagues (2006) explain that unsuccesful performance on card sorting tasks such as the WCST or the D-KEFS Sorting Test could be the result of poor working memory, inadequate set-shifting abilities, an inability to proces and incorporate fedback, visual procesing deficits, or low motivation, and future research should sek to control for these more basic proceses. The relative contribution of motivation (Strauss et al., 2006) is a point wel-taken as the imediate environment and stresors asociated with recent incarceration may have precluded acurate, motivated performance on testing within our sample. At a biological level, pathways in the PFC are shared betwen tasks such as executive functioning and motivation, thus integrating the two proceses. Increased motivation yields increased neural activity and improved performance. Motivation, therefore, can impact performance via behavioral engagement, but can also impact performance on a neurophysiological level (Taylor, Welsh, Wager, Phan, Fitzgerald, & Gehring, 2004). Thus, EF scores captured in this study may represent a restricted range or ! ! ! 52! limited variability in true performance abilities due to individual diferences in motivation to perform to the best of their abilities. Furthermore, sensitivity to ED may be additionaly blurred by the statistical properties of SEM. For one, the creation of latent factors can be misleading (Dick & Overton, 2010) as the provision of a concrete behavioral label can cause consumers to over-atribute neurocognitive proceses to behavior, as wel as to lose sight of isues with task impurity and ecological validity within the EF construct. Similarly, there is the possibility that the labels asigned to the established factor structure are misrepresentations of the measure?s true EF component (Dick & Overton, 2010) and may have led to premature conclusions about the adolescents? inhibition, conceptual flexibility, and monitoring abilities. Another limitation with using latent factors in EF research is that interpreters of the D-KEFS lose some sensitivity in their ability to understand the specific cognitive procesing deficits that contribute to ED. With the establishment of latent factors, the opportunity to analyze the unique contributions of fundamental and higher-order test conditions can be lost to the correlation and shared variance acounted for by the latent factor (Kline, 2011). Finaly, the factor structure of the D-KEFS utilized in this study was drafted after the development of the batery and thus does not represent strong, empiricaly validated theory. EF theorists suggest that research using factor analysis should be handled with caution as it may inflate the confidence we have in the notion of distinct but related constructs under the umbrela of EF (Bernstein & Waber, 2007). Therefore, the three-factor model (Latzman & Markon, 2010) utilized in this study may have been an inappropriate representation of the underlying constructs especialy in light of the high correlations betwen the EF factors of Inhibition, Conceptual Flexibility, and Monitoring. An alternative factor structure (e.g., unitary factor) may be ! ! ! 53! warranted and further confirmation of the appropriate factor structure for the D-KEFS should be sought. Alternatively, future researchers may wish to conduct analyses utilizing a more focused application of the D-KEFS variables and contrast scores. Clinical Aplication Beyond considerations of valid EF asesment, this study contributes to the literature regarding the best practices for treating children and adolescents who sexualy offend against others (ACAP, 1999; Burkhart & Cook, 2010). In particular, our findings emphasize the importance of using a developmental framework and prioritizing individual needs in the treatment of offenders, and fully support the decision to reject ?one size fits al? treatment protocols for delinquents (Wormith et al., 2007). Given that those individuals with an early trauma history are at particular risk for developing personal psychological distres, even within the corrections system, treatment should be individualized and trauma-focused for those who exhibit PTSD or should be trauma-informed for individuals with no current clinical PTSD symptomatology. Studies of evidence-based treatment for childhood maltreatment have often failed to analyze the eficacy and applicability of these interventions for adolescent clientele and for youths in the juvenile justice system (Mahoney, Ford, Ko, & Siegried, 2004; Saunders, Berliner, & Hanson, 2003). On the other hand, programs designed for use within the juvenile justice system such as the Multisytemic Therapy (MST; Henggeler, Schoenwald, Borduin, Rowland, & Cunningham, 1998)?which relies les on core cognitive principles and includes more family-based, behavioral interventions?have shown relatively good results within the penal system to help addres sub-clinical PTSD, self- regulation, and behavioral health with lower recidivism rates across time than was observed ! ! ! 