An Enhanced Mentoring Model’s Impact on Youth in Boys and Girls Clubs

Although federal funding has been provided to add mentoring to youth development programs for decades, we still lack knowledge about the impacts of mentoring on youth outcomes. This research seeks to fill a gap by documenting youth outcomes from an enhanced mentoring approach for urban Boys and Girls Clubs (BGC) in the Southeastern United States delivered by paid staff who serve as mentors through group activities and 1:1 interactions with youth. We perform logistic regressions of secondary data from a cohort of BGCs to understand the relationships between enhanced mentoring and youth outcomes related to program retention, behaviors, and academics. We find the presented approach has a significant relationship with retention with those mentored being 1.92 times more likely to return the following program year. Mentored youth also experienced higher expectations from staff and were less likely to be involved in a physical fight with peers. This study sought to determine the youth impacts of an enhanced mentoring approach implemented by a in a in Given limited research on this model, these findings advance youth outcome knowledge and lay groundwork for future studies.


Background
The U.S. Department of Justice Office of Juvenile Justice and Delinquency Prevention (OJJDP) has made significant mentoring investments as a prevention and early intervention strategy with at-risk youth. Despite these investments, youth outcome impacts are not fully understood.
This study examines OJJDP-funded mentoring's impact on youth in a Boys and Girls Clubs (BGC) network in a Southeastern U.S. metropolitan area.
Youth mentoring centers on adult mentor-youth mentee relationships. Although youth mentoring research has evolved with youth program growth, mixed results on youth outcomes continue to perplex program leaders and researchers Raposa et al., 2019;Rhodes & DuBois, 2006). Youth mentoring has shown encouraging impacts on areas like education and self-esteem, albeit with modest effect sizes Karcher, 2005;Raposa et al., 2019;). Effective mentoring during adolescence may also have positive effects into adulthood on college self-efficacy (McClain et al., 2021). Research indicates the mentoring relationship's duration positively impacts youth outcomes, especially if it is 12 months or greater (Grossman & Rhodes, 2002). Inversely, a sample of youth (n = 1,139) in Big Brothers Big Sisters programs report negative academic outcomes when mentoring relationships end abruptly (Grossman et al., 2012).
Youth mentoring research has expanded but the majority of literature focuses on one-to-one models, often delivered by volunteers, rather than youth development professionals. Studies have found comparable positive outcomes from models that utilized older peers, teams, and multiple adult mentors Rhodes & Dubois, 2006). More hybrid models, such as youth-initiated mentor selection, combine informal and formal strategies and indicate potential for positive youth outcomes (Van Dam et al., 2021). It is unknown if youth outcome impacts from traditional mentoring apply to hybrid models. BGC mentoring has been described as "collective mentoring," whereby the staff embody an allhands-on-deck approach to mentoring all youth (Hirsch et al., 2011). This study focuses on an enhanced BGC mentoring model delivered by paid staff mentors, where members receive oneto-one mentoring added to existing group activities. While all staff supported youth, mentors received informal and formal training and support. Staff-youth relationships are central to youth experiences; one study reported that high rates (96%) of BGC youth indicated that at least one adult staff had high expectations of them (Arbreton et al., 2009). Only one other study of the BGC approach addresses mentoring youth outcomes; however, the findings have limited generalizability given its focus on 3 evidence-based mentoring program designs (Mentzer et al., 2015). This study seeks to fill a literature gap by documenting youth outcomes from enhanced mentoring.
Previous BGC studies documented youth development program practices, youth and/or staff experiences, and programming's youth outcomes. Youth outcomes have primarily been analyzed in conjunction with attendance frequency (days per week). Previous BGC "hybrid" mentoring model research showed a strong relationship between attendance frequency and enhanced mentoring (Snyder et al., 2020). Anderson-Butcher et al. (2003) found several areas predictive of youth BGC attendance: unstructured games, recreation offerings, peer relationships, and parental buy-in. The authors also found BGC programs to be protective against delinquent behaviors like truancy regardless of attendance (Anderson-Butcher et al., 2003). Similarly, Mentzer et al. (2015) found that youth attending OJJDP mentoring-funded BGCs avoided delinquent behaviors throughout their tenure. Higher attendance frequency is associated with positive indicators particularly for teenagers, including decreased negative behaviors (Arbreton et al., 2009). This study describes the model's impact by answering the following question: How do youth outcomes, such as retention, club experience, and behaviors, vary according to participation in enhanced mentoring?

