J Youth Adolescence (2014) 43:884–896

DOI 10.1007/s10964-013-0085-4

EMPIRICAL RESEARCH

Academic Achievement in the High School Years: The Changing

Role of School Engagement
Paul A. Chase • Lacey J. Hilliard • G. John Geldhof •

Daniel J. A. Warren • Richard M. Lerner

Received: 12 September 2013 / Accepted: 20 December 2013 / Published online: 30 January 2014
Ó Springer Science+Business Media New York 2014

Abstract School engagement is an important theoretical Introduction

and practical cornerstone to the promotion of academic
accomplishments. This article used a tripartite—behav- For many American youth, the transition from middle
ioral, emotional, and cognitive—model of school engage- school to high school involves adapting to the outcomes-
ment to assess the relationship between school engagement oriented curriculum used in these latter school grades
and academic success among high school students, and to (Eccles and Midgley 1989). In addition to greater academic
determine whether a reciprocal relationship exists between expectations, youth must simultaneously cope with a new
these constructs. Data were derived from 710 youth (69 % school environment and social setting (Crosnoe et al.
female) who took part in Waves 6 through 8 (Grades 10 2004). In U.S. public schools, most incoming high school
through 12) of the 4-H study of positive youth develop- students must also adjust to a noticeably larger student
ment. Longitudinal confirmatory factor analyses confirmed population compared with the number of students in mid-
the invariance of the tripartite model of school engage- dle school (McNeeley et al. 2002). Taken together, these
ment. Results of a structural equation model showed that characteristics make the transition to high school chal-
the components of school engagement and academic lenging for even the most well-prepared and resilient
achievement were mutually predictive and that these pre- student.
dictions varied from grade to grade. Future possibilities for Many high school students are able to meet these
evaluating the relationship between school engagement and challenges successfully, however others fail to succeed
academic achievement, as well as the implications for academically. Students who struggle academically are
educational policy and practice, are discussed. more likely to experience unemployment, substance use,
and delinquency as adults (Chavez et al. 1994; Henry et al.
Keywords Academic achievement
School engagement
2012). These outcomes may occur because students with
Confirmatory factor analysis
Longitudinal structural poor academic outcomes have fewer opportunities for tra-
equation modeling ditional career success and therefore become less invested
in society, leading to the potential for a hefty burden on
themselves, their families, and their communities at large
(e.g., Balfanz et al. 2010). Therefore, it is important to
P. A. Chase
L. J. Hilliard
D. J. A. Warren
R. M. Lerner (&)
examine the potential determinants of students’ academic
Eliot-Pearson Department of Child Development, Institute for
Applied Research in Youth Development, Tufts University, 301 success or failure.
Lincoln Filene Building, Medford, MA 02155, USA
e-mail: iaryd.pubs@gmail.com Theoretical Framework
P. A. Chase
e-mail: paul.chase@tufts.edu A student’s academic success is the product of many fac-
tors, both individual and contextual (Li et al. 2010). Stu-
G. John Geldhof
Department of Human Development and Family Sciences, dents do not exist in a vacuum, and thus cannot succeed
Oregon State University, Corvallis, OR, USA academically without contextual supports such as strong

123
J Youth Adolescence (2014) 43:884–896 885

families, neighborhoods, and schools (Woolley and Gro- Rusk (2011) found that emotional engagement in school is
gan-Kaylor 2006). The positive youth development (PYD) related to intrinsic motivation. Larson and Rusk argued that
perspective provides a theoretical framework in support of an internal desire to succeed in school is as important as
this premise. PYD is centered on the notion that, in order to behavioral engagement because it gives students the
thrive, youth must capitalize on ecological assets. Con- motivation and agency to regulate themselves toward their
currently, adults who hope to serve youth, such as parents, goals of academic achievement.
teachers, and mentors, must build on the individual However, the link between school engagement and
strengths of youth. With such contextual support, youth academic achievement is neither simple nor universal.
acquire the knowledge and skill they need to become Early studies of school engagement indicated a positive
healthy and productive adults (Lerner et al. 2005). association between emotional engagement and achieve-
Several relevant strengths that youth must develop to be ment (Fredricks et al. 2004; Voelkl 1997). More recent
successful during adolescence include self evaluations, studies found that the strength of the association between
long-term planning, use of effective learning strategies, and engagement and academic achievement varied depending
prioritizing goals (Wigfield 1994). These skills all con- on how achievement was measured and the racial/ethnic
tribute to the meaningful engagement of students in their and socio-economic composition of the participants in the
own learning and in attaining their academic aspirations study (Shernoff and Schmidt 2008). These studies found
(Wigfield and Cambria 2010). Through practicing these that school engagement was less predictive of GPA for
skills, students can become cognitively, emotionally, and African-American students than European American stu-
behaviorally active participants in their own learning, dents in their sample. Similarly, school engagement was
which positively predicts academic achievement (Zim- less predictive of academic success for students from low
merman and Schunk 2001). income communities than for students from high income
Youth cannot achieve academic success on their own, communities.
however. Previous efforts aimed at enhancing academic Although the above studies attempted to explain the
achievement have focused on capitalizing on adolescents’ unidirectional predictive relationship of school engagement
positive social relationships in the school context (Green- on academic achievement, we must also consider the
berg et al. 2003). Accordingly, this article uses a PYD opposite direction of effects—the effect of academic
framework to examine how students relate to their aca- achievement on school engagement. Such a bidirectional
demic context, and how this person-context relation relationship would mean that the more students are
impacts academic achievement among high school stu- engaged, the more that they learn in school. The more that
dents. However, the focus here is not only on social rela- students are successful in school, the more efficacious they
tionships but, more broadly, on the construct of school feel, which, in turn, increases their engagement (Hauser-
engagement. School engagement (i.e., the degree to which Cram et al. 2006). Previous research using longitudinal
students are involved in and committed to the academic panel models found reciprocal effects of academic
and social activities in school) plays an influential role in achievement and academic self-concept (Marsh 1990;
preventing academic failure, promoting competence, and Marsh and O’Mara 2008). These reciprocal pathways will
influencing a wide range of adolescent outcomes (Eccles be tested in the current study to determine the direction of
and Roeser 2011). effect between three components of school engagement
and academic achievement.
School Engagement and Academic Achievement
The Tripartite Model of School Engagement
Several researchers have documented the effects of school
engagement on academic achievement. Balfanz and Byrnes Recent research has found that school engagement is a
(2006) investigated this relationship among Philadelphia multi-dimensional construct (Christenson et al. 2012).
public school students and found that aspects of behavioral Specifically, in this article, we focus on the tripartite—
engagement, such as student attendance, problem behavior, behavioral, emotional, and cognitive—components of
and indicators of effort, all independently and significantly school engagement and how engagement in the school
predicted academic achievement. Previous research has context relates to academic achievement throughout high
made a strong case that that behaviors related to school school. In order to elucidate the changing impact of a
engagement are critical in predicting academic outcomes student’s relationship with his or her school, we draw from
(Dotterer and Lowe 2011; Kindermann 2007). For exam- data collected in the 4-H study of PYD (Lerner et al. 2005).
ple, disengagement behaviors, such as being inattentive or The 4-H study of PYD is a longitudinal investigation that
disruptive in class at Grade 8, predicted lower grades in began assessing 5th Grade students and their parents in
high school (Finn and Rock 1997). Similarly, Larson and 2002. Data collection for Grade 12 was completed in 2011,

