Sample

The TCL 2.0 was archivally scored in 2020–2021 using data collected from juvenile offenders housed at a maximum-security juvenile correctional facility in New Mexico. These individuals had previously participated in a NIMH-funded research study (R01 MH071896); data collection for this study occurred from the years 2007 to 2011. The final sample (n = 237 males) consisted of participants with full data collection completed. Consistent with our prior report [39], we used only males in our current sample; however, this sample is different from our previous report (i.e., this report includes juveniles who were incarcerated at a maximum-security correctional facility in New Mexico, whereas our previous report included juveniles incarcerated at a treatment center in Wisconsin). A sample of incarcerated offenders was chosen for investigation, as incarcerated youth are characterized by considerably high rates of trauma. For example, one report found that 86% of adolescents in the New Mexico juvenile justice system experienced four or more ACEs [51], making this an ideal sample for archivally assessing trauma using the TCL 2.0. Participants included in the final sample ranged from 14.08 to 21.17 years of age (M = 17.79; SD = 1.15). Full-scale IQ was estimated from Vocabulary and Matrix Reasoning subtests using the Wechsler Adult Intelligence Scale—3rd Edition (WAIS-III; [52]) for participants sixteen years of age or older and from the Wechsler Intelligence Scale for Children—4th Edition (WISC-IV; [53]) for participants younger than sixteen years of age. IQ estimates for this sample ranged from 63 to 140 (M = 92.18; SD = 12.06). In our sample, 76.79% of participants reported to be Hispanic or Latino, 21.94% reported to be non-Hispanic or Latino, and 1.27% chose not to report their ethnicity. Racial breakdown of the sample was as follows: 11.39% American Indian/Alaskan Native, 3.38% Black or African American, 0.42% Native Hawaiian or other Pacific Islander, 60.76% White, 7.59% reported more than one race, and 16.46% chose not to report their race.

Study Procedures & Ethics

Research staff recruited participants by making announcements and sharing information about the study throughout the juvenile correctional facility. Research staff obtained informed consent from all study participants over the age of 18 and informed assent, along with their parent or legal guardian’s informed consent, for participants under the age of 18. Participants were paid at an hourly rate that was comparable to the current institutional wage for general labor. Procedures for the study were approved by the University of New Mexico Human Research Review Committee, the Office for Human Research Protections (OHRP), and by the staff at the juvenile correctional facility where the study was conducted.

Private rooms at the facility were used to conduct interviews and administer assessments to study participants. Two interviews were completed (i.e., the PCL: YV and K-SADS), and participants agreed to have these interviews videotaped. Video recordings were performed so that research staff working on the overall study could review the recordings for the purposes of training and double-rating. These interviews, in addition to other study assessments and the participant’s criminal and institutional records, were used to archivally score each participant on the TCL 2.0.

TCL 2.0

Trauma scoring was completed using the TCL 2.0, which contains several expansions compared to our original trauma scoring assessment (for more detailed information regarding the original trauma scoring completed by our research group, see [39]). A score of zero, one, or two was assigned to each trauma category (see “TCL 2.0 Scoring” section for more details). While both scoring methods used the same seven categories of trauma, expansions to the TCL 2.0 include more detailed definitions and examples under each category of abuse as well as raters using additional assessments to score the TCL 2.0. In addition, guidance was provided to raters on how to score commonly reported trauma information. Regarding use of institutional file information, our original TCL assessment relied heavily on reports available in the participant’s institutional file to confirm abuse history for scoring purposes (e.g., to receive a score of two, the institutional file had to confirm the participant’s account of the abuse). However, due to the varying level of detail regarding childhood trauma in each participant’s file and the fact that institutional records do not always include information about certain types of traumatic experiences, this criterion was removed from the TCL 2.0 (e.g., a participant could score a two even if reported abuse was not found in the participant’s institutional record). Finally, the original TCL scoring only assigned total trauma scores for each specific trauma category, whereas the TCL 2.0 includes scores for each age bin (i.e., from years 0–6, 7–12, and 13–18) under each trauma category in addition to the total scores for each trauma category. This allows raters to calculate the severity of trauma history (i.e., a chronicity score).

Training of Raters

Research staff with at least a bachelor’s degree in psychology or a related field scored the TCL 2.0 assessment. A research staff member who was experienced in scoring the TCL 2.0 assessment trained the raters. Staff were given the instructions and scoring sheet for the TCL 2.0, and they independently rated an assigned study participant on the TCL 2.0 using the provided instructions. After all staff had independently rated the assigned participant, an experienced TCL 2.0 rater extensively reviewed the scoring with all staff for the assigned participant. All TCL 2.0 scoring questions were answered during this session, and the experienced research staff member ensured that all raters were knowledgeable in the scoring of the TCL 2.0 before staff began to score participants.

