Do youth mentoring programs change the perspectives and improve the life opportunities of at-risk youth?

An overview of evaluation findings

Mentoring programs vary considerably in length, intensity, structure, training of mentors, and the composition of their mentee population, so it is not surprising that their effectiveness also varies widely. Rigorous studies of the effectiveness of mentoring programs find positive yet modest effects for some mentees [4], [5], [6], [7]. Mentoring programs tend to be better at improving youth’s noncognitive and social skills than their academic performance. The experimental Study of Mentoring in the Learning Environment (SMILE) found small, positive effects of mentoring on students’ connectedness to peers, as well as their self-esteem and social skills, but no effect on academic outcomes. Another study of community-based after-school programs found that they provide safe havens where young people feel comfortable expressing themselves, gain relief from the pressures of the streets, and receive guidance in how to engage in social and community activities. Other studies have found that mentoring programs reduce truancy and misbehavior and increase school attendance [8].

Two independent random-assignment evaluations of one of the most well-known mentoring programs in the US, Big Brothers Big Sisters of America, found that it reduced substance abuse and violence and improved parent and peer relationships and school attendance and performance [8], [9]. However, the positive effects were small and dissipated within a year. In addition, this is an exceptionally well-structured program that provides thorough training and continuous support to its mentors and has a high degree of quality control. “[T]his level of training and support is not available in many of the mentoring programs that have emerged in recent years. In fact, a survey of more than 700 mentoring programs found that 36% of volunteers received less than two hours of training and 22% received none at all. Similarly, 20% of volunteers almost never talk to staff people in their programs and 9% have no contact with staff at all” [2].

An important challenge in this literature is to understand the mechanisms through which mentoring works and affects academic outcomes. Evidence from Big Brothers Big Sisters of America suggest that the effects of mentoring on youth’s academic outcomes are mediated through improved parental relationships. Improved perceptions of parental relationships, although not the sole determinant, are important mediators of change in adolescents’ academic outcomes and behaviors. These results are consistent with other findings on the relevance of parents’ involvement in early childhood interventions.

Other evaluations have found that the positive results fade within a few months of program participation. An experimental study of the Across Ages mentoring program found that the encouraging outcomes (such as less substance abuse and fewer problem behaviors, as well as stronger attachments to school and family) were not sustained beyond the end of the school year.

Other studies of mentoring programs have found neutral effects. An experimental-design study of 32 US school mentoring programs for at-risk students in grades 4–8 whose objective was to improve student academic and behavioral outcomes through the guidance and encouragement of a volunteer mentor, found no statistically significant effects over one school year after accounting for multiple comparisons within the three domains examined: academic achievement and engagement, interpersonal relationships and personal responsibility, and high-risk or delinquent behavior [10]. The programs were part of the US Department of Education’s Student Mentoring Program.

Similarly, a different type of intervention that also aimed at improving non-cognitive skills has also found discouraging results. Although in this case, the authors present convincing evidence that unobservables may be biasing these results. Using a quasi-experimental design, the study evaluates the effects of a two-year intervention that targeted 14-year-old at-risk youth’s non-cognitive skills—such as self-confidence, locus of control, self-esteem and motivation—with the aim of improving students’ records of attendance, test scores, and end-of-compulsory-education (age 16) cognitive outcomes. The cross-sectional quasi-experimental estimates of the effect of the policy show a negative and significant impact on treated young people’s test scores at age 16, however, difference-in-differences and double-differences estimates suggest that negative selection into the program based on youth’s unobservables may be biasing these results.

Of greater concern, several studies have found detrimental effects of mentoring programs, especially in the long term. A randomized social experiment examined the Quantum Opportunity Program, which offers low-performing high school students in low-performing US schools mentoring, educational services, and financial rewards during high school (and for one additional year for students who fall behind one grade). The study found that the short-term educational successes were modest and that the program’s effects on risky behavior were detrimental in the long term, with male participants more likely to commit crimes and be arrested in their mid-20s than students who had not been in the program [1].

Based on psychologists and economists’ evidence of a strong link between substance abuse and school performance, a recent study explores whether the Quantum Opportunity Program worked best for those youth most (or least) at risk of engaging in risky behaviors. The author finds that the effectiveness of this five-year group mentoring program varied across groups, with the program benefiting ex-ante high-risk students and being detrimental to ex-ante low-risk students. The mentors were social workers trained to identify and deal with the many structural barriers facing youth. As such, they identified the youth at high risk and succeeded in curbing their risky behaviors during high school. By doing so, they improved high school graduation rates and post-secondary education enrollment at age 19, 21, and 24. In contrast, the program was unsuccessful among youth in the bottom half of the predicted drug-use distribution, given that it increased their engagement in risky behaviors (especially while in high school) and had no positive effects on educational outcomes in the short-, medium-, or long-term. The evidence presented is suggestive that negative peer effects might explain the lack of beneficial effects for this group (see following section).

A randomized experimental trial of the Cambridge-Somerville Youth Study, a community-based treatment program that sought to prevent delinquency, also found detrimental effects. Under the program, boys aged 5–13 and their families were visited twice a month for five years by a counselor, and the boys received services in a variety of areas, including tutoring, medical and psychiatric care, summer camps, Boy Scouts, YMCA, and other community programs. By contrast, children in the control group did not receive any referrals or visits from a counselor. The evaluation found that youth who had been in the program were more likely to be arrested and to have negative physical and psychological health impacts over the long term than were youth in the control group. The impacts were measured up to 30 years later, using official state records.

