Testing Predictors of Duration In Volunteer Youth Mentoring Relationships

Background and Purpose: Formal one-to-one mentoring programs like Big Brothers Big Sisters (BBBS) have been shown to improve social, academic, and behavioral outcomes for youth (Rhodes, 2008). Previous research suggests that most of the positive effects of mentoring occur in relationships that last more than a year (Grossman & Rhodes, 2002). Given the primary importance of facilitating long-term relationships, mentoring programs have a need to understand how individual and contextual factors may influence relationship duration. Yet many individual, dyadic, and family characteristics that are likely to impact mentoring relationship duration have never been tested. The current study tested a range of characteristics of youth participants and families, volunteer mentors, and mentoring dyads (or matches) as predictors of relationship duration in a large BBBS agency in the Northwest. Specifically, we hypothesized: 1) that school-based matches and school-referred mentors and youth would have shorter matches, 2) that younger children, children experiencing elevated environmental risk (e.g. living with a single parent), and children referred by their parents would have longer matches, 3) that older mentors, mentors who are employed, and mentors who are unmarried would have longer matches, and 4) that shorter processing time would predict longer matches for both children and mentors.

Methods: Using data from BBBS's Agency Information Management (AIM) system, information was collected for all matches that terminated between November 2010 and February 2011 (n=291). Due to the theoretical importance of the “one year mark” for mentoring relationships, relationship duration was measured as both a continuous and a binary (+ or – 12 months) dependent variable. Data analysis procedures included bivariate analyses of the hypothesized predictors followed by multiple linear and logistic regression models.

Results: The findings partially supported our hypotheses. For example, in the multiple regression model, school-based matches (B= -6.72, p < .01), child age (B = -.762, p < .05), mentor age (B = .195, p < .01), and child referred by parents (B = 4.12, p < .10) were significant or trend level predictors of relationship duration. In the multiple logistic regression model, mentor age (B = .032, p < .01), mentor processing time (B = -.140, p < .01), child living out of home (B = -1.31, p < .10), child having previous mentor (B = -.591, p < .10), and child having an incarcerated parent (B = .624, p < .10) were all significant or trend level predictors of matches reaching or failing to reach the one year mark.

Conclusions and Implications: These findings demonstrate the importance of considering the characteristics and experiences of children and mentors who enter into voluntary mentoring relationships, especially with regard to age, life stability, and environmental risk. The findings also offer the first evidence that youth with incarcerated parents may fare better in mentoring relationships than their peers. This study offers new insight into what makes mentoring relationships last, and can directly inform child and mentor recruitment, training, and agency support of mentoring relationships.