Methods
To compare retention rates, club experiences, and behaviors for youth in enhanced mentoring with those of non-mentees, data was obtained from multiples sources, including Boys and Girls Club of America (BGCA). Data sources were merged using a unique identifier. The Georgia State University Institutional Review Board determined informed consent was not required for the use of deidentified previously collected administrative data.

Administrative Data
Individual demographic and participation data were provided by regional BGC, representing 5,164 students attending 22 clubs in school year (SY) 2018-2019. Variables included member unique identifier, school year, age group (child or teen), gender, race/ethnicity, single parent household or a household living below the federal poverty level. Household characteristics were hypothesized to be proxies for greater mentoring need and transient youth. BGC calculated a school-year attendance variable, indicating average attendance 1, 2, or 3 days per week. A variable designating clubs as Teen Centers was included, as BGC observed these clubs with teen staff and space had greater retention and positive youth experiences.
Mentoring Data The regional BGC team compiled a list of members receiving enhanced mentoring from paper records and included a binary mentoring flag. There is no standard mentee selection process; mentors selected members they believed would benefit most from mentoring.

Outcome Variables
The analyses compared three categories containing a total of thirteen outcomes related to enhanced mentoring's impact. The first category is retention with one outcome (returning to BGC the following school year); the second is club experience with nine outcomes (sense of belonging, emotional safety, physical safety, overall safety, fun, adult connections, staff expectations, recognition, and overall club experience); and the third is youth behaviors with three outcomes (grades, truancy, fighting). Figure 1 shows the development of analytic samples to assess retention, club experience, truancy, grades, and fighting (asked only of teens age 13 and older). optimal (consistent very positive experiences), fair (not negative experiences but not consistently great), and needs improvement (negative or strongly lacking experiences; NYOI Guide to Measures, 2018). BGCA considers fair scores an opportunity for improvement, so fair and needs improvement were combined to not optimal in analyses (NYOI Primer, 2018). Club experience outcomes were dichotomized as a binary indicator: optimal or not optimal.

Youth Behaviors
Three youth self-reported survey items are included across the following areas: overall academic performance in the past year, number of school days lost due to truancy in the past month, and number of physical fights in the past year. Only teenagers respond to fighting questions.

Statistical Analysis
All thirteen outcomes were expressed as binary responses, therefore logistic regressions were used for analysis with odds ratios results. Separate regressions were run for all outcomes within the three categories. All regressions are fully adjusted controlling for age, gender, race/ethnicity, single head of household, household poverty status, attendance frequency, and Teen Center status. Chi-square tests were used to analyze whether OJJDP-mentored youth differed from non-mentees. Table 1 includes descriptive statistics of control variables with N indicating the number of youth who responded to each research question. Since youth attending frequently may have a higher chance of being selected for mentoring, an interaction term between mentored and attender type was included, but not found to be significant.
Because clubs with Teen Centers may affect teen outcomes differentially, an interaction term between Teen Center and age group was included, but not found to be significant. All analyses were conducted using Stata, version 16.1/MP. Table 1 presents demographic and participation characteristics across the three analytic samples by mentoring status with shaded values for mentored and non-mentored differences (.05 significance, Pearson chi-square test of independence). For the retention sample, mentored and non-mentored groups differ by attender type (members attending more are more likely to be mentored) and by age group (teens are more likely to be mentored), but are similar across gender, race, and household characteristics. For the NYOI survey sample, mentored and nonmentored groups differ by Teen Centers (mentored members are more likely to be from Teen Centers). For the fighting sample, mentored and non-mentored groups differ by attender type and single-parent households. Fully adjusted regression model results are presented in Table 2.  times more likely to report optimal levels of staff expectations than those not mentored.