Double Rating

Research staff were paired with another research staff member and given a list of study participants to score on the TCL 2.0. Each research staff member independently rated their list of participants and then met with their partner (i.e., another research staff member) to come to consensus on final scores for each study participant. For this sample, n = 233 study participants were independently scored by two raters and final TCL 2.0 scores were determined at a consensus meeting between the two raters assigned to that participant. Four remaining study participants were single-rated by an experienced staff member.

Categories of Abuse

Trauma was rated by categorizing experiences (as recorded in interviews, assessments, or the institutional file) into one of seven types of abuse. The seven trauma categories of the TCL 2.0, in addition to their definitions, are as follows:

Emotional Abuse

Adults in the family or someone acting in a caregiver role behaving in a way that implies that they do not care about the child. Statements intended to make the child feel bad, embarrassed, or humiliated were included under this item, as were instances of exploitation by the parent/caregiver, instances of the participant being threatened with physical violence or forced to do something against their will, and emotional manipulation.

Physical Abuse

Adults in the family, someone acting in a caregiver role, or a figure of authority physically harming the child. Reports of the participant being hit, pushed, kicked, or involved in any other physical altercation which left bruises or marks were included.

Sexual Abuse

Anyone forcing the child to do something sexual against their will. Due to high frequency of participants reporting that they had consensual sexual relationships in their teens with someone over the age of 18, relationships reported as consensual were not counted in this item. However, if the participant had a consensual sexual relationship with someone who was four or more years older than them, raters did make note of this.

Neglect/Poverty

Inability or refusal of the child’s caregiver to provide safety and care. Also included in this item was the inability or refusal of the caretaker to provide support and love to the child (e.g., refusal to visit child during incarceration and/or treatment, or caregivers making and breaking promises to the child). Poverty was included only if it resulted in a neglectful environment.

Community Violence

Exposure to neighborhood factors that might cause physical or psychological harm. Included in this item were things such as exposure to violence via gang affiliations, being the victim of a violent crime, physical abuse by peers, and frequent exposure to general violence in the child’s neighborhood or community.

Observed Trauma

Witnessing acts of violence against another person, including domestic and community violence. This item also includes witnessing a negative consequence of a parent/caregiver using illicit drugs or alcohol (e.g., child witnessing a parent/caregiver having a seizure or overdosing on a substance).

Traumatic Loss

Experiencing the death of a family member or close friend, willful abandonment, and/or traumatic separation from a parent or caregiver for an extended period of time (e.g., police arrest of a parent at gunpoint in front of the child, parent imprisonment where the other parent could not adequately care for the child).

TCL 2.0 Scoring

A score of zero, one, or two was used to rate experiences under each trauma category. Raters assigned a score of a zero when the participant denied having been abused and when no evidence of abuse was found during interviews and file review. A score of one was assigned when there was one or only a few minor instances of abuse detailed in the institutional file, assessments, or interviews. A score of two was assigned when either the institutional file, assessments, or interviews contained one serious instance of abuse or abuse that occurred over a long period of time.

Each trauma category was broken down into three age bins (0–6, 7–12, and 13–18 years old) as well as a total rating for each trauma category. A score of zero, one, or two was given for each age bin, as well as for each category’s total rating. If a score of zero was assigned to all three age bins for a particular category, a participant would receive a total score of zero for that specific trauma category. If a participant received a score of one in any age bin for a category, then they would receive a total score of one for that trauma category. If a participant received a score of two in any age bin for a category, then they would receive a total score of two for that trauma category, regardless of the presence of zeros or ones in other age bins. In instances where a participant received a one across all age bins for a category, raters gave a total trauma rating of a one or a two for that category, depending on abuse severity.

Calculation of Scores

For each participant, a total score and a chronicity score was calculated for the TCL 2.0. The TCL 2.0 total score was calculated by adding the total scores for each of the seven trauma categories. Possible TCL 2.0 total scores range from 0 to 14. The total chronicity score was calculated by summing the values in all age bins for each of the seven categories, with total chronicity scores ranging from 0–42.

Information Used for TCL 2.0 Scoring

The participant’s institutional file, as well as several other assessments, were used for scoring the TCL 2.0. Importantly, when assessments were used for TCL 2.0 scoring (e.g., PCL:YV and K-SADS), notes regarding traumatic experiences were used and this was separate from scoring of psychopathic traits and symptoms of psychopathology.