123
886 J Youth Adolescence (2014) 43:884–896

and the study involves about 7,000 youth and 3,500 of their Methods
parents from 42 states to date.
Using this data set, Li et al. (2010) sought to understand As noted above, the current study was part of a larger,
the ways youth engage with their school contexts and how ongoing longitudinal investigation of youth development in
school engagement related to both immediate and later the United States that began in 2002. The 4-H study of
developmental outcomes. Li et al. first used structural PYD is a longitudinal investigation that began by assessing
equation modeling to examine the role of school engage- 5th grade youth in the United States and their parents. Full
ment in predicting academic competence. Using data from details of the 4-H study of PYD have been presented
960 participants (45.6 % boys) from Waves 1 and 2 elsewhere (Lerner et al. 2005, 2009a, b, 2010, 2011;
(Grades 5 and 6) of the 4-H study of PYD, Li et al. (2010) Theokas and Lerner 2006; Phelps et al. 2007). Therefore,
found that emotional and behavioral school engagement we present here only the features of the methods relevant to
mediated the relationships between developmental assets the present research, which includes data from Waves 6
and self-reported academic competence. In addition, Li through 8. A discussion of the overall method of the 4-H
completed a longitudinal confirmatory factor analysis study is provided in the introductory article of this special
which established the tripartite model of school engage- issue (Bowers et al. in press).
ment from Waves 5–7 (Grades 9–11) of the 4-H study of
PYD (Li 2010). Using latent growth curve analysis, Li and Participants
Lerner also demonstrated that the mean levels of behav-
ioral engagement, as well as change in cognitive school The larger 4-H study examined various domains of indi-
engagement across Grades 9 through 11 were positively vidual development and ecological resources during ado-
associated with GPA at Grade 11 (Li and Lerner 2011). lescence. The current study focused on variables related to
More recently, Li and Lerner used these same waves of the school engagement and academic achievement. The sample
4-H study of PYD to demonstrate the longitudinal rela- for the 4-H study was recruited from schools and after-
tionships between the three distinct behavioral, emotional, school programs across the country in order to reflect the
and cognitive components of school engagement (Li and racial/ethnic and socioeconomic diversity of the United
Lerner 2013). The tripartite measure of school engagement States. Schools and districts were contacted across a wide
gained further support from a study by Scheidler (2012), array of cities, and targeted schools were recruited based on
which demonstrated that a tripartite measure of school professional contacts of the research team. Youth were also
engagement positively predicted standardized test scores recruited from after-school programs and 4-H-affiliated
among 8th Grade students. However, to date, these tripar- clubs and organizations, as well as other out-of-school-time
tite models have overwhelmingly emphasized the unidi- programs such as Boys and Girls Clubs. Once agreement
rectional effects of school engagement on academic from a school or youth organization was obtained, recruit-
achievement. The possibility of a reciprocal relationship ment information, including a flyer describing the study and
between the tripartite model of school engagement and a consent form, was sent or given out to youth and their
academic achievement has not been thoroughly explored parents. This information was distributed through school
and thus will be the focus of the current article. and site staff (e.g., participating after-school programs) to
youth and their parents.
Participants for the current analyses came from three
waves (Waves 6–8) of the larger study, principally repre-
Current Study senting youth in Grades 10 through 12, respectively
(Table 1). In the overall sample for Grade 10, data were
The tripartite model of behavioral, emotional, and cogni- collected from a total of 2,357 youth in Grade 10, 1,324 youth
tive school engagement captures how students feel, behave, in Grade 11, and 1,030 youth in Grade 12. For the longitu-
and think when interacting with their dynamic school dinal study, analyses were conducted with 710 adolescents
environment. Li and Lerner’s (2011) tripartite model of (mean age at Wave 6 = 15.7 years, SD = 0.73 years) all of
school engagement represents the basis for the current whom participated in at least two out of three waves of
study. We hypothesize that we will be able to replicate this assessment. This requirement for wave participation was
model in a different, older sample of participants in the 4-H implemented to minimize the threat to validity that occurs as
study of PYD in order to use longitudinal panel modeling a result of wave non-response error (Lindner et al. 2001).
procedures to parse out both the strength and direction of Participants in the longitudinal sample were predominantly
the relationship between behavioral, emotional, and cog- European American, with the following racial/ethnic back-
nitive school engagement and academic achievement grounds: 83.1 % were European American, 4.7 % were
among students through the end of high school. Latino/a, 4.7 % were African American, 2.8 % were Asian

123
J Youth Adolescence (2014) 43:884–896 887

Table 1 Sample characteristics Grade 10 Grade 11 Grade 12 Longitudinal

for cross-sectional and
longitudinal participants from Characteristics 2,357 1,324 1,030 710
Grades 10, 11, and 12 of the 4-H
Sex
study of PYD
Female (%) 1,487 (63.1) 898 (67.9) 770 (75.2) 493 (69.4)
Age (M, SD) 15.72 (1.37) 16.80 (1.42) 17.65 (1.18) 15.7 (0.73)
Race/ethnicity
European American (%) 1,790 (75.9) 1,054 (79.6) 770 (74.8) 587 (83.1)
African American (%) 131 (5.6) 68 (5.1) 53 (5.4) 33 (4.7)
Latino (%) 148 (6.3) 35 (2.6) 42 (4.3) 33 (4.7)
Asian (%) 37 (1.6) 29 (2.2) 29 (2.8) 20 (2.8)
Native American (%) 22 (0.9) 16 (1.2) 10 (1.1) 11 (1.6)
Multiethnic or multiracial (%) 52 (2.2) 17 (1.3) 11 (1.0) 18 (2.5)
Not available (%) 17 (0.7) 6 (0.5) 8 (0.8) 4 (0.6)
Urbanity
Urban (%) 202 (8.6) 281 (21.2) 172 (16.7) 134 (18.9)
Suburban (%) 291 (12.3) 324 (24.5) 283 (27.5) 156 (21.9)
Rural (%) 712 (30.2) 707 (53.4) 549 (53.3) 412 (58.1)
Not available (%) 1,152 (48.9) 12 (0.9) 14 (1.1) 8 (1.1)
Avg. Household income $69,363.90 $71,792.26 $70,508.08 $71,782.07
Avg. Household income for $41,873.03 $43,590.84 $44,201.91 $43,518.37
lower quartile—25 %)
Avg. Household income for $92,659.28 $94,811.20 $95,019.23 $95,142.76
upper quartile—75 %)
Mother’s education level (highest reported) 14.67 (2.26) 14.75 (2.30) 14.37 (2.29) 14.56 (2.31)

