UvA-DARE is a service provided by the library of the University of Amsterdam (http
s
://dare.uva.nl)
UvA-DARE (Digital Academic Repository)
The Effects of Youth Mentoring Programs: A Meta-analysis of Outcome Studies
Raposa, E.B.; Rhodes, J.; Stams, G.J.J.M.; Card, N.; Schwartz, S.; Yoviene Sykes, L.A.;
Kanchewa, S.; Kupersmidt, J.; Hussain, S.
DOI
10.1007/s10964-019-00982-8
Publication date
2019
Document Version
Final published version
Published in
Journal of Youth and Adolescence
License
Article 25fa Dutch Copyright Act
Link to publication
Citation for published version (APA):
Raposa, E. B., Rhodes, J., Stams, G. J. J. M., Card, N., Schwartz, S., Yoviene Sykes, L. A.,
Kanchewa, S., Kupersmidt, J., & Hussain, S. (2019). The Effects of Youth Mentoring
Programs: A Meta-analysis of Outcome Studies.
Journal of Youth and Adolescence
,
48
(3),
423-443. https://doi.org/10.1007/s10964-019-00982-8
General rights
It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s)
and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open
content license (like Creative Commons).
Disclaimer/Complaints regulations
If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please
let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material
inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter
to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You
will be contacted as soon as possible.
Download date:10 Aug 2022
Journal of Youth and Adolescence (2019) 48:423–443
https://doi.org/10.1007/s10964-019-00982-8
SYSTEMATIC LITERATURE REVIEW
The Effects of Youth Mentoring Programs: A Meta-analysis of
Outcome Studies
Elizabeth B. Raposa
1
●
Jean Rhodes
2
●
Geert Jan J. M. Stams
3
●
Noel Card
4
●
Samantha Burton
2
●
Sarah Schwartz
5
●
Laura A. Yoviene Sykes
2
●
Stella Kanchewa
2
●
Janis Kupersmidt
6
●
Saida Hussain
7
Received: 8 October 2018 / Accepted: 5 January 2019 / Published online: 19 January 2019
© Springer Science+Business Media, LLC, part of Springer Nature 2019
Abstract
Mentoring programs, which pair youth with caring, non-parental adults with the goal of promoting positive youth
development, are an increasingly popular strategy for early intervention with at-risk youth. However, important questions
remain about the extent to which these interventions improve youth outcomes. The presen t study involved a comprehensive
meta-analysis of all outcome studies of intergenerational, one-on-one youth mentoring programs written in the English
language between 1975 and 2017, using rigorous inclusion criteria designed to align with developmental theories of youth
mentoring. Analysis of 70 mentoring outcome studies, with a sample size of 25,286 youth (average age of 12 years old),
yielded a statistically significant effect of mentoring programs across all youth outcomes. The observed effect size fell within
the medium/moderate range according to empirical guidelines derived from universal prevention programs for youth, and
was consistent with past meta-analyses of youth mentoring. Moderation analyses indicated that programs serving a larger
proportion of male youth, deploying a greater percentage of male mentors or mentors with a helping profession background,
and requiring shorter meetings yielded larger effect sizes, as did evaluations that relied on questionnaires and youth self-
report. Taken together, these findings provi de some support for the efficacy of mentoring interventions, while also
emphasizing the need to remain realistic about the modest impact of these programs as currently implemented, and
highlighting opportunities for improving the quality and rigor of mentorin g practices.
Keywords Meta-analysis
●
Youth mentoring
●
Relational theory
Introduction
Youth mentoring progra ms show great promise as a low-
cost intervention for youth at risk for developing a range of
psychological, social, and behavioral problems. Recent
research has highlighted the positive impact of one-on-one
mentoring relationships for children and adolescents
showing externalizing behaviors such as aggression (Jolliffe
and Farrington 2007), substance use (Rhodes et al. 2005),
and other delinquent behaviors (Tolan et al. 2008). In
addition, one recent study assessed the influence of men-
toring relationships on a wide range of youth outcomes, and
showed particularly potent effects for mentoring on youth
depressive symptoms (Herrera et al. 2013). As a result,
youth mentoring programs have grown in popularity as a
strategy for intervening with youth at-risk for diverse pro-
blems (Blakeslee and Keller 2012), and an estimated 2.5
million U.S. children and adolescents are paired with caring
adults through mentoring programs each year (Raposa et al.
