The Youth Initiated Mentoring Approach for Juvenile Delinquents

Main aims and hypotheses

The main aim is to examine whether the YIM approach is an effective addition to the Halt-intervention. More specifically, researchers will examine whether the addition of a YIM to the Halt-intervention contributes to (1) strengthening adolescents' resilience (by stimulating protective factors), (2) decreasing or halting the need for formal support and (3) decreasing or halting the development of juvenile delinquency.

H1: It is expected that the YIM approach will strengthen youth resilience:

1. Improvement in social network quality as indexed by increased perceived social support, perceived mattering, and relatedness to others;
2. Improvement of social skills as indexed by increased social resourcefulness;
3. Increased feelings of autonomy and competence;
4. Decreased positive attitudes towards delinquency;
5. Stronger commitment to the social environment as indexed by increased school attachment, and structured free-time activities;
6. Stronger commitment to the social environment by increased school performance, school attachment and structured free-time activities.

H2: It is expected that YIM contributes to halting or decreasing the need for (formal) support.

H3: It is expected hat YIM contributes to halting or decreasing self-reported delinquency.

H4: It is expected that hypotheses three and four are mediated by an increase in resilience factors, such as perceived mattering, self-determination, and social control.

H5: It is expected that the effects of the YIM approach will work better for some than for others, therefore moderators such as demographic factors, psychosocial problems, mentor relationship quality, peer network quality, parental monitoring, and treatment characteristics (i.e., Halt-intervention characteristics and YIM approach characteristics) will be explored.

In addition, for parents the presence of a YIM could lead to an increase in family resilience, perceived social support and social resourcefulness, and a decrease in parental stress and need for (formal) support. It is also expected that they report less delinquent behaviour in their child.

Abovementioned youth and parent outcomes will be assessed as outcome measures, but exploratively also included in mediator analysis. Check measurements at "outcomes".

Methods/Design

Design

This quasi-experimental study was designed in co-creation with Halt professionals, policy officers and Halt management. Multiple design- and brainstorm sessions took place during the first phase of the study. Researchers presented several frameworks/ideas based on theory, which were then further developed, selected, combined, or adjusted together with Halt professionals. One part of the co-creation process was related to designing the participant flow of the study-to align this in an optimal way with the Halt working process. Several research steps (informed consent, pre-test, post-test, follow-up measurement, etc.) were printed on smaller post-its, which were then, in smaller groups, rearranged, organised, and pasted in an overview of the Halt working process. Ideas of the different groups were combined, strengths and pitfalls were assessed, and the most feasible design was selected. Another part of the co-creation process related to the selection of outcome measures. In a brainstorm session with Halt professionals, professionals selected outcome cards (i.e., change in which outcomes were expected after following the Halt-intervention and which after following the Halt-YIM intervention). In several adjacent brainstorm sessions about Halt's 'Theory of Change' with Halt policy officers and management, the outcome domains of Halt were further specified. Both the 'Theory of Change' discussion and the outcome cards discussion, then served as input for the researchers in the selection of outcome measures. These co-creation processes are in line with the Q-sort technique.

The study sample will consist of N = 300 Dutch adolescents (12-18 years) referred to Halt by a prosecutor, the police, a special investigating civil servant or a school attendance civil servant after committing an offense or crime. Common offenses are public inebriation, nuisance, and violation of the compulsory education law, common crimes are theft and vandalism. Adolescents will be non-randomly allocated to either a YIM trained Halt professional (N = 31) or a Halt professional not trained in the YIM approach (N = 31). These professionals will implement the Halt-plus-YIM-condition (N =150) or the care-as-usual (CAU) Halt-condition (N = 150), respectively. Despite non-random allocation to the Halt-plus-YIM and CAU, conditions will be comparable in terms of (1) the professionals who deliver the intervention and (2) case type and severity (see brief summary).

Researchers will perform a multi-informant measurement strategy. Youths are the primary informants, but parents of youths in both conditions are also approached to participate. Both youths and parents fill out questionnaires at baseline, 100 days after baseline (post-test) and nine months after baseline (follow-up). Participants with shortened cases (with less than three meetings) are excluded from post-test and follow-up measurements. Access is granted to relevant information about youth registered by Halt professionals in Halt's registration system.

Participant recruitment

Participants will be recruited by Halt professionals; data collection will be organised by the researchers. YIM-trained professionals will be responsible for recruiting participants for the Halt-YIM-condition (N = 150), non-trained Halt professionals will be responsible for recruiting participants for the CAU Halt-condition (N = 150). The procedure of recruiting participants will be the same in both conditions. The professionals will explicitly invite youths (and their parents) to participate in research with the aim to improve the quality of the Halt-intervention. This will be done (a) verbally during the preliminary conversation, and (b) by sending an invitation flyer with core information about the study immediately after the preliminary conversation. If youths and their parent(s) are interested in participating in the study, they can select a QR-code or link which refers them to the correct information video and an informed consent form corresponding to their age category (-15 years, 16+ years, parents). After giving active consent, youths, and parent(s) (from now on referred to as "participants") are directly referred to the first questionnaire. Participants will be instructed to fill out the first questionnaire before their initial meeting. The procedure for the second and third questionnaires will require no further action from the Halt professionals. All questionnaires will be administered in Dutch and will take approximately 30 minutes to complete per time point.

