Methods
The participants in this study are 166 females who were followed for the past 18–26 years, beginning in either 1997 (Cohort 1) or 2003 (Cohort 2). Participants were 13–17 years (M = 15.31, SD = 1.17) at study enrollment. At the final wave of data collection, participants were 29–42 years old, (M = 35.0, SD = 2.93). The self-reported racial and ethnic breakdown of the full sample was: 68.1% non-Hispanic White, 1.8% African American, 11.4% Hispanic, 0.6% Native American, 0.6% Asian, 16.9% multi-ethnic/racial heritage. At the most recent wave of data collection, the sample maintained similar self-reported racial and ethnic proportions: 68.9% non-Hispanic White, 1.6% African American, 11.6% Hispanic, 0.1% Native American, 0.1% Asian, 17.8% multi-ethnic/racial heritage. At the final assessment wave, participants self-reported gender identity and all participants who completed the assessment self-identified as women (n = 129). The participants reported their sexuality as heterosexual (66%), lesbian (4%), bisexual (26%), pansexual (2%), and 2% declined to answer.
Measures
Youth variables
Arrests and offenses
Age at first arrest and frequency counts of offenses, arrests, were all measured prior to age eighteen and are based on official records. Offenses include all types (i.e., status, misdemeanors, felonies) and are nested in arrests as youth can be charged with multiple offenses at one time of arrest (Chamberlain et al., 2007; Leve et al., 2022). Days in locked settings was based on participant self-report. Participants were asked at each wave of the study where they were and total days in locked settings at any point prior to the participant’s eighteenth birthday was computed (see Supplementary File A for details).
Foster care history
Number of foster care placements prior to age eighteen includes both official child welfare records and case worker report and, when child welfare records were not available, parent and self-report of all placement changes (unpublished measure, Oregon Social Learning Center; see Supplementary File A for details). It includes all out of home care experiences that were not related to delinquency court dispositions.
Adverse childhood experiences
A 10-item adverse childhood experiences (ACE) revised composite score was created using two items from youth self-report; seven items from caseworker report; and one coded item from official maltreatment records. All items were measured at baseline. Items were selected based on the original ACE measure (Felitti et al., 1998). All items were coded 0 (no) and 1 (yes) for the presence or absence of each risk factor. Items were summed to create a risk index ranging from 0 to 10, with higher scores reflecting greater ACEs. See Supplementary File B for the wording and scoring of items.
Mental health
Mental health diagnoses were measured through self-report at baseline using the Brief Symptom Inventory (BSI; Derogatis and Melisaratos, 1983) and included the proportion of those who met symptom threshold criteria for clinical levels of depression, anxiety, or both. If participants scored above 91 on T-scores for the depression and anxiety subscales, they were identified as having clinically elevated depression and anxiety symptoms, respectively. Cronbach’s alphas for the depression and anxiety subscales were 0.80 and 0.88, respectively.
Substance use
Participants were asked three questions at baseline regarding whether in the past year they had used alcohol, cannabis, and illicit drugs such as uppers, downers, or acid (i.e., drugs other than tobacco, alcohol, or cannabis). Responses were provided based on frequency of use as either, 1 = never, 2 = once or twice, 3 = occasionally, 4 = 1–6 times per week, 5 = 1 or more times per day. Thus, substance use variables were continuous variables, with lower scores indicating no alcohol, cannabis, or illicit substance use and higher scores indicating more severe substance use.
Adult variables
Adult criminal legal system involvement
Adult criminal legal system data were collected from official arrest records during each of the young adult and adult waves of the study. Records were collected for 100% of participants in both the young adult and adult phases of the study. The original study gathered residential history information (self-reported) from all participants, and then attempted to collect criminal record data from every county in which the participant lived for the duration of the study. We produced a complete record of each participant’s official criminal records by compiling the offenses at each assessment (Leve et al., 2022). We then removed any duplicate offenses that occurred because of the record compilation process. We computed the age of first adult arrest using participant date of birth and the date of the first arrest which occurred on or after the 18th birthday. If a participant self-reported an arrest, but no official arrest records data were found, we conservatively used the participant’s self-report of arrest.
Self-reported health
Self-reported health was measured using a modified 12-item version of the RAND 36-item Short Form Survey (Ware and Sherbourne, 1992). For self-reported general health, participants responded to five questions (e.g., “In general, would you say your health is:”) and rated their health from 1 – “excellent” to 5 – “poor.” Items were then reverse scored, with ratings of poor (5) being given a score of “0” and ratings of excellent (1) being given a score of “100,” and averaged. To assess chronic health conditions, participants reported yes or no as to whether they had ever been diagnosed or treated with 19 chronic health conditions. We provided data on the top five reported chronic health conditions and the proportion of participants who endorsed those conditions.
Mental health
To assess whether participants met clinical diagnostic thresholds for depression and anxiety, we used the Brief Symptom Inventory (BSI; Derogatis and Melisaratos, 1983). As with the youth measure, if participants scored above 91 on T-scores for the depression and anxiety subscales, they were identified as having clinically elevated depression and anxiety symptoms. Cronbach’s alphas for the anxiety and depression subscales were 0.79 and 0.85, respectively. Participants also self-reported as to whether they had ever been diagnosed or treated for depression or anxiety, for example, “Are you now or have you ever been treated for or diagnosed with depression [anxiety]?” To determine if participants met current diagnostic criteria for posttraumatic stress disorder (PTSD), the International Trauma Questionnaire was used (ITQ; Cloitre et al., 2018) The ITQ has nine symptom indicators of PTSD and participants are documented to have diagnosable PTSD if they endorse at least one symptom from symptoms clusters including (1) re-experiencing in the here and now, (2) avoidance, and (3) sense of current threat. Participants must also endorse functional impairment due to these symptoms by endorsing one of three items related to functional impairment.
