AN EXAMINATION OF THE RELATIONSHIP AMONG AFFECTIVE, COGNITIVE,
BEHAVIORAL, AND ACADEMIC FACTORS OF STUDENT ENGAGEMENT OF
9th GRADE STUDENTS
by
PETER 1. BURROWS
A DISSERTATION
Presented to the Department of Educational Methodology,
Policy, and Leadership
and the Graduate School of the University of Oregon
in partial fulfillment of the requirements
for the degree of
Doctor of Education
June 2010
11
University of Oregon Graduate School
Confirmation of Approval and Acceptance of Dissertation prepared by:
Peter Burrows
Title:
"An Examination of the Relationship among Affective, Cognitive, Behavioral, and Academic
Factors of Student Engagement of 9th Grade Students"
This dissertation has been accepted and approved in partial fulfillment of the requirements for
the Doctor of Education degree in the Department of Educational Methodology, Policy, and
Leadership by:
Edward Kameenui, Chairperson, Educational Methodology, Policy, and Leadership
Keith Zvoch, Member, Educational Methodology, Policy, and Leadership
Keith Hollenbeck, Member, Educational Methodology, Policy, and Leadership
Robert Davis, Outside Member, Romance Languages
and Richard Linton, Vice President for Research and Graduate Studies/Dean of the Graduate
School for the University of Oregon.
June 14,2010
Original approval signatures are on file with the Graduate School and the University of Oregon
Libraries.
III
An Abstract of the Dissertation of
Peter L. Burrows for the degree of Doctor of Education
in the Department of Educational Methodology, Policy, and Leadership
to be taken June 2010
Title: AN EXAMINATION OF THE RELATIONSHIP AMONG AFFECTIVE,
COGNITIVE, BEHAVIORAL, AND ACADEMIC FACTORS OF STUDENT
ENGAGEMENT OF 9th GRADE STUDENTS
Approved: _
Dr. Edward J. Kame'enui
Research has identified the construct of student engagement as an antecedent to
positive academic outcomes. In this study, the Student Engagement Instrument (SEI) was
administered to 371 9th grade students at a comprehensive high school to measure the
cognitive and affective engagement of students. Exploratory factor analyses were
conducted on the 35-item SEI with best model fit matching previous research in which a
five-factor model was found. Logistic and multiple regression analyses were then utilized
to explore the relationships among cognitive and aflective engagement and student
achievement and behavioral outcomes. Findings generally supported the significance of
the student engagement subtypes of cognitive and affective engagement in predicting
educational outcomes. Results suggest that further study of the affective and cognitive
subtypes and their development over the course of a student's education would enhance
the understanding of the student engagement construct and lead to the development of
interventions to mediate the effects of these subtypes.
IV
vCURRICULUM VITAE
NAME OF AUTHOR: Peter 1. Burrows
PLACE OF BIRTH: Stamford, CT
DATE OF BIRTH: 04/0911970
GRADUATE AND UNDERGRADUATE SCHOOLS ATTENDED:
University of Oregon, Eugene, Oregon
School for International Training, Brattleboro, VT
DEGREES AWARDED:
Doctor of Education, Educational Methodology, Policy, and Leadership, 2010, University
of Oregon
Master of Arts in Teaching, ESL, 1999, School for International Training
Bachelor of Arts, English, 1993, University of Oregon
AREAS OF SPECIAL INTEREST:
Secondary Literacy Interventions Using Response to Intervention (Rt!) in grades 9-12
School Inclusiveness and Peer Leadership Interventions
Equity Funding and Resource Allocation
PROFESSIONAL EXPERIENCE:
Assistant Principal, Willamette High School, 2008-2010
English Teacher, Willamette High School, 2004-2008
English Professor, Korea University, 2000-2002
English Professor, Universidad del Mar, 1999-2000
English Teacher, GSP International, 1998-1999
English Teacher, Gakushinkan, 1994-1997
VI
ACKNOWLEDGMENTS
I wish to thank the incredibly supportive College of Education for its support
throughout my studies. Specifically, I'd like to thank Dr. Edward Kame'enui and Dr.
Keith Zvoch for their unerring guidance as I've navigated the writing process. This
experience has been tremendously rewarding, and this is in large part due to their passion
for research and the advancement of sound educational practices. I'd also like to thank
Bethel School District for their support, particularly Drew Braun and Lori Smith, as I've
benefitted greatly from their openness and enthusiasm regarding this research.
I dedicate this project to my family, without whom this would not be.
Vll
Chapter
TABLE OF CONTENTS
Page
V111
I. INTRODUCTION 1
Conceptions of Student Engagement................................................... 1
Commonalities 2
Evolution of Subtype Definitions 3
Behavioral Engagement................................................................. 3
Emotional Engagement.................................................................. 3
Cognitive Engagement................................................................... 4
Development of the Four-Subtype Taxonomy.................................... 5
Multidimensionality of the Student Engagement Construct.......... 6
Cognitive and Affective Engagement.................................................. 6
Links to Student Achievement....................................................... 7
Links to Dropout............................................................................ 8
Measurement of Cognitive and Affective Engagement................. 10
Study Purpose 11
II. LITERATlTRE REVIEW............................................................................ 12
Measuring Cognitive and Psychological Engagement: Validation of
the Student Engagement Instrument (Appleton et al. 2006).......... 12
Rationale of the Study....... 13
Methodology.................................................................................. 15
Results............................................................................................ 16
Implications.................................................................................... 18
Engagement as Flourishing: The Contribution of Positive Emotions
and Coping to Adolescents' Engagement at School and with
Learning (Reschly et al. 2008)................................... 18
Rationale of the Study.................................................................... 18
Methodology.................................................................................. 20
Results............................................................................................ 21
Implications.................................................................................... 22
A Study of the Reliability and Construct Validity of the School
Engagement Instrument (SEI) across Multiple Grades
(Betts et al. 2009) 23
Rationale of the Study.................................................................... 24
Methodology.................................................................................. 24
Results............................................................................................ 26
Implications.................................................................................... 27
Research Questions 27
Hypotheses 28
III. METHODOLOGY 29
Setting and Participants........................................................................ 29
Measure.......................................................................................... 30
Chapter Page
IX
Predictor Variables......................................................................... 30
Criterion Measures (District Variables).. 31
SEI Administration Procedures............................................................ 32
Missing Data 33
Data Analysis 33
Exploratory Factory Analysis 33
Supplemental Analyses 34
Regression Analyses 34
IV. RESULTS 36
Descriptive Statistics............................................................................ 36
Exploratory Factor Analysis 36
Correlational Analysis 43
Regressional Analyses 45
Grade Point Average...................................................................... 46
Credit Completion.......................................................................... 47
Minor Behavioral Referrals 49
Major Behavioral Referrals............................................................ 49
V. DISCUSSION 52
Discussion of Analytical Results 53
Exploratory Factor Analysis 53
Logistic and Multiple Regression Analyses.................................. 54
Predictive Factor Strengths 55
Outcome Variables......................................................................... 58
Limitations 60
Future Implications.............................................................................. 61
Conclusion.................. 63
REFERENCES 65
Figure
LIST OF FIGURES
Page
x
1. Four-Part Typology of Student Engagement. Adapted from Appleton et al.
(2006, p.430)............................................................................................... 5
2. Scree Plot of Student Engagement Instrument Exploratory Factor Analysis
Loading onto Five Factors. 38
LIST OF TABLES
Table Page
4.1. Descriptive Statistics for the Sample of 9th Grade Students (N = 371) .. , 36
4.2. Item Loadings on SEI Factors................................................................... 40
4.3. Pearson Correlation Matrix between Factors and Outcomes.................... 45
4.4. Regression of GPA on Covariates and SEI Factors 47
4.5. Regression of Credit Completion on Covariates and SEI Factors............ 48
4.6. Regression of Minor Behavioral Referrals on Covariates and SEI
Factors 49
4.7. Regression of Major Behavioral Referrals on Covariates and SEI
Factors 51
Xl
1CHAPTER I
INTRODUCTION
The construct of student engagement has been identified as a key component in
understanding secondary students' complex psychological relationship with school
(Fredericks, Blumenfeld, & Paris, 2004).The scholastic engagement of students is also
considered a primary theoretical pathway to understand and respond to the thoughts,
behaviors, and feelings that lead to high school dropout (Christenson, Reschly, Appleton,
Berman, Spangers, & Varro, 2008; Finn, 1989). Yet, recent evidence demonstrates that
students are becoming more disengaged from school, both socially and academically
(Appleton, Christenson, & Furlong, 2008; Fredericks et aI., 2004; National Research
Council, 2004). The sharp decline in student engagement among high school students
(Fredricks & Eccles, 2002) suggests that individuals' cognitive, behavioral, and
emotional states put them at higher risk of dropping out without acquiring the basic skills
necessary to gain employment in the modem workplace (Furlong & Christenson, 2008;
National Research Council, 2004). As a result, the study of student engagement has
increased as researchers realize the potential of utilizing engagement-related strategies as
a tool to promote enthusiasm for learning and school reform (National Research Council,
2004).
Conceptions of Student Engagement
Although student engagement has been cited as an important mediator of
academic achievement, pro-social behavior, and educational persistence (Janosz,
2Archambault, Morizot, & Pagani, 2008), conceptions of the construct of student
engagement have varied. Researchers in the field of engagement have not reached a
consensus on the definition and measurement of student engagement (Appleton et aI.,
2008; Sharkey, You, & Schnoebelen, 2008). For example, a review of the research over
the last 25 years revealed that student engagement has been defined using at least eight
different terms: (a) engagement; (b) engagement in schoolwork; (c) academic
engagement; (d) school engagement; (e) student engagement; (f) student engagement in
academic work; (g) student engagement with school; and (h) participation identification
(Appleton et aI., 2008). Furthermore, researchers using the same terms have differed in
the use of these terms when referring to the student engagement construct, which has led
to a range of definitions that has made cross-study comparisons challenging (Fredericks
et aI., 2004). Researchers who study student engagement continue to struggle in gaining
agreement on the definitions and measures of the construct, as is apparent in recent
reviews of the literature that reveal an increase in the use oftelms related to student
engagement (Appleton et aI., 2008).
Commonalities. Despite the use of different terms and the varied
operationalizations of these terms, some conceptualizations of engagement share
common features (Jimerson, Campos, & Greif, 2003). For example, Appleton et aI.'s
(2008) meta-analysis found that all studies contained behavioral components, and most of
these components had emotional or psychological dimensions. Some studies combined
these components into singular constructs, although all three component dimensions were
arguably present in the studies (Fredericks et aI., 2004). The development of an
engagement typology has pushed conceptions of engagement towards a multidimensional
3meta-construct with multiple components (Anderson, Christenson, Sinclair, & Lehr,
2004; Guthrie & Wigfield, 2000), although components of the meta-construct have
differed (Appleton et aI., 2008). This typology has included two to four dimensions of
student engagement (Appleton et aI., 2008; Appleton, Christenson, Kim, & Reschly,
2006).
Evolution of Subtype Definitions
Behavioral engagement. Fredericks et al. (2004) identified three dimensions of
engagement--behavioral, emotional, and cognitive--in the literature and addressed the
multiple interpretations of each component. These researchers defined behavioral
engagement as (a) "positive conduct, such as following the rules and adhering to
classroom norms, as well as the absence of disruptive behaviors such as skipping school
and getting in trouble," (b) "involvement in learning and academic tasks and includes
behaviors such as effort, persistence, concentration, attention, asking questions, and
contributing to class discussion," and (c) "participation in school-related activities such as
athletics or school governance" (p. 62). A two-tiered conception of engagement
comprised of basic behaviors (e.g. participation, attendance) and higher-level behaviors
(e.g. effort to learn) is relatively common in the student engagement field (Glanville &
Wildhagen, 2007).
Emotional engagement. Emotional engagement is typically represented as
affective responses such as interest, excitement, stress, and attitude (Fredericks et aI.,
2004; Marks, 2000). Some conceptualizations have also tied emotional engagement to
students' sense of belonging and identification with school. However, some researchers
have argued that the components of belonging and value should be defined separately due
4to confounding antecedents, such as family, educational context, and cultural influences
(Finn, 1989; Finn, 1993; Glanville & Wildhagen, 2007). Emotional engagement has also
been characterized as representing students' feelings about the people, policies, and
practices of the school environment that include students' complex relationships to
school (Yazzie-Mintz, 2007).