Institutional File

Criminal files were reviewed in detail by raters in order to have objective information with which to score the TCL 2.0. While information within each participant’s file varied, all criminal files included at least some of the following reports: psychiatric reports, risk assessments, competency assessments, social services reports, social histories completed by social workers, child protective services reports, education reports, and law enforcement incident reports.

In addition to reviewing the participant’s institutional file, research staff also asked a series of questions after the participant consented to being part of the study. Information regarding prior surgeries and metal screening, health and medical history, and incarceration history was gathered. Review of any prior surgery and metal screening questions (e.g., Have you ever been injured with a metallic object?) were particularly useful for scoring the TCL 2.0.

Psychopathy Checklist: Youth Version (PCL:YV)

The PCL:YV was used to assess psychopathic traits. The PCL:YV consists of 20 psychopathic traits (e.g., pathological lying, lack of remorse, criminal versatility) which the interviewer rates on a scale of 0–2, depending on the degree to which the item applies to the participant [40]. Ratings are based on information collected during a semi-structured interview as well as an extensive review of the participant’s criminal record. Psychopathic traits fall into four facets, which are nested under two factors [54]. Facet 1, which consists of interpersonal traits (conning and manipulative behavior and grandiosity), and Facet 2, which consists of affective traits (callousness and shallow affect), both fall under Factor 1. Facet 3, which consists of lifestyle traits (irresponsibility and impulsivity), and Facet 4, which consists of antisocial/developmental traits from childhood through adulthood (juvenile delinquency and criminal versatility), fall under Factor 2.

Previous studies have reported associations between childhood trauma and psychopathic traits in juvenile and adult offenders [55,56,57]. Because the PCL:YV interview provides extensive details regarding the participant’s life, this assessment was used for TCL 2.0 scoring. Raters specifically reviewed the sections pertaining to family life, peer/sexual relationships, school history and adjustment, and childhood/adolescent antisocial behaviors for information relating to traumatic life events. Assessment notes and videotaped PCL:YV interviews were reviewed for TCL 2.0 scoring purposes.

Kiddie-Schedule for Affective Disorders and Schizophrenia (K-SADS)

The K-SADS was used to assess psychopathological diagnoses among study participants. This semi-structured interview is commonly used to diagnose psychopathology in children aged 6–18, including depressive disorders, psychotic disorders, trauma-related disorders, conduct disorders, and substance-related disorders [46]. TCL 2.0 raters reviewed the K-SADS assessment for endorsement of traumatic events included in the PTSD checklist, as well as notes taken during the interview by the K-SADS rater (e.g., notations regarding family death, community violence, and home life were often included in the K-SADS rater notes, which were in turn used for TCL 2.0 scoring). In addition, raters listened to the videotaped K-SADS interviews to obtain more detailed information regarding the participant’s childhood trauma. While the entire K-SADS was used for scoring, information from the mood and trauma-related sections were particularly useful with regard to TCL 2.0 scoring.

Post-Head Injury Symptoms Questionnaire (PHQ)

The PHQ asks about head injuries that participants have incurred and the symptoms associated with each injury [41]. TCL 2.0 raters looked at the events surrounding the head injury to gather information on trauma (e.g., if the participant reported they were hit over the head by a parent or guardian, this would count as physical abuse on the TCL 2.0).

Upsetting Events Survey (UES)

This assessment asks 17 questions regarding upsetting events that sometimes happen to people [43]. Potential responses for each question include: “No”, “Yes”, “More than one time”, or “I don’t know.” For two questions, the participants were asked to write down the event they were thinking of when they answered the question. Raters reviewed each participant’s endorsement of traumatic experiences to score the TCL 2.0.

Childhood PTSD Symptoms Scale (CPSS)

The CPSS assesses PTSD diagnosis and severity of symptoms in individuals aged 8–18. It asks about all DSM-IV PTSD symptoms and their prevalence as well as any functional impairment that resulted from the upsetting event [45]. For this assessment, raters reviewed information to score the TCL 2.0.

Childhood Trauma Questionnaire (CTQ)

This self-report questionnaire asks participants to respond to questions about their experiences growing up as a child and teenager on a 5-point scale [47]. Staff rating the TCL 2.0 reviewed answers to questions regarding emotional, physical, sexual, and neglectful experiences that the participant may have endorsed. Because the CTQ does not gather thorough detail on childhood trauma, additional trauma information was gathered from more detailed sources for TCL 2.0 scoring purposes (e.g., K-SADS, PCL:YV, institutional files). In the current sample, the CTQ was not administered while participants were incarcerated as youth. Instead, for a subsample of participants (n = 49), the CTQ was collected as part of a large follow-up study (R01 HD092331) later as adults.