American, 2.5 % were multiracial or multiethnic, and 1.6 % be made as readily about lower academically performing
were Native American. In addition, 1.2 % of youth either students.
reported their race/ethnicity inconsistently or did not report
it. Table 1 summarizes the demographic characteristics of Procedure
the cross-sectional samples for Grades 10–12, as well as our
longitudinal sample. In Waves 1 through 3 of the 4-H study, data collection
While no significant differences were found in racial/ from youth was conducted by trained study staff or, at more
ethnic composition, sex, or socio-economic status between distant locations, hired assistants. A detailed protocol was
the cross-sectional and longitudinal samples, as shown in used to ensure that data collection was administered uni-
Table 1, both the longitudinal sample and the three cross- formly and to ensure the return of all study materials. After
sectional samples (Grades 10–12) were slightly overrep- Wave 1, youth who were absent on the day of the survey or
resented by female participants. The longitudinal sample were from schools or programs that did not allow on-site
was comprised of 69.4 % females. Rural participants testing were contacted by e-mail, mail, or phone, and were
(58.1 %) and European American participants (83.1 %) asked to complete and return the survey to us. Beginning in
were also overrepresented in both the cross-sectional and Wave 5, youth completed the survey online unless they
the longitudinal sample, as compared with the U.S. popu- requested a paper survey. Parents completed online or
lation (U.S. Census Bureau 2010). In addition, participants paper surveys. Paper surveys were delivered to their homes
in our sample came from families from more advantaged by their children or through the mail (in the latter case,
socioeconomic backgrounds in terms of family income return postage was provided).
($71,782.07, compared with $50,054 in the overall U.S.
population) and levels of maternal education (42.7 % Measures
bachelor’s degree or higher, compared with 28.6 % in the
overall population) (U.S. Census Bureau 2010). The aver- The items used in our analysis came from the Student
age GPA of students in the longitudinal sample was 3.53 Questionnaire of the 4-H study of PYD. To measure the
(SD = 0.56), significantly higher than the overall sample tripartite school engagement constructs accurately, we used
3.31 (SD = 0.75), (p \ .05). This higher GPA in the lon- the fifteen-item Behavioral-Emotional-Cognitive School
gitudinal sample may indicate that generalizations cannot Engagement Scale (BEC-SES) (Li 2010).

123
888 J Youth Adolescence (2014) 43:884–896

Table 2 The Behavioral-Emotional-Cognitive School Engagement Scale (BEC-SES)

Item Response format

Behavioral school engagement

How often do you come to class unprepared (homework 0 = Never to 3 = Always
unfinished, forget to bring books or other materials, etc.)?*
How often do you complete homework on time? 0 = Never to 3 = Always
How often do you skip classes without permission?* 0 = Never to 3 = Always
How often do you actively take part in group (class) 0 = Never to 3 = Always
discussions?
How often do you work hard to do well in school? 0 = Never to 3 = Always
Emotional school engagement
I feel part of my school 0 = Strongly disagree to 3 = Strongly agree
I care about the school I go to 0 = Strongly disagree to 3 = Strongly agree
I am happy to be at my school 0 = Strongly disagree to 3 = Strongly agree
I don’t find school fun and exciting* 0 = Strongly disagree to 3 = Strongly agree
I enjoy the classes I am taking 0 = Strongly disagree to 3 = Strongly agree
Cognitive school engagement
I want to learn as much as I can at school 0 = Strongly disagree to 3 = Strongly agree
I think it is important to make good grades 0 = Strongly disagree to 3 = Strongly agree
I think the thing I learn at school are useful 0 = Strongly disagree to 3 = Strongly agree
I think a lot about how to do well in school 0 = Strongly disagree to 3 = Strongly agree
School is very important for later success 0 = Strongly disagree to 3 = Strongly agree
* Items are reverse-coded

Behavioral School Engagement used to measure three related yet distinct types of positive
affect. A sample item is: ‘‘I care about the school I go to.’’
The behavioral engagement subscale includes five items Items used to tap school connectedness, belonging, and
indicating shallow engagement (attendance) and items bonding was modified to assess different aspects of the
tapping deeper engagement (effort). More specifically, emotional relationships students have with their school and
items regarding contribution to class discussion, prepara- classes. The response format ranged from 0 (strongly dis-
tion, skipping class, and finishing homework on time are agree) to 3 (strongly agree). Items from the Psychological
included. The measure focuses on students’ voluntary Sense of School Membership (PSSM, Goodenow 1993)
behaviors within the school context to minimize possible and Add Health (McNeeley et al. 2002; Resnick et al.
confounding effects of non-student related variables. In 1997) were the major sources of items.
addition, the items are balanced in terms of valence, even
though indicators of active disengagement may not be Cognitive School Engagement
highly related to the items that denote varying levels of
active engagement. The response format ranged from 0 We measured cognitive engagement with five items
(never) to 3 (always). A sample item is: ‘‘How often do you designed to assess the extent to which students valued
actively take part in group (class) discussions?’’ Sources of education and things learned at school, as well as their
these items include the National Educational Longitudinal thoughts about learning. More specifically, goal orienta-
Study (NELS: 1988) (Finn and Rock 1997; Finn and Vo- tion, identification with school, and perceptions of the link
elkl 1993), Murray and Greenberg (2001), and Heaven between students’ lives and school were included as core
et al. (2002). indicators of cognitive engagement. One item, ‘‘I want to
learn as much as I can at school,’’ tapped participants’ goal
Emotional School Engagement orientation. Two items that asked whether school learning
is meaningful and important were used to measure stu-
The emotional engagement subscale included five items dents’ perceptions and beliefs. Another item was used to
that assessed students’ sense of belonging and their affect evaluate the extent to which a student is an intentional
toward school. Sense of belonging was measured by one learner. Given the scarcity of measures of cognitive
item asking the extent to which students feel like a part of engagement in the literature, the items were developed
their schools. Happiness, excitement, and enjoyment were primarily based on the definitions, instead of measures, of