Cognitive engagement. The construct of cognitive engagement temporally
follows the development of behavioral and emotional engagement (Appleton et aI.,
2008). Cognitive engagement has been defined as "self-regulation, relevance of
schoolwork to future endeavors, value of learning, personal goals and autonomy"
(Appleton et aI., 2008, p.372). Furthermore, cognitive engagement has been expressed by
"flexibility in problem solving, preference for hard work, and positive coping in the face
of failure" (Fredericks et aI., 2004, p.64), as well as the ability to use metacognitive skills
to evaluate task requirements (Connell & Wellborn, 1991; Reschly, Huebner, Appleton,
& Antaramian, 2008). The meta-analysis of Fredericks et al. (2004) fOlmd that cognitive
engagement has been presented with numerous competing conceptions and definitions.
For example, in Appleton et aI.'s (2008) meta-analysis of the engagement construct and
definitions, cognitive engagement was defined with wide variation in the 19 studies
selected. Fredericks et ai. (2004) found a split between literature that represented
psychological investments in learning - concentrated focus despite distraction, and the
study of cognition and strategic learning - represented by a student's effort exerted to
meet and exceed requirements (Como, 1993). Clearly, the research on student
engagement needs to reach agreement on the definition and measurement of cognitive
engagement to address the disparity between the psychological requirements of
5investment in learning and the cognitive framework of strategic learning (Fredericks et
aI., 2004).
Development of Four-Subtype Taxonomy
Citing the need to "empirically and theoretically refine and clarify [the student
engagement] construct" (p. 382), Appleton et aI. (2006,2008) attempted to incorporate
essential components of the multi-dimensional construct of student engagement research
into a taxonomy of student engagement. Figure 1 presents a graphic representation of the
engagement construct posited by Appleton et aI. (2006).
-Time 011 task
- Credithours towards
graduation
•HomeworkcOtllpletion
- Seif-regtlla~jQn
'Relevanceof~cho(ll to
. goals
":Value of learn'bi
-Strategizing ....
Academic
Engagement
Cognitive
Engagement
Behavioral
Engagement
Affective
Engagement
-Atten.Elal1ce
- Classroom participatiorj .
-Extracurricular participation
- btra cre<litriptions
Figure 1. Four-part typology of student engagement. Adapted from Appleton et aI. (2006,
p.430).
6Appleton et. aI.' s (2008) taxonomy of student engagement addresses both the absence of
academic engagement as well as the definitional ambiguity of the constructs of
behavioral, emotional, and cognitive engagement. The inclusion of academic engagement
in the taxonomy is important, as time on task and work completion have been correlated
with student achievement (Appleton et aI., 2008; Fredericks et aI., 2004).
lVIultidimensionality of the student engagement construct. Researchers have
argued that the construct of student engagement must be viewed as a meta-construct in
order to understand the interrelationship of the subtypes and their relationship with
student dropout behavior (Appleton et aI., 2008; Fredericks et aI., 2004; Glanville &
Wi1dhagen, 2007; Reschly et aI., 2008). Viewing the subtypes of student engagement
individually fails to address the "dynamically interrelated" (Fredericks et aI., 2004, p. 61)
nature of the cognitive, behavioral, and emotional experiences of students and
oversimplifies a complex set of interactions. The study of student engagement as a
multidimensional construct has the potential to reveal the interplay between subtypes of
engagement and the school environment. Such a study should reveal the components
important to the development of interventions to improve student engagement (Appleton
et aI., 2008; Archambault, Janosz, Fallu, & Pagani, 2009; Fredericks et aI., 2004).
Cognitive and Affective Engagement
In order to refine the various conceptualizations of cognitive engagement in the
literature, Appleton et aI. (2006) defined cognitive engagement as (a) "self-regulation,"
(b) "relevance of schoolwork to future endeavors," (c) "value of learning," and (d)
"personal goals and autonomy" (p. 429), which represented a narrowing of the extant
definitions of the components (Fredericks et aI., 2004; Jimerson et aI., 2003). For
7example, previous studies had constructed scale items for the cognitive engagement
subtype retroactively from national datasets without a taxonomy of the engagement
construct (Appleton et aI., 2006). In addition, studies had conflated cognitive and
behavioral engagement indicators to create weak subtype parameters (Fredericks et aI.,
2004). Affective engagement was defined as "feelings of identification or belonging, and
relationships with teachers and peers" (Appleton et aI., 2006, p. 429). These two
components--cognitive and affective-- represented what Appleton et aI. (2006)
characterized as the "internal forms of engagement," (p.429) signifying a student's sense
of connection to school and to the role of school in the student's future.
Links to student achievement. In the published research literature, investigators
have reported strong relations between cognitive and affective engagement and school
indicators of academic achievement (Appleton et ai., 2006; Appleton et ai., 2008;
Archambault et aI., 2009; Fan & Williams, 2009; Furlong & Christenson, 2008; Reschly
et ai., 2008). For example, in a study on school belonging, Goodenow (1993) found that
psychological membership in school was linked to school motivation, academic
achievement, and teacher-rated effort. In another study, Gutman and Midgley (2000)
found that school belonging was significantly correlated with grade point average. This
result was similar to the strong correlation that Goodenow and Grady (1993) found
between school belonging and school motivation and subsequent academic achievement.
Research on cognitive engagement has also demonstrated significant correlations
between student goals and self-efficacy and achievement (Appleton et aI., 2008;
Fredericks et aI., 2004; Greene et aI., 2004; Kortering & Braziel, 2008; McMahon et aI.,
2008). For example, in a study of high school students, Greene, Miller, Crowson, Duke,
8& Akey (2004) noted that perceptions of current class work which they found as being
instrumental to success were significantly associated with academic achievement.
Furthermore, indicators of self-efficacy were positively correlated with grades and
adaptive student motivation (Greene et aI., 2004). These indicators of self-efficacy were
also positively related to instructional interventions (Furlong & Christenson, 2008;
Marks, 2000).
Research on student engagement has increased as the construct has continued to
be linked to student academic outcomes and graduation rates (Reschly et aI., 2008). Finn
(1989, 1993) highlighted the separation between a student's demographic or historical
characteristics and behavioral risk factors such as participation or engagement that reduce
the likelihood of graduation and school success. Demographic factors include elements of
students' family or personal backgrounds that impact the students' success in school and
are unalterable, while behavioral risks were comprised of a set of behaviors that reduced
the likelihood that successful school outcomes would be realized (Finn, 1993). This
identification of alterable risk variables related to dropout provided a framework for the
development of interventions to address specific student needs (Finn, 1993; Reschly et
aI.,2008).
Links to dropout. Research has defined high school dropout as the cumulating
event of a "confluence of individual, social, family, cultural, socioeconomic, and
institutional factors" (Janosz et aI., 2008, p.22). The confluence has been tied together by
the degree to which students are engaged with the academic environment (Appleton et
aI., 2008; Reschly & Christenson, 2006). For example, controlling for socioeconomic
status, achievement test outcomes, and grade retention status, student engagement
9measures predicted the dropout status for students with and without disabilities in a study
utilizing data from the National Education Longitudinal Study (Reschly & Christenson,
2006; Sharkey et aI., 2008). In a longitudinal study of French-Canadian high school
students (N = 11,827), Archambault et aI. (2008) found that behavioral, affective, and
cognitive traits of engagement as measured in an 18-item self-report survey predicted
school dropout. In addition, Archambault et aI. (2009) surveyed 69 high schools in
Quebec (N = 13,330) in a cross-sectional study of affective, cognitive, and behavioral
dimensions of student engagement. Results of growth mixture modeling analyses
provided support for the significance of cognitive and affective engagement in predicting
dropout status. Furthermore, student engagement research has predicted high school
dropout based on student engagement measures in elementary school (Alexander et aI.,
1997; Reschly & Christenson, 2006). Students of color are particularly susceptible to
dropout, with graduation rates and measures of student engagement well below their
white counterparts (National Research Council, 2004; Yazzie-Mintz, 2007).
While student engagement and its effect on high school dropout have historically
been conceptualized in the literature as being difficult to change (Doll & Hess, 2001),
recent studies have begun to identify alterable influences that enable the development of
interventions that respond to students' gradual disengagement from school (Appleton et
aI., 2008). Furlong and Christenson (2008) drew the distinction between contextual
factors that were indicators of engagement and those that were facilitators of
engagement. Indicators of engagement, such as schoolwork completion, attendance, and
grades, measure a student's connection with learning and provide information to identify
disengagement. In contrast, facilitators of engagement, including parental supervision of
10
homework completion, school discipline systems, and student investment in learning,
represent factors that are amenable to intervention (Appleton et aI., 2008; Christenson &
Thurlow, 2004; Furlong & Christenson, 2008).
Measurement of cognitive and affective engagement. The high degree of
inference required to measure the internal forms of cognitive and affective engagement
has resulted in fewer studies on these subtypes of student engagement. However, the
studies that have been conducted demonstrate a positive relationship between cognitive
and affective engagement and student achievement (Appleton et aI., 2006; Reschly &
Christenson, 2006). As noted elsewhere (Appleton et aI., 2008; Kortering & Braziel,
2008; National Research Council, 2004; Reschly & Christenson, 2006), further research
is required to reveal the underlying cognitive and affective indicators of student
engagement that may be more elusive to measurement and intervention than the academic
and behavioral engagement subtypes. Difficulties in measuring and conceptualizing
cognitive and affective engagement have led to the haphazard development of research
on student engagement and have made comparisons among studies challenging.
In order to address the disparate approaches to the measurement of cognitive and
affective engagement, more research needs to pursue valid measures of cognitive and
affective engagement that employ self-report measures as opposed to inferred, objective
measures (Appleton et aI., 2006; Kortering & Braziel, 2008). Substantial evidence exists
that student self-report is a valid and reliable methodology (Assor & Connell, 1992;
Carini, Kuh, & Klein, 2006; Pace, 1984). Furthermore, Fredericks et aI. (2004) cited the
limitations of utilizing observation to assess cognition and the "deep-levelleaming
strategies" (p. 68) that reveal cognitive and affective engagement, and suggested that
---------------------
11
researchers work to develop self-report surveys to assess engagement. Some researchers
argue that self-report may be more reliable than inferred measures as a result of the
nature of the subtypes of cognitive and affective engagement and their measure of student
experiences, perceptions, and beliefs (Appleton et aI., 2006; Fredericks et aI., 2004;
Johnson, 2008). Although self-report measurement does not allow for cross-informant
perspectives (Sharkey et aI., 2008), it does control for the tendency for observers to infer
student thought and feeling (Johnson, 2008).
Study Purpose
The purpose of this study is to assess the cognitive and affective engagement of a
sample of high school freshmen and to examine the relationship among self-reported
cognitive and affective engagement and various academic and behavioral outcomes. Few
instruments have been developed to measure cognitive and affective engagement
(Appleton et aI., 2006; Appleton et aI., 2008; Betts et aI., 2009). This study utilizes the
Student Engagement Instrument (SEI), a self-report survey of cognitive and affective
engagement, as a means to explore the relationship among all four subtypes of the student
engagement construct. This study addresses the predictive strength of a five-factor model
of cognitive and affective engagement based on the SEI. In addition, this study addresses
the relative strength of each factor in predicting educational outcomes.
12
CHAPTER II
LITERATURE REVIEW
Appleton et aI. (2006) cited the absence of "theoretically sound and empirically-
based measures of cognitive and affective engagement" (p. 431) as the main limitation to
the development of a research base from which to develop an understanding of the multi-
dimensional construct of student engagement. Because of a perceived overemphasis in
the student engagement literature on indicators of behavioral and academic engagement
(Appleton et aI. 2006), the development of instrumentation to accurately measure the
student perspective has been limited (Appleton et aI., 2008; Furlong & Christenson,
2008). To this end, the Student Engagement Instrument (SEI) was developed to provide
valid data on cognitive and affective engagement by assessing a students' perception and
experience with the educational setting (Appleton et aI. 2006; Betts et aI., 2009; Reschly
et aI., 2008).
Measuring Cognitive and Psychological Engagement: Validation of the Student
Engagement Instrument (Appleton et aI., 2006)
Appleton et aI. (2006) utilized existing research to develop a survey instrument to
measure cognitive and affective engagement. Drawing on the divergent approaches to the
conceptualization and measurement ofthe construct, Appleton and colleagues (2006)
attempted to examine the construct of student engagement by establishing a rationale and
an empirical basis for the creation of four construct subtypes: academic, behavioral,
cognitive, and affective engagement.
