Child Development, xxxx 2017, Volume 00, Number 0, Pages 1–16

The title for this Special Section is Positive Youth Development in Diverse and
Global Contexts, Edited by Emilie Phillips Smith, Anne C. Petersen and Patrick Leman

Promoting Positive Youth Development Through School-Based Social and
Emotional Learning Interventions: A Meta-Analysis of Follow-Up Effects
Rebecca D. Taylor

Eva Oberle

Collaborative for Academic, Social, and Emotional Learning
(CASEL)

The University of British Columbia

Joseph A. Durlak

Roger P. Weissberg

Loyola University Chicago

Collaborative for Academic, Social, and Emotional Learning
(CASEL) and The University of Illinois at Chicago

This meta-analysis reviewed 82 school-based, universal social and emotional learning (SEL) interventions
involving 97,406 kindergarten to high school students (Mage = 11.09 years; mean percent low socioeconomic
status = 41.1; mean percent students of color = 45.9). Thirty-eight interventions took place outside the United
States. Follow-up outcomes (collected 6 months to 18 years postintervention) demonstrate SEL’s enhancement
of positive youth development. Participants fared significantly better than controls in social-emotional skills,
attitudes, and indicators of well-being. Benefits were similar regardless of students’ race, socioeconomic background, or school location. Postintervention social-emotional skill development was the strongest predictor of
well-being at follow-up. Infrequently assessed but notable outcomes (e.g., graduation and safe sexual behaviors) illustrate SEL’s improvement of critical aspects of students’ developmental trajectories.

Positive youth development (PYD) focuses on
enhancing young people’s strengths, establishing
engaging and supportive contexts, and providing
opportunities for bidirectional, constructive youth–
context interactions (Larson, 2000; Lerner, Phelps,
Forman, & Bowers, 2009; Snyder & Flay, 2012).
Interventions that are grounded in the PYD framework, therefore, must move beyond a problemoriented focus and address protective and risk factors across family, peer, school, and community
environments that affect the successful completion
of youths’ developmental tasks (Catalano, Berglund, Ryan, Lonczak, & Hawkins, 2002).
This article was supported, in part, by grants from the Einhorn
Family Charitable Trust, 1440 Foundation, Lucille Packard Foundation for Children’s Health, NoVo Foundation, Robert Wood
Johnson Foundation, William T. Grant Foundation, the University of Illinois at Chicago, and the Social Sciences and Humanities Research Council of Canada. We also thank Mark Lipsey
and David Wilson for providing the macros used to calculate
effects and conduct the statistical analyses.
Correspondence concerning this article should be addressed to
Roger P. Weissberg, The University of Illinois at Chicago, 1007
West Harrison Street Chicago, IL60607. Electronic mail may be
sent to rpw@uic.edu.

Operational definitions of PYD’s key constructs
vary—for example, the five Cs model (Lerner et al.,
2009) or the external and internal developmental
assets model (Benson, Leffert, Scales, & Blyth,
1998). However, they share a common focus on
building young people’s positive personal competencies, social skills, and attitudes (i.e., asset development) through increased positive relationships,
social supports, and opportunities that strengthen
assets and help youth flourish within their environments (i.e., environmental enhancement).
A systematic review of 25 PYD program evaluations indicated that PYD interventions operating in
family, school, and community settings are indeed
effective in promoting positive development in a
broad range of outcome domains (Catalano et al.,
2002). For example, the authors found that PYD
interventions were successful in improving young
people’s self-control, interpersonal skills, problem
© 2017 The Authors
Child Development © 2017 Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/2017/xxxx-xxxx
DOI: 10.1111/cdev.12864

 2

Taylor, Oberle, Durlak, and Weissberg

solving, the quality of their peer and adult relationships, commitment to schooling, and academic
achievement. Although these examples of asset
development are the key outcomes of interest for
PYD, some interventions have also decreased substance use, risk taking, and problem behaviors.
PYD interventions, therefore, appear to foster positive outcomes and also be able to protect against
negative ones. A variety of specific intervention
strategies are compatible with the broad asset
development and environmental enhancement orientation of PYD, such as service learning, mental
health promotion, and social and emotional learning (SEL; Catalano et al., 2002; Tolan, Ross, Arkin,
Godine, & Clark, 2016). School-based SEL is the
focus of this review.
Similar to the goals of PYD, school-based SEL
involves implementing practices and policies that
help students and adults acquire and apply knowledge, skills, and attitudes that enhance personal
development, social relationships, ethical behavior,
and effective, productive work (Elias et al., 2015;
Greenberg et al., 2003; Weissberg & O’Brien, 2004).
SEL interventions promote asset development by
enhancing five interrelated cognitive, affective, and
behavioral competencies considered to be important
for success in school and life: self-awareness (e.g., recognizing emotions, strengths and limitations, and
values), self-management (e.g., regulating emotions
and behaviors), social awareness (e.g., taking the perspective of and empathizing with others from
diverse backgrounds and cultures), relationship skills
(e.g., establishing and maintaining healthy relationships), and responsible decision making (e.g., making
constructive choices across varied situations; Weissberg, Durlak, Domitrovich, & Gullotta, 2015).
Previous research has shown that the assets promoted within SEL interventions are associated with
positive developmental trajectories. Specifically, longitudinal analyses have shown links between social
and emotional competencies assessed in childhood
and health, education, and well-being later in life
(Hawkins, Kosterman, Catalano, Hill, & Abbott,
2008; Jones, Greenberg, & Crowley, 2015). There is
also a substantial research base indicating that
school-based SEL interventions have been effective
in promoting targeted social and emotional competencies, which results in both enhanced social and
academic adjustment and reduced levels of conduct
problems and emotional distress (Durlak, Domitrovich, Weissberg, & Gullotta, 2015; Durlak, Weissberg, Dymnicki, Taylor, & Schellinger, 2011; Sklad,
Diekstra, Ritter, Ben, & Gravesteijn, 2012). In other
words, SEL interventions are a form of PYD asset

development that focuses primarily on positive outcomes including school, career, and life success
while also showing evidence of effective protection
from negative outcomes (Bonell et al., 2016; Mendelez-Torres et al., 2016; Weissberg et al., 2015).
The social and emotional competency domains
that are the defining focus of SEL intervention are
conceptually aligned with many of the outcomes of
interest within PYD’s asset development objective.
SEL interventions promote personal strengths in
youth that overlap substantially with the PYD internal assets of social competencies, positive values,
positive identity, and commitment to learning (Benson et al., 1998; Theokas et al., 2005). The five core
SEL competency domains are also reflected in the
15 core PYD objectives identified in the previously
mentioned review of PYD interventions, which
include promoting social, emotional, behavioral,
and cognitive competencies (Catalano et al., 2002).
The close alignment between PYD and SEL has
been emphasized recently in a review of the major
theoretical frameworks for positive development
(Tolan et al., 2016). The authors argue that the theoretical frameworks for PYD and SEL share substantial overlap and recommend a move toward
integrating the approaches to advance the study
and implementation of approaches that aim to
enhance positive development in youth.
One of the important issues regarding PYD, in
general, and SEL, in particular, involves examining
how intervention affects youths’ subsequent development. In their review of 46 meta-analyses and
narrative reviews involving hundreds of studies
and more than a half million students, Weare and
Nind (2011) indicated that school-based universal
promotion programs produced positive impact
immediately following intervention but that the
long-term effects of such interventions were in need
of further study. In their review, Catalano et al.
(2002) also pointed out that further research is
needed to determine whether PYD interventions
can sustain their initial positive findings.
The main purpose of this study was to fill this
gap in research by conducting a meta-analysis of
the follow-up effects of school-based universal SEL
interventions. This research is an extension of a previous meta-analytic review of SEL programs that
found significant positive effects at post on a range
of outcomes (Durlak et al., 2011). The interventions
included in this review vary considerably in duration, location, participants, and other features
(Table 1), but they all collected follow-up assessments at 6 months or more postintervention. In
addition, the current review aims to identify and