Substance use
Alcohol use was measured in this study through measures of number of drinking days in the past 6 months (“How many days did you drink beer, wine or hard liquor in the last six months?”) and number of drinks in a drinking day (“Thinking about all types of alcohol, like beer, wine, or hard liquor, how many drinks have you usually had at one time in the past 6 months?”). Cannabis use, opioid use, and other illicit drug use was also reported on for the past 6 months. In the illicit drug use category, participants were asked about use of the following illicit drugs: inhalants, hallucinogens, club drugs, prescription drugs for non-prescription use; stimulants, and downers. If participants indicated they had used any of the listed illicit drugs, opiates, or cannabis in the past six months, they were indicated as having that specific drug use. Participants self-reported whether they had ever been diagnosed or treated for a substance use disorder (“Are you now or have you ever been treated for or diagnosed with alcohol or drug abuse?”). All questions for substance use were drawn from an unpublished measure (Rhoades et al., 2014) but are aligned with typical retrospective reports of substance use similar to the widely used Timeline Follow-Back procedure (Sobell and Sobell, 1992).
Educational and socioeconomic outcomes
Participants were asked one question about educational attainment (“What is the highest level of education you have completed?”). Participants were asked about their household income (“What is your annual (gross) household income?”) and how many people the household income supported (“How many people does that income support?”). To assess the poverty level of the sample, we computed poverty threshold by dividing household income by the federal poverty level in relation to the number of individuals in the household. Next, we transferred this number into a percent poverty level by multiplying the number by 100. To assess subjective socioeconomic status (SES), we used the MacArthur Scale of Subjective Social Status – Adult Version (Adler et al., 1994). Participants were shown a ladder and asked where they placed themselves on the ladder in relation to other people in terms of money, education, and jobs.
Family and relational outcomes
Pregnancy and childbirth
Data on pregnancy and childbirth were obtained prospectively at each wave of the study. During adolescence, participants and caregivers were separately interviewed regarding the youth’s pregnancies and pregnancy outcomes (i.e., miscarriage, induced abortion, live birth, stillbirth) that had occurred at each assessment wave. Caregiver report was used only when adolescent report was unavailable (Kerr et al., 2009). In adulthood, participants self-reported this information. The presence or absence of a pregnancy, stillbirth, induced abortion, miscarriage, and live birth was coded as 1 = yes and 0 = no to indicate presence or absence of a pregnancy or pregnancy outcome that occurred at any wave of the study (Cioffi et al., 2022). The reported date of the first pregnancy and childbirth was used along with participant date of birth to compute the first pregnancy and childbirth variables, only for participants with histories of pregnancy and childbirth.
Child welfare contact
To determine the number of children placed into child welfare custody, official child welfare records were obtained in the young adult assessment (Leve et al., 2015). Participants also self-reported custody status at each wave of the study, and in cases where the child was not in the custody of the study participant, the custodial caregiver data was collected, including whether custody was held by “the state, SCF, child welfare, or DHS.” In the final assessment, if official child welfare records were not available for a newly born child, self-report of child custody status was used, but in all other cases, child welfare involvement was indexed through official records.
Intimate partner relationships
Current relationship status was assessed by three questions asking if participants were currently in a relationship, cohabitating with their current partner, and currently married. Intimate partner violence was assessed through the participant’s report on the Revised Conflict Tactics – Short Form (CTS2S; Straus and Douglas, 2004). The CTS2S is a 20-item instrument that assesses negotiation, physical and sexual assault, and psychological aggression in couple relationships. Participants reported on the frequency of each behavior, including their own behaviors, on a scale from 1 (“once in the past year”) to 8 (“this has never happened”). In this study, participants were categorized into having experiences of intimate partner violence if they reported being a victim, perpetrator, or both a victim and perpetrator on the mutuality scores of psychological aggression, injury, or physical or sexual assault in their current relationship. The CTS2S authors note that internal consistency is not appropriate for the measure and therefore it is not presented here.
Couple adjustment was assessed through the Dyadic Adjustment Scale (DAS; Spanier, 1976). The DAS is a 32-item questionnaire that comprises four subscales (Consensus, Satisfaction, Cohesion, and Affectional Expression). The full-scale score (range = 0 to 151) was used in the present study, and higher scores reflect better dyadic adjustment and scores above 114.8 reflecting better adjustment and a happier couple. Conversely, participants in this study were indicated to be in a distressed relationship if they were at or below the cut-score of 98. Cronbach’s alpha for the total scale was 0.94.
CC BY 4.0. "The long-term collateral consequences of juvenile justice involvement for females" by Schweer-Collins ML, Dierkhising CB, Leve LD, Frontiers in psychology (2023). https://doi.org/10.3389/fpsyg.2023.1321355. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Peer-reviewed and open access; full text from PubMed Central (PMC10800427). Headings and layout adapted for reading; figures and tables are in the original.