13
Rationale of the study. The taxonomy proposed by Appleton et aI. (2006) argued
for the existence of multiple indicators of student engagement associated with each of the
four subtypes. These indicators are considered to be representative of essential contexts
of learning (e.g., value of learning, support from family members). This taxonomy placed
equal importance on the less observable indicators that make up cognitive engagement
(e.g., self-regulation, relevance of schoolwork to future endeavors, value ofleaming, and
personal goals and autonomy) and affective engagement (e.g., feelings of identification or
belonging, relationships with teachers and peers), which have been linked to improved
school performance. For example, it appears that a strong relationship exists between
cognitive engagement and personal goal orientation and investment in learning, which in
tum is positively associated with academic achievement (Greene & Miller, 1996; Greene
et aI., 2004; Miller, Greene, Montalvo, Ravindran, & Nichols, 1996; as cited in Appleton
et aI., 2006; Pokay & Blumenfeld, 1990). Likewise, adaptive school behaviors such as
participation and attendance have revealed strong associations with affective engagement
(Goodenow, 1993; as cited in Appleton et aI., 2006).
Furthermore, Appleton et aI. (2006) cited a growing body of literature that
demonstrated that indicators of cognitive and affective engagement are: (a) associated
with increased learning outcomes, (b) related to motivation, and (c) responsive to targeted
teaching strategies (Fredericks et aI., 2004; Marks, 2000; National Research Council &
Institute of Medicine, 2004; Reeve, Jang, Carrell, Jeon, & Barch, 2004). Appleton et aI.
(2006) posited that the understudied subtypes of cognitive and affective engagement
provided a more complete understanding of student engagement. These researchers cited
..------------------- --------
14
research in which students who felt more connected to staff, autonomously engaged in
positive school-related behaviors (Ryan, Stiller, & Lynch, 1994).
Appleton et al. (2006) argued that the limitations in measuring cognitive and
affective engagement have led to a dearth of research and difficulties in establishing
comparisons of the subtypes across studies. Historically, many studies have established
subtypes retroactively, which has arguably led to weak conceptual frameworks, according
to Appleton et al. (2006). Additionally, Appleton et al. (2006) criticized previous studies
that grouped subtypes together into an "amalgamation" (PA30) that failed to provide
clarity to the construct. The selection of participants in engagement studies was also
criticized.
According to Appleton et al. (2006), these limitations led to an overemphasis on
the more easily measured indicators of academic and behavioral engagement, which were
unable to measure a student's experience and the impact of the environment on student
engagement. Appleton et al. (2006) contended that the measurement of cognitive and
affective engagement from the student perspective would result in a more valid
understanding of student experience and meaning. Citing empirical studies on the validity
of self-report measures (Connell & Wellborn, 1991; Finn, 1989), Appleton et al. (2006)
concluded that the measurement of students' perspectives on their own experiences,
perceptions, and beliefs about education was necessary to understand the complete
construct of student engagement. The purpose ofthe work of Appleton et al. (2006) was
the development and validation of a psychometrically sound instrument designed to
measure students' cognitive and engagement from the student perspective.
15
Methodology. Appleton et al. (2006) conducted the first research study on the
Student Engagement Instrument (SEI) with 1,931 9th graders in a diverse, urban
Midwestern school district. The research division of the school district randomly selected
2,577 students from the 9th grade class (N = 3104), 75% (N = 1940) to complete the
survey.
The sample was 40.4% African American, 35.1 % White, 10.8% Asian, 10.3%
Hispanic, and 3.5% American Indian. Twenty-three percent of respondents came from
homes in which languages other than English were spoken, 61.4% were eligible for a free
or reduced priced lunch, and 7.6% were received special education services. Chi-square
tests to examine differences between respondents and non-respondents found that
significant differences existed between students selected to take the SEI and those
students that completed the survey only on the free or reduced lunch variable (V2=.156, p
< .001), which ostensibly serves as a proxy for SES and warrants attention.
The SEI was developed from reviews of literature on student engagement, and
was designed to measure the cognitive and affective subtypes of engagement. The SEI
was developed by documenting the varied conceptualizations of cognitive and affective
engagement present in the literature through a search of archival databases and selected
journal articles. A search of the terms (a) engagement, (b) belonging, (c) identification
with school, (d) self regulation, (e) academic engagement, (f) behavioral engagement, (g)
cognitive engagement, and (h) affective engagement identified several pre-existing
engagement scales. Using this work as a basis, probes and items were used to create a
preliminary scale. The SEI contained three subscales of cognitive engagement (i.e.,
Control and Relevance of School Work, Extrinsic Motivation, and Future Aspirations and
16
Goals) along with three subscales of affective engagement (i.e., Family Support for
Learning, Peer Support for Learning, and Teacher-Student Relationships).
An initial pilot of 31 8th graders examined the initial scale in two groups and
provided feedback that led to structural and semantic changes to the items and increased
the clarity and relevance of the instrument. The complete version of the SEI contained 30
items designed to measure student cognitive engagement and 26 items designed to
measure affective engagement. A four-point Likert scale was utilized (1 =strongly agree,
2=agree, 3=disagree, and 4=strongly disagree). Student demographics, including gender,
ethnicity, free or reduced lunch status, special education status, documented suspensions,
and scores on the Northwest Achievement Levels Test were obtained from the district.
Passive consent was used in the administration of the SEI to increase student
participation in the research and thus, provided greater variation of student perspective.
Reverse-keying of six items within the study to reduce acquiescence response and
monitoring of students were employed to dissuade careless responding. To avoid
confounds with reading ability deficiencies, the SEI was administered orally. Surveys
were inspected for missing data and any SEIs with five or more data points were
reviewed. Responses with more than one answer were coded as missing, and no evidence
of systematic responses for missing responses was found.
Results. Polychoric correlations were computed to define item associations.
Factor analyses were conducted using principal axis factoring to extract factors, utilizing
halfthe dataset. Confirmatory factor analyses (CFAs) were used to determine model fit
for plausible factor models with the other half of the dataset. The Chi-square test, the
Comparative Fit Index (CFI), the Tucker-Lewis Index (TLI), and the root mean square
17
error of approximation (R1\1SEA) were used to address limitations within each index and
to provide a comprehensive examination of the SEI.
Findings from exploratory methods revealed support for conceptualizations of
factors that underlie cognitive and affective engagement. Polychoric correlations of the
first half of the dataset and principal axis factoring were used to determine a model of
between four and six factors. Further exploratory factor analyses (EFAs) of the four, five,
and six-factor models were conducted with items loaded onto cognitive or affective
engagement subtypes. The six-factor model included items associated with extrinsic
motivation.
CFAs were conducted on the other half of the dataset, producing x2/df ratios for
the five- and six-factor models that indicated acceptable model fit (the four-factor model
did not fit the data) with all of the model fit indices (CFI, TLI, RMSEA, Chi-Square, x2/df
, and Dv2). Change statistics comparing the fit between the five- and six-factor models
revealed the importance of the sixth factor, with Chi-Square difference test results that
were significant (L1 x2 = .203.71, df= 60, p <.001).
The convergent and discriminant validity of the six-factor structure were
supported based on the relationships between the cognitive and affective factors. The
factors --Student-Teacher Relationships, Peer Support for Learning, Future Aspirations
and Goals, Family Support for Learning, and Extrinsic Motivation and school indicators--
evinced correlational relationships that were in the expected positive (e.g., GPA) and
negative (e.g., absences) directions.
Variance attributed to cognitive and affective subtypes was unique to each factor,
and positive inter-correlations suggested that each factor measured a different aspect of
18
cognitive and affective engagement. Findings substantiated previous research that
reported positive peer relationships were more affected by intrinsic, as opposed to
extrinsic, motivation. Most SEI factors exhibited positive relationships in expected
directions with school indicators. The control/relevance factor correlated negatively with
math/reading scores and GPA, which could have been a result of the perceived
irrelevance that students' arguably harbor about work that does not connect to long-term
goals, or the perceived lack of benefit of achievement on a state examination.
Implications. Appleton et al. (2006) suggested that future research needs to
obtain data from a cross-section of social and ethnic groups to aid in the generalizability
of the SEI. In addition, the authors noted that relation of cognitive and affective
engagement to academic and behavioral engagement required further examination to
reveal the degree of relationship among the constructs. Finally, they suggested that
further research was warranted to investigate whether specific (academic, behavioral) or
general (cognitive, psychological) forms of engagement are more responsible for
important outcomes. In addition, Appleton et al. (2006) recommended further study of
the interaction between the specific and general forms of engagement to assess the
interrelationship of the entire construct of student engagement.
Engagement as Flourishing: The Contribution of Positive Emotions and Coping to
Adolescents' Engagement at School and with Learning (Reschly et aI., 2008)
Rationale of the study. Building upon the work of Appleton et al. (2006) and the
initial validity study of the SEI, Reschly et al. (2008) utilized the SEI in a study of the
relationship of positive emotions and coping mechanisms to student engagement in
school. The authors argued that positive emotions were central to the process of
19
expanding human thought and behavior that led to greater adaptability and heightened
learning opportunities (Fredrickson, 2001; as cited in Reschly et al., 2008).
Frederickson's (1998, 2001) "broaden and build theory," which was the basis of this
study, posited that the accrual of frequent positive emotions builds emotional resources
that create an upward spiral toward well-being. Reschly et aI. (2008) hypothesized that
positive emotions produced future as well as present well-being, and enabled individuals
to build resources that strengthened responses to challenges. In contrast to the narrowing
of thought and behavior during frequent negative emotions, positive emotions were
linked to flourishing adaptive responses.
Examinations of the research literature provided support for the role of positive
emotions in broadening "attention, cognition, and behavior, as well as build[ing]
physical, intellectual, and social resources" (Reschly et aI., 2008, p. 19). Further evidence
linked positive emotions to success across life activities, such as marriage, work success,
friendship, and physical health (Lyubomirsky, King, & Diener, 2005). Reschly et aI.
(2008) hypothesized that the broadening of positive emotions and their effect on
resiliency were linked to greater coping resources (Frederickson, 2001; Frederickson &
Joiner, 2002). These positive emotions were purported to improve broad-minded coping,
which has been defined as the broadening of one's thinking and heightened ability to
cope with challenges due to exposure of positive emotions. Coping and positive
emotions were posited to interact reciprocally, spiraling toward well-being (Frederickson,
2001). In a similar process, student engagement and supportive learning contexts were
hypothesized to interact reciprocally, moving towards a cyclical increase in the
perception of support and student engagement.
20
The broaden and build theory had been utilized in research with adults, yet few
studies had examined adolescents in educational contexts. Gaps in the research of the
broaden and build theory were addressed by reviewing existing literature and focusing on
the measurement of affect, coping, and student engagement as they relate to educational
contexts. Reschly et al. (2008) tested the broaden and build theory by building
hypotheses from previous research that predicted that (a) affect would relate to student
engagement, (b) broad-minded coping would be positively linked to student engagement,
(c) positive but not negative affect would be linked to coping strategies, and (d) positive
affect and engagement would be mediated by broad-minded coping.
Methodology. A review of the research led Reschly et al. (2008) to include
demographic variables that had evinced strong relationships with the variables linked to
positive emotions and affective and cognitive engagement. Three demographic variables
were included in the study: grade level, gender, and socioeconomic status (SES). Citing
research on gender and engagement (Furrer & Skinner, 2003; Goodenow, 1993; Marks,
2000), Reschly et al. (2008) argued that the relationship between relatedness and
engagement is stronger for boys, while girls have been found to be more engaged
academically even when accounting for engagement variables. SES was chosen over
racial/ethnic group membership because previous research had found that engagement
variables were strong predictors within groups of students based on demographic/risk
factors (Finn & Rock, 1997).
Reschly et al. (2008) administered the three measures of positive emotions and
coping as part of a 3-year longitudinal study, which utilized active consent. The sample
was comprised of 293 students (178 females, 115 males) from grade 7 (N = 4), grade 8
21
(N = 110), grade 9 (N = 87), and grade 10 (N = 92) in the rural southeastern United
States. The sample was 47.8% Caucasian, 41.2% African American, 2.4% Asian/Indian,
1.0% Hispanic, and 5.2% Other. Nearly half of all students in the sample (47.8%)
indicated that they received free and reduced lunch.
Reschly et al.'s (2008) study utilized three measures: the (a) Positive and
Negative Affect Schedule-Children (PANAS-C), (b) the Self-Report Coping Scale
(SRCS), and (c) the Student Engagement Instrument (SEI). The PANAS-C is a 27-item
survey with 12 items on a Positive Affect (PA) subscale (frequency of emotions such as
excited, cheerful, energetic, etc.) and 15 items on an Negative Affect O\TA) subscale
(frequency of emotions such as sad, nervous, ashamed, etc.). Reschly et al. (2008) cited
prior studies that found both subscales exhibited strong internal consistency.
The SRCS is a 34-item scale that measures five coping strategies (Causey &
Dubow, 1992), although two subscales were used within this study: Seeking Social
Support (8 items) and Self-Re1iancelProblem Solving (8 items). The authors stated that
the SRCS had adequate internal consistency reliability (0.68 - 0.94), and adequate
reliability and validity based on the study of Causey and Dubow (1992). Causey and
Dubow (1992) found that these two subscales were correlated with positive coping
behaviors and strong self-esteem.
