You are on page 1of 11

Teaching and Teacher Education 89 (2020) 102936

Contents lists available at ScienceDirect

Teaching and Teacher Education


journal homepage: www.elsevier.com/locate/tate

Teaching practices and student engagement in early adolescence: A


longitudinal study using the Classroom Assessment Scoring System
Sarah E. McKellar*, Kai S. Cortina, Allison M. Ryan
University of Michigan, United States

h i g h l i g h t s

 Quality Feedback was the strongest predictor of behavioral engagement.


 In addition to Quality Feedback, other practices matter for behavioral engagement.
 Regard for Student Perspective was the only predictor of emotional engagement.
 No gender and race differences were found among students for changes in engagement.

a r t i c l e i n f o a b s t r a c t

Article history: Student engagement in schoolwork is crucial for positive academic adjustment, particularly during early
Received 13 January 2019 adolescence. We investigated how observations of teaching practices predicted change in student
Received in revised form behavioral and emotional engagement. In the fall, we examined teacher behavior in 54 fifth and sixth
9 September 2019
grade classrooms through external observers' reports of 11 dimensions of teachers’ practices (Classroom
Accepted 18 September 2019
Available online 6 December 2019
Assessment Scoring System, CLASS). Students reported on their behavioral and emotional engagement in
the fall and spring. We found quality feedback was the strongest predictor of behavioral engagement and
regard for student perspective was the strongest emotional engagement. Our findings were more nuanced
Keywords:
Engagement
for what predicts behavioral engagement, as there is evidence that other teaching practices work in
Adolescence conjunction with quality feedback to predict behavioral engagement.
Teachers © 2019 Published by Elsevier Ltd.
Classroom observations

1. Introduction Hemphill, & Heerde, 2017). However, teachers may seek to tap
into different facets of student engagement through taking
Student engagement in schoolwork is crucial for positive aca- different approaches in their teaching practices and interactions
demic adjustment, particularly during early adolescence (Lawson & with students (Pianta, Hamre, & Allen, 2012). The aim of the current
Lawson, 2013; Lei, Cui, & Zhou, 2018). The construct of student study was to examine a range of observed teaching practices in
engagement has received considerable attention over the past few early adolescent classrooms and assess which are most strongly
decades because it robustly predicts numerous outcomes and is linked with changes in students’ emotional and behavioral
responsive to intervention (Eccles, 2016; Finn & Zimmer, 2012; engagement.
Ulmanen, Soini, Pyh€ alto€, & Pietarinen, 2014). In particular, early Student engagement is viewed as a multifaceted construct most
adolescent engagement has been linked with academic achieve- often assessed by asking students about what they do, think, and/or
ment, psychosocial adjustment, and long-term occupational suc- feel in relation to academic tasks (Eccles, 2016; Fredricks, Reschly, &
cess (Abbott-Chapman et al., 2014; Chase, Warren, & Lerner, 2015; Christenson, 2019). According to Finn’s (1989) participation-
Reyes, Brackett, Rivers, White, & Salovey, 2012). Student reports of identification model, students who engage in learning are those
teacherestudent relationships have been consistently associated who are emotionally invested and participate in actions that align
with students’ enhanced engagement in school (Quin, 2017; Quin, with their investment. In the present study and as in previous
studies (see, for example, Gonida, Voulala, & Kiosseoglou, 2009;
Salmela-Aro & Upadaya, 2012; Skinner, Kindermann, & Furrer,
2009), student engagement was theorized as active behavior
* Corresponding author. exhibited by the students in the classroom and as positive emotion
E-mail address: smckella@umich.edu (S.E. McKellar).

https://doi.org/10.1016/j.tate.2019.102936
0742-051X/© 2019 Published by Elsevier Ltd.
2 S.E. McKellar et al. / Teaching and Teacher Education 89 (2020) 102936