2017). Important questions remain, however, about the
* Elizabeth B. Raposa
ebraposa@wm.edu
1
Department of Psychological Sciences, College of William and
Mary, 540 Landrum Dr., Williamsburg, VA 23188, USA
2
Department of Psychology, University of Massachusetts,
Boston, MA, USA
3
Department of Child Development and Education, University of
Amsterdam, Amsterdam, Netherlands
4
Department of Human Development and Family Studies,
University of Connecticut, Storrs, CT, USA
5
Department of Psychology, Suffolk University, Boston, MA, USA
6
Innovation Research & Training, Durham, NC, USA
7
Department of Psychology, University of Virginia,
Charlottesville, VA, USA
Supplementary information The online version of this article (https://
doi.org/10.1007/s10964-019-00982-8) contains supplementary
material, which is available to authorized users.
1234567890();,:
1234567890();,:
extent to which mentoring interventions influence youth
outcomes, and the conditions under which they are most
effective (e.g., Aos et al. 2004). Many of the existing meta-
analyses designed to summarize the effectiveness of men-
toring programs have focused on subsets of youth or par-
ticular program approaches (e.g., Tolan et al. 2008).
Moreover, the most recent meta-analysis of mentoring
outcomes included studies that were conducted between
1999 and 2010 (DuBois et al. 2011), and therefore did not
capture the recent proliferation of mentoring studies
designed to influence an increasingly diverse set of youth
outcomes (Blakeslee and Keller 2012).
The current study aims to address these gaps in the
existing literature by conducting a comprehensive meta-
analysis of all mentoring outcome studies written in the
English language to-date, with a focus on intergenerational,
one-on-one mentoring programs that are consistent with a
developmental conceptual model of youth mentoring. These
meta-analytic findings have relevance to researchers and
practitioners invested in preventing mental health problems
and promoting positive youth development.
A Developmental Conceptual Model of Youth
Mentoring
Although there is considerable diversity in the structure and
purpose of adult-youth mentoring interventions, most pro-
grams are grounded in a substantial literature showing the
importance of supportive intergenerational relationships for
promoting positive youth development and preventing a
host of psychosocial problems, such as depression and
delinquent behavior (DuBois and Karcher 2013). Present
analyses draw from a developmental model of youth men-
toring relationships (Rhodes et al. 2002; Rhodes 2005)asa
guiding conceptual framework. This developmental model
posits an interconnected set of three processes (i.e., social-
emotional, cognitive, and identity formation processes)
through which the establishment of close, caring relation-
ships with non-parental adults are expected to promote
positive develop mental trajectories.
First, by modeling prosocial skills and providing a con-
sistent and safe relational context, ment ors are thought to
enhance youth’s perceptions of social support and to facil-
itate more positive connections with others. In particular,
relationships with caring adults may become a context for
helping youth interpret and manage interpersonal dif ficul-
ties, improving peer and adult relatedness, and increasing
youth’s receptivity to adult values, advice, and perspectives
(Ruzek et al. 2016). The basis for expecting that positive
mentoring relationships can modify youths ’ perceptions of
other relationships is derived largely from attachment the-
ory, which posits that children construct cognitive repre-
sentations of relationships through their early experiences
with primary caregivers, which in turn influence inter-
personal behavior (Bowlby 1988). Although these
experience-based and generalized expectations around
social interactions, or working models of attachment, are
relatively stable over time, they remain flexible to mod-
ification in response to changing life circumstances, such as
engagement in supportive relationships (Belsky and Cas-
sidy 1994). Empirical research on mentoring has provided
consistent evidence for such processes, indicating that high-
quality mentoring relationships are associated with
improvements in social and emotional functioning, includ-
ing perceptions of relationships with parents, peers, and
teachers (Cavell et al. 2013; Kanchewa et al. 2016 ; Karcher
et al. 2002).