 Follow-Up Effects of SEL Programs
Table 1
Descriptives of 82 School-Based Universal Interventions With FollowUp of at Least 6 Months
N
General publication features
Date of initial intervention report
1981–1989
9
1990–1999
12
2000–2009
39
2010–2014
22
Source of report
Published article/books
81
Unpublished dissertation
1
Methodological features
Randomization
Yes
52
No
28
Implementation
Monitored without significant
50
problems reported
Significant problems reported
15
Reliable outcome measures at follow-up (of 426)
Yes
310
No/not reported
116
Valid outcome measures at follow-up (of 426)
Yes
236
No/not reported
190
Source of outcome data at follow-up (of 426)
Child
308
Other (parent, teacher, observer,
118
school records)
Participant features
Developmental level during intervention
Childhood (age 5–10; grades K-5)a
31
Early adolescence (age 11–13; grades 6–8)
37
Adolescence (age 14–18; grades 9–12)
11
Race of participants
Predominantly White
21
Predominantly students of color
14
No predominant race
19
Race not reported
28
Socioeconomic status of participants
Predominantly low and working class
14
Predominantly upper and middle class
9
No predominant SES
28
SES not reported
31
Median total initial sample size
438
Sample size sum across all interventions
97,406
Mean total attrition at follow-up
17.7%
Mean differential attrition at follow-up
1.1%
Intervention features
Intervention format
Classroom by school personnel
32
Classroom by nonschool personnel
27
Multicomponent
23

3

Table 1
Continued
N

%

73
9

89.0
11.0

%

11.0
14.6
47.6
26.8
98.8
1.2

63.4
34.1
61.0
18.3
72.9
27.1
55.5
44.5
72.2
27.8

37.8
45.1
13.4
25.6
17.1
23.2
34.1
17.1
11.0
34.1
37.8

39.0
32.9
28.0

Use of recommended training procedures
Intervention rated as SAFE
Intervention not rated as SAFE
Number of sessions
Mean number of sessions
Median number of sessions
Weeks until follow-up averaged at study level
24 weeks to < 52 weeks (1 year)
52 weeks to < 104 weeks (2 years)
104 weeks to < 208 weeks (4 years)
208 weeks to 780 weeks (15 years)
Mean number of weeks-outcome level
Median number of weeks-outcome level
Locale of intervention
United States
Outside the United States
General area of schools
Urban
Suburban
Rural
Combination of areas

20
15
50
13
11
6
115
52.0

61.0
15.8
13.4
7.3

44
38

53.7
46.3

41
8
9
5

50.0
9.8
11.0
6.1

Note. The Ns do not always add up to 82 due to missing data on
some variables. SES = socioeconomic status; SAFE = sequenced,
active, focused, and explicit. aMostly middle childhood; only six
in K-3rd grade.

test the specific theory of change that underlies SEL
interventions. A recent theoretical review indicated
that empirical data are needed to support the theory of change guiding PYD interventions and identify which assets provide the greatest support for
positive outcomes and buffer against negative outcomes (Bonell et al., 2016). The current meta-analysis empirically tests one such theory of PYD
articulated in the SEL framework: fostering social
and emotional skills and positive attitudes provides
students with assets that will promote well-being
and protect against negative outcomes.
The theory of change behind the SEL approach is
presented in Figure 1. SEL interventions focus on
student-centered competence development. In some
multicomponent approaches, an environmental
focus enhances effectiveness by integrating SEL into
school curriculum and practices or fostering climates that are safe, well-managed, caring, and participatory (Zins, Weissberg, Wang, & Walberg,
2004). As Figure 1 illustrates, the two primary
social and emotional assets targeted by SEL interventions are social and emotional skills in the five
competency domains and positive attitudes toward

 4

Taylor, Oberle, Durlak, and Weissberg

SEL Intervention
Student-Centered SEL
Competencies Instruction
Environmental Focus:
Integration With
Curriculum or Practices
Improvement of
Classroom, School, or
Family Environment

Social and Emotional
Assets

Positive and Negative
Indicators of Well-Being

Social and Emotional
Skill Acquisition

Positive Social Behavior

Improved Attitudes
About Self, Others,
and School

Academic Success
Fewer Conduct Problems
Less Emotional Distress
Less Drug Use

Figure 1. Social and emotional learning (SEL) framework for positive youth development, with SEL interventions fostering assets
within youth, which promote the development of positive behavioral, academic, and mental health outcomes.

the self (e.g., more self-confidence), others (e.g.,
prosocial attitudes that disdain violence), and the
school or education in general (e.g., feeling connected to teachers). The effective promotion of these
social and emotional assets (i.e., enhanced skills and
improved attitudes) is then expected to lead to better short- and long-term developmental outcomes
that include more prosocial behavior, enhanced academic performance, fewer conduct problems, lower
levels of emotional distress, and reduced substance
abuse (Collaborative for Academic, Social, and
Emotional Learning, 2013; Elias et al., 1997; Farrington et al., 2012; Zins et al., 2004). The current metaanalysis evaluates whether SEL interventions that
encourage the development of social and emotional
assets through school-based interventions yield significant effects at follow-up on multiple positive and
negative indicators of well-being.
A final aim of the current review was to examine
whether SEL interventions were effective in promoting positive developmental trajectories across
diverse and global populations (Torrente, Alimchandani, & Aber, 2015). The assets promoted
within SEL have the potential to enhance positive
development for all youth, and the goal of universal school-based approaches is to reach all students
rather than targeting specific subgroups. Positive
benefits have been reported for SEL interventions
conducted outside the United States and for students from various racial and socioeconomic backgrounds, although whether demographic subgroups
of students benefit differentially from intervention
is still unclear. The Promoting Alternative Thinking
Strategies (PATHS) intervention found positive
effects for both Black and White student participants (Conduct Problems Prevention Research
Group, 1999). Some universal SEL intervention
results indicate that students from ethnic minority

groups or low socioeconomic status actually benefit
more from intervention. Stronger intervention
effects have been found for ethnic minority youth
on the development of assets like emotional coping
skills (Kraag, Van Breukelen, Kok, & Hosman,
2009) and for students from poor families on school
attachment and achievement (Hawkins, Catalano,
Kosterman, Abbott, & Hill, 1999). However, there is
also empirical evidence that some universal SEL
programs have been less effective in promoting
social competence within high-poverty schools
(Conduct Problems Prevention Research Group,
2010) or boosting optimism to reduce depression
for African American youth (Miranda et al., 2005).
The current meta-analytic review allows us to
undertake a quantitative, systematic examination of
whether SEL is an effective strategy within and outside the United States and with students from
diverse racial and socioeconomic backgrounds.
We had three main hypotheses. First, we predicted that significant effects for outcomes assessed
at follow-up periods of 6 months or longer would
significantly favor SEL program participants over
controls in seven distinct outcome categories, which
cover both the social and emotional assets that are
the focus of SEL intervention as well as several positive and negative indicators of well-being. Second,
we predicted that SEL interventions would be an
effective PYD approach with diverse racial and
socioeconomic populations. That is, we expected to
find similar positive effects for interventions conducted within and outside of the United States; for
student participants who were predominantly
White, predominantly students of color, or racially
diverse; and, finally, for students from low- or working-class and from upper- or middle-class families.
Third, based on the SEL framework, we tested the
relative benefits of enhancing social and emotional

 Follow-Up Effects of SEL Programs

skills and positive attitudes at postintervention on
positive and negative indicators of well-being at follow-up. Based on the accumulating empirical evidence base in SEL and PYD, we predicted that social
and emotional asset development at post would predict positive long-term outcomes at follow-up.
In addition, follow-up studies conducted over a
long-time period afford the opportunity to examine
important indices of development that may not be
relevant or available for immediate postassessments
or that may not be collected frequently enough
across studies to be amenable to formal meta-analysis. For example, does intervention in the early elementary years lead to higher rates of high school
graduation or college attendance, to stronger social
relationships, or to a reduction in serious social or
mental health problems later in life? Based on our
knowledge of the literature we knew some researchers had collected these less frequently reported but
important outcomes, and we summarize these findings to present a more complete picture of how participation in SEL programs has affected some critical
subsequent developmental outcomes.

Method
The overview of methods provided below is greatly
expanded upon in Supporting Information, including more detailed descriptions of the literature
search, inclusionary criteria, coding procedures, and
variables analyzed.