experienced during class work. While behavioral and emotional predict engagement and achievement outcomes (Wang & Eccles,
engagement are conceptually and empirically correlated, they are 2012).
nevertheless theoretically distinct (Skinner et al., 2009). This means Skinner and Belmont’s (1993) seminal study employed moti-
that while students who are enjoying activities in math are likely to vational theory to assess the reciprocal relationship between
be putting forth effort into these activities, some may be more likely teacher involvement, structure, and autonomy support on the one
to report that they are enjoying it without showing much active hand and student engagement on the other hand. This approach
engagement, and others are visibly engaged but are not particularly was based on the understanding that engagement is being opti-
fond of the activities they are engaged in. Given this distinction in mized when the social context of the classroom fulfills children’s
two correlated types of engagement, further investigation is and adolescents’ basic psychological needs (Connell & Wellborn,
needed to understand the extent to which specific teacher practices 1991). These are just a few examples of measures of teaching
promote behavioral and/or emotional engagement. practices that have been developed or aligned with those targeting
While both facets of engagement tend to be associated with the students’ psychological needs, engagement, and achievement.
same student outcomes, they may differ in the way they respond to In order to better understand the relationship between teaching
the learning context. Less is known about the extent to which practices and student engagement, both must be conceptually and
specific observed teaching practices e or teacher-student in- empirically defined. In the past few decades, several common
teractions e differentially predict student behavioral and emotional conceptualizations of engagement have emerged, which include
engagement. According to Brophy (2010), teachers often over- academic, agentic, affective, social, and cognitive engagement
emphasize catering instruction to student enjoyment, interest, or (Jimerson, Campos, & Greif, 2003; Reschly & Christenson, 2012).
emotions as a means of engaging students in learning. Instead, he However, with greater specificity in measuring engagement and a
asserted that the appropriate amount of instructional rigor, sup- proliferation of constructs, Eccles (2016) cautions that researchers
port, and feedback over time would foster students’ investment in may lose sight of engagement as a more holistic concept. We
learning, even if students do not initially see tasks as relevant or therefore adopted Finn’s (1989) parsimonious two-dimensional
interesting. In contrast, recent work has stressed how teachers’ approach to understanding engagement and its antecedents. Ac-
emotional support, or facilitating of student enjoyment in class- cording to this model and subsequent empirical conceptualizations
room activities, drives positive student outcomes (Ruzek et al., (Skinner et al., 2009), engagement has two dimensions: behavioral
2016; Weyns, Colpin, De Laet, Engels, & Verschueren, 2018). A and emotional. Behavioral engagement refers to students’ active
teacher might be able to enlist students in class activities through engagement (e.g., a student pays attention, participates, listens, and
subtle pressures that cause negative or mixed emotions; another is involved in class activities), while emotional engagement refers
teacher might grab students’ attention and make them feel to the internal state (e.g., a student is interested, having fun, and
welcome using an entertaining teaching style at the expense of enjoying class activities). While adolescent engagement may also
academic focus and pace. be assessed via teacher or observer reports, students’ self-report of
There is a wealth of studies looking at adolescents’ reports of engagement taps at the students’ state of mind which include
general teachers’ support in relation to their engagement (Quin, cognitive engagement and emotional state that is not necessarily
2017), but studies are just beginning to use standardized observa- expressed in observable behavior. Consistently, student self-reports
tion of these practices (Rimm-Kaufman, Baroody, Larsen, Curby, & tend to correlate with observer rating but are better predictors of
Abry, 2015). Trained observers of real-time or videotaped teach- students' learning and academic achievement (Lei, Cui, & Zhou,
ing practices assess teachers’ behavior apart from perception biases 2018). In line with theoretical and empirical work on student
that are inherent in student reports (Pianta et al., 2012; Pianta & motivation and engagement, different teaching practices align with
Hamre, 2009). These third-person observations can capture as- different students’ adaptive classroom behaviors (Pianta, Belsky,
pects of teaching largely outside of teacher and student awareness et al., 2008; Pianta et al., 2012). While it is assumed that positive
in real time. This allows for more effective teacher training and teaching practices support all types of engagement, theory suggests
coaching because educators, particularly novice teachers, can bet- that certain teaching strategies are more effective at engendering
ter understand what concrete practices are helpful for student specific types of student engagement (Pianta et al., 2012; Rimm-
engagement regardless of their own perception (Rimm-Kaufman Kaufman et al., 2015).
et al., 2015). The current study examined whether systematic ob-
servations of specific teaching practices predict student behavioral 1.2. Defining teacher practices
and emotional engagement.
Similar to student engagement, there have been attempts to
1.1. Defining student engagement create frameworks and standardized measures surrounding
teacher practices that predict positive outcomes. Building on prior
The study of motivation and engagement in the classroom has work, Pianta and Hamre (2009) developed observational measure
contributed to our understanding of effective teaching practices at and a corresponding frameworkdTeaching Through Interactions
the intersection of research in psychology and education (Brophy & frameworkdfor assessing teacher-student interactions or teaching
Good, 1986; Pianta et al., 2012). Specifically, motivational theories practices. In line with other categories of teaching practices from
such as self-determination theory (Ryan & Deci, 2000), expectancy- educational and psychological literature (e.g., Eccles & Roeser,
value theory (Eccles et al., 1993), and goal theory (Ames, 1992; 2011; Pressley et al., 2003; Roehrig, Johnson, Moore, & Bryan,
Covington, 2000) have been used to frame the ways in which 2015), the Teaching Through Interactions model describes three
teachers can engage students. For example, self-determination broad domains teaching practices: Emotional Support, Instructional
theory posits that teachers’ autonomy supportive practices, mes- Support, and Classroom Organization (Hafen et al., 2015; Hamre
sages about competence, and facilitation of student belongingness et al., 2013). Each of these domains describes teaching practices
in the classroom can engage and motivate students (see Pianta & more generally, and contains dimensions within the larger domains
Hamre, 2009, for a review). According to expectancy-value theory that outline more specifically what teachers are doing to support
and goal theory, teachers’ messages about students’ abilities and students. The Emotional Support domain encompasses four mean-
feedback about learning shape students’ beliefs (i.e., learning goals ingful dimensions: Positive Climate, Negative Climate, Teacher
and expectations about their success); in turn, students’ beliefs Sensitivity, and Regard for Student Perspective. The Instructional
S.E. McKellar et al. / Teaching and Teacher Education 89 (2020) 102936 3