54! among those boys who received standard community care (Borduin, Schaefer, & Heiblum, 2009; Mahoney et al., 2004; Timons-Mitchel, Bender, Kishna, & Mitchel, 2006). Ultimately, more research should be conducted on the specific treatment modalities that counteract some of the adverse consequences of child maltreatment, especialy within the corrections system and unique population of juvenile delinquents. Similarly, the mechanisms by which trauma leads to undesired outcomes should be examined further in order to identify individuals who may be at particular risk and to help interrupt the pathway to delinquency, PTSD, or any other negative consequence. As identified in this study, early victimization, familial-based trauma, and combined physical/sexual abuse may be particularly potent. Therefore, clinicians should strive to help children and adolescents describe the nature and characteristics of their abuse, including their relationship with the perpetrator (Ryan, 2002), because it is clear that for those individuals who experience sexual and/or physical maltreatment at a young age, early intervention is necesary to help protect against a host of negative outcomes?including delinquency and risk of incarceration. Additionaly, despite the fact that pervasive group diferences in EF performance were not observed in our study, clinicians and other mental health profesionals should continue to consider the role of executive functioning for those individuals who demonstrate significantly weakened executive skils. While there may be no identifiable trends within the population, the relationship betwen delinquency and executive dysfunction should then be considered on a case-by-case basis to identify those individuals who may benefit from EF skils training. Treatment focusing on EF might highlight the importance of seting goals, identifying pathways to goal achievement, planning for barriers and roadblocks, and understanding the consequences of actions as they relate to long-term goals and values (Dawson & Guare, 2010). However, ! ! ! 55! behavior health and executive skils training should likely be secondary treatment goals relative to addresing more salient pathology and developmental concerns. Limitations and Future Directions While interpreting our results, several limitations warrant consideration and future atention. Primarily, several limitations were identified as they related to study design and statistical analysis including: smal sample size, limited power, reliance on self-report of information, reliance on performance based measures of EF exclusively, and lack of a non- delinquent control group. Idealy, future studies seking to examine the relationships betwen trauma, EF, and delinquency wil use larger samples with more equaly distributed group membership and wil also include non-delinquent peers. A larger sample size would have yielded greater power and model flexibility and thus may have been more likely to identify true significant efects. Future studies should also be designed to cross-validate self-report of trauma and EF with multiple informants, including parents, teachers, school records, and records on file with child and family protective services. Cross-validation is also necesary for the final overal model to determine the appropriatenes of modifications made to the model throughout data analysis. Furthermore, the final models are quite complex, estimating many possible pathways. Researchers should strive to achieve parsimony in their designated models. In our analyses, several residual covariances indicated over- or under-estimation of specific pathways; these residuals should be examined more closely and pathways should be trimed appropriately in order to develop a more parsimonious model. Subsequent models should, of course, be cross-validated in independent samples. ! ! ! 56! Furthermore, model complexity is problematic given the vast possibility of equivalent models. For example, in the current study, correlations betwen latent factors could be replaced with direct efects, indicating that one EF construct caused another. This may be consistent with some researchers? theories of EF such that basic cognitive components and EF skils (e.g., WM) must be intact before an individual can develop and utilize more advanced neurocognitive skils (e.g., set-shifting; Best & Miler, 2010). In our model, Monitoring (i.e., Updating Working Memory) may predict Inhibition, which may in turn predict Conceptual Flexibility. While such a change would alter the model specification, theory and supporting literature, it would result in a mathematicaly equivalent model. Many other possible and plausible equivalent models may also exist. To avoid such a problem in the future, models should be generated following longitudinal design and data collection to ases for causal relationships. It is possible that a diferent EF model and D-KEFS factor structure would have been more sensitive to identifying specific EF deficits. As discussed previously, the field of EF theory and measurement is controversial. Given that no gold standard has been identified, researchers have some latitude and should strive to have breadth and depth when measuring EF, including both performance and behavioral rating scales. Additionaly, researchers should provide a strong justification for their selected EF measurement model. Future research should also explore various definitions of delinquency. In our study, delinquency was specified as a binary variable depending on whether the adolescent was adjudicated for a sexual or non-sexual offense. Unique relationships betwen trauma, EF, and other characteristics of delinquent offending may exist such as acts of violence or aggresion. Ultimately, researchers should continue to examine the pathways that intercede the relationship betwen a history of victimization and juvenile delinquency. Researchers should ! ! ! 57! examine a variety of proximal, distal, physiological, behavioral, and environmental variables (e.g., social cognition, IQ, atachment, psychopathology) in order to determine their relative protective or detrimental efects on the social and psychological wel-being of children and adolescents who experience maltreatment. By understanding the long-term efects of childhood trauma and the etiology of delinquent behavior, practitioners wil be beter able to focus and tailor interventions for specific at-risk children and populations. ! ! ! 