Results. Reschly et al. (2008) computed descriptive statistics and correlations,
and conducted regression analyses using the study variables. Descriptive analyses from
affect scores indicated that people reported more positive than negative emotions.
Problem-solving and social support coping strategies were both used moderately, while
22
mean scores for Future Aspirations and Goals and Family Support for Learning were
high.
Correlational analyses uncovered a positive relationship between gender and two
other subtypes--peer and family support for learning and seeking social support. A
negative association was found between grade level and positive affect (PA) and the
engagement subscales, although a positive correlation was discovered between PA and
the engagement subscales and PA and coping strategies. The two coping scales were
strongly correlated with the engagement measures.
The role of seeking social support in the mediation of positive affect and
engagement was investigated through regression analyses. Results indicated that seeking
social support exhibited significant meditational effects on teacher-student relationships,
control and relevance, peer support, and family support. This effect was further
demonstrated by the decrease in the correlations among these four engagement measures
and positive affect. Additionally, the role of problem solving in the mediation of
engagement was explored, and findings indicated meditational effects on Teacher-
Student Relationships, Control and Relevance of Schoolwork, Future Aspirations and
Goals, and Family Support for Learning.
Implications. Reschly et al. (2008) concluded that evidence from this study
supported the hypothesis that frequent positive emotions in school are related to
cognitive, behavioral, and coping strategies. These findings support the hypotheses of the
broaden and build theory (Fredrickson, 1998) that positive emotions exhibit strong
mediational relationships with student engagement and personal and environmental
resources. The authors stressed the importance of these findings in providing support for
23
the mediation of alterable student engagement variables that are crucial factors to
individual success. The malleability and responsiveness of positive emotions were cited
as a means to developing intervention strategies aimed at increasing positive emotions
that subsequently influence student engagement and coping strategies.
Research on the predictive qualities of positive emotions to future well-being was
cited as a logical progression of this study. Additionally, further inquiry into the complex
relationship between positive emotions and the two internal subtypes of engagement
(cognitive, psychological) as they relate to the two external, overt subtypes (academic,
behavioral) of engagement was suggested. Reschly et al. (2008) stressed the need to
investigate individual differences in personality, ability, and interests to develop
interventions that address unique personal and environmental resources.
The authors identified generalizability as a limitation of the study, with the SES
and race/ethnicity of the sample as being non-representative of student populations
nationwide. Furthermore, the decision to use active versus passive consent may have
been a possible source of engagement bias, as students that do not return surveys could be
those that are most disengaged with school. Finally, data were collected cross-sectionally,
which may not have provided sufficient support for the broaden and build theory and its
central focus on the development of positive emotions and greater personal and
environmental resources.
A Study of the Reliability and Construct Validity of the School Engagement Instrument
(SEI) across Multiple Grades (Betts et a!., 2009)
Betts et al. (2009) furthered research on the reliability and validity of the SEI that
Appleton et al. (2006) had initiated. Citing the potential of the instrument to measure the
24
alterable aspects of risks of dropping out and academic failure, the authors argued for
centralizing research on student engagement on the measurement of cognitive and
affective engagement. Yet, difficulties in the measurement of cognitive and affective
engagement and broad conceptualizations of these subtypes have resulted in disparate
approaches to the study of these traits of student engagement (Fredericks et aI., 2004; as
cited in Betts et aI., 2009).
Rationale of the study. To provide more evidence for the efficacy of the SE1,
Betts et aI. (2009) extended the maturational level of respondents and utilized a broader
sampling plan than Appleton et aI. (2006). Additionally, the study built upon the
construct validity for the SE1 by (a) confirming the factor structure, (b) estimating scale
reliabilities, and (c) measuring invariance of parameter estimates across grade levels.
Latent variable modeling was used in the investigation of the SEI. Participants (N
= 2,416) in this study ranged from grades 6-12, with the intention oftesting the internal
construct validity of the initial study with samples below and above grade 9. Two school
districts in South Carolina (N = 418) and Minnesota (N = 1,998) were sampled, with a
relatively equal number of males (N = 1,197) to females (N = 1,219). The study sample
was also 87% European American (N = 2,073), 9% African American (N = 220), 1%
Asian American (N = 33), 1% Hispanic American (N = 29), 1% Native American (N =
19), and 1 % Other (N = 27).
Methodology. Betts et al. (2009) did not utilize the factor of extrinsic motivation
from the six-factor SE1, as its two-item composition confounded measurement. The five-
factor model contained the following aspects of student engagement: Teacher-Student
Relationships (TSR), Control and Relevance of School Work (CRSW), Peer Support for
25
Learning (PSL), Future Aspirations and Goals (FG), and Family Support for Learning
(FSL). The five- and six-factor models exhibited the same fit. Reviews of student
engagement literature were used to create items utilizing student focus groups. The SEI
contained 19 items measuring affective engagement and 16 items surveying cognitive
engagement.
An analysis of the SEIs factorial structure was conducted. Exploratory factor
analyses (EFAs) were conducted to identify factor-specific items. A series of
Confirmatory Factor Analytic (CFA) models were then estimated. Model fit indices were
used to make model comparisons, and factors were chosen using two stringent statistical
criteria to address over- and under-extraction of factors-the Comparative Fit Index
(CFI) and the Root Mean Square Error of Approximation (RMSEA). The model with a
CFI of .95 or higher and an RMSEA at or below .05 was chosen, providing it contained
the lowest number of factors.
The indicator variables' patterns ofloadings on the latent factors were analyzed
using Hoyle and Duvall's (2004) methodology. The Wald test was used to place
restrictions on the model, and a baseline for significance was set between the
hypothesized value and parameter estimate. Item indicator salience of .30 was set as a
minimum level and items with significant loadings on factors below this minimum were
set to zero.
The reliability of the scales was computed utilizing the composite reliability for
congeneric measures model (CRCMM, Raykov, 1997; as cited in Betts et aI., 2009),
which estimated true reliability of sample scales. A multiple group confirmatory factor
analysis with nested models was used (Joreskog, 1971; Bollen, 1989; as cited in Betts et
26
aI., 2009). Finally, model fit statistics were used to measure evidence of invariance across
grades.
Results. Both four~ and five-factor models were found to be good fitting,
although only the five-factor model met CFI and RMSEA criteria. All models had
statistically significant Chi square tests. The similarity between indicator items and the
number of factor grades in model fit indices suggested a good model fit. Results from
covariance matrices and unstandardized factor loadings indicated measurement
invariance across the sample.
Betts et al. (2009) removed the sixth factor from the analysis. Reservations
regarding the fit of a six-factor model had been noted (Appleton et aI., 2006), as the
wording of the two items of the sixth factor that gauged extrinsic motivation were
reverse-coded, and measured the extent to which students were not extrinsically
motivated (Appleton et aI., 2008). Although these two items exhibited a high reliability
coefficient (Appleton et aI., 2006; Betts et aI., 2009), it appeared difficult to examine the
extent to which the coefficient was affected by the negative wording of the items.
Furthermore, there was a strong similarity of fit between the 5- and 6-factor models
(Betts et aI., 2009). Betts et aI. (2009) suggested that subsequent studies of the SEI utilize
a five-factor model representing cognitive and affective engagement.
Betts et aI. (2009) found that all factors were correlated (r = .45 - .79). The five-
factor model replicated the intended factor structure and indicators exhibited significant
relationships with intended factors. In contrast to previous research, a single item with
significant cross-loading was discovered, as well as a more complex error structure. Six
pairs of indicator variables contained correlated residuals, which provided evidence for
27
possible construct irrelevant factors. Betts et aI. (2009) suggested that correlated errors
may have been a product of this sample and stressed the need for further replication of
the study.
Implications. Findings indicated replication of the intended factor structure, and
provided data to suggest that the SEI measured both cognitive and affective factors across
maturational levels in grades 6-12 with similar reliability. The authors cited the potential
of the SEI to be used as an instrument to focus interventions and address the aspects of
cognitive and affective engagement that are ostensibly linked to dropout. Correlated
residuals were identified with six pairs of items (1 & 5; 10 & 12; 11& 13; 14 & 15; 22 &
24; 25 & 26), which suggested the possibility of "mini-factors" (Betts et aI., 2009, p. 14)
that may be more relevant items within each factor.
The authors cited limitations that need to be addressed in future studies of
cognitive and affective student engagement. Betts et aI. (2009) suggested that more work
needs to address equivalent measurement across demographic variables such as ethnicity
and special education. Furthermore, tests of the convergent and divergent validity of the
SEI with hypothesized constructs and relationships should be established. Further
research should also include an investigation of the item pool and the deletion of pairs
with correlated residuals from the scale along with the computation of scale reliabilities.
Research Questions
The purpose of this study was to build upon the work of Appleton et aI. (2006),
Reschly et aI. (2008), and Betts et aI. (2009) and examine the relationship among
cognitive and affective engagement and student academic and behavioral outcomes as
measured by the SEI with a cohort of 9th grade students. Specifically, this study
28
addressed the following questions: Controlling for demographic and risk variables, what
is the direction and strength of the relationship between measures of cognitive and
affective engagement gathered from the SEI self-report measure and (1) student
achievement outcomes as measured by 9th grade, first semester GPA, (2) number of
behavioral discipline referrals, and (3) credit completion?
Hypotheses
This study investigated the following hypotheses utilizing self-report data from
the SEI: (a) self-report data representing cognitive and affective engagement will have
strong positive relations with student engagement academic outcome variables (grade
point average and credit completion) measured at the end of the first semester for 9th
grade students; and (b) behavioral discipline referrals will have significant negative
association with scores on the SEI at the end of the first semester for 9th grade students.
..--------------
29
CHAPTER III
METHODOLOGY
Setting and Participants
The Student Engagement Instrument (SEI) was administered in a school district in
Oregon with a total population of 5,500 students. The district maintains one
comprehensive high school, which resides in a suburban community with a
predominately working class population. The ethnic makeup of the district is (a) 75 %
Caucasian, (b) 14 % Hispanic, (c) 3 % Native American, (d) 3% Asian, (e) 3 % African
American, (f) <1 % Pacific Islander, and (g) 2% declined. The percentage of students on
free and reduced lunch (FRL) status is reported as 34% district-wide. English language
learners make up 3% (N = 180) of the student population, while 19% (N = 1092) of
students receive special education services in the district.
The survey was administered to all 9th grade students (N = 413) at the only
comprehensive high school in the school district. Ninety percent of the 9th grade students
completed the survey (N = 371). The remaining students had either moved away or were
absent on the day of administration. The sample was split almost evenly between male (N
= 174) and female (N =197). The sample was (a) 80% Caucasian (N= 295), (b) 14%
Hispanic (N= 52), (c) 3% American Indian/Alaskan Native (N = 10), (d) 2% African
American (N = 7), (e) 2% Asian (N = 6), and (f) less than 1% Native HawaiianlPacific
Islander (N = 1). The percentage of students identified as receiving free or reduced lunch
(FRL) was 45% (N = 168). Students receiving English language development services
30
made up 1% (N = 4) of the sample, and 13% (N = 48) of the sample received special
services.
Measure. The SEI (Appleton et aI., 2006) is a scale that measures cognitive and
affective engagement with school. It is comprised of 35 items and is administered orally
through a self-report survey. Respondents provide degree of agreement on a four-point
scale, from Strongly Disagree to Strongly Agree. The SEI measures five subscales of
student engagement that represent cognitive (Control and Relevance of School Work
(CRSW), Future Aspirations and Goals (FG» and affective (Teacher-Student
Relationships (TSR), Peer Support for Learning (PSL), Family Support for Learning
(FSL» engagement.
Findings from previous validity studies (Appleton et aI., 2006; Betts et aI., 2009)
indicated strong model fit for all items, with similar reliability in measuring cognitive and
affective factors across maturational levels. The internal consistency of the SEI factors
has ranged from 0.72 to 0.88, and correlations in expected directions with demographic
and academic variables support the instrument's validity (Appleton et aI., 2006).
Predictor variables. Both demographic and district outcome data were utilized in
the study. Demographic variables included Free-Reduced Lunch (FRL) status, gender,
special education status, English language development status, and ethnicity. Each of the
student demographic variables was dummy coded utilizing SPSS 16.0. The FRL variable
ostensibly serves as a proxy for socioeconomic status. Ethnicity was dummy coded into a
dichotomous variable with "white" coded as 0, and all other ethnic identifications coded
as 1. Dummy codes were also used to identify free and reduced lunch recipients, special
education students, and English language learners. These variables served as control
31
variables in the analyses. English language development status was dropped due to lack
of variance in the sample.