Data collection

Both recruitment and data collection will start in September 2022. Recruitment will stop after 15 months, because researchers strive to complete pre- and post-tests within 18 months. Follow-up measurements can take place up to 24 months after the start of the data collection. Youths and parents will be asked to fill out three questionnaires: before the intervention, 100 days after the start of the intervention, and 6 months after completion of the intervention. Hundred days is used as a point in time for the post-test because Halt strives to complete all interventions within 100 days.

Aside, from data collected via questionnaires, pseudonymised data from Halt's registration system will be securely transferred to the researchers. These data concerns relevant information regarding the youths, as registered by Halt professionals. For example, offense and referral information (type of crime, prior referrals to Halt, and referral organisation) and Halt-intervention characteristics (confession to the crime, parental consent to Halt, number of meetings between Halt professional and adolescent, and parental presence during meetings). In addition, relevant documents, like the Halt-SI and the YIM-document, are extracted from this system. The Halt-SI, which is administered by all Halt professionals, is a risk assessment tool that assesses various dynamic risk factors (e.g., alcohol use, drug use, and impulsivity) and psychosocial problems (e.g., mood and victim of neglect). In the YIM-document, YIM-trained Halt professionals register information regarding the YIM approach (e.g., the reaction of both youths and parents on the explanation of the YIM approach, which materials are used when motivating the YIM approach, the relationship of the youth with the chosen YIM, etc.). These data will be transferred every three months from Halt to the responsible researcher.

(Statistical) analysis plan

To test similarity between conditions, the two groups of Halt professionals (YIM-trained versus not trained in YIM) will be compared on several demographic factors (age, biological sex, gender identity, educational level, number of years of work experience, number of years working at Halt, and registration at the youth quality register), expectancy outcomes and personality. Second, youths will be compared on demographic factors, offense and referral information, prosocial behaviour and psychosocial problems, and dynamic risk factors. If differences are detected between conditions, propensity score matching (PSM) techniques will be used to match participants from the Halt-plus-YIM condition to the participants in the CAU Halt-condition based on the above-mentioned variables.

For the effect analysis on the different outcome variables, as well as for the moderator and mediator analyses, a series of multiple regression analyses (with baseline levels of the outcome variable as covariate, and condition as predictor variable) will be performed. To correct for the clustering of participants within Halt professionals, the regression analyses will be performed in a multilevel model. To control for chance capitalisation, researchers will implement a Benjamini Hochberg False Detection Rate (FDR) for correction on the p-values for the respective outcome tests.

Power analysis and sample size assessment

To assess the sample size per condition for the primary and secondary outcomes, researchers performed an a priori power analysis in G*Power 3.1. To attain a power level of 0.80 with a significance level of α = .05, a total sample size of N = 100 is necessary to detect a small effect size (f = .10) in a linear multiple regression analysis. For moderator and mediator analysis, a larger sample size is required. Schoemann et al. state that "approximately N = 150 participants are required to ensure statistical power is at least 80% for detecting the hypothesized indirect effect". This would be an indication for a sample size of N = 75 participants per condition. To be able to perform PSM techniques to obtain two comparable groups, researchers will include a doubled ratio in both conditions (N =150 per condition).

A small effect size (f = .10) is expected based on prior knowledge. Prior meta-analyses that explored the effects of formal mentoring programs on delinquent behaviour and reoffending revealed small overall effects, both d = .21 (~f =.10). Even though these effect sizes are small, these demonstrated effect sizes are comparable to other interventions aimed at decreasing delinquent behaviour. It is worthwhile to note that small to medium effect sizes were demonstrated for natural mentoring programs, such as the YIM approach, on positive youth outcomes (g = 0.30). These effects were larger as compared to formal mentoring programs (g = 0.21) and simply the presence of a natural mentor (g = 0.22).

Plan for missing data

Researchers will try to minimise the occurrence of missing data. First, the responsibility of inviting participants for the post-test and follow-up test will lie with the researchers. This will be independent from the intervention adolescents receive at Halt. Thus, even if participants discontinue their intervention at Halt, they will still be approach by the researchers for the post-test and follow-up measurements. Second, the researchers designed the study in way that it relates as much as possible to the participating youth. They are asked how they prefer to be approached and the questionnaires can be easily accessed via their smartphones. This also applies to participating parents.

Missing data is quite unavoidable. Researchers will therefore follow the Intention-to-Treat (ITT) principle for handling missing data on post-test and follow-up measurements. This means that researchers will analyse all missing data of participants according to their assigned condition, regardless of whether the intervention was received or completed. Missing data points will be estimated by performing Full Information Maximum Likelihood (FIML) methods. Researchers will take the percentage of missing data in consideration, and patterns of missing data (e.g., missing completely at random).