123
J Youth Adolescence (2014) 43:884–896 889

Fig. 1 Hypothesized model of

school engagement and
academic achievement in
Grades 10–12 of the 4-H study.
(Dotted lines are correlations,
solids lines are regression
coefficients)

cognitive engagement (Skinner et al. 2009). A sample item from Waves 6 through 8. Each school engagement con-
is ‘‘School is very important for later success.’’ The struct was operationalized by five manifest indicators per
response format was also a four-point Likert scale, ranging wave of measurement. Academic Achievement was op-
from 0 (strongly disagree) to 3 (strongly agree). erationalized by a single item, self-reported GPA, at each
All scales of school engagement demonstrated adequate of the three waves of measurement (see the theoretical
reliability, with Cronbach’s alphas across the scales rang- model in Fig. 1). In order to address missing data in our
ing from .76 to .90. For the complete BEC-SES see analysis, we used full information maximum likelihood
Table 2. (FIML) procedures (see Wothke 2000 for a more detailed
description of FIML). To verify the tripartite model of
Academic Achievement behavioral, emotional, and cognitive school engagement
within the current sample, a confirmatory factor analysis
We operationalized academic achievement in our model by was conducted using Mplus Version 6.11 (Muthén and
self-reported grade point average (GPA), with the single Muthen 2010).
item, ‘‘What grades do you earn in school?’’ The scale We began by attempting to demonstrate that the load-
ranged from 1 ‘‘(0.5) = Mostly below Ds’’ to 8 ings and intercepts of each indicator were equivalent across
‘‘(4.0) = Mostly As’’. This single item was deemed a suf- waves. Establishing this invariance gave us a basis to
ficient proxy for academic achievement, as GPA has been assume that, because the constructs were defined in the
shown to be a highly reliable self-report measure among same operational manner for each group, the construct’s
adolescents (Cassady 2001). variance, correlations, and mean differences could be
compared meaningfully and with quantitative accuracy
Analysis Technique (Little 1997, 2013). After we established a measurement
model through longitudinal confirmatory factor analyses,
For our longitudinal confirmatory factor analyses, we we proceeded with a structural equation model to deter-
analyzed the factor structure of behavioral, emotional, and mine the most parsimonious relationship between the latent
cognitive school engagement at each of the three waves variables of behavioral, emotional, and cognitive school
of the 4-H study of PYD within our longitudinal sample engagement and academic achievement, specifically the

123
890 J Youth Adolescence (2014) 43:884–896

direction and magnitude of effects as measured by cross- the latent mean structure across waves of measurement.
lag relationships across high school. Once again, the model was found to be invariant across the
3 waves of data [v2(1,017) = 2,038.35, p \ .001,
RMSEA = .04, 90 % CI (.04, .04); CFI = .93,
Results TLI = .92].
The factorial invariance of the behavioral, emotional,
The purpose of this study was to determine how a tripartite and cognitive school engagement model had been dem-
model of behavioral, emotional, and cognitive school onstrated in previous studies of school engagement that
engagement related to academic achievement, operational- used the 4-H data set (Li 2010). The findings in the current
ized by GPA, in later waves of the 4-H study of PYD (Grades sample verify the previous findings of Li (2010), in that a
9 through 12). Originally, we intended to use four waves in tripartite model of school engagement remained strong in
our model, Waves 5–8 (Grades 9–12). This original mea- later high school years. However, there is a caveat that this
surement model failed to demonstrate sufficient model fit model did not include Wave 5 (Grade 9) as originally
[v2(1746) = 4,966.02, p \ .001, RMSEA = .08, 90 % CI designed due to poor fit in the longitudinal sample (see
(.07, .08); CFI = .73, TLI = .71]. After evaluating the Tables 3, 4, 5 for correlations, intercepts and fit indices,
model, we determined that Wave 5 (Grade 9) failed to respectively).
demonstrate strong model fit on its own [v2(95) = 217.51, Having established weak and strong invariance, we next
p \ .001, RMSEA = .09, 90 % CI (.08, .09); CFI = .91, tested a structural equation model based on our original
TLI = .91], and contributed to poor fit in our original Wave configural measurement model (see Fig. 1). In this longitu-
5–8 longitudinal model. Therefore, Wave 5 was dropped dinal structural equation model, we hypothesized direct
from the remainder of our analyses, and our final measure- associations from the behavioral, emotional, and cognitive
ment model consisted of data from participants from Waves school engagement constructs in Grades 10 and 11 to the
6 through 8. This measurement model demonstrated suffi- single-item construct of academic achievement (self-repor-
cient fit [v2(972) = 2,047.69, p \ .001, RMSEA = .04, ted GPA) in Grades 11 and 12, controlling for the constructs
90 % CI (.04, .04); CFI = .91, TLI = .92]. in the prior grades. To saturate the latent correlation matrix
Two further modifications were made to the model. fully, we additionally estimated correlations between all
First, constraints on residual variances were relaxed other pairs of latent constructs within waves. In order to
between two cognitive school engagement items (Items 2 maximize model parsimony, non-significant latent regres-
and 5), allowing their residuals to covary across waves sion paths were removed one at a time, until a significant
(please see Table 2 for item numbers). This modification change in Chi square values was obtained, and only signifi-
was deemed justifiable, as these items were worded simi- cant latent regression paths remained in the final model.
larly. In addition, residual variances for two emotional Our final model after pruning nonsignificant paths dem-
school engagement items (Items 1 and 2) were allowed to onstrated adequate fit [v2(981) = 1,946.79, p \ .001,
covary in Wave 7 only. This was also deemed justifiable, RMSEA = .04, 90 % CI (.04, .04); CFI = .93, TLI = .92]
given that Items 1 and 2 were worded similarly as well. (see Fig. 2). The final model showed significant relationships
After these adjustments to the model, the new mea- among aspects of Grade 10 through 12 behavioral, emotional,
surement model demonstrated good model fit, and suffi- and cognitive school engagement and Grade 10 through 12
cient configural invariance in the longitudinal sample GPA, as hypothesized. However, the nature of the relationship
[v2(969) = 1,933.64, p \ .001, RMSEA = .04, 90 % CI was more complex than our theoretical model. As seen in
(.04, .04); CFI = .93, TLI = .92]. We next tested the Fig. 2, the final model included significant regression coeffi-
model for loading invariance. This test was conducted to cients from Grade 10 Behavioral School Engagement to
determine whether items load onto their factors similarly Grade 12 GPA, as well as Grade 10 Emotional School
across waves of measurement and was necessary to ensure Engagement to Grade 11 GPA. Grade 10 GPA significantly
that meaningful comparisons could be made across multi- predicted Grade 11 Cognitive School Engagement, as well as
ple waves of testing (Brown 2006). The loading (weak) all three aspects of Grade 12 School Engagement. The two
invariance model had sufficient fit [v2(993) = 1,968.85, strongest relationships were found between Grade 10
p \ .001, RMSEA = .04, 90 % CI (.04, .04); CFI = .93, Behavioral School Engagement and Grade 12 GPA (b = .19)
TLI = .92], Therefore, the model met the requirements for and Grade 10 GPA on Grade 12 Behavioral School Engage-
loading (weak) invariance, with a change of CFI of \.01 ment (b = .21; see Fig. 2 for additional regression coeffi-
(Cheung and Rensvold 2002). The next model tested for cients). In addition, strong auto-regressive relationships were
intercept (strong) invariance in the longitudinal model. found within all latent constructs across waves. All reported
Testing for intercept invariance is important to determine regressions were found to be significant, at p \ .05.