5

targeted, and what outcomes were used to assess
program impact. In terms of procedures, the programs were most often classroom-based interventions; the majority of these sought to promote
competencies through a series of structured group
lessons lasting between 30 and 45 min. A few incorporated the development of competencies as part of
regular academic instruction, and a minority also
expanded the classroom intervention with additional components such as efforts to enhance classroom or school climate, various school-wide
initiatives, or parent involvement. Several of the
reviewed programs (e.g., PATHS, Positive Action,
Life Skills) have achieved recognition as effective
interventions by various organizations and agencies. Each included program had to target at least
one of the five SEL competency domains (e.g., selfmanagement, relationship skills) to be included,
and some focused on all five. Eighty-nine percent
of the interventions were rated as having
sequenced, active, focused, and explicit (SAFE)
practices (Durlak et al., 2011).
The sample included kindergarten through high
school students, with 37.8% in kindergarten to 5th
grade, 45.1% in 6th to 8th grade, and 13.4% in 9th
to 12th grade. Students represented ethnically,
socioeconomically, and regionally diverse samples.
Forty-four of the intervention studies were conducted within and 38 outside of the United States.
Forty-one interventions occurred in urban school
districts, eight in suburban school districts, nine in
rural settings, and five were in a combination of
these locations.

Study Sample
We used procedures similar to an earlier metaanalysis to search for, select, and code studies (Durlak et al., 2011). Reports had to describe a schoolbased universal SEL program for kindergarten to
12th-grade students that collected follow-up data
from intervention and control groups 6 months or
more postintervention, contained sufficient data to
calculate an effect size (ES) on at least one outcome,
and appeared by December 2014. Eighty-two interventions constituted the final sample; although our
search covered all types of reports, all but one of
the qualifying interventions came from a published
report. Descriptive information on the 82 interventions involving a total of 97,406 students is shown
in Table 1. A majority of studies used randomized
designs, monitored implementation, and used reliable and valid outcome measures.
The SEL interventions varied in general procedures, which of the core SEL competencies were

Socioeconomic Characteristics
Only 51 of the 82 interventions reported any
information on socioeconomic status (SES); only 26
of those provided an actual percentage of students
from low SES households (M = 41.1, SD = 33.8). In
order to create comparison groups for meta-analysis, we identified interventions where 75% or more
of the students were of a specified SES status as
being “predominantly” of that group and thus
roughly representative of SEL intervention impact
on that demographic. Fourteen of the interventions
included children predominantly (i.e., at least 75%)
from poverty level and working-class families, and
those interventions served as our lower income
comparison group. Nine studies included children
predominantly from middle- and upper-class families, and in an additional 28 studies, the student
sample was a socioeconomically mixed group with
no predominant SES.

 6

Taylor, Oberle, Durlak, and Weissberg

Participant Race
Only 54 studies reported any information on
race; of those, 46 studies provided enough data to
calculate the percentage of students of color (i.e.,
non-White students; M = 45.85, SD = 35.53). Similar
to the procedure for SES, we also identified interventions where 75% or more of the students were
of a specified racial background. Twenty-one studies had a predominately White student sample and
14 studies involved predominantly students of
color, with seven of the latter having a predominately Black student population. An additional 19
studies involved racially diverse student samples
that were not predominantly of any group.
Dependent Variables
The outcomes of interest from the interventions
were limited to those measures that reported
changes in students. Effects on caregivers or teachers were not included in the analyses. Outcomes
were sorted into seven distinct categories assessing
positive social and emotional assets (social and emotional skills; attitudes toward self, others, and school)
and positive (positive social behaviors; academic performance) and negative (conduct problems; emotional distress; substance use) indicators of well-being.
Social and emotional skills.
This outcome consisted of such skills as identifying emotions, perspective taking, self-control, interpersonal problem
solving, conflict resolution and coping strategies,
and decision making, depending on the specific targets and developmental levels of the participating
samples. All these outcomes were measured in a
hypothetical situation or using structured tasks or
questionnaires (e.g., feeling word questionnaires,
conflict resolution role plays, or interviews). Any
reports of general behaviors or observations of students’ skills occurring during daily situations were
instead placed in the positive social behavior category.
Attitudes toward self, others, and school.
This
outcome assessed students’ attitudes about the self,
others, and school. Self-perceptions included measures of self-esteem, self-efficacy, or self-concept.
Attitudes about others reflected prosocial beliefs
such as disapproval of substance abuse and violent
behavior or endorsements related to understanding
and helping others. Finally, attitudes related to
school included both beliefs about the teacher,
learning, or education in general as well as school
bonding, connectedness, or belonging. All the attitude outcomes came from student self-reports.

Positive social behavior.
Positive behaviors
included measures that represent the use of social
skills in naturalistic settings. This category reflects
prosocial behaviors outside of the practice context
of the intervention typically measured by teacher or
students reports (e.g., cooperation, use of problemsolving skills, or efforts to help others).
Academic performance.
This category included
data from school records of either grades or
achievement test scores. Students’ self-reports of
their academic performance were not included.
Conduct problems.
This outcome included
reports of problem behaviors, such as violence,
aggression, bullying, classroom disruption, or noncompliance. Disciplinary referrals or suspensions
were also included in this category as indications of
these behaviors. These measures of externalizing
behaviors could either be self-reported or observed
by others, but all data on referrals or suspensions
came directly from official school records.
Emotional distress.
This category included primarily symptoms of internalizing difficulties, such
as depression, anxiety, and stress, which were typically based on student reports.
Substance use.
All measures of initiation, use,
and misuse of intoxicating substances, including
both legal and illegal drugs, were included in this
outcome category. Only self-reports of drug use
were included. Intentions, attitudes, or perceptions
of drug use were not included.
Additional outcomes.
Finally, a few studies
examined important developmental outcomes that
did not fit into the above seven outcome categories.
These include positive variables, such as high
school graduation, income/employment, relationships, and safe sexual behaviors, and negative outcomes, such as mental health problems and arrests.
These data were too infrequent to include in the
meta-analyses but are presented separately as additional measures of program impacts.
ES Calculations
For each outcome, an ES was calculated as
Hedge’s g (Hedges & Olkin, 1985). If results for an
outcome were only reported as “nonsignificant”
(6.6% of the 361 follow-up outcomes), the ES was set
to zero. Whenever pre-intervention differences
between the control and intervention groups were
reported for an outcome, both post and follow-up ES
were adjusted by subtracting the pre-ES from the
post or follow-up effect. ESs were calculated so that
positive values indicated a more beneficial outcome
favoring the intervention group over the controls.

 Follow-Up Effects of SEL Programs

Prior to any analyses, outlier values for follow-up
ESs, post-ESs, and sample sizes for control and
experimental groups within each outcome category
were winsorized according to common meta-analytic
practices (Lipsey & Wilson, 2001). This process
recodes extreme outliers to a more normally distributed value instead of removing them altogether,
which prevents any individual study with large ES
from having an undue influence on the analyses. All
outliers were winsorized to a value representing
3 SD of the mean ES of their respective outcome category with the outliers removed. In all analyses, ESs
were also weighted by their inverse variance to give
a greater weight to studies with larger sample sizes.
For each analysis, ESs were aggregated based on
the dependent variable of interest, so that each intervention contributed only one ES per outcome. Thus,
when multiple outcomes from the same outcome category were assessed within a single study, these outcomes were averaged to obtain a single ES.
All of the meta-analyses used a random effects
model with a maximum likelihood estimation procedure to arrive at ES and 95% confidence intervals
(Lipsey & Wilson, 2001). A mean ES is considered
significantly different from zero at p < .05 when its
confidence intervals do not include zero. When conducting group comparisons within a meta-analytic
analysis of variance, the weighted mean ES and
95% confidence intervals are calculated for each
group, and those ES are considered to be significantly different from each other if their confidence
intervals do not overlap.
General Analytic Procedures
Our first set of analyses evaluated the mean ESs at
follow-up for each of the seven outcome categories to
test our hypothesis that all of the social and emotional
assets and positive and negative indicators of wellbeing would be significant. Our second set of analyses
tested our second hypothesis regarding the effectiveness of SEL interventions across diverse global, racial,
and socioeconomic groups using meta-analytic analyses of variance. Finally, we tested the contribution of
postintervention assets in predicting students’ followup outcomes. These analyses employed a series of
meta-regressions as explained below.