Second, youth engagement in shared activities and
meaningful conversations with more sophisticated thinkers
is thought to scaffold and advance cognitive skills (Rogoff
1978). Key cognitive processes, such as information pro-
cessing and self-regulation, strengthen during adolescence,
particularly in the context of supportive interactions with
caring adults (Parra et al. 2002). Research on the role of
social support in fostering cognitive development under-
scores the social nature of learn ing and, specifically, the
potential contributions of adults in mentoring roles. Feel-
ings of closeness with teachers, for example, have been
associated with greater cognitive engagement and executive
functioning, as well as more positive academic adjustment
for children and adolescents (Spilt et al. 2012).
Finally, mentors are thought to promote identity devel-
opment by servi ng as concrete models of success, demon-
strating qualities that youth might wish to emulate and
exposing youth to new contexts and resources for interest
exploration (Sánchez et al. 2006). Markus and Nurius
(1986) have referred to possible selves: individuals’ ideas of
what they might become, what they would like to become,
and what they fear becoming. Such possibilities, which
often emerge as youth observe and compare the adults they
know, can inform decisions and desired behaviors. Along
these lines, mentors can open doors to activities, resources,
and educational or occupational opportunities that youth
can draw on to construct their sense of identity (Darling
et al. 2002). Indeed, findings regarding mentors’ protective
influence on risk behaviors, and related improvements in
physical health and well-being, are suggestive of a more
positive future orientation in youth (Herrera et al. 2013).
It is important to note that, according to this develop-
mental model of youth ment oring, improvements across
these three domains (i.e., social-emotional, cognitive, and
identity formation) are interconnected, and the long-term
impact of these improvements on more distal youth out-
comes is often non-specific. For example, the use of a
mentor as a role model, and the ability to entertain multiple
possible selves in the service of identity development, may
424 Journal of Youth and Adolescence (2019) 48:423–443
be fostered by the ability of the youth to enter into a secure,
trusting relationship, as well as a growing capacity to
understand the world from the perspective of others. Like-
wise, growth in certain cognitive abilities, such as infor-
mation processing and self-regulation, can enhance the
capacity of youth to regulate complicated emotions. As a
result, mentoring programs often assess, and see improve-
ments in, a wide variety of youth outcomes, even if the
proposed intervention mechanism targets only one of these
developmental processes.
Effectiveness of Youth Mentoring
Several meta-analytic studies have advanced the field’s
understanding of the effect of youth mentoring on various
outcomes. Some meta-analyses have focused on specific
subsets of youth or particular program model s. For exam-
ple, meta-analyses with youth at risk for delinquent or
aggressive behavior have found impacts of mentoring on
juvenile reoffending (Cohen’s d = .21; Jolliffe and Far-
rington 2007) and delinquency (Cohen’s d = .23; Tolan
et al. 2008). Another meta-analysis of three large-scale,
school-based mentoring evalua tions showed positive effects
of mentoring on a range of school-related outcomes
(Cohen’s d ranging from .07 to .18; Wheel er et al. 2010). In
addition, when mentoring programs were evaluated across
youth, academic, and workplace settings, the effect of
mentoring was, again, statistically significant, with youth
mentoring programs showing somewhat smaller effects on
most outcomes (sample-size weighted corrected correlations
(r
c
) ranging from .03 to .14) than mentoring programs for
adults implemented in wor kplace or higher education set-
tings (r
c
ranging from .03 to .36; Eby et al. 2008).