Results
We found one youth behavior outcome to be statistically significant (

Discussion
This study sought to determine the youth impacts of an enhanced mentoring approach implemented by a BGC network in a major metropolitan area in the Southeastern United States.
Given limited research on this model, these findings advance youth outcome knowledge and lay groundwork for future studies.
Mentoring goals should be tailored to each youth but generally focus on reducing or preventing negative outcomes, building life and leadership skills, improving academic achievement, and strengthening social and emotional wellness. For mentoring to have its desired effects, youth must engage with mentors to build trusting relationships. Mentoring approaches can be one-toone or group formats, or a combination of both and can be delivered by paid professionals or volunteers. However, youth outcomes based on receiving one-to-one or group mentoring show little difference (Haddock et al., 2020). A sample of young girls' (n = 113) self-reported outcomes from a combined one-to-one and group mentoring point to positive social outcomes (Deutsch et al., 2017), yet combinations of one-to-one and group mentoring approaches delivered by professionals are not well-documented in the literature. The BGC model analyzed in this study is an enhanced mentoring approach, where paid professional staff offer one-to-one mentoring integrated into youth development programming. Gaps exist in the literature to both evaluate youth outcomes and define this approach. This study examines an enhanced mentoring approach where select BGCs received funds for mentoring a subset of youth.
The results for enhanced mentoring are consistent with published studies of BGC's youth development model that show a positive relationship with retention rates, reduced negative behaviors, positive adult-youth relationships, and positive academic performance (Anderson-Butcher et al., 2003;Arbreton et al., 2009). Enhanced mentoring correlates with certain youth outcomes in our study, including year-over-year retention, increased positive perception of staff expectations, and decreased physical fighting. A similar correlation between mentoring and decreased fighting has been found especially for at-risk youth with environmental risk factors, such as single-parent households (Cheng et al., 2008;DuBois & Silverthorn, 2005). Mentoring has been shown to be an effective strategy for youth violence prevention due to mentoring relationships providing youth increased protective factors, such as connection to supportive adults (Thornton et al., 2002).
This study analyzed year-over-year retention, whereas prior BGC studies have emphasized weekly attendance rates within a school year (Anderson-Butcher et al., 2003;Arbreton et al., 2009;Mentzer et al., 2015). When compared to non-mentored youth, mentees were found to have higher average weekly participation rates within the school year (Snyder et al., 2020) and more likely to return the following year. The findings translate to the BGC retaining 3 out of 4 mentored youth versus 2 out of 4 non-mentees. The higher participation and retention rates among mentees have implications for other youth outcomes and warrant additional study.
Some BGCs have adopted attendance targets of one to three times per week based on the prior finding of a positive link between more frequent attendance and teen outcomes across delinquency, character and citizenship, and healthy lifestyle choices (Arbreton et al., 2009).
Higher program dosage in a school year and across years has the potential to impact youth growth and development. Youth retained in BGC programming and enhanced mentorship receive greater exposure to BGC's positive youth development opportunities during critical times. Quality relationships with caring adults have been shown to buffer negative socioenvironmental and familial experiences (Cavell & Elledge, 2013;Herrera et al., 2013).
Because of their participation level, mentored youth can create and maintain relationships with caring adults.
The BGC-enhanced mentoring model occurs in a collective mentoring context, so understanding staff-youth relationships is essential, especially those with additional OJJDP mentoring.
Relationships are cited in qualitative studies of BGC's youth development model as a primary driver of youth and family engagement (Arbreton et al., 2009;Carruthers & Busser, 2000).
Regardless of mentoring status, this study's members reported a high level of adult connectedness. A significant finding shows that mentees experienced higher expectations from staff than non-mentored individuals. All members interact with adult staff and create connections; however, our study points to a differentiation for mentees. Enhanced mentees were more likely to have "optimal" staff expectations, which points to the model's impact on quality relationships. Enhanced mentorship also influenced youth behavior-mentees were less likely to report fighting within the prior year. These outcomes align with overarching BGC program goals and could justify continued investment in prosocial interventions that reduce delinquency.

Limitations
Limitations exist due to analyzing historical secondary BGC data, most notably a small sample size for some survey questions. Because of COVID-19, the NYOI survey was not collected in 2019-2020, which limited the power to detect additional outcome differences and study outcomes longer than a year. The NYOI survey is self-reported and may be impacted by socialdesirability bias, an effect where respondents tend to over-report perceived good behaviors and under-report perceived bad behaviors. Youth complete the survey on site with peers and staff.
Selection bias could not be ruled out during mentee selection since staff subjectively chose youth mentees rather than random assignment. There are no standard mentee selection criteria. The NYOI is also completed by attending youth at the school year's end which biases the sample toward those who remained in the program longer.
Validating the BGC's enhanced mentoring model in a controlled environment was not this study's focus. This study was not prospective experimental research but a retrospective analysis of collected data to inform practice change. Future research collecting qualitative data on youth experiences and mentee selection may reduce selection bias to an extent; however, the BGC collective mentoring environment cannot be changed.

Conclusions
Several positive youth outcomes were associated with the enhanced mentoring model. Future research should explore multi-year mentee outcomes to understand longer-term impacts. More research is needed to understand mentee selection criteria for BGC's enhanced mentoring and whether selection bias is an attenuating factor. More robust research would inform the development of best practices, policy, and training for this mentoring approach, and provide further insights to funders like OJJDP.