123
J Youth Adolescence (2014) 43:884–896 891

Table 3 Latent correlations of 1 2 3 4 5 6 7 8 9 10 11 12

variables included in model
testing G10BSE 1.00
G10ESE 0.47* 1.00
G10CSE 0.68* 0.54* 1.00
G11BSE 0.76* 0.41* 0.56* 1.00
G11ESE 0.42* 0.68* 0.38* 0.47* 1.00
G11CSE 0.65* 0.44* 0.60* 0.72* 0.58* 1.00
G12BSE 0.67* 0.42* 0.56* 0.74* 0.42* 0.63* 1.00
BSE = Behavioral school
G12ESE 0.45* 0.50* 0.31* 0.38* 0.63* 0.38* 0.52* 1.00
engagement. ESE = Emotional
school engagement. G12CSE 0.56* 0.34* 0.48* 0.52* 0.41* 0.52* 0.75* 0.67* 1.00
CSE = Cognitive school G10GRAD 0.68* 0.31* 0.40* 0.55* 0.26* 0.39* 0.52* 0.24* 0.28* 1.00
engagement. GRAD = Self- G11GRAD 0.58* 0.30* 0.32* 0.60* 0.31* 0.39* 0.55* 0.26* 0.28* 0.75* 1.00
reported GPA
G12GRAD 0.58* 0.27* 0.30* 0.57* 0.26* 0.35* 0.57* 0.27* 0.35* 0.65* 0.78* 1.00
* p \ .05

Discussion self-reported GPA and behavioral, emotional, and cogni-

tive school engagement across 3 years of high school.
School engagement plays an influential role in promoting As hypothesized, bidirectional relationships were found
academic competence and positive social relations with between the three aspects of school engagement and GPA.
teachers and students (Eccles and Roeser 2011; Fredricks behavioral school engagement at Grade 10 predicted GPA
and Eccles 2006; Li and Lerner 2011). The influence of at Grade 12, and emotional school engagement at Grade 10
school engagement is both logical and intuitive. Adoles- predicted GPA at Grade 11. GPA at Grade 10 positively
cents who perceive that they are in a positive and sup- predicted Cognitive School Engagement at Grade 11, as
portive educational atmosphere may have a greater ability well as all three aspects of School Engagement at Grade
to capitalize on that support, leading to academic 12. These findings support a bidirectional, reciprocal rela-
achievement (Klem and Connell 2004; Meehan et al. tionship between school engagement and academic
2003). Although most students receive enough contextual achievement (Finn and Rock 1997; Li et al. 2010). This
support and show sufficient effort and ability to succeed in bidirectional relationship demonstrates that GPA is not
school, many young people become disengaged, decreasing only an outcome, but can also predict the degree to which
their likelihood of academic success. students are engaged in high school.
Disengaged high school students who do not experience We found that the greatest predictor of GPA in the
academic success are more likely to experience unem- model was behavioral school engagement at Grade 10 on
ployment, substance use, and delinquency as adults (Cha- Grade 12 GPA. This finding is consistent with previous
vez et al. 1994; Henry et al. 2012). Balfanz and Byrnes studies that demonstrated the importance of behavioral
(2006) documented the effects of several indicators of school engagement (Dotterer and Lowe 2011). Youth
school engagement on academic achievement among stu- who display appropriate behaviors of successful students
dents in Philadelphia schools in which student attendance, (such as doing homework and participating in class) earn
behavior, and effort all independently and significantly higher GPAs. In addition, emotional engagement at
predicted multiple measures of academic achievement. In Grade 10 significantly predicted students’ GPA at Grade
order to understand and more widely generalize the rela- 11, which supports previous studies related to emotional
tionship between school engagement and academic engagement (e.g., Larson and Rusk 2011). However, the
achievement, researchers have attempted to parse out the regression coefficient for Grade 10 emotional engage-
roles of various aspects of school engagement in predicting ment on Grade 11 GPA was not as strong as Grade 10
academic achievement (Li 2010; Li and Lerner 2013; behavioral engagement on GPA at Grade 12. Previous
Scheidler 2012). Despite such interest in the academic studies supported this finding as well, such that when
effects of school engagement, researchers have not mea- behavioral engagement is included in the model, emo-
sured whether the relationship between the tripartite model tional engagement’s relationship with GPA accounts for
of school engagement (i.e., one that examines the unique a smaller share of the variance (Li 2010). Significant
contributions of behavioral, emotional, and cognitive latent correlations (shown in Table 3) between emotional
components) and academic achievement is, indeed, bidi- and cognitive school engagement and GPA give support
rectional. The current study sought to determine the mag- to an underlying relationship between these three latent
nitude and direction of the relationships among students’ constructs.