Results
Effects of SEL Interventions at Follow-Up
As shown in Table 2, the hypothesized statistically significant positive effects of SEL interventions

7

were found at follow-up for each of the seven outcome categories. Mean ESs ranged from .13 to .33,
with SEL program participants benefiting significantly more than controls across all of the social
and emotional assets and positive and negative
indicators of well-being. The mean follow-up period
varied from 56 to 195 weeks depending on the particular outcome category.
The possible impact of publication bias on these
effects was examined using trim-and-fill analyses
for each of the seven outcome categories. Following
procedures described by Duval and Tweedie (2000),
we first identified and “trimmed” the studies with
smaller sample sizes and larger significant ES until
a symmetrical funnel plot stabilized. The number of
trimmed studies is an estimate of the missing studies for each outcome, which ranged from 3 to 14.
ES values for the missing studies were estimated
and “filled” into the funnel plot by mirroring the
trimmed effects around the center. These trimmed
and filled ES were added into the analyses to create
an adjusted estimate of mean ES in each category.
Effects in every category remained significant after
including these adjustments for possible publication
bias. Complete details can be found in Supporting
Information.
Although the effects at postintervention were
not of primary interest in this meta-analytic
review, they provided a baseline context and were
used in the prediction of follow-up effects. For the
82 studies, measures of social and emotional
assets at post showed significant positive impacts
of the intervention, with participants having stronger SEL skills (n = 36, ES = .17, 95% CI [.11, .24])
and improved attitudes (n = 25, ES = .17, 95% CI
[.09, .24]) compared with controls. Program participants also faired significantly better than controls
at post on academic performance (n = 8, ES = .22,
95% CI [.07, .36]), emotional distress (n = 38,
ES = .12, 95% CI [.06, .19]), and drug use (n = 21,
ES = .12, 95% CI [.04, .19]). However, postintervention mean ESs were not significant for either
positive social behaviors (n = 28, ES = .06, 95% CI
[ .01, .13]) or conduct problems (n = 30, ES = .07,
95% CI [.00, .14]). Greater details on the postintervention analyses can be found in Supporting
Information.
Effects of SEL Interventions With Diverse Populations
In order to examine the effectiveness of SEL
interventions across demographic groups, it was
necessary to collapse the outcome categories into a
single intervention level ES to obtain sufficient

 8

Taylor, Oberle, Durlak, and Weissberg

Table 2
Mean Effect, Confidence Intervals, Follow-Up Periods, and Improvement Indices for Total Sample
Follow-up ES by outcome category
Social and emotional
assets

ES
95% CI
N
Mean follow-up (weeks)
Improvement index, %

Positive and negative indicators of well-being

SEL skills

Attitudes

Positive
social
behavior

.23a
.15, .31
29
56
9.09

.13a
.05, .21
26
103
5.17

.13a
.05, .21
28
89
5.17

Academic
performance

Conduct
problems

Emotional
distress

.33a
.17, .49
8
195
12.93

.14a
.07, .21
34
113
5.56

.16a
.08, .23
35
88
5.64

Drug use
.16a
.09, .24
28
139
5.64

Note. ES = effect size; SEL = social and emotional learning. aMean effect is significantly different from zero at the .05 level.

sample sizes. Significant positive effects for SEL
program participants were found at follow-up
across all demographic subgroups. That is, there
was no significant difference in the impact of SEL
at 6 months or more postintervention between
interventions involving predominately White students (n = 21, ES = .23, 95% CI [.14, .32]), predominately students of color (n = 13, ES = .18, 95% CI
[.06, .30]), or interventions containing a diverse student population (n = 19, ES = .17, 95% CI [.08,
.27]). There was also no significant difference in follow-up ES between interventions involving predominately low- and working-class students
(n = 13, ES = .21, 95% CI [.08, .33]) compared with
those of another SES status (i.e., either predominately middle and upper class or diverse SES samples; n = 36, ES = .23, 95% CI [.15, .30]). Finally, a
comparison of follow-up effects for interventions
conducted in the United States (n = 43, ES = .20,
95% CI [.14, .26]) and abroad (n = 38, ES = .16, 95%
CI [.09, .22]) revealed comparable positive effects in
both contexts.
What Predicts Follow-Up Effects?
Before examining whether skills, attitudes, or
both predicted follow-up effects, we conducted
meta-regressions to examine the possible influence
of 21 alternative predictors on the combined mean
ES across all outcomes. The alternative predictors
included methodological variables (i.e., randomization, validity of outcome measures, reliability of
outcome measures, source of outcome data, quality
of implementation, length of follow-up, total attrition, and relative attrition), features of the intervention (i.e., SAFE practices, intervention format,
duration, number of sessions, and tailored content),

and characteristics of the participants (i.e., average
age, developmental level, percentage of White students, percentage of Black students, percentage
female, total sample size, school community location, and domestic or international population).
Only significant findings for these alternative predictors are reported here; complete information
about these variables and the analyses are in Supporting Information.
Two variables emerged as significant predictors.
Higher total sample attrition at follow-up was associated with lower ES (B = .30; b = .24, p < .05).
Participant age was also significant and negatively
related to follow-up ES (B = .02; b = .21,
p < .05). We examined the age finding further by
dividing age into three developmental levels. The
31 interventions with student participants in childhood (ages 5–10) had the largest follow-up effects
(ES = .27, 95% CI = [.19, .34]); their effects at follow-up were significantly higher than the 37 interventions with early adolescent students (ages 11–13;
ES = .12, 95% CI = [.06, .18]). Only 11 interventions
focused on adolescent populations (ages 14–18),
and these students did not differ significantly at follow-up from either of the other age groups
(ES = .18, 95% CI = [.05, .31]).
In order to create sufficient sample size for the
meta-regression analyses predicting follow-up
effects, it was necessary to average the ESs across
two positive (i.e., prosocial behaviors and academic
performance) and three negative (i.e., conduct problems, emotional distress, and drug use) indicators
of well-being to produce a single dependent outcome variable indicative of follow-up effects. This
composite mean effect for these five outcomes was
also significant at follow-up (ES = .18, 95%
CI = [.13, .23]).

 Follow-Up Effects of SEL Programs

First, to explore the relative influence of attrition
and age, a meta-regression was conducted with
both predictors of the indicators of well-being
added simultaneously. With their shared variance
removed, age was no longer a significant predictor
(B = .01; b = .11, ns), but total attrition remained
significant (B = .36; b = .32, p < .01). Next, we
conducted a hierarchical meta-regression to test the
hypothesized relationship between postintervention
ES for social and emotional assets and the mean follow-up ES for positive and negative indicators of
well-being. Hierarchical regression was used to
demonstrate the strength of the relationship
between postintervention assets and follow-up
well-being after controlling for any variance
explained by attrition. Attrition was entered alone
as a control into the first block of a hierarchical
meta-regression, and it explained 9% of the variance (R2 = .09, B = .37; b = .32, p < .01). The
combined postintervention ES for assets was added
in the second block. As we predicted, a significant
relationship between assets at post and indicators
of well-being at follow-up (R2 change = .15; B = .29;
b = .35, p < .01) was found in the 42 studies with
relevant data for all variables. Higher levels of
social and emotional assets at post were associated
with higher levels of well-being at follow-up, predicting an additional 15% of the variance after controlling for attrition.
Finally, to examine possible differences in the
predictive power of skills versus attitudes, hierarchical meta-regressions were run again with the
postintervention ES for either social-emotional skills
or attitudes entered separately in the second block.
Social-emotional skills had a significant positive
relationship (R2 change = .16; B = .33; b = .39,
p < .01; n = 31), such that better skills at postintervention predicted higher follow-up effects on positive and negative indicators of well-being,
predicting an additional 16% of the variance after
controlling for attrition. For the interventions measuring attitudes at postintervention, however, the
relationship was not significant (R2 change = .05;
B = .19; b = .24, ns; n = 20).
Additional Follow-Up Outcomes
Table 3 presents data on several additional and
critical developmental outcomes assessed by some
investigators at various follow-up periods. Twentythree of the 29 ES in Table 3 are positive. Only four
are zero and two are negative in sign favoring the
control group. In most cases, investigators have
reported positive results for outcomes such as