Criterion measures (district variables). District outcome data included the
following variables from the first semester of grade 9: (a) grade point average (GPA), (b)
credit completion, (c) documented minor behavioral referrals, and (d) documented major
behavioral referrals. Data were accessed through archival databases maintained by the
Bethel School District, and student identifying information was removed to maintain
confidentiality. Credit completion was a variable derived to track individual student
progress towards graduation during the first 9th grade semester. In SPSS, a dichotomous
dummy variable was computed, utilizing the two variables provided in the district
dataset: "total credits attempted" and "total credits earned." Students with at least one
failure were coded as 0, and students whose credits attempted matched credits earned
were coded as 1. This variable provided data on how many students had failed at least
one class in the first semester. Both credit completion and minor and major behavioral
referrals were categorized as a result of the skewed distribution of the sample.
Behavioral referrals were identified as either minor or major referrals. Minor
referrals included all behavioral infractions that did not result in a suspension, lunch
detention, or Tuesday School, which is an alternative to suspension. Major referrals
included those behavioral infractions that resulted in a suspension, lunch detention, or
Tuesday School (alternative to suspension). Referral data was limited to the first semester
of the 2009-2010 academic year, and did not include district K-8 behavioral data. Both
minor and major behavioral referrals were dummy coded into dichotomous variables.
32
Students with no referrals were coded as 0, and students with one or more referrals were
coded as 1.
SEI Administration Procedures
The school district's instruction department oversaw the administration and
scoring of the SEI. The survey was group administered in classroom settings. Students
marked their responses on a scanable bubble sheet. Social studies instructors were
selected to administer the SEI in their classes, as all 9th grade students take the same
social studies class. Students were given 45 minutes to complete the survey, and those
students that were absent were not tested. The SEI was also administered orally to control
for variation in reading ability among the students.
As noted in Reschly et al.'s (2008) study utilizing the SEI, active consent may
create an engagement bias by excluding students from the survey that do not exhibit the
engagement with school to tum in consent forms. Therefore, the administration of the
survey occurred at regular class meeting times and was not announced to students prior to
the assessment date. Students were instructed that the survey was voluntary and were
given the opportunity to decline participation in the survey.
A script providing survey instructions was used in the administration of the
survey and was read to students in each classroom. The purpose of the script was to
instruct students to answer each item response and to respond honestly about each item.
Students were instructed that survey data were anonymous and that they would not be
identified by name. Random numbers generated by SPSS were assigned to each student.
Surveys had a sheet of paper affixed to them with the student's name on it. Students tore
33
off and recycled this identification sheet, and were left with a survey that had a random
generated number that identified them.
Missing Data
Median substitution was used to replace missing values on specific items that students
failed to mark. As a result of the small proportion of students that provided missing data
on the survey, data estimation techniques such a single/multiple imputation (Schafer &
Graham, 2002) and listwise deletion were not utilized. In the entire survey
administration, 30 item responses were not marked out of a total of 12,985 item
responses, which equates to .2% of the responses collected. In SPSS, a frequency
distribution was computed on each item. Item 31 had the highest number of missing
values, with 4 missing values out of371 total values on the item (1%).
Median substitution was performed using SPSS to replace the missing values of
each variable with the median of all values in that variable. A new variable was computed
for each of the 35 items on the SEI, and all missing values were replaced with the series
median of valid surrounding values, rather than the series mean. The series median was
chosen as a result of the ordinal Likert items on the SEI and the skewed distribution of
the data. The span of nearby points was set to "all."
Data Analysis
Exploratory factor analysis. An exploratory factor analysis (EFA) was used to
examine the shared variance among items and explore the underlying factor structure of
the SEI. The EFA was estimated using principal axis factoring with a promax oblique
rotation. An oblique rotation was used in the estimation as it was expected that the
hypothesized dimensions or factors describing the structure would be intercorrelated.
34
Items that loaded less than .275 and items that cross loaded with other items were
removed from the analysis. The level of .275 was selected as a minimum level for item
indicator salience. To test the hypothesis of a five-factor model of the SEI (Appleton et
al., 2006; Betts et al., 2009; Reschly et al., 2008), the EFA was restricted to five factors.
Supplemental analyses. Reliability coefficients were computed on each factor to
evaluate the consistency of student responses on the cognitive and affective engagement
scales. Correlation coefficients were also computed to index the bivariate relationships
between variables.
Regression analyses. Sequential logistic and multiple regression analyses were
used to assess the relations between measures of cognitive and affective engagement on
the SEI and student academic and behavioral outcome measures. A sequential multiple
regression analysis was used to examine relationships with GPA, the one continuous
outcome variable in the study. Sequential logistic regression analyses were used to
examine relationships with the three dichotomous outcome variables (credit completion,
minor behavioral referral, and major behavior referral).
In the first step of each analysis, all five control variables (FRL status, gender,
special education status, English language development status, and ethnicity) were added
to the model to control for the effects of demographic and risk factor variables. The
English language development status variable was later removed from the analysis as a
result of the absence of English language learners on the day of administration. In the
second step of the analysis, cognitive and affective engagement factors, as measured by
the SEI, were added to determine if student SEI responses independently added to the
explanation of academic achievement and behavioral outcomes. These factors included
35
Teacher-Student Relationships (TSR), Control and Relevance of Schoolwork (CIR), Peer
Support at School (PEER), Future Aspirations and Goals (ASP), and Family Support for
Learning (Family). A separate regression analysis was performed for each outcome
variable. The percent of variance associated with cognitive and affective engagement
factors was used to represent the unique relationship between the set of predictors and
each outcome variable.
36
CHAPTER IV
RESULTS
Descriptive Statistics
Table 4.1 presents descriptive statistics on each of the outcome variables. For the
dichotomous outcomes, the mean represents the proportion of the sample with a
particular type of referral or a class failure. Overall, 66% of the sample passed all of their
classes. However, 15% of the sample received a minor behavioral referral while 13%
received a major behavioral referral. The mean GPA was 2.609, indicating an average
grade ofB-. However, GPAs ranged from 0 to 4.0 with a standard deviation of 1.169.
Table 4.1
Descriptive Statistics for the Sample of9th Grade Students (N = 371)
GPA
Credit Completion!
Minor Behavioral Referrals!
Major Behavioral Referrals!
Mean
2.609
0.666
0.150
0.130
Standard Deviation
1.169
!For the dichotomous outcomes, the mean represents the proportion of the sample with
referrals or failures (e.g. 13% of the sample had a major behavioral referral).
Exploratory Factor Analysis
The exploratory factor analysis was performed using SPSS 16 for Windows.
Eigenvalues, scree plots, and factor loading patterns were used to determine the number
37
of factors to retain. Using Kaiser's rule, the analysis extracted eight factors accounting
for 49.6% of the variance ofthe 35 items. Post-extraction item communalities were
moderate to high, ranging from .295 to .799. Item 10 showed the lowest communality (h 2
= .230). Because the item fit well in a later interpretation, however, it was retained in the
analysis.
Inspection of the pattern matrix revealed moderate to high loadings for items on
their respective factors. The pattern of factor loadings suggested that six of the eight
extracted factors were uniquely defined. However, the sixth factor, labeled Extrinsic
Motivation in previous studies of the SEI and associated items, was not retained. Both
items 18 and 32 were reverse-coded, with students responding to the extent that they
were not engaged. Previous studies revealed difficulty in determining the extent to which
the coefficient was affected by the negative wording of these items (Appleton et aI.,
2006; Betts et aI., 2009). In addition, there is a strong similarity of fit between the five-
and six-factor models (Betts et aI., 2009). Furthermore, items 18 and 32, which
represented Extrinsic Motivation, are not subsumed under other factors (Appleton,
February 10,2010: personal correspondence). Therefore, the EFA was rerun with items
18 and 32 excluded.
Five factors were extracted from the EFA after the removal of items 18 and 32.
Consistent with previous studies of the SEI (Appleton et aI., 2006; Betts et aI., 2009), the
identified factors were labeled as follows: 1) Teacher-Student Relationships (TSR), 2)
Future Aspirations and Goals (ASP), 3) Peer Support at School (PEER), 4) Family
Support for Learning (Family), and 5) Control and Relevance of Schoolwork (C/R).
Visual inspection of the scree plot also confirmed that a five-factor model was
38
appropriate. Figure 2 illustrates the scree plot elbow, which conforms to results predicted
using eigenvalues of a five-factor model.
10
8 \
g" G
.e
~;;-
::
':'i
ell
....
\.>.l
4-
o
:~~o 0 0 00 0 o () 0 0 --B--+)
Factor NnmlHll'
Figure 2. Scree plot of Student Engagement Instrument exploratory factor analysis
loading onto five factors
The first factor, Teacher-Student Relationships, contained 8 items (3, 5, 10, 13,
16,21,22,31) and accounted for 27.84% of the variance. Items loading on this factor
were representative of traits of affective engagement and a student's psychological
connection to teachers. The second factor, Future Aspirations and Goals, contained 5
items (8, 11, 17, 19,30), and accounted for 7.06% of the variance. These items
represented cognitive engagement. The third factor, Peer Support at School, contained 6
items (4, 6, 7, 14,23,24) and accounted for 5.85% of the variance. These items
39
represented affective engagement. The fourth factor, Family Support for Learning,
represented affective engagement. This factor contained 4 items (1, 12,20,29) and
accounted for 3.50% of the variance. The fifth factor, Control and Relevance of
Schoolwork, contained 6 items (15,25,26,33,34,35) and accounted for 1.94% of the
variance. These six items represented cognitive engagement.
Contrary to previous studies, however, it should be noted that three items cross
loaded on two or more factors. Items 27 and 28 loaded onto both Teacher-Student
Relationships (.295 and .211, respectively) and Peer Support at School (.390 and .217,
respectively). Item 9 loaded onto both Future Aspirations and Goals and Control and
Relevance of Schoolwork (.270 and .291, respectively). In addition, item 2 did not load
onto any of the factors, with coefficients ranging from .170 to .226 across Teacher-
Student Relationships, Future Aspirations and Goals, and Control and Relevance of
Schoolwork. Thus, items 27, 28, 9, and 2 were not retained in the final model.
In Table 4.2, the sorted factor loadings associated with the final model are
presented. As can be seen in the Table, all retained items loaded uniquely onto the five-
factor model. However, standardized factor loadings varied. Family Support for Learning
had the strongest loadings (> .573) but only contained four items. Control and Relevance
of Schoolwork had the weakest loadings (> .285). Item 17, which loaded onto Future
Aspirations and Goals, was the strongest individual item loading (.968). Table 4.2 also
presents Cronbach's alpha values for each of the factors. The size of the coefficients
indicates the reliability of scores on each of the SEI factors was adequate.
Table 4.2
Item Loadings on SEI Factors
40
Item *TSR ASP Peer Family C/R
**(AE) (CE) (AE) (AE) (CE)
Item Description
5 .776 -.036 .045 -.110 -.084 Adults at my school listen to the
students.
13 .645 -.074 .078 .031 .042 Most teachers at my school are
interested in me as a person, not
just as a student.
21 .634 .079 -.050 .120 -.069 Overall, adults at my school
treat students fairly.
31 .632 -.167 -.037 .054 .253 At my school, teachers care
about students.
16 .586 .007 -.056 .039 .074 Overall, my teachers are open and
honest with me.
3 .570 .107 -.034 .058 -.081 My teachers are there for me
when I need them.
22 .510 .041 -.108 .005 .295 I enjoy talking to the teachers
here.
10 .378 .151 -.004 -.016 .104 The school rules are fair.
17 .060 .968 .069 -.078 -.205 I plan to continue my education
following high school.
--~-~-.
Table 4.2 (continued)
41
Item *TSR ASP Peer Family C/R
**(AE) (CE) (AE) (AE) (CE)
Item Description
11 .202 .726 -.016 .025 -.104 Going to school after high school
is important.
19 -.056 .637 -.038 -.017 .223 School is important for achieving
my future goals.
30 -.173 . .536 .021 .093 .174 I am hopeful about my future.
8 -.044 .531 -.004 .070 .201 My education will create many
future opportunities for me.
6 .036 -.108 .778 .000 .109 Other students here care about
me.
4 -.086 .040 .763 -.033 .042 Other students here like me the
way I am.
7 .037 .029 .662 -.009 .082 Students at my school are there
for me when I need them.
23 -.040 .059 .630 .064 -.038 I enjoy talking to the students
here.
14 .235 -.033 .595 -.031 -.031 Students here respect what I have
to say.
24 -.180 .064 .477 .086 -.042 I have some friends at school.
Table 4.2 (continued)
42
Item *TSR ASP Peer Family C/R
**(AE) (CE) (AE) (AE) (CE)
Item Description
20 .075 .017 .041 .789 -.082 When I have problems at school,
my familyIguardian(s) are willing
to help me.