123
892 J Youth Adolescence (2014) 43:884–896

Table 4 Means and variances of indicator variables Table 4 continued

Mean Variance Mean Variance

G10Grade 3.508 .025 G12CEI5 2.514 .029

G11Grade 3.539 .023
BE = Behavioral school engagement. EE = Emotional school
G12Grade 3.548 .024 engagement. CE = Cognitive school engagement. Grade = Self-
G10BEI1 2.353 .026 reported GPA. G10, G11, and G12 represent Grades 10 through 12
G10BEI2 2.338 .031 respectively
G10BEI3 2.855 .017
G10BEI4 1.897 .036
G10BEI5 2.436 .031
Nevertheless, behavioral school engagement emerged as
G10EEI1 2.085 .030
the strongest predictor of GPA in the high school years.
This may reinforce the notion that study skills and effort in
G10EEI2 2.137 .030
school, key components of behavioral engagement, could
G10EEI3 2.083 .032
contribute to academic success. It is important to note that
G10EEI4 1.818 .036
GPA is not determined by mastery alone. In many schools,
G10EEI5 2.056 .029
the effort that a student shows toward completing the
G10CEI1 2.371 .026
required tasks, such as homework and participation, are
G10CEI2 2.584 .023
included in a students’ GPA.
G10CEI3 2.215 .028
Our findings show that, even if students think that school
G10CEI4 2.312 .029
is important (demonstrating cognitive engagement), if they
G10CEI5 2.590 .024
fail to behave in a way that allows them to do what is
G11BEI1 2.366 .025
required of them, they tend not to succeed academically.
G11BEI2 2.250 .032
Conversely, a student could have little understanding of
G11BEI3 2.794 .021 how school may affect his or her future, but still work hard
G11BEI4 1.966 .035 in school and get good grades anyway. The finding that
G11BEI5 2.373 .031 GPA predicts all three aspects of school engagement sug-
G11EEI1 2.116 .031 gests that students who thrive in school academically may
G11EEI2 2.108 .032 be encouraged by their successes, which may influence all
G11EEI3 2.111 .031 three components of engagement in the school context.
G11EEI4 1.744 .034 A possible explanation for the lack of influence of
G11EEI5 2.051 .028 cognitive school engagement on GPA is a ceiling effect,
G11CEI1 2.321 .027 and overall stability of academic achievement in the high
G11CEI2 2.536 .026 school years. Heckman and Masterov (2006) argued that,
G11CEI3 2.130 .030 by the time a student has reached high school, his or her
G11CEI4 2.282 .029 study habits and attitudes towards school are entrenched to
G11CEI5 2.550 .025 such a degree that very little will alter that student’s aca-
G12BEI1 2.364 .030 demic achievement outcomes so late in their academic
G12BEI2 2.293 .034 career. In the current study, this phenomenon of stable
G12BEI3 2.720 .027 attitudes and performance may have manifested itself as
G12BEI4 1.975 .038 well, in that students who demonstrated high levels of
G12BEI5 2.349 .035 school engagement had already demonstrated high aca-
G12EEI1 2.054 .036 demic ability. For this reason, a change in academic suc-
G12EEI2 2.097 .035 cess may not have been as readily predictable as a function
G12EEI3 2.085 .035 of earlier school engagement in late high school students.
G12EEI4 1.814 .040 Indeed, the longitudinal sample had an extremely high and
G12EEI5 2.100 .031 consistent mean GPA across the 3 years of measurement
G12CEI1 2.347 .031 (Grades 10–12). This consistency and potential ceiling
G12CEI2 2.513 .030 effect may have diminished cognitive school engagement’s
G12CEI3 2.200 .033 impact on academic achievement in the current study, due
G12CEI4 2.228 .035 to the fact that the GPAs were already quite high for many
participants.

123
J Youth Adolescence (2014) 43:884–896 893

Table 5 Fit indices for the confirmatory factor analyses and structural equation modeling
Model v2 df p Dv2 Ddf Dp RMSEA RMSEA CFI
90 % CI

Configural model before modifications 2,047.69 972 \.001 0.040 0.037–0.042 0.925
Configural model after modifications 1,933.64 969 \.001 0.038 0.035–0.040 0.932
Weak invariance model 1,968.85 993 \.001 35.21 24 \.001 0.037 0.035–0.040 0.932
Strong invariance model 2,038.35 1,017 \.001 104.71 48 \.001 0.038 0.035–0.040 0.929
Full path model 1,933.64 969 \.001 0.038 0.035–0.040 0.932
Final model 1,946.79 981 \.001 13.15 12 \.001 0.037 0.035–0.040 0.932

Fig. 2 Final model of Grades

10 through 12 school
engagement and academic
achievement (dotted lines are
correlations, solids lines are
regression coefficients).
Correlational values were
omitted for clarity. v2 test of
model fit
(df = 981) = 1,946.79.
RMSEA = .037 (.035–.040).
CFI = 0.932

To more fully understand the relationship between the The question still remains, however, about the impor-
three aspects of school engagement and academic tance of the freshman year (Wave 5) on engagement. Since
achievement, future studies may consider additional sour- these Wave 5 data were removed from the model because
ces of information regarding academic achievement to of poor longitudinal representativeness and misfit from the
parse out potential social effects. Including standardized original longitudinal model, it is uncertain whether school
test scores such as college admissions assessments (e.g., engagement levels during the first high school year have
the SAT) would give a more diverse measure of student the same, or potentially greater, implications than
achievement (Scheidler 2012). Although the SAT and other engagement during the 10th Grade. Exploring this possi-
standardized tests have their own biases and limitations, bility further, and highlighting the transition to high school,
such measures may show achievement in students who may allow for a better understanding of the relationship
were not necessarily engaged in the practices of a school between the various aspects of engagement and academic
and yet have strong regard for and connection to learning success over the high school years. Future studies may also
(Cameron 1989). Furthermore, it may show that even benefit from focusing on participants with lower academic
though a student may not enjoy his or her current context, achievement at early grades and evaluating their changes
he or she is making efforts that would increase his or her with regard to academic achievement and school engage-
future academic opportunities. ment to better recognize patterns that may emerge in lower