9

positive relationships with peers and family, school
attendance, safe sexual behaviors, graduation rates,
college attendance, arrests, and various indices of
mental health adjustment. To reflect their practical
importance, we have translated the ES for the
dichotomous outcomes in Table 3 in two ways. First,
we used the Binomial Effect Size Display (Rosenthal
& Rubin, 1982) to indicate how the obtained ES
reflects the percentage differences between the
intervention and control groups. For example, the
ES of .12 and .22 reported, respectively, on high
school and college graduation at long-term followup (Bradshaw, Zmuda, Kellam, & Ialongo, 2009;
Hawkins et al., 2008) reflects 6% more students succeeding in high school, and 11% more students
completing college among the intervention group
compared with the controls. The ES of .37 reported
by Eddy, Reid, Stoolmiller, and Fetrow (2003) at a
120-week follow-up reflects 18.5% fewer arrests
among the intervention group compared with the
controls, and the effect of .12 for placement in special education (Bradshaw et al., 2009) indicates 6%
fewer placements for the intervention group.
In addition, we have applied the findings of several authors who have conducted cost analyses to
calculate the lifetime monetary benefit or cost per event
of achieving certain outcomes (Carnevale, Rose, &
Cheah, 2011; Chambers, Parrish, & Harr, 2004;
Chesson, Blanford, Gift, Tao, & Irwin, 2004; Cohen,
Piquero, & Jennings, 2010). When they can be calculated, the figures for these outcomes are listed in
the last column of Table 3, and they have been
updated to 2015 U.S. dollars. For example, the estimated benefit in lifetime earnings for graduating
from high school compared to dropping out is
worth over $367,000 to each graduating student. Each
teenage pregnancy costs nearly $150,000, and the
savings in reducing arrest and delinquency rates
are also considerable. Reducing the incidence of
drug abuse and conduct disorder by one person
can potentially save between one to nearly four million dollars. Although only a minority of the studies in this review is represented in Table 3, the
results add support for the ability of some SEL programs to influence critical developmental outcomes
during follow-up.

Discussion
There are five important findings from the current
review. The first involves the durability of impacts
from PYD programs. Students in school-based SEL
interventions continued to demonstrate significant,

 10

Taylor, Oberle, Durlak, and Weissberg

Table 3
Additional Follow-Up Outcomes

Outcome
category
Relationships

School status

Report
Harnett and Dadds (2004)
Gesten et al. (1982)
Sawyer et al. (1997)
Sarason and Sarason (1981)
Elias, Gara, Schuyler,
Branden-Muller, and Sayette (1991)a
Elias et al. (1991)a
Gottfredson, Jones, and Gore (2002)
Bradshaw et al. (2009)
Hawkins et al. (1999)
Felner et al. (1993)
Hawkins et al. (2008)
Bradshaw et al. (2009)
Bradshaw et al. (2009)
Hawkins et al. (2008)

Sexuality

Income/
employment
Criminality

Mental health

Harrington, Giles, Hoyle,
Feeney, and Yungbluth (2001)a
Harrington et al. (2001)a
Hawkins et al. (2008)
Hill et al. (2014)
Hawkins et al. (2008)
Hawkins et al. (2008)
Hawkins et al. (2008)
Cook and Hirschfield (2008)
Eddy et al. (2003)
Hawkins et al. (2008)
Ialongo, Poduska,
Werthamer, and Kellam (2001)
Ialongo et al. (2001)
Riggs and Pentz (2009)
Hawkins et al. (2008)
Hawkins et al. (2008)

ES

Family cohesion
Peer acceptance
Peer relationships
Attendance
Attendance

144
52
52
52
312

.19
.29
.11
.47
0

Attendance
Attendance
Placement in special
education class
Repeating a grade
High school dropout
High school dropout
High school graduate
College attendance
College degree
(Assoc or BA)
Safe sexual behaviors

312
26
572

.44
.28
.12

6

312
156
780
572
624
780

.23
.51
.15
.12
.21
.22

12.5
25.5
7.5
6
10.5
11

$367,687b
$637,621b
$1,138,054b

38.5
6.5

$9,940
$147,351

6

$240,221
$175,702

Outcome

Safe sexual behaviors
Safe sexual behavior
STD diagnosis
Pregnancy and births
Income
Socioeconomic status
Juvenile justice
involvement
Arrests
Arrests
Ever used mental
health services
Diagnosis of conduct
disorder
Adult mental health
services use
Clinical disorder
Substance abuse
diagnosis

52

% Advantage
for intervention
group

Lifetime
monetary benefit
or cost saving
per event

Follow-up
period
(in weeks)

$93,781

0

52
780
936
780
780
780
208

0
0
.76
.13
.06
.22
.12

120
780
260

.37
.04
.18

18.5
2
9

260

.2

10

728

.13

6.5

780
780

.27
.03

13.5
1.5

$3,950,000

$1,051,688

Note. Monetary benefits or costs could not be estimated for all outcomes. ES = effect size. aFor Elias et al. (1991) and Harrington et al.
(2001), the articles each contributed two unique interventions to the meta-analysis, resulting in the report of two effect sizes in the outcome category, one for intervention A and one for intervention B. bThis is the incremental value in lifetime earnings of continuing education compared to not finishing high school.

positive benefits in seven outcomes collected, on
average, from 56 weeks and up to 195 weeks (i.e.,
3.75 years) following program participation. These
results are impressive; although at first glance, the
follow-up mean ES may seem quite modest. However, Cohen’s (1988) suggestions for judging the

magnitude of effects as small (.20), medium (.50), or
large (.80) are not applicable for universal promotion or prevention studies nor are they relevant for
interpreting follow-up data. For example, the mean
effect of .33 on academic performance (based on
grades and test scores drawn from school records,

 Follow-Up Effects of SEL Programs

obtained at a mean follow-up period of 195 weeks)
compares favorably to the post effects obtained by
many educational interventions (Hill, Bloom, Black,
& Lipsey, 2008). Although based on only eight
studies, these long-term academic outcomes are
notable. For the other outcomes, there are no current empirical standards for judging the magnitude
of follow-up effects for interventions designed to
promote youth development. Thus, current data are
new to the fields of SEL and PYD. Therefore, the
mean effects reported here may serve as initial
benchmarks that other researchers can use to compare the success of their efforts.
A second important finding involves the dual
benefits of SEL interventions in terms of affecting
both positive and negative indicators of well-being.
The main purpose of PYD is to set young people
on a positive developmental trajectory so that they
are prepared to fully realize their potential and are
resilient to the challenges they may face. By fostering social and emotional skills and positive attitudes in students, the school-based, universal SEL
interventions reviewed in this study achieved these
ends during follow-up in terms of significantly
improving skills, positive attitudes, prosocial behavior, and academic performance. These programs
were also able to serve as a protective factor against
the development of subsequent problems (i.e., conduct problems, emotional distress, and drug use).
In other words, the enhancement of asset development through PYD approaches like SEL can have
both promotion and preventive impact (National
Research Council and Institute of Medicine, 2009).
Third, the SEL approach to PYD was beneficial
for all demographic groups that we were able to
examine in this review. Consistent positive effects
at follow-up were found for SEL interventions with
student populations from different racial groups
and socioeconomic statuses, and for both domestic
and international student bodies. This finding is in
alignment with the conceptual perspective that the
social and emotional assets promoted in SEL can
support the positive development of students from
diverse family backgrounds and geographic contexts (Greenberg et al., 2003). However, although
we did not find differential effects among groups,
this should not be interpreted as an endorsement
that “one size fits all” when it comes to SEL intervention. It is critical that program developers and
researchers examine strategies to design and implement interventions in culturally competent ways
(Hecht & Shin, 2015; Hoffman, 2009).
The fourth important finding concerns the positive relationship between stronger social and