It should be noted that, according to traditional conven-
tions for the interpretation of the magnitude of effect sizes
(Cohen 1988), a standardized mean difference of 0.20 or
lower is considered small (while 0.50 is considered med-
ium/moderate and 0.80 is considered large), which would
suggest a fairly limited impact of these youth mentoring
interventions. However, it is more informative to interpret
effect sizes using guidelines derived empirically from this
particular intervention area. One recent review of meta-
analyses within the field of universal youth prevention
programs showed that the median average effect of pro-
grams tended to fall within the range of 0.07 to 0.20 stan-
dard deviations, with different effects depending on the
outcome assessed (Tanner-Smith et al. 2018). These find-
ings suggest that the average observed effect for youth
mentoring in these meta-analyses tends to be medium/
moderate and fairly consistent with other programs
designed to improve a range of youth outcomes, including
externalizing and internalizing behaviors, social compe-
tence, drug use, and academic achievement.
DuBois and colleagues have published the most com-
prehensive meta-analyses of youth mentoring programs to
date, both of which showed similarly sized effects of
mentoring across outcomes (Hedge’s g ranging from .18 to
.21; DuBois et al. 2002, 2011). Youth who received men-
toring, on average, showed functioning on academ ic, psy-
chosocial, and career outcomes that was about one fifth of a
standard deviation higher than non-mentored youth. The
most recent of these meta-analyses included studies that
were conducted between 1999 and 2010 (DuBois et al.
2011). Although this meta-analysis provided important
direction to the field, a recent proliferation of mentoring
studies designed to influence an increasingly diverse set of
youth outcomes (Blakeslee and Keller 2012) warrants an
updated, comprehensive meta-anal ysis. The current meta-
analysis examined all relevant studies of intergenerational,
one-on-one youth mentoring programs, which included
studies conducted from 1975 through 2017.
In line with the theoretical framework described above,
the current study aimed to improve on past meta-anal ytic
findings by using selection criteria that adhered more
strictly to a relationship-focused model of mentoring. To
this end, analyses included only studies that evaluated a
program aimed at improving youth outcomes through a one-
on-one, intergenerational mentoring relationship. The pre-
sent sample therefore consisted of studies that examined a
relationship between a younger mentee and an older, non-
parental mentor, rather than programs that involved only
group mentoring or peer-to-peer mentoring (although some
programs did include high school or college students
mentoring elementary school students). Likewise, programs
that were structured around mentors serving largely
instructional roles or administering specific curricula were
excluded. Finally, studies in which mentoring was not a
primary, or even secondary component, of the intervention
were also excluded. These guidelines ensured that analyses
were examining mentoring programs designed to improve
youth outcom es through a caring, supportive relationship
with a non-parental adult.
These more stringent selection criteria led to a different
sample of youth mentoring studies than included in pre-
vious meta-analyses. As an example, one study that was
included in severa l past meta-analyses (Barnet et al. 2007)
was excluded from the present meta-analysis because it
evaluated a curriculum-based home visitation program
through which intensively-trained home visitors, with
caseloads of up to 15 youth, provided structured skills
training to pregnant teenag ers. Similarly, the present ana-
lyses excluded other previously-included studies that eval-
uated the effectiveness of a structured “cha racter
development” curriculum (Everhart 2000) or academic
tutoring as the primary intervention (e.g., Burns et al. 2003;
Morrow-Howell et al. 2009), with the mentoring component
Journal of Youth and Adolescence (2019) 48:423–443 425
largely considered secondary and not consistently delivered
to all participating youth. Finally, a number of other eva-
luations were excluded because they focused on group as
opposed to one-to-one mentoring (e.g., Cummings 2010;
Gatto 2006; Hanlon et al. 2009; Jent and Niec 2009).
Examining the Influence of Mentoring Across Youth
Outcomes
The current meta-analysis also takes a novel approach to
answering the important question of which specific youth
outcomes are most strongly influenced by mentoring. Youth
mentoring has been deployed to address a wide range of
emotional, health, behavioral, academic, and vocational
outcomes. Past meta-analyses have therefore attempted to
evaluate the effects of mentoring on certain subsets of
outcomes and have found varying effect sizes depending on
the outcome domain. For example, DuBois and colleagues
(2002) found that the effect size for mentoring was only .10,
on average, for emotional/psychological outcomes, but
doubled in size to .22 for career/employment outcomes.