123
894 J Youth Adolescence (2014) 43:884–896

achieving students. These youth with lower GPAs may engagement, coupled perhaps, at least initially, with
have the most to gain from an engaging academic extrinsic incentives for such engagement, may be the best
environment. approach for capitalizing on the reciprocal relations
An additional limitation is demonstrated by the demo- between school engagement and academic achievement.
graphic composition of the participants in our sample, Such efforts may serve to create a developmental cascade
which was significantly different from the U.S. population of positive, bidirectional relations between a young person
as a whole, most noticeably with regard to sex, rural–urban whose actions involve engagement and whose levels of
neighborhood, and SES, as noted in the description of achievement reflect high competence in his or her educa-
participants. We must therefore temper our findings with tional context.
the acknowledgement that our longitudinal sample may
have unidentified differences in model fit and structure as Acknowledgments This research was supported in part by grants
from the National 4-H Council, the Altria Corporation, the Thrive
compared with the overall sample in the 4-H study, as well Foundation for Youth, and the John Templeton Foundation.
as in the population. For example, our minimal urban
sample within the current sample of 4-H study participants Authors’ contributions Paul A. Chase participated in the study’s
limited the generalizability of our findings. These demo- design and coordination, performed the statistical analyses, and
drafted the manuscript; Lacey J. Hilliard participated in the coordi-
graphic differences relating to the ethnic and socio eco- nation and editing of the manuscript; G. John Geldhof participated in
nomic composition of the present sample may also impact the design and interpretation of the data analyses; Daniel J. A. Warren
generalizability, as previous studies have demonstrated participated in the coordination and editing of the manuscript; Rich-
their relationship with academic achievement (Shernoff ard M. Lerner conceived of the study, participated in its design and
coordination, and helped to draft the manuscript. All authors read and
and Schmidt 2008). Finally, as noted earlier, the strength of approved the final manuscript.
the association between school engagement and GPA has
been found to vary significantly depending on how
engagement is measured, for instance, involving whether References
school engagement is self-reported (as in the current study)
or teacher reported (Shernoff and Schmidt 2008). Future Balfanz, R., & Byrnes, V. (2006). Closing the mathematics achieve-
studies may consider a triangulation of measures from both ment gap in high poverty middle schools: Enablers and
constraints. Journal of Education for Students Placed at Risk,
teachers and students in order to evaluate these constructs 11, 143–159.
in adolescence more effectively. Balfanz, R., Bridgeland, J. M., Moore, L. A., & Fox, J. H. (2010).
Building a grad nation: Progress and challenge in ending the
high school dropout epidemic. Washington DC: America’s
Promise Alliance, Civic Enterprises Everyone Graduates Center
Conclusion at Johns Hopkins University.
Bowers, E. P., Geldhof, G. J., Johnson, S. K., Lerner, J. V., &, Lerner,
The current study demonstrated that the tripartite model of R. M. (in press). Thriving across the adolescent years: A view of
school engagement and academic achievement are recip- the issues. Journal of Youth and Adolescence.
Brown, T. A. (2006). Confirmatory factor analysis for applied
rocally linked, and school engagement positively predicts research. New York: Guilford Press.
the GPA of high school students. In addition, GPA plays a Cameron, R. G. (1989). The common yardstick: A case for the SAT.
role in subsequent behavioral, emotional, and cognitive New York: College Entrance Examination Board.
school engagement. Regardless of the existing academic Cassady, J. C., (2001). Self-reported GPA and SAT: A methodolog-
ical note. Practical Assessment, Research, and Evaluation,
skills of the student, researchers must continue to 7(12). http://ericae.net/pare/getvn.asp?v57&n512.
acknowledge the importance of student engagement within Chavez, E. L., Oetting, E. R., & Swaim, R. C. (1994). Dropout and
the context of learning, and the relationship that students delinquency: Mexican-American and Caucasian non-Hispanic
develop with their educational context, including the youth. Journal of Clinical Child Psychology, 23(10), 47–55.
Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit
behavioral, emotional, and cognitive connections to their indexes for testing measurement invariance. Structural Equation
academic environment. Modeling, 9, 233–255.
As such, educational policy makers and school person- Christenson, S. L., Reschly, A. L., & Wylie, C. (Eds.). (2012).
nel should develop initiatives to promote enhanced school Handbook of research on student engagement. New York:
Springer Science.
engagement, focused perhaps on the strongest predictor of Crosnoe, R., Johnson, M. K., & Elder, G. H. (2004). Intergenerational
GPA in our model (behavioral engagement), in order to bonding in school: The behavioral and contextual correlates of
promote academic achievement. Policies and programs student-teacher relationships. Sociology of Education, 77, 60–81.
may be aimed at ‘‘winning the hearts and minds’’ (the Dotterer, A. M., & Lowe, K. (2011). Classroom context, school
engagement, and academic achievement in early adolescence.
emotional and cognitive facets of school engagement) of Journal of Youth and Adolescence, 40(12), 1649–1660.
high school students. Nevertheless, it may be that initia- Eccles, J. S., & Midgley, C. (1989). Stage/environment fit: Devel-
tives that promote intrinsic motives for behavioral opmentally appropriate classrooms for early adolescents. In R.