11

emotional assets at post and higher levels of wellbeing at follow-up. Our meta-regressions support
the conceptual model of SEL in which the targeting
of various social and emotional assets will be associated with significant improvement in students’
long-term adjustment (Durlak et al., 2015). When
we examined the differential associations of socialemotional skills and positive attitudes, we found
that enhanced skills, rather than attitudes, predicted
long-term follow-up effects. The impact of skills in
our analysis is consistent with the growing literature documenting that improving children’s intrapersonal and interpersonal competencies—such as
self-regulation, problem solving, and relationship
skills—enhances children’s academic performance
and behavior (Domitrovich, Staley, Durlak, &
Weissberg, 2016; Sorensen, Dodge, & Conduct
Problems Prevention Research Group, 2015). In
addition, the finding that attitudes achieved at post
were not a significant predictor of follow-up effects
bears further consideration. The result could be
seen to offer empirical support to an extensive
research base in the promotion and prevention literature emphasizing the particular importance of skill
training (i.e., competency enhancement) in improving the adjustment of youth (e.g., Durlak, 2014; Wilson & Lipsey, 2007). Alternatively, because attitude
outcomes varied considerably among the studies
assessed in the model, it is possible that some attitudes are effective predictors of later well-being,
but their impact is washed out when combined
with other less impactful attitudes. Recent reviews
point to the potential of enhancing personal and
social attitudes to improve academic performance
and behavior (e.g., Farrington et al., 2012; Yeager &
Walton, 2011). Identifying which attitudes might be
effective for enhancing later well-being and how to
better coordinate SEL programming that fosters
improved skills, attitudes, and behavioral functioning are important priorities for future research.
A fifth and final set of findings involve the positive effects on several additional important developmental outcomes, collected up to 936 weeks (i.e.,
18 years) postintervention, which were reported in
a subsample of studies. For example, improving
future social relationships, increasing high school
graduation rates and college attendance, and reducing later negative outcomes such as arrests or the
presence of clinical disorders are notable achievements. These are the type of outcomes for which
seemingly small ESs can nevertheless reflect practical advantages for the intervention. For example,
based on an ES of “only” .12, most educators
would welcome an intervention that could reduce

 12

Taylor, Oberle, Durlak, and Weissberg

special education placements or increase high
school graduation rates by 6%. The outcomes noted
in Table 3 are not only indices of positive developmental trajectories for program participants that
appear at follow-up, but they are also evidence of
how much both participating students and society
can profit from the sometimes substantial monetary
benefits and cost savings that can be achieved by
SEL programs. These findings build on a recent
study examining the economic value of six SEL
interventions that found for every dollar invested
there was a return of 11 dollars (Belfield et al.,
2015). Only a small group of the 82 interventions
collected data on these later critical developmental
outcomes, and we encourage other investigators to
include such outcomes in their follow-up studies.
Limitations and Some Future Research Directions
Six limitations suggest directions for future
research. First, it is noteworthy that most interventions in this meta-analysis incorporated the following four SAFE program features that have been
suggested as best practices for SEL intervention
(Durlak et al., 2011): Sequenced: The program had a
coordinated progression of activities or practices to
build competencies; Active: Participatory elements
such as role plays involved students in active learning of SEL competencies; Focused: There was a dedicated time or specific program element that was
focused on developing SEL competencies; and
Explicit: The program identified specific SEL competencies that it was trying to develop within the
intervention. Due to the low incidence of programs
that did not meet the SAFE criteria (n = 9), it was
not possible to evaluate whether non-SAFE SEL
programming also might lead to improved longterm adjustment.
Second, in the examination of alternative predictors and within our meta-regressions, we had to
combine all follow-up outcomes into a single
dependent variable to establish sufficient cell sizes
for the analyses. Therefore, we could not assess
which variables were significant predictors of each
separate indicator of positive or negative wellbeing. Furthermore, not all reports contained
assessments of either social and emotional skills or
attitudes at postintervention, which limited our predictive power in the regression analyses. This is a
particular oversight given that all interventions
included in this study targeted competency building in at least one of the five SEL core competency
areas. Future research should consistently assess
these social and emotional assets, so that mediators

of SEL interventions on key positive developmental
outcomes can be more rigorously evaluated.
Third, almost three quarters of the studies (i.e.,
72.2%) relied on self-report measures to evaluate
student outcomes. Two of the outcome categories,
prosocial attitudes and drug use, were intentionally
limited to self-report measures. However, although
it is important to include young people’s perspectives regarding their skills, attitudes, and behaviors,
future research should also incorporate additional
measures from the perspectives of others (e.g.,
teachers, parents, observers) and public record data
(e.g., graduation rates, employment, income). Examining these kinds of follow-up data (Table 3) can
provide a foundation for a more rigorous exploration of the return on investment of SEL programming.
Fourth, the alternative predictors examined in
this meta-analysis do not allow us to draw conclusions about what specific features make SEL interventions more or less effective. Aside from
postintervention social and emotional skills, only
attrition was a significant predictor in the hierarchical regression analysis. Attrition can be a challenge
in longitudinal studies because it is logical to
assume that more participants may be lost as the
time span lengthens, so researchers should evaluate
how this possibility may affect outcomes. In addition, we encourage future research to also examine
how environmental supports (e.g., parenting, teacher instructional practices) influence long-term outcomes (Weissberg et al., 2015). Greater attention to
the environmental enhancement components of SEL
intervention would allow a more complete understanding of how PYD trajectories are fostered
within these interventions. However, studying
those variables was beyond to scope of the current
meta-analysis.
Fifth, although age was significantly negatively
related to follow-up effects when examined as an
individual predictor, we urge caution in interpreting the finding as an indication that SEL intervention is more appropriately targeted in childhood
than early adolescence. For one, the effect of age
was reduced to nonsignificance when both attrition
and age were entered together into the first step of
the regression predicting indicators of well-being.
In addition, age has significant covariates in this
sample that could be considered as alternative
explanations. For example, the average intervention
duration is significantly negatively correlated with
participant age (r = .34, p < .01) so that interventions targeted at younger children were also delivered over a longer period of time.

 Follow-Up Effects of SEL Programs

Sixth, although we found consistent positive
effects for SEL interventions with students from
diverse racial and socioeconomic demographics,
these analyses were limited by the lack of data in
many studies. More than 40% (34 of 82) of the studies did not report any specific percentages of student ethnicity, and only a third (26 of 82) reported
the percentage of students in poverty. It is critical
in future research to assess if students from diverse
socioeconomic and racial and ethnic groups
respond differently to interventions on a variety of
outcomes. To do so, authors must provide complete
demographic data on program participants, and
conduct and report subgroup analyses whenever
possible.
Concluding Comments
School-based SEL represents an important set of
approaches to promote the positive academic
growth, behavior, and development of youth (Durlak et al., 2011, 2015; Sklad et al., 2012). Building
on the short-term benefits that hundreds of studies
have documented, this review provides evidence
for long-term positive effects that school-based SEL
programs can foster across diverse geographic contexts and age groups. Findings also add to the
growing literature on the potential economic and
societal return on investment for SEL programming
(Belfield et al., 2015). It is noteworthy that these
findings lend empirical support to the opinions of a
large majority of teachers in the United States, who
indicated in a recent national survey that they
believe that (a) students from all types of backgrounds, both poor and wealthy, would benefit
from SEL in school; (b) SEL programming can prepare students to move successfully through school
and college, and to be productive workers and
good citizens; and (c) they (teachers) should play a
key role in promoting the positive social, emotional, and academic growth of students (Bridgeland, Bruce, & Hariharan, 2013). However, for
school-based SEL to be an effective approach to
fostering PYD, educators need support to implement and appropriately adapt interventions such as
those in the current meta-analysis. Without quality
implementation, the potential positive impact of
SEL programming is reduced (Durlak et al., 2011).
With the support of sound federal and state policies, district and school leaders, quality professional
preparation, and ongoing, embedded professional
learning, it will be possible to enhance the positive
development of many more students through SEL.