Similarly, a more recent meta-analysis (DuBois et al. 2011)
found an effect size close to zero for physi cal health out-
comes, versus approximately .20 for conduct problems.
Although these findings suggest that youth mentoring
can different ially affect youth outcomes, the specific areas
of impact of mentoring on youth are difficult to discern due
to the fact that constructs have been inconsistently defined
and assessed across studies. As a result, outcomes across
categories are sometimes more conceptually similar than
those within a particular category. In one study, for exam-
ple, prosocial attitudes were grouped with achievement
motivation in an “ attitudinal/motivational” category
(DuBois et al. 2011), rather than with social skills and peer
relationships in the “social/interpersonal” category. At the
same time, externalizing behaviors and internalizing
symptoms are often placed in separate categories (DuBois
et al. 2002, 2011), despite the close association and frequent
comorbidity of these types of psychopathology in the
clinical and dev elopmental psychopathology literatures
(Bornstein et al. 2010). As mentoring programs increasingly
seek to target specific clinical outcomes, there has been a
call for greater precision and conceptual clarity in youth
outcomes (Arnold and Cater 2016).
To address this issue, the present study utilized a two-tier
system to examine the differential impact of mentoring on
youth outcomes. In keeping with the typical targets of
mentoring interventions, youth outcomes were grouped into
five broad categories, including school functioning, social
relationships, health, cognition, and psychological symp-
toms. In addition, sub-categories were created within each
broad category. For example, school functioning outcomes
were also coded as relevant to one of three sub-categories:
school engagement, academic achiev ement, or extra-
curricular activities (see Methods section for more details).
These categories were derived from an expert review of the
developmental psychopathology and positive youth devel-
opment literatures, and they allowed for assessment of the
effectiveness of mentoring on constructs that are aligned
with more recent research on the etiology and prevention of
clinical issues, as well as the promotion of well-being, in
youth.
In addition to re-coding outcomes using a two-tiered
system, the current study also used three-level meta-analy-
sis, which accounts for the statistical dependency among
effect sizes within studies, and therefore allows for the
inclusion of more than one effect size per study. This
approach increases statistical power, accounts for both
within- and between-study variability, and facilitates ana-
lyses of moderators that might explain either within- or
between-study variance (Van den Noortgate et al. 2014).
The present three-level meta-analysis was designed to
account for the nesting of three types of outcome data (i.e.,
narrow outcome domains within broad outcome domains
within overal l study effect sizes), while also allowing for
estimates of multiple between-study (e.g., mentoring pro-
gram characteristics, publication type) and within-study
(e.g., questionnaire versus interview assessment)
moderators.
Moderators of Mentoring Effectiveness
The current meta-analysis examined a wide range of youth,
mentor, and program characteristics that were considered
potential moderators of program effects, given increasing
evidence that certain individual and program factors might
significantly influence the impact of mentoring. As men-
toring programs continue to multiply and use a wide array
of program practices, it is crucial to identify which of these
practices are most helpful to youth, as well as whether any
might actually be harmful.
Youth characteristics
A number of youth demographic characteristics have been
shown to moderate the effects of mentoring. There is some
evidence that youth mentoring may be more effective with
mid- to late-elementary school-aged children, while men-
toring relationships are less close and enduring with ado-
lescent mentees (Kupersmidt et al. 2017a). Youth gender
may also influence the impact of mentoring relationships.
Male and female mentees tend to be referred to mentoring
programs for different reasons, with male referrals more
commonly stemming from the need for a male role model,
while female mentees are more often referred because of
relational challenges with their primary caregivers (Rhodes
426 Journal of Youth and Adolescence (2019) 48:423–443