1 .044 .024 .081 .713 -.143 My family/guardian(s) are there
for me when I need them.
29 -.038 -.007 -.053 .701 .099 My family/guardian(s) want me to
keep trying when things are tough
at school.
12 .053 -.012 .016 .573 .085 When something good happens at
school, my family/guardian(s)
want to know about it.
26 .072 -.078 .065 -.009 .519 The tests in my classes do a good
job of measuring what I'm able to
do.
33 .159 .143 .061 -.159 .498 Learning is fun because I get
better at something.
35 .114 .120 -.026 .042 .469 The grades in my classes do a
good job of measuring what I'm
able to do.
43
Table 4.2 (continued)
Item *TSR ASP
**(AE) (CE)
Peer
(AE)
Family C/R
(AE) (CE)
Item Description
34 .055 .307 -.005 -.104 .452 What I'm learning in my classes
will be important in my future.
25 -.135 .208 -.015 .226 .398 When I do well in school it's
because I work hard.
15 .052 .096 .058 .116 .285 When I do schoolwork I check to
see whether I understand what I'm
doing.
Variance 27.84 7.06 5.85 3.50 1.94
Explained
Cronbach's a .846 .844 .827 .802 .742
Post-Rotation 5.70 5.51 4.07 5.05 5.42
Eigenvalues
*TSR: Teacher-Student Relationships **CE: Cognitive Engagement
ASP: Future Aspirations and Goals AE: Affective Engagement
Peer: Peer Support at School
Family: Family Support for Learning
C/R: Control and Relevance of Schoolwork
Correlational analysis. Correlations among the five factors of the SEI and the
district outcome variables are presented in Table 4.3. As expected, both credit completion
and GPA correlated in expected directions across all five SEI factors, with credit
completion ranging from .037 to .368, and GPA from .029 to .412. These findings are
consistent with the hypothesis that cognitive and affective engagement are positively
44
related to academic outcomes. Future Aspirations and Goals had the strongest
correlations with credit completion and GPA, with coefficients of .368 and .412,
respectively. Peer Support at School had the weakest correlation with credit completion
and GPA, with coefficients of .037 and .029, respectively. Alternatively, minor and major
behavioral referrals correlated negatively across the five factors, except for two positive
associations between minor behavioral referrals and Peer Support at School and Family
Support for Learning (.033 and .015, respectively). These negative correlations were not
statistically significant, ranging from -.075 to -.248.
The five SEI factors were all positively related, with correlations ranging from
.302 to .625. These correlations suggested a moderate to strong relationship between
factors.
45
Table 4.3
Pearson Correlation Matrix between Factors and Outcomes
Major
Factors
TSR C/R Peer Asp Family Credit GPA Minor
(AE) (CE) (AE) (CE) (AE)
Teacher-Student
Relationships (AE)
Control/Relevance (CE) .593
Peer Support (AE) .326 .343
Aspirations (CE) .448 .625 .302
Family Support (AE) .443 .480 .386 .486
Credit .175 .158 .037 .368 .154
GPA .213 .227 .029 .412 .219 .818
Minor -.112 -.102 .033 -.105 .015 -.100 -.117
Major -.154 -.142 -.075 -.248 -.094 -.255 -.338 .286
Regression Analyses
Sequential logistic regression analyses were performed on each dichotomous
outcome variable (credit completion, minor behavioral referral, and major behavioral
referral). A sequential multiple regression was performed on GPA, a continuous outcome
variable. These analyses were used to estimate the unique relationship among cognitive
and affective engagement and student achievement and behavioral data. The regression
analyses addressed the three research questions:
46
1. Controlling for demographic and risk variables, what is the direction and strength
of the relationship between measures of cognitive and affective engagement gathered
from the SEI self-report measure and student achievement outcomes as measured by 9th
grade, first semester GPA?
2. Controlling for demographic and risk variables, what is the direction and strength
of the relationship between self-report data on cognitive and affective engagement as
obtained from the SEI and number of behavioral discipline referrals?
3. Controlling for demographic and risk variables, what is the direction and strength
of the relationship between self-report data on cognitive and affective engagement as
obtained from the SEI and credit completion?
Grade point average. Table 4.4 presents the results of the sequential multiple
regression analysis. After controlling for demographic and risk factors (ethnicity, special
education status, gender, and FRL status) in Step 1, SEI factors were inputted in Step 2.
The control variables accounted for 16% ofthe variance in GPA (R2 = .16,p < .001). The
Free and Reduced Lunch (FRL) variable (b = -.83,p < .001) had the strongest negative
relationship with GPA, while ethnicity (b = -.04, p = .354) showed the weakest relation.
The five engagement factors accounted for an additional 13% of the variance in GPA (R2
= .29,p < .001) and was statistically significant. Future Aspirations and Goals were
uniquely associated with GPA (b = .15, p < .001). Peer Support at School was also
statistically related to GPA (b = -.05, p < .05) but in a negative direction. Family Support
for Learning (b = .05,p = .124), Teacher-Student Relationships (b = .03,p = 138), and
Control and Relevance of Schoolwork (b = -.01, p = .893) were not statistically related to
GPA.
47
Table 4.4
Regression ofGPA on Covariates and SEI Factors
Modell Model2
Predictors Std. Std.b Error
p p b Error p p
Gender .26 ,II .11 ,021 .25 ,10 .11 ,015
FRL Status -,83 .12 -.35 ,000 -,64 ,11 -,27 ,000
Ethnicity -.05 .14 -,02 .724 -,IS ,13 -.05 .251
Special Education Status
-.43 .17 -.12 ,011 -.46 .16 -,13 ,003
Teacher-Student
,03 .02 ,09 .138
Relationships (AE)
Control/Relevance (CE) -.01 ,03 -,01 ,893
Peer Support (AE) -.05 .02 -,13 ,010
Aspirations (CE) .15 ,03 .32 ,000
Family Support (AE) .05 .03 .09 .124
Credit completion. Table 4.5 presents the results of the logistic regression
analysis for credit completion. Odds ratios were computed to estimate the increase in the
odds of credit completion associated with a one unit increase in each predictor variable.
Modell consisted of the covariates inputted as controls (ethnicity, special education
status, gender, and FRL status), The covariates were statistically significant predictors of
credit completion. The strongest predictor was FRL status, with students classified as
receiving free and reduced lunch having 67% less likely odds of attaining full credit than
48
those without FRL status. Students with special education status had 55% lower odds of
attaining full credit. Females had 39% higher odds of attaining full credit than males.
Ethnic identified non-white students were 21 % less likely to attain full credit than
students identified as white.
Future Aspirations and Goals was the only statistically significant factor in the
logistic regression on credit completion (b = .36,p < .001). Each unit increase in Future
Aspirations and Goals was associated with an increase of 44% in the odds of attaining
full credit.
Table 4.5
Regression ofCredit Completion on Covariates and SEI Factors
Predictors
Gender
FRL Status
Ethnicity
Special Education Status
Teacher-Student
Relationships (AE)
Control/Relevance (CE)
Peer Support (AE)
Aspirations (CE)
Family Support (AE)
Modell Model 2
b SEb p Exp(B) b SEb p Exp(B)
.33 .23 .162 1.386 .39 .26 .127 1.472
-1.10 .25 .000 .329 -.81 .27 .003 .445
-.23 .28 .411 .793 -.43 .30 .159 .654
-.81 .33 .014 .445 -.93 .36 .009 .396
.08 .05 .106 1.078
-.09 .06 .163 .917
-.08 .05 .129 .928
.36 .07 .000 1.437
.03 .07 .695 1.029
.--------~------
49
Minor behavioral referrals. Table 4.6 presents the results associated with the
minor referrals logistic regression analysis, None of the demographic covariates that were
entered on Step 1 were statistically significant. The SEI factors were inputted in Step 2 of
the analysis. Findings indicated that none of the engagement factors were statistically
associated with minor behavioral referrals.
Table 4,6
Regression ofMinor Behavioral Referrals on Covariates and SEI Factors
Predictors
Gender
FRL Status
Ethnicity
Special Education Status
Teacher-Student
Relationships (AE)
ControllRelevance (CE)
Peer Support (AE)
Aspirations (CE)
Family Support (AE)
Modell Model 2
b SEb P Exp(B) b SEb P Exp(B)
.08 .30 .795 1,079 .10 .30 ,745 1.103
.26 ,30 .384 1.301 .16 ,32 ,626 1.169
-.23 .39 ,543 ,791 -.19 .39 .633 ,828
-.30 .47 ,529 .745 -,29 .48 .633 ,749
-,09 ,05 .091 .915
-,05 ,07 .475 .950
.08 .06 .178 1,083
-,11 ,08 .177 .896
.16 .10 .084 1.178
Major behavioral referrals. Results associated with the logistic regression of
major behavioral referrals on student risk factors and SEI factors are presented in Table
4.7. In Step 1, covariates were again entered as controls. Gender (b =-.76,p < .05) and
------------------------
50
FRL status (b =.82, P < .05) were statistically significant. Males were 53% more likely to
receive a major behavioral referral than females. Students with FRL status were two and
a halftimes more likely than non-FRL students to receive a major referral. Non-whites
were also 41 % less likely to receive a major behavioral referral while special education
students were 88% more likely to be referred for a major behavioral infraction.
Findings in Step 2 of the analysis were similar to the other regression analyses.
Future Aspirations and Goals was the only factor statistically related to the outcome (b =-
.24,p < .01). The odds of receiving a major behavioral referral were 21 % lower with each
unit increase in this factor. Teacher-Student Relationships was predictive of major
behavioral referrals, decreasing a student's likelihood of receiving a referral by 10%, but
was not statistically significant (p > .05).
------------------- --------
51
Table 4.7
Regression ofMajor Behavioral Referrals on Covariates and SEI Factors
Modell Model 2
Predictors
b SEb p Exp(B) b SEb p Exp(B)
Gender -.76 .33 .020 .469 -.83 .34 .016 .438
FRL Status .82 .35 .018 2.274 .54 .37 .149 1.707
Ethnicity -.54 .43 .204 .583 -.38 .44 .382 .681
Special Education
.63 .40 .114 1.878 .74 .42 .078 2.093Status
Teacher-Student
-.11 .06 .072 .896Relationships (AE)
Control/Relevance .06 .08 .471 1.061
(CE)
Peer Support (AE) .01 .06 .831 1.013
Aspirations (CE) -.24 .09 .005 .787
Family Support (AE) .05 .10 .615 1.049
52
CHAPTER V
DISCUSSION
The purpose of the study was to investigate the factor structure of the SEI and to
examine the relationship among affective, cognitive, behavioral, and academic factors of
student engagement of 9th grade students. Using factor analytic procedures, this study
appears to add support to previous studies of the SEI (Appleton et aI., 2006; Betts et al.,
2009; Reschly et aI., 2008) that demonstrated good model fit and internal consistency of
the five-factor model. Results from the logistic regression analyses supported conceptions
of cognitive and affective engagement as important mediators in a student's academic
achievement (Christenson et aI., 2008). Overall, this study confirmed the importance of
measuring cognitive and affective engagement and the impact these student values and
beliefs have on educational outcomes (Appleton et aI., 2008).
This study examined the hypothesis that cognitive and affective engagement have
strong positive relations with academic engagement variables and significant negative
association with behavioral engagement variables. Results generally supported the
direction of the associations between cognitive and affective engagement and academic
and behavioral engagement variables, although specific factor relationships within the
cognitive and affective subtypes did not always follow the hypothesized direction.
Regression analyses revealed unique positive relationships between cognitive
engagement and the factor Future Aspirations and Goals in predicting academic and
behavioral engagement outcomes.
53
Overall, the predictability of cognitive and affective engagement was not as
strong as expected. Although previous studies have cited the importance of examining the
multidimensionality ofthe engagement construct (Appleton et aI., 2008; Fredericks et aI.,
2004; Glanville & Wildhagen, 2007; Reschly et aI., 2008), findings from this study
suggest that the student engagement typology should address the specific strengths of
each factor and its effect on the relationship of the four subtypes. In particular, the
importance of cognitive engagement relative to affective engagement in predicting GPA,
credit completion, and minor and major behavioral referrals supports the conception of
the engagement construct as being comprised of factors with unequal degrees of
importance.