58! References Alesandri, S.M. (1991). Play and social behavior in maltreated preschoolers. Development and Psychopathology, 3, 191-205. Alexander, P.C., & Anderson, C.L. (1997). Incest, atachment, and developmental psychopathology. In D. Cicheti & S. Toth (Eds.), Rochester symposium on developmental psychopathology: Vol 8. Developmental perspectives on trauma: Theory, research, and intervention (pp. 343-377). Rochester, NY: University of Rochester Pres. American Academy of Child and Adolescent Psychiatry (AACAP; 1999). Practice parameters for the asesment and treatment of children and adolescents who are sexualy abusive of others. Journal of American Academy of Child and Adolescent Psychiatry, 38(12 Supplement), 55-76. Appelof, E. S., & Augustine, E. A. (1985). Prefrontal functions in juvenile delinquents. Journal of Clinical and Experimental Neuropsychology, 7, 79-109. Barkley, R. A. (1997a). Behavioral inhibition, sustained atention, and executive functions: Constructing a unifying theory of ADHD. Psychological Bulletin, 121(1), 65-94. Barkley, R. A. (1997b). ADHD and the nature of self-control. New York: Guilford Pres. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182. Barroso, C. S., Peters, R. J., Kelder, S., Conroy, J., Murray, N., & Orpinas, P. (2008). Youth exposure to community violence: Asociation with aggresion, victimization, and risk ! ! ! 59! behaviors. Journal of Aggresion, Maltreatment, and Trauma, 17, 141-155. Becker, S. P., & Kerig, P. K. (2011). Postraumatic Stres symptoms are asociated with the frequency and severity of delinquency among detained boys. Journal of Clinical Child & Adolescent Psychology, 40, 765-771. Beers, S. R., & De Belis, M. D. (2002). Neuropsychological function in children with maltreatment-related Postraumatic Stres Disorder. American Journal of Psychiatry, 159, 483-486. Bergeron, T. K. & Valiant, P. M. (2001). Executive function and personality in adolescent and adult offenders vs. non-offenders. Journal of Ofender Rehabilitation, 33, 27-45. Bernstein, J. H., & Waber, D. P. (2007). Executive capacities from a developmental perspective. In L. Meltzer (Ed.), Executive function in education: From theory to practice (pp. 39-54). New York: Guilford Pres. Best, J. R. & Miler, P. H. (2010) Developmental perspective on executive function. Child Development, 81, 1641-1660. Blakemore, S. J., & Choudhury, S. (2006). Development of the adolescent brain: Implications for executive function and social cognition. Journal of Child Psychology and Psychiatry, 47, 296-312. Borduin, C. M., Schaefer, C. M., & Heiblum, N. (2009). A randomized clinical trial of multisystemic therapy with juvenile sexual offenders: Efects on youth social ecology and criminal activity. Journal of Consulting and Clinical Psychology, 77, 26-37. Brocki, K. C. & Bohlin, G. (2004). Executive functions in children aged 6 to 13: A dimensional and developmental study. Developmental Neuropsychology, 26, 571-593. Brown, T. A. (2006). Confirmatory factor analysis for applied research. New York: Guilford ! ! ! 60! Pres. Burkhart, B. B. & Cook, P. K. (2010). Adolescent sexual offending. In J. C. Thomas & M. Hersen (Eds.), Handbook of clinical psychology competencies (pp. 1541-1562). New York: Springer. Burton, D. L., Miler, D. L., & Shil, C. T. (2002). A social learning theory comparison of the sexual victimization of adolescent sexual offenders and nonsexual offending male delinquents. Child Abuse and Neglect, 26, 893-907. Carrion, V. G., Garret, A., Menon, V., Wems, C. F., & Reis, A. L. (2008). Postraumatic stres symptoms and brain function during a response-inhibition task: An fMRI study in youth. Depresion and Anxiety, 25, 514-526. Cicheti, D., & Howes, P. W. (1991). Developmental psychopathology in the context of the family: Illustrations from the study of child maltreatment. Canadian Journal of Behavioural Science, 23, 257-281. Curran, P.J., West, S.G., & Finch, J. (1996). The robustnes of test statistics to non-normality and specification error in confirmatory factor analysis. Psychological Methods, 1, 16-29. Dawson, P., & Guare, R. Executive skils in children and adolescents: A practical guide to assesment and interventions (2nd ed.). New York, NY: Guilford Pres. Delis, D. C. (2012). Delis-Rating of Executive Function (D-REF). San Antonio, TX: Pearson Education, Inc. Delis, D. C., Kaplan, E., & Kramer J. H. (2001). Delis-Kaplan Executive Function System (D- KEFS). San Antonio, TX: Pearson Education, Inc. DePrince, A. D., Weinzierl, K. M., & Combs, M. D. (2009). Executive function performance and trauma exposure in a community sample of children. Child Abuse & Neglect, 33, ! ! ! 61! 353-361. Dick, A. S., & Overton, W. F. (2010). Executive function: Description and explanation. In B. Sokol, U. Muller, J. Carpendale, A. Young, and G. Iarocci (Eds.), Self and social regulation: Social interaction and the development of social understanding and executive functions (pp. 7-34). New York: Oxford University Pres. Donnelan, M. B., Ge, X., & Wenk, E. (2000). Cognitive abilities in adolescent-limited and life- course persistent criminal offenders. Journal of Abnormal Psychology, 109, 396-402. Enders, C. K. (2010). Applied mising data analysis. New York, NY: Guilford Pres. Farris, K. L. (2007). Developmental perspective on juvenile sex offenders: The role of childhood abuse in the development of psychopathology and sex offending behavior. (Unpublished doctoral disertation). Auburn University, Alabama. Finkelhor, D. (1995). The victimization of children: A developmental perspective. American Journal of Orthopsychiatry, 65, 177-193. Flora, D. B., & Curran, P. J. (2004). An empirical evaluation of alternative methods of estimation for confirmatory factor analysis with ordinal data. Psychological Methods, 9, 466-491. Fonagy, P., & Target, M. (1997). Atachment and reflective function: Their role in self- organization. Development and Psychopathology, 9, 679-700. Freyd, J. J. (1996). Betrayal trauma: The logic of forgeting childhood abuse. Cambridge, MA: Harvard University Pres. Garcia-Barrera, M. A., Kamphaus, R. W., & Bandalos, D. (2011). Theoretical and statistical derivation of a screener for the behavioral asesment of executive functions in children. Psychological Asesment, 23(1), 64-79. ! ! ! 