Why might mentoring programs backfire?

A study of the unanticipated negative effects of mentoring and counseling programs proposed several potential explanations [7]. One of them, the deterrence hypothesis, postulates that overprotection by mentors reduces the costs to mentees of engaging in criminal and other risky behaviors by enabling them to avoid internalizing the full costs of engaging in such behaviors. This view is consistent with Becker’s economic model of crime deterrence through punishment and sanctions. Opposing this view is a quasi-experimental study suggesting that youth are not responsive to sanctions.

Another explanation is that mentors may displace parents and other family members in a mentee’s life, thus breaking important social bonds and weakening a source of informal mentoring. An alternative and related explanation is that because mentees’ parents trust that another adult (the mentor) is also watching over their children, they end up investing less time with their children and paying less attention to possible warning signs than the parents of youth in the control group. In essence, the program may have led to a substitution effect away from the parents’ attention, which could explain these detrimental findings.

Some researchers contend that mentoring programs may make mentees more aware of their relative disadvantage, causing feelings of disappointment, inadequacy, or envy that lead them into risky behaviors.

“Deviancy training” is another proposed explanation for unexpected negative effects of mentoring programs. Mentees’ peers may reinforce deviant conduct by responding with approval and attention, especially in programs that assign mentors to work with several mentees at the same time. Research has found that increased exposure to negative peer influences via deviancy training is associated with increased substance use, delinquency, and violence. Furthermore, interventions serving high-risk students may generate a greater deviancy training effect as the population is more likely to actively model and encourage deviance. Indeed, research suggests that grouping high-risk youth together for intervention services can increase subsequent delinquency.

A recent study of the Quantum Opportunity Program, a group mentoring program, finds that the low-risk students provided positive peer effects to the high-risk students, and vice-versa. Under this interpretation, group mentoring might not be effective if only targeted at high-risk students. Instead, it would be effective if implemented only for low-risk students, if no high-risk students were involved as evidence suggests that the program was detrimental for youth in the bottom half of the risk distribution through their interaction with the ex-ante-high-risk peers. Should there have been no high-risk students in the program, the low-risk students would not have performed badly. Conversely, the program might not benefit high-risk students if low-risk students are not involved.

A final possibility is that unintentional negative effects of mentoring emerge, particularly in cases where mentoring relationships are disrupted or terminated. A rigorous experimental evaluation finds negative effects of school-based mentoring on cooperation of high school-aged boys that is consistent with this possibility.

Differential effects by gender

Mentoring programs, just as other youth interventions aiming at improving their educational outcomes, tend to work better for females than for males. Possible explanations for this differential impact include the fact that young women may have more self-discipline, be more likely to delay gratification or have lower discount rates than young men. As the Illustration shows, the Quantum Opportunity Program improved outcomes to a greater extent for the female mentees than the male ones. Note that this is consistent with the fact that males are more likely to engage in risky behaviors and thus have more high-risk peers than females, especially during high school. The Illustration shows that the program was only effective at improving the high school graduation rate in the short term for the whole group, and that this effect was mainly driven by females’ higher graduation rate. It is also interesting to note that this positive effect dies off soon after the end of the program.

Figure 1 shows the effects of the program on being arrested. While we observe that the program reduced arrest rates among males when they were in their late-teens, this effect fades away two years later. Perhaps more concerning, Figure 1 shows that by the time youth were in their mid-20s, males who had participated in the program where 7 percentage points more likely to have been arrested than those in the control group. This effect is statistically significantly different from zero and from the effect of the program on females.

Another study to find heterogeneous effects by gender is the experimental evaluation of the Student Mentoring Program, where two important differences across genders are found. While the program improved females’ academic performance, it found detrimental effects on males’ non-cognitive skills, such as interpersonal relationships, personal responsibility, and community involvement. Both of these estimates were statistically significantly different from zero and the authors rejected the hypothesis that male and female estimates were equivalent to each other.

Best practices and recommendations

Experts agree that the benefits of mentoring programs are greater for the most disadvantaged and at-risk youth and increase with the quality of the mentor-mentee relationship. Studies have emphasized that activities aiming to improve youth’s social and emotional skills are most effective among younger children, who are more receptive and malleable, and youth’s whose individual or environmental circumstances place them most at-risk. However, while it may be effective to intervene with youth considered most at-risk on the basis of environmental characteristics or both individual and environmental characteristics, one study cautions that few benefits have been found in programs that identify youth for intervention based solely on individual characteristics (such as academic failure) since few benefits from mentoring have been identified for such a group.

Another key element in determining effectiveness is program quality. When mentors build strong personal relationships with mentees, the positive effects tend to endure, as the benefits of greater socialization and integration into mainstream society foster further personal and emotional development. Indeed, secondary analyses of both school- and community-based Big Brothers Big Sisters of America interventions find that outcomes depend on the quality of the mentoring relationships, with greater benefits for mentees in stronger relationships and neutral or even negative outcomes for mentees with less effective mentoring relationships [8]. Beneficial outcomes are more likely when programs provide adequate support and structure to mentoring relations throughout the formative stages of their development and when programs build strong relationships between mentors and mentees through frequent contact, emotional closeness, and enduring ties [4].