Discussion of Analytical Results
Exploratory factory analysis. Previous SEI studies (Appleton et aI., 2006; Betts
et aI., 2009; Reschly et aI., 2008) estimated a five-factor model of cognitive and affective
engagement consisting of 33 items. Results of the current study also support a five-factor
model. However, the current study of 9th grade students at a comprehensive high school
(N = 371) provided the best model fit with a five-factor structure that consisted of only
29 items. Although each SEI item retained matched factor item groupings aligned with
published factor keys (Appleton, February 3,2010, personal correspondence), the cross
loading of items 9 (most of what is important to know you learn in school), 27 (I feel safe
at school), and 28 (I feel like I have a say about what happens to me at school) may point
to potential overlap in factor definitions. Item 2 (after finishing my schoolwork I check it
over to see ifit's correct) was also not related to any factor. Future research with a
54
broader cross-section of students will likely be necessary to determine whether this item
should be retained.
Research on the measurement of cognitive and affective engagement is in a
nascent stage. The EFA results highlight the need to further examine the SEI factor
structure. Cross-loadings suggest that some items may be representative of either
cognitive or affective engagement as a whole rather than a single factor under each of the
subtypes. Furthermore, three of the four items that were dropped (9, 27, 28) were
identified under the Control and Relevance of Schoolwork factor in an SEI identification
key that identifies the five factors (Appleton, February 3, 2010, personal
correspondence). The weak predictive power of the Control and Relevance of
Schoolwork factor in this study suggests the need for refinement of the items and further
research on the factor's identification as representative of cognitive engagement.
Logistic and multiple regression analyses. Despite some contrasts with previous
studies, findings from the regression analyses generally supported the notion that
cognitive engagement is a more relevant predictor of student achievement and behavior
than affective engagement. Similar to previous research of cognitive engagement using
the SEI (Appleton et aI., 2006; Betts et aI., 2009; Reschly et aI., 2008), Future Aspirations
and Goals was a unique predictor of academic success. In the two regressions of GPA
and credit completion, this factor showed the strongest unique relationship with the
outcome. Future Aspirations and Goals was also a unique negative predictor of the major
behavioral referral variable. These findings highlight the importance of Future
Aspirations and Goals in understanding the overall engagement construct and the unique
contribution of this factor to the subtype of cognitive engagement.
r--------------~------------------
55
In addition, the predictive strength of the Future Aspirations and Goals factor in
contrast to the other four factors suggests that the five factors that represent cognitive and
affective subtypes may not be equally important predictors of student engagement. This
finding counters findings in the extant literature. Researchers have posited that all of
these factors are important components in understanding the values and beliefs that
students bring to the educational environment (Appleton et aI., 2008; Archambault et aI.,
2009; Reschly et aI., 2008). However, it may be that some factors are more relevant in
understanding cognitive and affective engagement and their relationship with educational
outcomes. Findings suggest that cognitive engagement is a more significant subtype in
predicting academic and behavioral engagement than affective engagement.
Predictive Factor Strengths
Previous studies have linked all five factors of cognitive and affective
engagement to educational outcomes and personal well-being (Appleton et aI., 2006;
Betts et aI., 2009; Reschly et aI., 2008). However, the substantial variation in the
predictive strength of each factor of cognitive and affective engagement across the four
outcome variables in this study underscores the importance of improving the assessment
of cognitive and affective engagement. Only one factor, Future Aspirations and Goals,
was statistically significant in predicting students' academic and behavioral engagement.
The findings of this study suggest that future development of the SEI should focus on
replication of the initial validation study (Appleton et aI., 2006). Furthermore, research
needs to determine if the Future Aspirations and Goals factor is the most significant
predictor of educational outcomes.
-------------------
56
Results from this study counter previous SEI findings on the strength of specific
factors and their associations with student outcomes. Family Support for Learning was a
weak predictor in all of the regression analyses and did not relate to outcomes to the
extent measured in other studies (Betts et aI., 2009; Reschly et aI., 2008). Furthermore,
the exploratory factor analysis only identified four items (l, 12, 20, 29) for this factor. In
contrast to findings of the current study, Reschly et aI.' s (2008) study of student
engagement with a sample of 293 students in grades 7 to 10 found that Family Support
for Learning was statistically related to all three engagement subscales in that study.
However, those engagement subscales represented social support and did not include
academic outcome measures. The current study suggests that the Family Support for
Learning factor may not be predictive of academic achievement indicators, although it
may be more significant in measuring a student's well being and sense of belonging
(Reschly et aI., 2008).
The Control and Relevance of Schoolwork factor was not a significant predictor
of academic achievement in any of the regression analyses, which countered the findings
of previous studies (Appleton et aI., 2006; Betts et aI., 2009). Furthermore, Appleton et
aI. (2006) identified 9 items on the SEI that represent the Control and Relevance of
Schoolwork factor. However, three of those items (2, 9, 27) did not fit the five-factor
model of cognitive and affective engagement in this study's EFA. Correlations were in
the expected directions, with positive associations with GPA and credit completion, and
negative associations with minor and major behavioral referrals. The absence of
significant relationships between the Control and Relevance of Schoolwork factor and
academic and behavioral engagement variables suggests that findings from previous
57
studies (Appleton et aI., 2006; Betts et aI., 2009) may not be generalizable across
demographic groups. Future study should test the link between this factor and academic
engagement to determine its significance in the cognitive engagement subtype. It may be
that the Control and Relevance of Schoolwork factor is more predictive of social well
being than academic outcomes (Reschly et aI., 2008).
Much research supports the relationship between students' peer influences and
academic outcomes (Christenson et aI., 2008; Finn, 1989; Osterman, 1998). Although
Appleton et aI. (2006) found that Peer Support at School was a significant factor in
student academic and behavioral outcomes, findings from this study do not support
previous work on the significance of peer influence on student engagement (Appleton et
aI., 2008; Fredericks et aI., 2003). Contrary to expectations, Peer Support at School had a
negative relationship with GPA and credit completion.
The explanations for the contrast between this study and previous research on the
predictive significance of peer support at school are not clear. The context of the
classroom administration of the SEI may have confounded responses to Peer Support at
School factor items and dissuaded students from making truthful responses. The negative
association between the Peer Support at School factor and GPA may be the result of
disengaged students and their hesitation to admit an absence of friendship support while
sitting among their peers. Administering this survey to students in private, away from
peers, may have led to responses that better reflected student thought and aligned with the
body of research on student engagement that identifies peer support as a powerful
predictor of student academic success (Finn, 1989; Osterman, 1998).
58
The relative weakness of the Teacher-Student Relationships factor in predicting
student outcomes was unexpected, especially on GPA and credit completion. Substantial
research cites the development of supportive relationships between teachers and students
as predictive of both student academic and behavioral outcomes (Klem & Connell, 2004;
Ladd et aI., 1999; Reeve et aI., 2004). Although the Teacher-Student Relationships factor
was a stronger predictor overall than Control and Relevance of Schoolwork, Peer Support
at School, and Family Support for Learning, results did not reflect previous findings.
Both Appleton et aI. (2006) and Reschly et aI. (2008) identified the Teacher-Student
Relationships factor as significant in predicting academic and emotional outcomes. The
weak associations of Teacher-Student Relationships with academic outcomes may be the
result of timing of this study. The first semester administration of the SEI provided
students with a five-month period to develop relationships with their teachers.
However, Teacher-Student Relationships did have a stronger relationship with
behavioral engagement in this study, which reflects previous findings (Fredericks et aI,
2004; Reschly et aI. 2008; Skinner & Belmont, 1993). The link between teacher support
for learning and positive emotions is well documented (Osterman, 1998; Reschly et aI.,
2008), and further studies utilizing the SEI should attempt to develop more refined
outcome measures of student behavior to test the Teacher-Student Relationships factor
and its ability to predict behavioral and emotional engagement.
Outcome variables. The only continuous outcome variable, GPA, exhibited
relationships in expected directions for most of the cognitive and affective engagement
factors. In addition, the distribution of the sample was symmetric. However, the relative
......----------- -----
59
variability of GPA as a measure of academic achievement limited the overall strength of
this variable in assessing relationships with predictor variables.
All three dichotomous outcome variables (credit completion, minor behavioral
referral, major behavioral referral) provided limited predictability ofthe relationship
between cognitive and affective engagement factors and educational outcomes. The
dummy-coding ofthese three variables constrained the substantial differentiation in the
data (e.g. the range of major behavioral referrals and course failures was from 1 - 7). The
development of other continuous outcome variables, such as course-taking patterns or
scores on state assessments, would allow for greater refmement in assessing the
relationships of cognitive and affective engagement on student outcomes and provide
data to make stronger inferences about the degree to which a student is disengaged from
school. In addition, other variables that have been used to assess academic (time on task,
homework completion) and behavioral (attendance, classroom participation, extra credit
options) engagement (Appleton et aI., 2006) would provide more meaningful data to
support the unique predictive strengths of cognitive and affective engagement.
Based on the findings in this study, there remains little evidence that using the
minor behavioral referral variable in future engagement studies will yield meaningful
data on cognitive and affective engagement. None ofthe cognitive or affective
engagement factors was statistically significant in predicting minor behavioral referrals (p
= .084 - .475). Previous studies on student engagement have used the dichotomous
variable of suspension, but have not differentiated among other behavioral referrals
(Appleton et aI., 2006; Caraway et aI., 2003; Fredericks et aI., 2004; Furlong et al., 2003).
The behavioral incidents that make up the minor behavioral referral variable in this study
60
may not be of sufficient magnitude (i.e. cell phone infraction, tardiness) to capture a
student's values and beliefs as measured by the SEI.
Limitations
This study had several limitations. First, this study was limited to a single
freshman cohort at one suburban high school, and findings may not be relevant for
students in other educational settings or locations. The relatively small (N = 371) and
select sample limits the generalizability of this study. In addition, the sample included all
9th grade students at the end of the first semester (February, 2010), after substantial
attrition had occurred among the freshmen class from September to February due to
dropout and school enrollment changes. Many of those disengaged students lost to
attrition or dropout may have had substantial impact on the findings of this study. The
potential significance of the relationships among cognitive and affective engagement and
academic and behavioral engagement outcomes may have been more supported had these
students been examined prior to dropping out before the end of the first semester.
Furthermore, the sample was predominately Caucasian and did not represent extensive
cultural diversity. Therefore, findings may not be generalizable to other ethnic groups.
This study would be strengthened by replication in larger, more culturally diverse
educational institutions with all grade levels.
Second, this study only provides insight into a singular moment in the high school
experience, and may not be representative of the cognitive and affective engagement of
all students across grades 9-12. This is particularly important as a result of the
maturational changes a student undergoes (LaNasa, Cabrera, & Trangsrud, 2009). As a
result of the limited research on cognitive and affective engagement, it is difficult to infer
61
whether the findings in this study apply to other grade levels or are specific to this age
group.
Finally, the subjective nature of self-report measures may have created biased
findings. Although the strength of the design of the SEI is the measurement of a student's
own perspective, this subjectivity could create biases, such as wanting to present oneself
in a more positive or negative framework. A more comprehensive perspective of a
student's cognitive and affective engagement may have been achieved through the use of
additional forms of assessment, such as teacher- or parent-report instruments in addition
to the self-report measure. Multiple measures would provide data from the internal,
cognitive reality of a student, as well as the objective, observable phenomena which are
the outward manifestations of a student's cognitive and affective engagement with school
(Fredericks et aI., 2004).
Future Implications
Although the study of freshmen dropout is essential given the high risk of dropout
in the transition to high school (Finn, 2006; Zvoch, 2006), further research needs to
approach the maturation of cognitive and affective engagement over the course of a
student's career in school, particularly from grades 9-12 where dropout is highest (Finn,
2006). This is particularly important in light of college and career readiness standards and
the challenges of preparing students for higher education (Conley, 2010). Longitudinal
approaches would provide more relevant data regarding the growth and variability of
cognitive and affective engagement related to a myriad of physical, social, and
educational influences. Thus far, research utilizing the SEI has been confined to isolated
r
I
62
cross-sectional studies, and has failed to capture the longitudinal development of student
engagement through different educational institutions and maturational levels.
In addition, the use of cross-sectional approaches in the study of cognitive and
affective engagement limits the ability to create strong predictive connections between
the subtypes of student engagement (Gutman & Midgley, 2000). The periodic
measurement of a student's values and beliefs and their relationship to educational
outcomes is important in understanding the influence of cognitive and affective
engagement on dropout, which is a gradual process that may take years to become fully
realized (Finn, 1989; Janosz et aI., 2008). Furthermore, longitudinal research is critical in
order to develop interventions that mediate the effects of low cognitive and affective
engagement that lead to dropout (Fredericks et aI., 2004)
There is a dearth of studies examining the amenability of cognitive and affective
engagement to interventions (Christenson & Thurlow, 2004). Significantly, however,
studies on high school dropout have identified characteristics of dropout that respond to
intervention (Barton, 2004; Zvoch, 2006; Appleton et aI., 2008) and are associated with
engagement. Many of these studies have been limited to the holistic construct of student
engagement and have not isolated the subtypes of cognitive and affective engagement. To
address these gaps in the engagement literature, substantial research needs to test the
amenability of these factors of cognitive and affective engagement to intervention. A link
is clearly established in the literature between cognitive and affective engagement and
educational outcomes, but specific research-based programs to effect widespread change
in students' experiences and improved student outcomes is in its nascence. Future
research needs to pursue the development of intervention systems for disengaged students
63
that are constructed from data on the specific factors of cognitive and affective
engagement (Furlong & Christenson, 2008).