62! Gioia, G. A., Isquith, P. K., Guy, S. C., & Kenworthy, L. (2000). The Behavioral Rating Inventory of Executive Function profesional manual. Odesa, FL: Psychological Asesment Resources. Gioia, G. G., Isquith, P. K., Retzlaf, P. D., & Epsy, K. A. (2002). Confirmatory factor analysis of the Behavior Rating Inventory of Executive Function (BRIEF) in a clinical sample. Child Neuropsychology, 8, 249-257. Goldberg, M. C., Mostofsky, S. H., Cutting, L. E., Mahone, E. M., Astor, B. C., Denckla, M. B., & Landa, R. J. (2005). Subtle executive impairment in children with autism and children with ADHD. Journal of Autism and Developmental Disorders, 35, 279-293. Grabel, A. S., & Knight, R. A. (2009). Examining childhood abuse paterns and sensitive periods in juvenile sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 21, 208-222. Henggeler, S. W., Schoenwald, S. K., Borduin, C. M., Rowland, M. D., & Cunningham, P. B. (1998). Multisystemic treatment of antisocial behavior in children and adolescents. New York, NY: Guilford Pres. Howard, A. L., Kimonis, E. R., Munoz, L. C., & Frick, P. J. (2012). Violence exposure mediates the relation betwen calous-unemotional traits and offending paterns in adolescents. Journal of Abnormal Child Psychology, 40, 1237-1247. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55. IBM Corp. (Released 2010). IBM SPSS Statistics for Windows, Version 19.0. Armonk, NY: IBM Corp. Keiley, M. K., Howe, T. R., Dodge, K. A., Bates, J. E., & Petit, G. S. (2001). The timing of ! ! ! 63! child physical maltreatment: A cross-domain growth analysis of impact on adolescent externalizing and internalizing problems. Development and Psychopathology, 13, 891- 912. Kely, T., Richardson, G., Hunter, R., & Knapp, M. (2002). Atention and executive function deficits in adolescent sex offenders. Child Neuropsychology, 8, 138-143. Kendal-Tacket, K. A., Wiliams, L. M., & Finkelhor, D. (1993). Impact of sexual abuse on children: A review and synthesis of recent empirical studies. Psychological Bulletin, 113, 164-180. Kennedy, T. D., Burnet, K. F., & Edmonds, W. A. (2011). Intelectual, behavioral, and personality correlates of violent vs. non-violent juvenile offenders. Aggresive Behavior, 37, 1-11. Kerig, P. K., & Becker, S. P. (2010). From internalizing to externalizing: Theoretical models of the proces linking PTSD to juvenile delinquency. In S. J. Egan (Ed.), Post-traumatic stres disorder (PTSD): Causes, symptoms, and treatment (pp. 33-78). Hauppauge, NY: Nova Science Publishers. Kline, R. B. (2011). Principles and practice of structural equation modeling (3rd ed.). New York: Guilford Pres. Krupnick, J. L., Green, B. L., Stockton, P., Goodman, L., Corcoran, C., & Pety, R. (2004). Mental health efects of adolescent trauma exposure in a female college sample: Exploring diferential outcomes based on experiences of unique trauma types and dimensions. Psychiatry, 67, 264-279. Latzman, R. D., & Markon, K. E. (2010). The factor structure and age-related factorial invariance of the Delis-Kaplan Executive Function System (D-KEFS). Asesment, 17, ! ! ! 64! 172-184. Lewis, D. O., Pincus, J. H., Bard, B., & Richardson, E. (1986). Neuropsychiatric, psychoeducational, and family characteristics of 14 juveniles condemned to death in the United States. American Journal of Psychiatry, 145, 584-589. MacCalum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis and ! determination of sample size for covariance structure modeling. Psychological Methods, ! 1, 130-149.! MacKinnon, D. P., Fritz, M. S., Wiliams, J., & Lockwood, C. M. (2007). A comparison of methods to test mediation and other intervening variable efects. Behavior Research Methods, 39, 384-389. Mahoney, K., Ford, J. D., Ko, S. J., & Siegried, C. B. (2004). Trauma-focused interventions for youth in the juvenile justice system. National Child Traumatic Stres Network. McCafrey, R. J., Lynch, J. K., & Westervelt, H. J. (2011). Clinical neuropsychology. In D. H. Barlow (Ed.), The Oxford handbook of clinical psychology (pp. 680-700). New York: Oxford University Pres. Mendola, P., Selevan, S. G. Gutter, S., & Rice, D. (2002). Environmental factors asociated with a spectrum of neurodevelopmental deficits. Mental Retardation and Developmental Disabilities Research Reviews, 8, 188-197. Moffit, T. E. (1990). The neuropsychology of juvenile delinquency: A critical review. In M. Tonry and N. Morris (Eds.), Crime and justice: A review of research (pp. 99-169). Chicago: University of Chicago Pres. Moffit, T. E., & Henry, B. (1989). Neuropsychological asesment of executive functions in self- reported delinquents. Development & Psychopathology, 1, 105-118. ! ! ! 65! Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T. D. (2000). The unity and diversity of executive functions and their contributions to complex ?frontal lobe? tasks: A latent variable analysis. Cognitive Psychology, 41, 49?100. Muth?n, L. K., & Muth?n, B. O. (2011). Mplus User's Guide (6th ed.). Los Angeles, CA: Muth?n & Muth?n. Ozonoff, S. (2001). Advances in the cognitive neuroscience of Autism. In C. A. Nelson & M. Luciana (Eds.), Handbook of developmental cognitive neuroscience, (pp. 537-548). Cambridge, MA: MIT Pres. O?Brien, N., Langhinrichsen-Rohling, J., & Sheley-Tremblay, J. (2007). Reading problems, atentional deficits, and current mental health status in adjudicated adolescent males. The Journal of Correctional Education, 58, 293-315. Palone, N. J., & Voelbel, G. T. (1998). Limbic system dysfunction and inventoried psychopathology among incarcerated pedophiles. Current Psychology, 17, 57-74. Pennington, B. F., & Ozonoff, S. (1996). Executive functions and developmental psychopathology. Journal of Child Psychology and Psychiatry, 37, 51-87. Roberts, R. J., & Pennington, B. F. (1996). An interactive framework for examining prefrontal cognitive proceses. Developmental Neuropsychology, 12(1), 105-126. Ryan, G. (2002). Victims who go on to victimize others: No simple explanations. Child Abuse and Neglect, 26, 891-892. Samuelson, K. W., Krueger, C. E., Burnet, C., & Wilson, C. K. (2010). Neuropsychological functioning in children with Postraumatic Stres Disorder. Child Neuropsychology, 16, 119-133. Santila, P., & Hapasalo, J. (1997). Neurological and psychological risk factors among young ! ! ! 66! homicidal, violent and non-violent offenders in Finland. Homicide, 1, 1-20. Satler, J. M. (2008). Asesment of children: Cognitive foundations (5th ed.). San Diego: Jerome M. Satler, Publisher, Inc. Saunders, B.E., Berliner, L., & Hanson, R.F. (Eds.). (2003). Child Physical and Sexual Abuse: Guidelines for Treatment (Final Report: January 15, 2003). Charleston, SC: National Crime Victims Research and Treatment Center. Schaaf, K. K., & McCanne, T. R. (1998). Relationship of childhood sexual, physical, and combined sexual and physical abuse to adult victimization and postraumatic stres disorder. Child Abuse and Neglect, 22, 1119-1133. Seguin, J. R., Boulerice, B., Harden, P. W., Tremblay, R. C., & Pihl, R. O. (1999). Executive functions and physical aggresion after controlling for Atention Deficit Hyperactivity Disorder, general memory, and IQ. Journal of Child Psychology and Psychiatry, 40, 1197-1208. Sergeant, J. A., Geurts, H., & Oosterlan, J. (2002). How specific is a deficit of executive functioning for Atention-Deficit/Hyperactivity Disorder? Behavioural Brain Research, 130, 3-28. Skoff, B. F, & Libon, J. (1987). Impaired executive functions in a sample of male juvenile delinquents. Journal of Clinical and Experimental Neuropsychology, 9, 60. Strauss, E., Sherman, E. M. S., & Spreen, O. (2006). A compendium of neuropsychological tests (3rd ed.). New York: Oxford University Pres. Taylor, S. F., Welsh, R. C., Wager, T. D., Phan, K. L., Fitzgerald, K. D., & Gehring, W. J. (2004). A functional neuroimagining study of motivation and executive function. NeuroImage, 21, 1045-1054. ! ! ! 67! Timons-Mitchel, J., Bender, M. B., Kishna, M. A., & Mitchel, C. C. (2006). An independent efectivenes trial of multisystemic therapy with juvenile justice youth. Journal of Clinical Child and Adolescent Psychology, 35, 227-236. Ulman, S. (2007). Relationship to perpetrator, disclosure, social reactions, and PTSD symptoms in child sexual abuse survivors. Journal of Child Sexual Abuse, 16, 19-36. van der Kolk, B. A. (2003). The neurobiology of childhood trauma and abuse. Child and Adolescent Psychiatric Clinics of North America, 12, 293-317. Veneziano, C., Veneziano, L., LeGrand, S., & Richards, L. (2004). Neuropsychological executive functions of adolescent sex offenders and nonsex offenders. Perceptual and Motor Skils, 98, 661-674. Walker, J. L., Carey, P. D., Mohr, N., Stein, D. J., & Sedat, S. (2004). Gender diferences in the prevalence of childhood sexual abuse and in the development of pediatric PTSD. Archives of Women?s Mental Health, 7, 111-121. Walter, K. H., Palmieri, P. A., & Gunstad, J. (2010). More than symptom reduction: Changes in executive function over the course of PTSD treatment. Journal of Traumatic Stres, 23, 292-295. Wechsler, D. (1999). Wechsler Abbreviated Scale of Inteligence (WASI). San Antonio, TX: Harcourt Asesment. Weis, R. (2008). Introduction to abnormal child and adolescent psychology. Thousand Oaks, CA: Sage Publications. Wind, T. W. & Silvern, L. (1992). Type and extent of child abuse as predictors of adult functioning. Journal of Family Violence, 7, 261-281. Wilcutt, E. G., Doyle, A. E., Nigg, J. T., Faraone, S. V., & Pennington, B. F. (2005). Validity of ! ! ! 68! the executive theory of atention-deficit/hyperactivity disorder: A meta-analytic review. Biological Psychology, 57, 1336-1346. Wormwith, J. S., Althouse, R., Simpson, M., Reitzel, L. R., Fagan, T. J., & Morgan, R. D. (2007). Rehabilitation and reintegration of offenders: The current landscape and some future directions for correctional psychology. Criminal Justice and Behavior, 34, 879- 892. Yates, P. M., & Prescott, D. (2011). Building a beter life: A good lives and self-regulation workbook. Brandon, VT: Safer Society Pres. Yu, C. Y. (2002). Evaluating cutoff criteria of model fit indices for latent variable models with ! binary and continuous outcomes. (Unpublished doctoral disertation). University of California, Los Angeles.! Zelazo, P. D., Muller, U., Frye, D., & Marcovitch, S. (2003). The development of executive function in early childhood. Monographs of the Society for Research in Child Development, 68(3), Serial No. 274. ! ! ! 69! Table 1 Psychometric Properties of Select D-KEFS Variables Note. *No information provided in D-KEFS Technical Manual regarding the Internal Consistency for indicated items due to item- interdependences and ability for examinees to adjust their performance acording to fedback and rehearsal on previous components. ! D-KEFS Test/Variable Type of Measure & Latent Variable Label Internal Consistency for Ages 8-19 Test-Retest r 12 for Ages 8-19 Trail Making Test Condition 4: Number-Leter Switching Inhibition Low (! .59) Low (.20) Verbal Fluency Leter Fluency Monitoring High (.68-.81) Marginal (.67) Category Fluency onitoring Marginal (.53-.75) Adequate (.70) Category Switching Total Monitoring arginal (0.53-.76) Marginal (.65) Category Switching Acuracy onitoring Low (.37-.62) Low (.53) Color Word Interference Condition 3: Inhibition Inhibition * Very High (.90) Condition 4: Inhibition/Switching Inhibition * High (.80) Sorting Test Category 1: Free Sorting Conceptual Flexibility Adequate (.55-.82) Low (.49) Category 2: Free Sorting Description Conceptual Flexibility Adequate (.55-.80) Marginal (.67) Category 3: Sort Recognition Conceptual Flexibility Adequate (.62-.74) Low (.56) ! ! ! 