13

References
Belfield, C., Bowden, B., Klapp, A., Levin, H., Shand, R.,
& Zander, S. (2015). The economic value of social and emotional learning. New York, NY: Center for Benefit-Cost
Studies in Education, Teachers College, Columbia
University.
Benson, P. L., Leffert, N., Scales, P. C., & Blyth, D. A.
(1998). Beyond the “village” rhetoric: Creating healthy
communities for children and adolescents. Applied
Developmental Science, 2, 138–159. https://doi.org/10.
1207/s1532480xads0203_3
Bonell, C., Hinds, K., Dickson, K., Thomas, J., Fletcher,
A., Murphy, S., . . . Campbell, R. (2016). What is positive youth development and how might it reduce substance use and violence? A systematic review and
synthesis of theoretical literature. BMC Public Health, 16
(135), 1–13. https://doi.org/10.1186/s12889-016-2817-3
Bradshaw, C. P., Zmuda, J. H., Kellam, S. G., & Ialongo,
N. S. (2009). Longitudinal impact of two universal preventive interventions in first grade on educational outcomes in high school. Journal of Educational Psychology,
101, 926–927. https://doi.org/10.1037/a0016586
Bridgeland, J., Bruce, M., & Hariharan, A. (2013). The
missing piece: A national teacher survey on how social and
emotional learning can empower children and transform
schools. Washington, DC: Civic Enterprises.
Carnevale, A. P., Rose, S. J., & Cheah, B. (2011). The college payoff: Education, occupations, lifetime earnings. Washington, DC: The Georgetown University Center on
Education and the Workforce.
Catalano, R. F., Berglund, L. M., Ryan, J. A. M., Lonczak,
H. S., & Hawkins, J. D. (2002). Positive youth development in the United States: Research findings on evaluations of positive youth development programs.
Prevention and Treatment, 5, 15a. https://doi.org/10.
1037/1522-3736.5.1.515a
Chambers, J. G., Parrish, T. B., & Harr, J. J. (2004). What
are we spending on special education services in the United
States, 1999–2000? Washington, DC: American Institutes
for Research.
Chesson, H. W., Blanford, J. M., Gift, T. L., Tao, G., &
Irwin, K. L. (2004). The estimated direct medical cost of
sexually transmitted diseases among American youth,
2000. Perspectives on Sexual and Reproductive Health, 36,
11–19. https://doi.org/10.1363/3601104
Cohen, J. (1988). Statistical power analysis for the behavioral
sciences (2nd ed.). Hillsdale, NJ: Erlbaum.
Cohen, M. A., Piquero, A. R., & Jennings, W. G. (2010).
Estimating the costs of bad outcomes for at-risk youth
and the benefits of early childhood interventions to
reduce them. Criminal Justice Policy Review, 21, 391–434.
https://doi.org/10.1177/0887403409352896
Collaborative for Academic, Social, and Emotional Learning. (2013). CASEL guide: Effective social and emotional
learning programs: Preschool and elementary school edition.
Chicago IL: Author.

 14

Taylor, Oberle, Durlak, and Weissberg

Conduct Problems Prevention Research Group. (1999).
Initial impact of the Fast Track prevention trial for conduct problems: I. The high-risk sample. Journal of Consulting and Clinical Psychology, 67, 631–647. https://doi.
org/10.1037/0022-006X.67.5.631
Conduct Problems Prevention Research Group. (2010).
The effects of a multiyear universal social–emotional
learning program: The role of student and school characteristics. Journal of Consulting and Clinical Psychology,
78, 156–168. https://doi.org/10.1037/a0018607
Cook, T. D., & Hirschfield, P. J. (2008). Comer’s school
development program in Chicago: Effects on involvement with the juvenile justice system form the late elementary through the high school years. American
Educational Research Journal, 45, 38–67. https://doi.org/
10.3102/0002831207308648
Domitrovich, C. E., Staley, K., Durlak, J. A., & Weissberg,
R. P. (2016). Social and emotional learning: The importance of social-emotional competencies and school climate in fostering positive student outcomes.
Manuscript submitted for publication.
Durlak, J. A. (2014). Programming for effective primary
prevention and health promotion. In T. P. Gullotta &
M. Bloom (Eds.), Encyclopedia of primary prevention and
health promotion (2nd ed., pp. 102–111). New York, NY:
Springer.
Durlak, J. A., Domitrovich, C. E., Weissberg, R. P., &
Gullotta, T. P. (Eds.). (2015). Handbook of social and emotional learning: Research and practice. New York, NY:
Guilford.
Durlak, J. A., Weissberg, R. P., Dymnicki, A. B., Taylor,
R. D., & Schellinger, K. B. (2011). The impact of
enhancing students’ social and emotional learning: A
meta-analysis of school-based universal interventions.
Child Development, 82, 405–432. https://doi.org/10.
1111/j.1467-8624.2010.01564.x
Duval, S., & Tweedie, R. (2000). Trim and fill: A simple
funnel-plot- based method of testing and adjusting for
publication bias in meta-analysis. Biometrics, 56, 455–
463. https://doi.org/10.1111/j.0006-341X.2000.00455.x
Eddy, M., Reid, J. B., Stoolmiller, M., & Fetrow, A. (2003).
Outcomes during middle school for an elementary
school-based preventive intervention for conduct problems: Follow-up results from a randomized trial. Behavior Therapy, 34, 535–552. https://doi.org/10.1016/
S0005-7894(03)80034-5
Elias, M. J., Gara, M. A., Schuyler, T. F., Branden-Muller,
L. R., & Sayette, M. A. (1991). The promotion of social
competence: Longitudinal study of a preventive schoolbased program. American Journal of Orthopsychiatry, 61,
409–417. https://doi.org/10.1037/h0079277
Elias, M. J., Leverett, L., Duffell, J. C., Humphrey, N.,
Stepney, C., & Ferrito, J. (2015). Integrating SEL with
related prevention and youth development approaches.
In J. A. Durlak, C. E. Domitrovich, R. P. Weissberg, &
T. P. Gullotta (Eds.), Handbook for social and emotional
learning: Research and practice (pp. 33–49). New York,
NY: Guilford.

Elias, M. J., Zins, J. E., Weissberg, R. P., Frey, K. S.,
Greenberg, M. T., & Haynes, N. M. (1997). Promoting
social and emotional learning: Guidelines for educators.
Alexandria, VA: ASCD.
Farrington, C. A., Roderick, M., Allensworth, E.,
Nagaoka, J., Keyes, T. S., Johnson, D. W., & Beechum,
N. O. (2012). Teaching adolescents to become learners The
role of noncognitive factors in shaping school performance:
A critical literature review. Chicago, IL: University of
Chicago Consortium on Chicago School Research.
Felner, R. D., Brand, S., Adan, A. M., Mulhall, P. F., Flowers, N., Sartain, B., & Dubois, D. L. (1993). Restructuring
the ecology of the school as an approach to prevention
during school transitions: Longitudinal follow-ups and
extensions of the School Transitional Environment Project (STEP). Prevention in Human Services, 10, 103–136.
https://doi.org/10.1300/J293v10n02_07
Gesten, E. L., Rains, M. H., Rapkin, B. D., Weissberg, R.
P., de Apocada, R. F., Cowen, E. L., Bowen, R., (1982).
Training children in social problem-solving competencies: A first and second look. American Journal of Community Psychology, 10, 95–115. https://doi.org/10.1007/
BF00903307
Gottfredson, G. D., Jones, E. M., & Gore, T. W. (2002).
Implementation and evaluation of a cognitive-behavioral intervention to prevent problem behavior in a
disorganized school. Prevention Science, 3, 43–56.
https://doi.org/10.1023/A:1014671310038
Greenberg, M. T., Weissberg, R. P., O’Brien, M. U., Zins,
J. E., Fredericks, L., Resnik, H., & Elias, M. J. (2003).
Enhancing school-based prevention and youth development through coordinated social, emotional, and academic learning. American Psychologist, 58, 466–474.
https://doi.org/10.1037/0003-066X.58.6-7.466
Harnett, P. H., & Dadds, M. R. (2004). Training school
personnel to implement a universal school-based prevention of depression program under real-world conditions. Journal of School Psychology, 42, 343–357. https://
doi.org/10.1016/j.jsp.2004.06.004
Harrington, N. G., Giles, S. M., Hoyle, R. H., Feeney, G. J.,
& Yungbluth, S. C. (2001). Evaluation of the All Stars
character education and problem behavior prevention
program: Effects on mediator and outcome variables for
middle school students. Health Education and Behavior, 28,
533–546. https://doi.org/10.1177/109019810102800502
Hawkins, J. D., Catalano, R. F., Kosterman, R., Abbott, R.,
& Hill, K. G. (1999). Preventing adolescent health-risk
behaviors by strengthening protection during childhood. Archives of Pediatrics and Adolescent Medicine, 153,
226–234. https://doi.org/10.1001/archpedi.153.3.226
Hawkins, J. D., Kosterman, R., Catalano, R. F., Hill, K.
G., & Abbott, R. D. (2008). Effects of social development intervention in childhood 15 years later.
Archives of Pediatrics and Adolescent Medicine, 162,
1133–1141.
https://doi.org/10.1001/archpedi.162.12.
1133
Hecht, M. L., & Shin, Y. (2015). Culture and social and
emotional competencies. In J. A. Durlak, C. E.