My Worst Experience Scale

This two part self-report questionnaire asks the participant about the worst experience they have ever had [44]. Part I asks the participant six questions to ascertain more details about the specific experience, and Part II asks participants 105 questions to describe what happened after their worst experience (responses on a six-point scale, with additional questions about frequency). Raters reviewed Part I of the assessment, which describes the event in question, for TCL 2.0 scoring.

Socioeconomic Status (SES) Questionnaire

This questionnaire asks the participant about marital status, education and work history, living arrangements, and parental education and work history [48]. Raters used this assessment to gather information on poverty levels as well as presence of caregivers in the participant’s life.

Data Analyses

Unless otherwise specified, SPSS (version 20) software was used for analyses. Cronbach’s alpha was used to assess the internal consistency of the TCL 2.0. In addition, intraclass correlation coefficients (ICCs) were used to examine inter-rater reliability. ICC estimates were calculated using a two-way random effects model on average measures with absolute agreement. In line with our previous trauma report [39], principal component analysis (PCA) with Varimax rotation was used to examine the dimensionality of the TCL 2.0. Trauma categories that had loadings of ≥ 0.5 were assigned to each principal component (PC; [58]). In addition, this PCA was followed by exploratory structural equation modeling (ESEM) to rigorously test how well a two-component solution accounted for the TCL 2.0 data. This was done for a few reasons. First, the PCA approach, one of the most frequently employed statistical methods, is often used to summarize a main component in a dataset, but it assumes perfect measurement (i.e., does not model residual error/unique variance). ESEM, like confirmatory factor analysis, estimates common factor variance separately from error/unique variance, and therefore it provides a riskier test of model fit. In other words, PCA is simply used to reduce or summarize the data, whereas ESEM is designed to account for the data [59]. Also, PCA can produce problematic solutions when the number of variables per factor are low [60]. Therefore, subsequent ESEM after PCA provides a robust test of the theoretical model (i.e., two different latent variables reflecting trauma experiences). Mplus software was used for ESEM analyses [61], and standard fit criteria (i.e., CFI > 0.90; RMSEA < 0.08) were used to assess ESEM fit (see [62] for a discussion on ESEM approach). Given the ordinal nature of the TCL domains, the robust weighted least squares estimator was used for ESEM and efficient rotation was done via GEOMIN (oblique rotation; Mplus default). Using standardized factor loadings and residual error/unique variance, Omega reliability estimates were calculated in terms of common factor variance over total variance [63, 64]. More specifically, Omega was calculated from Mplus standardized model parameters in terms of the common sources of unit-weighted total score variance (i.e., sum of the squared loadings), divided by the unit-weighted total score variance (common sources of total score variance plus error/unique variance). Additionally, we used Pearson correlations to examine the relationships between the TCL 2.0 total and component scores with a self-reported trauma measure (i.e., the UES), measures of psychopathology obtained via the K-SADS (i.e., anxiety, mood, and PTSD), and psychopathic traits assessed via the PCL:YV.

Finally, multivariate Cox proportional-hazard regressions were run using the survival package [65] in R [66] to test whether TCL 2.0 scores predicted time (in months) to first felony-related re-offense. New Mexico recidivism data were obtained from the Center for Science and Law’s Criminal Record Database (CRD; [67]). More details are provided in [68]; briefly, re-offense data was extracted from the CRD of criminal court records for offenders in New Mexico. Data in the CRD were matched to current participants via four separate identifiers (i.e., first and last name, date of birth, and social security number). Extensive online searches including social media, White Pages, Been Verified, county records, New Mexico Corrections Department offender search, and out–of–state inmate databases were conducted for the entire sample. This enabled us to compile recidivism data for subjects who were not found in the CRD. Felony recidivism was operationally defined as any felony re-offense following the participant’s release from the juvenile correctional facility. This recidivism data was examined continuously (i.e., time [in months] to first felony re-offense following release outcomes) via multivariate Cox proportional-hazard regressions. Consistent with other studies from our research group [68, 69], time at risk was defined as the time period between an individual’s release date and their felony re-offense date or the end of the follow up window (August 31, 2019) for those who did not commit a felony offense. In participants with all available data (i.e., n = 221 offenders), ~ 74% of our participants (i.e., n = 164) committed a felony offense following their release from the juvenile correctional facility. One-tailed statistical tests were performed, given that trauma has been previously associated with recidivism in existing studies [20, 21]. In each model, TCL 2.0 scores (either the total trauma score or the two PC scores) were entered along with IQ scores and PCL:YV factor scores, to see if TCL 2.0 scores remained associated with time to felony re-offense. Because IQ and PCL:YV scores have been previously linked to criminal behavior [49, 50], we accounted for shared variance with these relevant covariates in the Cox proportional-hazard regression analyses.