70! Table 2.1 Correlation Matrix for D-KEFS Variables as a Function of Sample Note. Correlations for victimized subsample of participants (n = 92) are presented above the diagonal, and correlations for the full sample of juvenile delinquents (n = 188) are presented below the diagonal. TMT_CT4 = Trail Making Test, Condition 4: Number- Leter Switching; VF_leter = Verbal Fluency, Leter Fluency; VF_catflu = Verbal Fluency, Category Fluency; VF_catsco = Verbal Fluency, Category Switching Total; VF_catswiac = Verbal Fluency, Category Switching Acuracy; CWIT3 = Color-Word Interference, Condition 3: Inhibition; CWIT4 = Color-Word Interference, Condition 4: Inhibition/Switching; Sort1a = Sorting Test, Category 1a: Free Sorting; Sort1b = Sorting Test, Category 1b: Free Sorting Description; Sort2 = Sorting Test, Category 2: Sort Recognition. *p < .05, **p < .01 Subtest 1 2 3 4 5 6 7 8 9 10 1. TMT_CT4 ? .24* .34** .26* .34** .39** .34** .37** .38** .43** 2. VF_leter .24** ? .55** .30** .27** .32** .20 .32** .33** .34** 3. VF_catflu .31** .49** ? .44** .35** .25* .28** .38** .39** .40** 4. VF_catsco .23** .31** .42** ? .86** .25* .20 .18 .22* .25* 5. VF_catswiac .25** .25** .30** .87** ? .31** .23* .30** .34** .37** 6. CWIT3 .36** .23** .19** .26** .29** ? .55** .23* .21* .21 7. CWIT4 .30** .11** .15* .20** .25** .53** ? .17 .19 .24* 8. Sort1a .34** .23** .33** .22** .29** .25** .18* ? .96** .69** 9. Sort1b .34** .27** .35** .25** .32** .24** .20** .95** ? .74** 10. Sort2 .33** .24** .35** .26** .32** .25** .20** .72** .77** ? ! ! ! 71! Table 2.2 Correlation Matrix for Delinquent Ofending and Trauma Variables Note. Correlations for victimized subsample of participants (n = 92) are presented above the diagonal, and correlations for the full sample of juvenile delinquents (n = 188) are presented below the diagonal. FreqOne = Frequency of Traumatic Experience, Once; FreqTwo = Frequency of Traumatic Experience, Twice or More. *p < .05, **p < .01 Characteristic 1 2 3 4 5 6 7 8 9 10 11 1. JSO vs. NJSO ? -.41 -.31** -.00 -.00 -.01 .19 -.19 -.13 -.02 .14 2. Age at Intake -.38 ? .08 -.03 .03 .04 -.05 .05 .10 -.04 -.06 3. Age First Trauma .09 .02 ? .34** -.34** -.36** -.52** .52** -.06 .41** -.40** 4. FreqOne .17* -.03 .70** ? -.27** -.36** .36** .20* .57** -.82** 5. FreqTwo .12 .01 .22** -.31** ? .27** .36** -.36** -.20* -.57** .82** 6. Duration of Trauma .10 .02 .15* .10 .41** ? .37** -.37** .21* -.38** .21* 7. Incestuous .27** -.04 .39** .31** .56** .53** ? .23* -.53** .36** 8. Non-Incestuous -.01 .02 .64** .52** -.07 -.13 -.30** ? -.23* .53** -.36** 9. Type Physical .03 .06 .29** .40** -.03 .34** .43** -.07 ? -.54** -.37** 10. Type Sexual .13 -.04 .65** .72** -.15* -.06 .09 .65** -.21** ? -.58** 11. Type Combo .18* -.05 .12 -.27** .85** .35** .52** -.12 -.15* -.22** ? *p < .05, **p < .01! ! ! ! 72! Table 2.3 Correlation Matrix for D-KEFS x Ofending and Trauma Variables for Full Sample Variable JSO vs. NJSO Age at Intake Age First Trauma FreqOne FreqTwo Duration of Trauma Incestuous Non- Incestuous Type Physical Type Sexual Type Combo TMT_CT4 .09 .03 .07 -.08 .19** .10 .08 -.00 .04 -.10 .17* VF_leter -.02 .00 .03 -.06 .07 .00 .00 .01 .08 -.12 .07 VF_catflu .04 -.01 -.04 -.20** .20** .06 .03 -.10 .07 -.21** .14 VF_catsco .07 -.05 .00 -.08 .01 -.02 -.09 .04 .00 -.07 -.02 VF_catswiac .04 .01 .01 -.07 .00 .03 -.04 -.00 .04 -.06 -.05 CWIT3 -.06 .12 -.00 .02 -.03 .07 -.04 .06 .10 -.04 -.05 IT4 -.07 .11 .03 -.02 -.02 .01 -.09 .09 .06 -.01 -.08 Sort1a .18* -.07 -.04 -.05 .02 -.06 -.01 -.01 .10 -.12 .02 Sort1b .14 -.04 -.03 -.08 .06 -.10 -.01 -.01 .05 -.12 .04 Sort2 .09 -.02 -.02 -.05 .05 -.06 .02 -.02 .03 -.06 .04 Note. TMT_CT4 = Trail Making Test, Condition 4: Number-Leter Switching; VF_leter = Verbal Fluency, Leter Fluency; VF_catflu = Verbal Fluency, Category Fluency; VF_catsco = Verbal Fluency, Category Switching Total; VF_catswiac = Verbal Fluency, Category Switching Acuracy; CWIT3 = Color-Word Interference, Condition 3: Inhibition; CWIT4 = Color-Word Interference, Condition 4: Inhibition/Switching; Sort1a = Sorting Test, Category 1a: Free Sorting; Sort1b = Sorting Test, Category 1b: Free Sorting Description; Sort2 = Sorting Test, Category 2: Sort Recognition. FreqOne = Frequency of Traumatic Experience, Once; FreqTwo = Frequency of Traumatic Experience, Twice or More. *p < .05, **p < .01 ! ! ! 73! Table 2.4 Correlation Matrix for D-KEFS x Ofending and Trauma Variables for Subsample. Note. TMT_CT4 = Trail Making Test, Condition 4: Number-Leter Switching; VF_leter = Verbal Fluency, Leter Fluency; VF_catflu = Verbal Fluency, Category Fluency; VF_catsco = Verbal Fluency, Category Switching Total; VF_catswiac = Verbal Fluency, Category Switching Acuracy; CWIT3 = Color-Word Interference, Condition 3: Inhibition; CWIT4 = Color-Word Interference, Condition 4: Inhibition/Switching; Sort1a = Sorting Test, Category 1a: Free Sorting; Sort1b = Sorting Test, Category 1b: Free Sorting Description; Sort2 = Sorting Test, Category 2: Sort Recognition. FreqOne = Frequency of Traumatic Experience, Once; FreqTwo = Frequency of Traumatic Experience, Twice or More. *p < .05, **p < .01 Variable JSO vs. NJSO Age at Intake Age First Trauma FreqOne FreqTwo Duration of Trauma Incestuous Non- Incestuous Type Physical Type Sexual Type Combo TMT_CT4 .09 .03 .01 -.25* .25* .10 .04 -.04 .01 -.20 .21* VF_leter -.02 .00 .09 -.12 .12 .01 .01 -.01 .12 -.21* .11 VF_catflu .04 -.01 -.02 -.31** .31** .11 .10 -.10 .11 -.29** .22* VF_catsco .07 -.05 .14 -.06 .06 .01 -.09 .09 .04 -.05 .01 VF_catswiac .04 .01 .13 -.05 .05 .07 -.03 .03 .09 -.05 -.04 CWIT3 -.06 .12 .00 .04 -.04 .10 -.08 .08 .14 -.06 -.08 IT4 -.07 .11 .12 .00 -.00 .04 -.18 .18 .10 .01 -.12 Sort1a .18* -.07 -.02 -.06 .06 -.08 .00 -.00 .15 -.18 .05 Sort1b .14 -.04 -.02 -.12 .12 -.10 .00 -.00 .11 -.17 .09 Sort2 .09 -.02 -.03 -.10 .10 -.10 .02 -.02 .60 -.11 .07 ! ! ! 