Interventions that address cognitive and affective engagement should target the
underlying cognitive and affective factors that are associated with students' persistence
and commitment to education. Students need to know that there is someone that they can
rely on when they begin to become distracted from school. Interventions such as the
Check and Connect program have been created to establish formalized, individual
relationships between staff and students in order to foster an increased commitment to
education (Christenson & Thurlow, 2004). The makeup of interventions that address low
levels of cognitive and affective engagement have included sustained, personalized
programs that build supportive relationships between students and adults along with
explicit instruction that focuses on building students' confidence and persistence
(Brophy, 2004).
Conclusion
Although the findings of this study lend credence to the view of student
engagement as a multidimensional construct, future study needs to more precisely gauge
the effects, over time, of each subtype and their dynamic interactions (Fredericks et aI.,
2004; Glanville & Wildhagen, 2007; Reschly et aI., 2008). The strength ofthe Future
Aspirations and Goals factor and the relative weakness of the Peer Support at School,
Family Support for Learning, Control and Relevance of Schoolwork, and Teacher-
Student Relationships factors provide data to support a view of the subtypes of cognitive
and affective engagement as being more isolated in their effects. This disparity in the
strength of each factor was unexpected and reveals the challenge of gauging the strength
64
of the interplay of each of the four subtypes of engagement - cognitive, affective,
behavioral, and academic - and the factors that represent them. Future studies should
determine the relative strengths of each subtype and the specific significance that
cognitive and affective engagement exhibit in the multi-dimensional construct of student
engagement.
65
REFERENCES
Alexander, K. L., Entwisle, D. R., & Horsey, C. S. (1997). From first grade
forward: Early foundations of high school dropouts. Sociology ofEducation,
70,87-107.
Anderson, A. R., Christenson, S. L., Sinclair, M. F., & Lehr, C. A. (2004). Check &
Connect: The importance of relationships for promoting engagement with school.
Journal ofSchool Psychology, 42(2),95-113.
Appleton, J. J., Christenson, S. L., & Furlong, M. 1. (2008). Student engagement with
school: Critical conceptual and methodological issues of the construct.
Psychology in the Schools, 45(5), 369-386.
Appleton, J. J., Christenson, S. L., Kim, D., & Reschly, A. L. (2006). Measuring
cognitive and psychological engagement: Validation of the Student Engagement
Instrument. Journal ofSchool Psychology, 44(5),427-445.
Archambault, 1., Janosz, M., Fallu, J., & Pagani, L. (2009). Student engagement and its
relationship with early high school dropout. Journal ofAdolescence, 32(3),651-
670.
Assor, A., & Connell, 1. P. (1992). The validity of students' self-reports as measures of
performance affecting self appraisals. In D. H. Schunk & J. L. Meece (Eds.),
Student perceptions in the classroom (pp. 25-47). Hillsdale, NJ: Lawrence
Erlbaurn Associaties, Inc.
Betts, J., Appleton, J. J., Reschly, A., Christenson, S. L., & Huebner, E. S. (2009). A
study of the reliability and construct validity of the School Engagement
Instrument (SEI) across grades. Manuscript submitted for publication.
Bollen, K. (1989). Structural equations with latent variables. New York: Wiley.
Brophy, J. (2004). Motivating students to learn. Mahwah, NJ: Lawrence Erlbaurn.
Bryson, C., & Hand, L. (2007). The role of engagement in inspiring teaching and
learning. Innovations in Education & Teaching International, 44(4), 349-362.
Carini, R. M., Kuh, G. D., & Klein, S. P. (2006). Student engagement and student
learning: Testing the linkages. Research in Higher Education, 47(1), 1-32.
66
Causey, D. L., & Dubow, E. F. (1992). Development ofa self-report coping measure for
elementary school children. Journal o/Clinical Child Psychology, 21,47-59.
Christenson, S. L., Reschly, A. L., Appleton, J. J., Berman, S., Spangers, D., & Varro, P.
(2008). Best practices in fostering student engagement. In A. Thomas & J. Grimes
(Eds.), Best practices in school psychology (Vol. 5, pp. 1099-1120). Washington,
DC: National Association of School Psychologists.
Christenson, S. L., & Thurlow, M. L. (2004). School dropouts: Prevention considerations,
interventions, and challenges, Current Directions in Psychological Science, 13,
36-39.
Conley, D. (2010) College and career ready. San Francisco, CA: Jossey-Bass.
Connell, J. P., & Wellborn, J. G. (1991). Competence, autonomy, and relatedness: A
motivational analysis of self-system processes. In M. R. Gunnar & L. A. Sroufe
(Eds.), Selfprocesses and development (Vol. 23, pp. 43 -77). Hillsdale, NJ:
Lawrence Erlbaum.
Como, L. (1993). The best-laid plans: Modern conceptions of volition and educational
research. Educational Researcher, 22, 14-22.
Doll, B., & Hess, R. S. (2001). Through a new lens: Contemporary psychological
perspectives on school completion and dropping out of high school. School
Psychology Quarterly, 16,351-356.
Finn, J. D. (1989). Withdrawing from school. Review o/Educational Research, 59,117-
142.
Finn, J. D. (1993). School engagement and students at risk. Washington, DC: National
Center for Education Statistics.
Finn, J. D.,& Rock, D. A. (1997). Academic success among students at risk for school
failure. Journal 0/Applied Psychology, 82,221-234.
Fredricks, J. A., Blumenfeld, P. C., & Paris, A. H. (2004). School engagement: Potential
of the concept, state of the evidence. Review o/Educational Research, 74,59-109.
Fredricks, J. A., & Eccles, J. S. (2002). Children's competence and value beliefs from
childhood through adolescence. Developmental Psychology, 38(4),519-553.
Fredrickson, B. L. (1998). What good are positive emotions? Review o/General
Psychology, 2,300-319.
67
Fredrickson, B. 1. (2001). The role of positive emotions in positive psychology: The
broaden-and-build theory of positive emotions. American Psychologists, 58, 218-
226.
Fredrickson, B. 1., & Joiner, T. (2002). Positive emotions trigger upward spirals toward
emotional well-being. Psychological Science, 13, 172-175.
Furlong, M. J., & Christenson, S. 1. (2008). Engaging students at school and with
learning: A relevant construct for all students. Psychology in the Schools, 45(5),
365-368.
Furrer, c., & Skinner, E. (2003). Sense of relatedness as a factor in children's academic
engagement and performance. Journal ofEducational Psychology, 95, 148-162.
Glanville, J. 1., & Wildhagen, T. (2007). The measurement of school engagement:
Assessing dimensionality and measurement invariance across race and ethnicity.
Educational & Psychological Measurement, 67(6), 1019-1041.
Goodenow, C. (1993). Classroom belonging among early adolescent students:
Relationship to motivation and achievement. Journal ofEarly Adolescence, 13,
21-43.
Goodenow, c., & Grady, K. E. (1993). The relationship of school belonging and
friends' values to academic motivation among urban adolescent students. Journal
ofExperimental Education, 62, 60-71.
Greene, B. A., & Miller, R. B. (1996). Influences on course performance: Goals,
perceived ability, and self-regulation. Contemporary Educational Psychology, 21,
181-192.
Greene, B. A., Miller, R. B., Crowson, H. M., Duke, B. 1., & Akey, K. 1. (2004).
Predicting high school students' cognitive engagement and achievement:
Contributions of classroom perceptions and motivation. Contemporary
Educational Psychology, 29(4),462-482.
Guthrie, 1. T., & Wigfield, A. (2000). Engagement and motivation in reading. In M. 1.
Kamil, P.B. Mosenthal, P. D. Pearson, & R. Barr (Eds.), Handbook ofreading
research (3rd ed., pp. 403-422). New York: Longman.
Gutman, 1. M., & Midgley, C. (2000). The role of protective factors in supporting the
academic achievement of poor African American students during the middle
school transition. Journal ofYouth and Adolescence, 29(2),223-148.
Janosz, M., Archambault, 1., Morizot, J., & Pagani, 1. S. (2008). School engagement
trajectories and their differential predictive relations to dropout. Journal ofSocial
Issues, 64,21-40.
68
Jimerson, S. R, Campos, E., & Greif, J. L. (2003). Toward an understanding of
definitions and measures of school engagement and related terms. California
School Psychologist, 8, 7-27.
Joreskog, K. (1971). Simultaneous factor analysis in several populations. Psychometrika,
36, 409-426.
Klem, A. M., & Connell, J. P. (2004). Relationships matter: Linking teacher support to
student engagement and achievement. Journal o/School Health, 74, 262-273.
Kortering, L., & Braziel, P. (2008). Engaging youth in school and learning: The emerging
key to school success and completion. Psychology in the Schools, 45(5),461-465.
Ladd, G. W., Birch, S. H., & Buhs, E. S. (1999). Children's social and scholastic lives in
kindergarten: Related spheres of influence. Child Development, 70, 1373-1400.
LaNasa, S. M., Cabrera, A. F., & Trangsrud, H. (2009). The construct validity of student
engagement: A confirmatory factor analysis approach. Research in Higher
Education, 50(3),315-332.
Lyubomirsky, S., King, L., & Diener, E. (2005). The benefits of frequent positive affect:
Does happiness lead to success? Psychological Bulletin, 131, 803-855.
Maehr, M. L., & Meyer, H. A. (1997). Understanding motivation and schooling: Where
we've been, where we are, and where we need to go. Educational Psychology
Review, 9, 371-408.
Miller, R. B., Greene, B. A., Montalvo, G. P., Ravindran, B., & Nichols, J. D. (1996).
Engagement in academic work: The role of learning goals, future consequences,
pleasing others, and perceived ability. Contemporary Educational Psychology, 21,
388-422.
McMahon, S. D., Parnes, A. L., Keys, C. B., & Viola, 1. J. (2008). School belonging
among low-income urban youth with disabilities: Testing a theoretical model.
Psychology in the Schools, 45(5),387-401.
Osterman, K. F. (1998, April). Student community within the school context: A research
synthesis. Paper presented at the annual meeting of the American Educational
Research Association, San Diego, CA. Retrieved from ERIC database
(ED425519).
Pace, C. R. (1984). Measuring the Quality o/College Student Experiences. Higher
Education Research Institute. Los Angeles: University of California.
69
Pokay, P., & Blumenfeld, P. C. (1990). Predicting achievement early and late in the
semester: The role of motivation and use of learning strategies. Journal of
Educational Psychology, 82,41-50.
Raykov, T. (1997). Estimation of composite reliability for congeneric measures. Applied
Psychological Measurement, 21, 173-184.
Reeve, J., Jang, H., Carrell, D., Jeon, S., & Barch, J. (2004). Enhancing students'
engagement by increasing teachers' autonomy support. k/otivation and Emotion,
28(2), 147-169.
Reschly, A. L., & Christenson, S. L. (2006). Prediction of dropout among students with
mild disabilities: A case for the inclusion of student engagement variables.
Remedial and Special Education, 27,276-292.
Reschly, A. L., Huebner, E. S., Appleton, J. J., & Antaramian, S. (2008). Engagement as
flourishing: The contribution of positive emotions and coping to adolescents'
engagement at school and with learning. Psychology in the Schools, 45(5),419-
431.
Ryan, R. M., Stiller, J., & Lynch, J. H. (1994). Representations of relationships to
teachers, parents, and friends as predictors of academic motivation and self-
esteem. Journal ofEarly Adolescence, 14,226-249.
Sharkey, J. D., Sukkyung, Y., & Schnoebelen, K. (2008). Relations among school assets,
individual resilience, and student engagement for youth grouped by level of
family functioning. Psychology in the Schools, 45(5),402-418.
Skinner, E. A., & Belmont, M. J. (1993). Motivation in the classroom: Reciprocal effect
of teacher behavior and student engagement across the school year. Journal of
Educational Psycholgoy, 85, 571-581.
Yazzie-Mintz, E. (2007). Voices ofStudents on Engagement: A Report on the 2006 High
School Survey ofStudent Engagement: Center for Evaluation and Education
Policy, Indiana University.
Zvoch, K. (2006). Freshman year dropouts: Interactions between student and school
characteristics and student dropout status. Journal ofEducation for Students
Placed at Risk, 11, 97-117.