 Follow-Up Effects of SEL Programs
Domitrovich, R. P. Weissberg, & T. P. Gullotta (Eds.),
Handbook for social and emotional learning: Research and
practice (pp. 50–64). New York, NY: Guilford.
Hedges, L. V., & Olkin, I. (1985). Statistical methods for
meta-analysis. Orlando, FL: Academic Press.
Hill, K. G., Bailey, J. A., Hawkins, J. D., Catalano, R. F.,
Kosterman, R., Oesterle, S., & Abbott, R. D. (2014). The
onset of STI diagnosis through age 30: Results from the
Seattle Social Development Project intervention. Prevention Science, 15, S19–S32. https://doi.org/10.1007/
s11121-013-0382-x
Hill, C. J., Bloom, H. S., Black, A. R., & Lipsey, M. W.
(2008). Empirical benchmarks for interpreting effect
sizes in research. Child Development Perspectives, 2, 172–
277. https://doi.org/10.1111/j.1750-8606.2008.00061.x
Hoffman, D. M. (2009). Reflecting on social and emotional
learning: A critical perspective on trends in the United
States. Review of Educational Research, 79, 533–556.
https://doi.org/10.3102/0034654308325184
Ialongo, N., Poduska, J., Werthamer, L., & Kellam, S.
(2001). The distal impact of two-first-grade preventive
interventions on conduct problems and disorder in
early adolescence. Journal of Emotional & Behavioral
Disorders,
9,
146–161.
https://doi.org/10.1177/
106342660100900301
Jones, D. E., Greenberg, M., & Crowley, M. (2015). Early
social-emotional functioning and public health: The
relationship between kindergarten social competence
and future wellness. American Journal of Public Health,
e1–e8. https://doi.org/10.2105/AJPH.2015.302630
Kraag, G., Van Breukelen, G. J. P., Kok, G., & Hosman,
C. (2009). “Learn Young, Learn Fair,” a stress management program for fifth and sixth graders: Longitudinal
results from an experimental study. Journal of Child Psychology and Psychiatry, 50, 1185–1195. https://doi.org/
10.1111/j.1469-7610.2009.02088.x
Larson, R. W. (2000). Toward a psychology of positive
youth development. American Psychologist, 55, 170–183.
Lerner, J. V., Phelps, E., Forman, Y., & Bowers, E. P.
(2009). Positive youth development. In R. M. Lerner &
L. Steinberg (Eds.), Handbook of adolescent psychology
(3rd ed., pp. 524–558). Hoboken, NJ: Wiley.
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Applied social research methods series (Vol. 49).
Thousand Oaks, CA: Sage.
Mendelez-Torres, G. J., Dickson, K., Fletcher, A., Thomas,
J., Hinds, K., Campbell, R., . . . Bonell, C. (2016). Positive youth development programmes to reduce substance use in young people: Systematic review.
International Journal of Drug Policy, 36, 95–103. https://
doi.org/10.1016/j.drugpo.2016.01.007
Miranda, J., Bernal, G., Lau, A., Kohn, L., Hwang, W.-C.,
& LaFromboise, T. (2005). State of the science on psychosocial interventions for ethnic minorities. Annual
Review of Clinical Psychology, 1, 113–142. https://doi.
org/10.1146/annurev.clinpsy.1.102803.143822
National Research Council and Institute of Medicine.
(2009). Preventing mental, emotional, and behavioral

15

disorders among young people: Progress and possibilities.
Committee on Prevention of Mental Disorders and Substance Abuse Among Children, Youth, and Young
Adults: Research Advances and Promising Interventions. M. E. O’Connell, T. Boat, & K. E. Warner (Eds.).
Board on Children, Youth, and Families, Division of
Behavioral and Social Sciences and Education. Washington, DC: The National Academies Press.
Riggs, N. R., & Pentz, M. A. (2009). Long term effects of
adolescent marijuana use prevention on adult mental
health services utilization: The Midwestern prevention
project. Substance Use and Misuse, 44, 616–631. https://
doi.org/10.1080/10826080902809691
Rosenthal, R., & Rubin, D. B. (1982). A simple general
purpose display of magnitude and experimental effect.
Journal of Educational Psychology, 74, 166–169.
Sarason, I. G., & Sarason, B. R. (1981). Teaching cognitive
and social skills to high school students. Journal of Consulting and Clinical Psychology, 49, 908–919.
Sawyer, M. G., Macmullin, C., Graetz, B., Said, J. A.,
Clark, J. J., & Baghurst, P. (1997). Social skills training
for primary school children: A 1-year follow-up study.
Journal of Pediatric Child Health, 33, 378–383. https://
doi.org/10.1111/j.1440-1754.1997.tb01624.x
Sklad, M., Diekstra, R., Ritter, M. D., Ben, J., & Gravesteijn, C. (2012). Effectiveness of school-based universal social, emotional, and behavioral programs: Do they
enhance students’ development in the area of skill,
behavior, and adjustment? Psychology in the Schools, 49,
892–909. https://doi.org/10.1002/pits.21641
Snyder, F. J., & Flay, B. R. (2012). Positive youth development. In P. M. Brown, M. W. Corrigan, & A. HigginsD’Allessandro (Eds.), Handbook of prosocial education (Vol,
2, pp. 415–443). New York, NY: Rowman & Littlefield.
Sorensen, L. C., & Dodge, K. A.; Conduct Problems
Prevention Research Group. (2015). How does the fast
track intervention prevent adverse outcomes in young
adulthood? Child Development, 87, 429–445. https://doi.
org/10.1111/cdev.12467
Theokas, C., Almeigi, J., Lerner, R. M., Dowling, E. M.,
Benson, P. L., Scales, P. C., & von Eye, A. (2005). Conceptualizing and modeling individual and ecological
asset components of thriving in early adolescence. Journal of Early Adolescence, 25, 113–143. https://doi.org/10.
1177/0272431604272460
Tolan, P., Ross, K., Arkin, N., Godine, N., & Clark, E.
(2016). Toward an integrated approach to positive
development. Applied Developmental Science, 20, 214–
236. doi:10.1080/10888691.2016.1146080.
Torrente, C., Alimchandani, A., & Aber, J. L. (2015). International perspectives on SEL. In J. A. Durlak, C. E.
Domitrovich, R. P. Weissberg, & T. P. Gullotta (Eds.),
Handbook for social and emotional learning: Research and
practice (pp. 566–587). New York, NY: Guilford.
Weare, K., & Nind, M. (2011). Mental health promotion
and problem prevention in schools: What does the evidence say? Health Promotion International, 26, i29–i68.
https://doi.org/10.1093/heapro/dar075

 16

Taylor, Oberle, Durlak, and Weissberg

Weissberg, R. P., Durlak, J. A., Domitrovich, C. E., & Gullotta, T. P. (2015). Social and emotional learning: Past,
present, and future. In J. A. Durlak, C. E. Domitrovich,
R. P. Weissberg, & T. P. Gullotta (Eds.), Handbook for
social and emotional learning: Research and practice (pp. 3–
19). New York, NY: Guilford.
Weissberg, R. P., & O’Brien, M. U. (2004). What works in
school-based social and emotional learning programs
for positive youth development. The ANNALS of the
American Academy of Political and Social Science, 591(1),
86–97. https://doi.org/10.1177/0002716203260093
Wilson, S. J., & Lipsey, M. W. (2007). School-based interventions for aggressive and disruptive behavior:
Update of a meta-analysis. American Journal of Preventive Medicine, 33, 130–143. https://doi.org/10.1016/j.a
mepre.2007.04.011
Yeager, D. S., & Walton, G. M. (2011). Social-psychological interventions in education: They’re not magic.
Review of Educational Research, 81, 267–301. https://doi.
org/10.3102/0034654311405999
Zins, J. E., Weissberg, R. P., Wang, M. C., & Walberg, H.
J. (2004). Building academic success on social and emotional
learning: What does the research say? New York, NY:
Teachers College Press.

Supporting Information
Additional supporting information may be found in
the online version of this article at the publisher’s
website:
Table S1. Postintervention Mean Effect Sizes,
Confidence Intervals, Average Intervention Duration, and Improvement Indices
Table S2. Trim-and-Fill Results: Adjusted Mean
Effect and Confidence Intervals
Table S3. Methodological Control Analyses for
the Total Sample Using a Study Level Effect Size
(ES) Collapsed Across All Outcome Categories
Table S4. Intervention Feature Analyses at Follow Up for the Total Sample, Using a Study Level
Effect Size (ES) Collapsed Across All Outcome Categories
Table S5. Participant Feature Analyses at Follow
Up for the Total Sample