74! Table 3 Descriptive Statistics as a Function of Delinquent Group Status for Full and Subsample of Delinquents Note. Values mising from table are inapplicable. Full Subsample Variable Overal M (SD) JSO M (SD) NJSO M (SD) Overal M (SD) JSO M (SD) NJSO M (SD) JSO vs. NJSO .68 (.47) 1.00 (0.00) 0.00 (0.00) .79 (.41) 1.00 (0.00) 0.00 (0.00) Age at Intake 194.07 (17.65) 189.42 (18.85) 203.75 (9.14) 193.66 (19.11) 189.66 (19.23) 209.05 (7.18) Age First Trauma 8.30 (4.21) 7.64 (3.91) 10.84 (4.48) FreqOne .31 (.46) .36 (.48) .20 (.40) .63 (.49) .63 (.49) .63 (.50) FreqTwo .18 (.39) .21 (.41) .11 (.32) .37 (.49) .37 (.49) .37 (.50) Duration of Trauma 34.24 (54.51) 34.03 (53.07) 35.05 (61.26) Incestuous .35 (.48) .43 (.50) .16 (.37) .70 (.46) .74 (.44) .53 (.51) Non-Incestuous .14 (.35) .14 (.35) .15 (.36) .30 (.46) .26 (.44) .47 (.51) Type Physical .13 (.34) .13 (.34) .11 (.32) .26 (.44) .23 (.43) .37 (.50) Type Sexual .22 (.42) .26 (.44) .15 (.36) .46 (.50) .45 (.50) .47 (.51) Type Combo .14 (.35) .18 (.39) .05 (.22) .28 (.45) .32 (.47) .16 (.38) TMT_CT4 6.63 (3.41) 6.83 (3.41) 6.20 (3.41) 6.89 (3.49) 7.04 (3.50) 6.32 (3.50) VF_leter 8.37 (2.76) 8.34 (2.59) 8.44 (3.09) 8.37 (2.63) 8.32 (2.78) 8.58 (2.04) VF_catflu 8.74 (2.98) 8.83 (3.11) 8.56 (2.69) 8.64 (3.31) 8.64 (3.40) 8.63 (3.06) VF_catsco 8.12 (3.19) 8.27 (3.24) 7.82 (3.07) 7.89 (3.48) 7.92 (3.52) 7.79 (3.39) VF_catswiac 9.02 (2.94) 9.09 (3.04) 8.85 (2.73) 8.84 (3.29) 8.78 (3.33) 9.05 (3.24) CWIT3 7.88 (2.90) 7.76 (3.05) 8.11 (2.58) 7.87 (3.09) 7.67 (3.29) 8.63 (2.09) IT4 8.28 (2.98) 8.13 (3.06) 8.59 (2.82) 8.20 (2.96) 7.85 (3.00) 9.53 (2.44) Sort1a 7.68 (2.82) 8.02 (2.79) 6.97 (2.91) 7.59 (2.91) 7.69 (2.94) 7.17 (2.83) Sort1b 7.67 (2.98) 7.96 (2.94) 7.07 (3.01) 7.58 (3.09) 7.58 (3.18) 7.56 (2.81) Sort2 6.43 (3.18) 6.62 (3.12) 6.03 (3.30) 6.40 (2.95) 6.41 (2.89) 6.39 (3.29) ! ! ! 75! Table 4.1 Unstandardized and Standardized Path Estimates: Age of First Traumatic Experience Note. Unstandardized loadings could not be provided for some indicators due to handling scale dependency and are indicated by ?na.? Pathway Estimate SE StdYX p-value Inhibition by TMTCT4 na na .68 na by CWIT3 .88 .28 .68 < .01 by CWIT4 .74 .24 .60 < .01 Conceptual Flexibility by Sort1a na na .78 na by Sort2 1.15 .26 .89 < .001 Monitoring by VFleter na na .61 na by VFcatflu 1.50 .34 .73 < .001 by VFcatsco 1.11 .30 .52 < .001 by VFcatswiac 1.17 .32 .57 < .001 VFcatsco with VFcatswiac 6.31 1.29 .80 < .001 Direct Efects Group Membership (NJSO vs. JSO) on AgeAbuse -.11 .04 -.31 < .01 Group Membership on Age at Intake -.05 .02 -.66 .02 Indirect Efects Inhibition on AgeAbuse .03 .07 .06 .66 Flexibility on AgeAbuse -.02 .06 -.03 .78 Monitor on AgeAbuse .05 .05 .12 .37 Group Membership on Inhibition -.14 .18 -.22 .44 Group Membership on Flexibility < .01 .15 .01 .37 Group Membership on Monitoring .15 .32 .16 .64 Latent Factor Correlations Inhibition with Flexibility 2.70 1.08 .50 .01 Flexibility with Monitoring 2.30 .66 .64 < .001 Monitoring with Inhibition 2.50 .94 .67 < .01 ! ! ! 76! Table 4.2 Unstandardized and Standardized Path Estimates: Duration of Traumatic Victimization Note. Unstandardized loadings could not be provided for some indicators due to handling scale dependency and are indicated by ?na.? ! Pathway Estimate SE StdYX p-value Inhibition by TMTCT4 na na .69 na by CWIT3 .87 .27 .68 < .01 by CWIT4 .75 .24 .61 < .01 Conceptual Flexibility by Sort1a na na .79 na by Sort2 1.14 .27 .89 < .001 Monitoring by VFleter na na .62 na by VFcatflu 1.48 .33 .74 < .001 by VFcatsco 1.07 .29 .50 < .001 by VFcatswiac 1.13 .30 .56 < .001 VFcatsco with VFcatswiac 6.60 1.22 .81 < .001 Direct Efects Group Membership (NJSO vs. JSO) on Duration < .01 < .01 .02 .87 Group Membership on Age at Intake -.05 .02 -.67 < .01 Indirect Efects Inhibition on Duration < .01 < .01 .13 .35 Flexibility on Duration < -.01 < .01 -.10 .42 Monitor on Duration < .01 < .01 .14 .55 Group Membership on Inhibition -.15 .16 -.26 .35 Group Membership on Flexibility .06 .13 .10 .65 Group Membership on Monitoring .08 .27 .09 .77 Latent Factor Correlations Inhibition with Flexibility 2.75 1.06 .52 < .01 Flexibility with Monitoring 2.36 .66 .65 < .001 Monitoring with Inhibition 2.52 .96 .66 < .01 ! ! ! 77! Table 4.3 Unstandardized and Standardized Path Estimates: Frequency of Traumatic Experiencing Note. Unstandardized loadings could not be provided for some indicators due to handling scale dependency and are indicated by ?na.? ! Pathway Estimate SE StdYX p-value Inhibition by TMTCT4 na na .67 na by CWIT3 .90 .20 .69 < .001 by CWIT4 .73 .18 .55 < .001 Conceptual Flexibility by Sort1a na na .83 na by Sort2 1.17 .19 .86 < .001 Monitoring by VFleter na na .55 na by VFcatflu 1.33 .25 .69 < .001 by VFcatsco 1.21 .24 .58 < .001 by VFcatswiac 1.07 .22 .56 < .001 VFcatsco with VFcatswiac 5.18 .74 .82 < .001 Direct Efects Group Membership (NJSO vs. JSO) on Freq Once .80 .27 .29 < .01 Group Membership (NJSO vs. JSO) on Freq Twice+ .84 .32 .25 .01 Group Membership on Age at Intake -.04 < .01 -.54 < .001 Indirect Efects Inhibition on Freq Once -.06 .44 -.01 .89 Flexibility on Freq Once -.29 .43 .03 .50 Monitor on Freq Once -.52 .33 -.16 .11 Inhibition on Freq Twice+ .44 .58 .08 .45 Flexibility on Freq Twice+ .17 .57 .03 .77 Monitor on Freq Twice+ .30 .38 .08 .43 Group Membership on Inhibition -.03 .09 -.05 .76 Group Membership on Flexibility .11 .07 .20 .11 Group Membership on Monitoring -.02 .15 -.03 .90 Latent Factor Correlations Inhibition with Flexibility 2.53 .71 .49 < .001 Flexibility with Monitoring 1.93 .43 .56 < .001 Monitoring with Inhibition 1.98 .58 .60 < .01 ! ! ! 