123
J Youth Adolescence (2014) 43:884–896 895

E. Ames & C. Ames (Eds.), Research on motivation in education Lerner, R. M., von Eye, A., Lerner, J. V., Lewin-Bizan, S., & Bowers,
(Vol. 3, pp. 139–186). San Diego, CA: Academic Press. E. P. (2010). The meaning and measurement of thriving in
Eccles, J., & Roeser, R. W. (2011). Schools as developmental adolescence: Findings from the 4-H study of positive youth
contexts during adolescence. Journal of Research on Adoles- development. Journal of Youth and Adolescence, 39(7),
cence, 21, 225–241. 113–143.
Finn, J. D., & Rock, D. A. (1997). Academic success among students Lerner, R. M., von Eye, A., Lerner J., & Lewin-Bizan, S. (Eds.)
at risk for school failure. Journal of Applied Psychology, 82(2), (2009b). Foundations and functions on thriving in adolescence:
221–234. Findings from the 4-H study of positive youth development.
Finn, J. D., & Voelkl, K. E. (1993). School characteristics related to Journal of Applied Developmental Psychology, 30(5), 567–648.
school engagement. Journal of Negro Education, 62, 249–268. Li, Y., (2010). School engagement in adolescence: Theoretical
Fredricks, J. A., Blumenfeld, P., & Paris, A. H. (2004). School structure, measurement equivalence, and developmental trajec-
engagement: Potential of the concept, state of the evidence. tories. (Doctoral dissertation, Tufts University).
Review of Educational Research, 74, 59–109. Li, Y., & Lerner, R. M. (2011). Trajectories of school engagement
Fredricks, J. A., & Eccles, J. S. (2006). Is extracurricular participation during adolescence: Implications for grades, depression, delin-
associated with beneficial outcomes: Concurrent and longitudi- quency, and substance use. Developmental Psychology, 47(1),
nal relations? Developmental Psychology, 42, 698–713. 233–247.
Goodenow, C. (1993). Classroom belonging among early adolescent Li, Y., & Lerner, R. M. (2013). Interrelations of behavioral,
students: Relationship to motivation and achievement. Journal of emotional, and cognitive school engagement in high school
Early Adolescence, 13, 21–43. students. Journal of Youth and Adolescence, 42(1), 20–32.
Greenberg, M. T., Weissberg, R. P., O’Brien, M. U., Zins, J. E., Li, Y., Lerner, J. V., & Lerner, R. M. (2010). Personal and ecological
Fredericks, L., Resnik, H., et al. (2003). Enhancing school-based assets and academic competence in early adolescence: The
prevention and development through coordinated social, emo- mediating role of school engagement. Journal of Youth and
tional, and academic learning. American Psychologist, 58, Adolescence, 39(7), 801–815.
466–474. Lindner, J. R., Murphy, T. H., & Briers, G. E. (2001). Handling
Hauser-Cram, P., Warfield, M. E., Stadler, J., Sirin, S. R., Huston, A. nonresponse in social science research. Journal of Agricultural
C., & Ripke, M. N. (2006). School environments and the Education, 42(4), 43–53.
diverging pathways of students living in poverty (pp. 198–216). Little, T. D. (1997). Mean and covariance structures (MACS)
Developmental contexts in middle childhood: Bridges to analyses of cross-cultural data: Practical and theoretical issues.
adolescence and adulthood. Multivariate Behavioral Research, 32, 53–76.
Heaven, P. C. L., Mak, A., Barry, J., & Ciarrochi, J. (2002). Little, T. D. (2013). Longitudinal structural equation modeling. New
Personality and family influences on adolescent attitudes to York, NY: Guilford Press.
school and self-rated academic performance. Personality and Marsh, H. W. (1990). Causal ordering of academic self-concept and
Individual Differences, 32, 453–462. academic achievement: A multiwave, longitudinal panel ana-
Heckman, J. J., & Masterov, D. V. (2006). The productivity argument lysis. Journal of Educational Psychology, 82, 646–656.
for investing in young children, discussion paper: Early Child- Marsh, H. W., & O’Mara, A. (2008). Reciprocal effects between
hood Research Collaborative, University of Minnesota. academic self-concept, self-esteem, achievement, and attainment
Henry, K. L., Knight, K. E., & Thornberry, T. P. (2012). School over seven adolescent years: Uni-dimensional and multidimen-
disengagement as a predictor of dropout, delinquency, and sional perspectives of self-concept. Personality and Social
problem substance use during adolescence and early adulthood. Psychology Bulletin, 34, 542–552.
Journal of Youth and Adolescence, 41, 156–166. McNeeley, C. A., Nonnemaker, J. M., & Blum, R. W. (2002).
Kindermann, T. A. (2007). Effects of naturally existing peer groups Promoting school connectedness: Evidence from the National
on changes in academic engagement in a cohort of sixth graders. Longitudinal Study Of Adolescent Health. Journal of School
Child Development, 78, 1186–1203. Health, 72, 138–146.
Klem, A. M., & Connell, J. P. (2004). Relationships matter: Linking Meehan, B. T., Hughes, J. N., & Cavell, T. A. (2003). Teacher-student
teacher support to student engagement and achievement. Journal relationships as compensatory resources for aggressive children.
of School Health, 74(7), 262–273. Child Development, 60, 981–992.
Larson, R. W., & Rusk, N. (2011). Intrinsic motivation and positive Murray, C., & Greenberg, M. (2001). Relationships with teachers and
development. In R. M. Lerner, J. V. Lerner, & J. B. Benson bonds with school: Social emotional adjustment correlates for
(Eds.), Advances in child development and behavior: Positive children with and without disabilities. Psychology in the Schools,
youth development (pp. 89–130). Oxford, UK: Elsevier. 38, 25–41.
Lerner, R. M., Lerner, J. V., Almerigi, J., Theokas, C., Phelps, E., Muthén, L. K., & Muthen, B. (2010). Mplus 6.0. Los Angeles, CA:
Gestsdóttir, S., et al. (2005). Positive youth development, Muthén & Muthén.
participation in community youth development programs, and Phelps, E., Balsano, A., Fay, K., Peltz, J., Zimmerman, S., Lerner, R.
community contributions of fifth-grade adolescents: Findings M., et al. (2007). Nuances in early adolescent development
from the first wave of the 4-H study of positive youth trajectories of positive and of problematic/risk behaviors:
development. Journal of Early Adolescence, 25(1), 17–71. Findings from the 4-H study of positive youth development.
Lerner, R. M., Lerner, J. V., von Eye, A., Bowers, E. P., & Lewin- Child and Adolescent Clinics of North America, 16(2), 473–496.
Bizan, S. (Eds.) (2011). Individual and contextual bases of Resnick, M. D., Bearman, P. S., Blum, R. W., Bauman, K. E., Harris,
thriving in adolescence: Findings from the 4-H study of positive K. M., Jones, J., et al. (1997). Protecting adolescents from harm:
youth development. Journal of Adolescence, 34(6), 1107–1228. Findings from the National Longitudinal Study on Adolescent
Lerner, J. V., Phelps, E., Forman, Y., & Bowers, E. P. (2009a). Health. JAMA, 278, 823–832.
Positive youth development. In R. M. Lerner & L. Steinberg Scheidler, M. J. (2012). The Relationship Between Student Engage-
(Eds.), Handbook of adolescent psychology, Individual bases of ment and Standardized Test Scores of Middle School Students:
adolescent development (3rd ed., Vol. 1, pp. 524–558). Hoboken, Does Student Engagement Increase Academic Achievement?
NJ: Wiley. (Doctoral dissertation, University of Minnesota).

123
896 J Youth Adolescence (2014) 43:884–896

Shernoff, D. J., & Schmidt, J. A. (2008). Further evidence of an Paul A. Chase is a doctoral student at the Institute for Applied
engagement-achievement paradox among U.S. high school Research in Youth Development at Tufts University. His research
students. Journal of Youth and Adolescence, 37, 564–580. interests involve the study of positive youth development and
Skinner, E. A., Kindermann, T. A., & Furrer, C. J. (2009). A intentional self regulation.
motivational perspective on engagement and disaffection: Con-
ceptualization and assessment of children’s behavioral and Lacey J. Hilliard is a research assistant professor at the Institute for
emotional participation in academic activities in the classroom. Applied Research in Youth Development at Tufts University. Her
Educational and Psychological Measurement, 69, 493–525. work integrates the study of positive youth development and gender
Theokas, C., & Lerner, R. M. (2006). Observed ecological assets in roles.
families, schools, and neighborhoods: Conceptualization, mea-
surement and relations with positive and negative developmental G. John Geldhof is an assistant professor in Human Development
outcomes. Applied Developmental Science, 10, 61–74. and Family Sciences at Oregon State University. His research
U.S. Census Bureau (2010). American fact finder. http://factfinder2. emphasizes the integration of diverse theoretical approaches to self-
census.gov/. regulation and examines how self-regulation develops across the life
Voelkl, K. E. (1997). Identification with school. American Journal of span.
Education, 105, 204–319.
Wigfield, A. (1994). Expectancy-value theory of achievement moti- Daniel J. A. Warren is a doctoral student at the Institute for Applied
vation: A developmental perspective. Educational Psychology Research in Youth Development at Tufts University. His research
Review, 6, 49–78. interests involve the study of positive youth development and
Wigfield, A., & Cambria, J. (2010). Students’ achievement values, educational programs.
goal orientations, and interest: Definitions, development, and
relations to achievement outcomes. Developmental Review,
Richard M. Lerner is the Bergstrom Chair in Applied Develop-
30(1), 1–35.
mental Science and the Director of the Institute for Applied Research
Woolley, M., & Grogan-Kaylor, A. (2006). Protective family factors
in Youth Development at Tufts University. He received his Ph.D. in
in the context of neighborhood: Promoting positive school
developmental psychology from the City University of New York.
outcomes. Family Relations, 55, 93–104.
His work integrates the study of public policies and community-based
Wothke, W. (2000). Longitudinal and multigroup modeling with
programs with the promotion of positive youth development and
missing data. In T. D. Little, K. U. Schnabel, & J. Baumert
youth contributions to civil society.
(Eds.), Modeling longitudinal and multilevel data (pp. 219–240).
Mahwah, NJ: Erlbaum.
Zimmerman, B. J., & Schunk, D. H. (2001). Self-regulated learning
and academic achievement: Theoretical perspectives (2nd ed.).
Mahwah, NJ: Erlbaum.

123