78! Table 4.4 Unstandardized and Standardized Path Estimates: Relationship to Perpetrator of Traumatic Experience Note. Unstandardized loadings could not be provided for some indicators due to handling scale dependency and are indicated by ?na.? ! Pathway Estimate SE StdYX p-value Inhibition by TMTCT4 na na .66 na by CWIT3 .88 .19 .68 < .001 by CWIT4 .72 .17 .55 < .001 Conceptual Flexibility by Sort1a na na .83 na by Sort2 1.17 .19 .86 < .001 Monitoring by VFletter na na .55 na by VFcatflu 1.34 .24 .58 < .001 by VFcatsco 1.22 .24 .58 < .001 by VFcatswiac 1.08 .22 .56 < .001 VFcatsco with VFcatswiac 5.12 .77 .82 < .001 Direct Efects Group Membership (NJSO vs. JSO) on Non-Incest .44 .32 .12 .17 Group Membership (NJSO vs. JSO) on Incest 1.02 .28 .33 < .001 Group Membership on Age at Intake -.04 < .01 -.53 < .001 Indirect Efects Inhibition on Non-Incest .44 .57 .07 .44 Flexibility on Non-Incest -.13 .52 -.02 .81 Monitor on Non-Incest -.14 .43 -.03 .74 Inhibition on Incest .02 .45 < .01 .96 Flexibility on Incest -.01 .44 < -.01 .98 Monitor on Incest -.14 .31 -.05 .65 Group Membership on Inhibition -.02 .10 -.03 .84 Group Membership on Flexibility .10 .07 .18 .14 Group Membership on Monitoring -.03 .16 -.03 .87 Latent Factor Correlations Inhibition with Flexibility 2.57 .72 .49 < .001 Flexibility with Monitoring 1.96 .42 .56 < .001 Monitoring with Inhibition 2.03 .61 .60 < .01 ! ! ! 79! Table 4.5 Unstandardized and Standardized Path Estimates: Type of Traumatic Victimization Note. Unstandardized loadings could not be provided for some indicators due to handling scale dependency and are indicated by ?na.? Pathway Estimate SE StdYX p-value Inhibition by TMTCT4 na na .75 na by CWIT3 .80 .22 .74 < .001 by CWIT4 .51 .18 .49 < .01 Conceptual Flexibility by Sort1a na na .76 na by Sort2 1.44 .36 .98 < .001 Monitoring by VFleter na na .53 na by VFcatflu 1.12 .29 .54 < .001 by VFcatsco 1.51 .45 .64 < .01 by VFcatswiacc 1.24 .39 .60 < .01 VFcatsco with VFcatswiac 4.04 1.14 .73 < .001 Direct Efects Group Membership (NJSO vs. JSO) on Physical .93 .59 .18 .11 Group Membership (NJSO vs. JSO) on Sexual .86 .49 .22 .08 Group Membership (NJSO vs. JSO) on Combo < .01 < .01 .15 .26 Group Membership on Age at Intake -.06 .02 -.69 .01 Indirect Efects Inhibition on Physical .96 1.40 .11 .49 Flexibility on Physical .57 .98 .08 .56 Monitor on Physical 1.07 .66 .23 .11 Inhibition on Sexual .63 .97 .09 .52 Flexibility on Sexual -.49 .61 -.09 .42 Monitor on Sexual -.58 .48 -.16 .23 Inhibition on Combo < .01 < .01 .14 .32 Flexibility on Combo < .01 < .01 .08 .48 Monitor on Combo < .001 < .01 .02 .88 Group Membership on Inhibition -.09 .12 -.16 .47 Group Membership on Flexibility .05 .09 .06 .59 Group Membership on Monitoring .16 .23 .14 .49 Latent Factor Correlations Inhibition with Flexibility 2.58 1.15 .46 .03 Flexibility with Monitoring 1.03 .50 .37 .04 Monitoring with Inhibition 2.28 1.01 .66 .02 EXAMINING EXECUTIVE FUNCTIONING ! ! ! 80! Figure 1.1. Age of first traumatic experience. Final model with standardized (StdYX) loadings. Latent variable item loadings, disturbance terms, and latent variable correlations not depicted. Estimates of latent factor correlations: Inhibition with Conceptual Flexibility StdYX = .504*, Conceptual Flexibility with Monitoring = .642***, Monitoring with Inhibition = .665**. For estimates of latent variable loadings, refer to Table 4.1. *p < .05, **p < .01, or ***p < .001. EXAMINING EXECUTIVE FUNCTIONING ! ! ! 81! Figure 1.2. Duration of traumatic victimization. Final model with standardized (StdYX) loadings. Latent variable item loadings, disturbance terms, and latent variable correlations not depicted. Estimates of latent factor correlations: Inhibition with Conceptual Flexibility StdYX = .517**, Conceptual Flexibility with Monitoring = .647***, Monitoring with Inhibition = .662**. For estimates of latent variable loadings, refer to Table 4.2. *p < .05, **p < .01, or ***p < .001. EXAMINING EXECUTIVE FUNCTIONING ! ! ! 82! Figure 1.3. Frequency of traumatic experiencing. Final model with standardized (StdYX) loadings. Latent variable item loadings, disturbance terms, and latent variable correlations not depicted. Estimates of latent factor correlations: Inhibition with Conceptual Flexibility StdYX = .489***, Conceptual Flexibility with Monitoring = .555***, Monitoring with Inhibition = .601**. For estimates of latent variable loadings, refer to Table 4.3. *p < .05, **p < .01, or ***p < .001. EXAMINING EXECUTIVE FUNCTIONING ! ! ! 83! Figure 1.4. Relationship to perpetrator of traumatic experience. Final model with standardized (StdYX) loadings. Latent variable item loadings, disturbance terms, and latent variable correlations not depicted. Estimates of latent factor correlations: Inhibition with Conceptual Flexibility StdYX = .491***, Conceptual Flexibility with Monitoring = .558***, Monitoring with Inhibition = .603**. For estimates of latent variable loadings, refer to Table 4.4. *p < .05, **p < .01, or ***p < .001. EXAMINING EXECUTIVE FUNCTIONING ! ! ! 84! Figure 1.5. Type of traumatic victimization. Final model with standardized (StdYX) loadings. Latent variable item loadings, disturbance terms, and latent variable correlations not depicted. Estimates of latent factor correlations: Inhibition with Conceptual Flexibility StdYX = .462*, Conceptual Flexibility with Monitoring = .374*, Monitoring with Inhibition = .655*. For estimates of latent variable loadings, refer to Table 4.5. *p < .05, **p < .01, or ***p < .001.