Support domain captures four teachers’ practices that support is linked with efforts to identify “best,” highly effective, or quality
cognitive and academic development, including the promotion of teaching practices (Cochran-Smith, 2016; Doyle, 1977). The CLASS,
Problem Solving, Content Understanding, Substantive Dialogue, developed at the University of Virginia, has become a widely used
and teachers’ Quality of Feedback. The Classroom Organization protocol aimed at measuring aspects of instruction and teaching
domain includes teachers’ abilities to manage classroom structures quality (Pianta et al., 2012).
and routines to maximize students actively participating, specif- Growing use of the CLASS has also been accompanied by
ically Behavior Management, Productivity, and Instructional research and calls for additional work to further assess the pre-
Learning Formats. While teaching practices are defined by these dictive validity of the domains and dimensions (Bihler, Agache,
three larger domains, efforts to support teachers in improving their Kohl, Willard, & Leyendecker, 2018; Hafen et al., 2015; Hamre
practices have leveraged the specificity of the 11 teaching di- et al., 2013; Westergård, Ertesvåg, & Rafaelsen, 2018). Some work
mensions of the CLASS that fall under the three broader domains revealed moderate to high correlation across the three domains,
(Hafen et al., 2015). including discussions of multicollinearity among its three larger
domains when predicting student engagement (See Rimm-
1.3. Assessing teaching practice: standardized observational Kaufman et al., 2015). Despite support for a three-domain struc-
protocols ture for the 11 teaching dimensions of the framework, a more
practical focus, in particular from a teacher training perspective,
While student reports of teaching practices are the most com- favors a more differentiated model because the 11 dimensions
mon way to assess teaching practices, this method has limitations. (three to four dimensions per domain) are clearly defined and can
The vast majority of studies have used aggregated student per- be isolated in video footage for training purposes.
ceptions to understand teaching practices due to proximity to
student outcomes and the ease of collection compared with other 1.4. Observational assessment of teaching practices (the CLASS) and
sources (Burniske & Meibaum, 2012). Overreliance on student re- engagement
ports to assess teaching practices may be problematic for several
reasons. First, the validity of student self-reports has been called In our review of the literature, we identified three studies to
into question, among others, for halo effects (Nisbett & Wilson, date that examined student engagement during early adolescence
1977; Wetzel, Wilson, & Kort, 1981) and social desirability bias in relation to teaching practices using at least one of the three
(Fisher, 1993). In other words, relying solely on student reports of domains of the Classroom Assessment Scoring System (CLASS
their teacher may cause issues because students have individual Emotional Support, Instructional Support, and Classroom Organiza-
biases that affect their general preferences and the extent to which tion). The first of these studies, by Reyes and colleagues (2012),
they can parse out the effectiveness of teachers’ specific practices. examined the extent to which the three domains of the CLASS
In predicting other questionnaire variables, common method bias predicted engagement e focusing solely on emotional engagement
becomes a major threat to the validity of the findings (Podsakoff, e and achievement in 5th and 6th grade English and Language Arts
MacKenzie, Lee, & Podsakoff, 2003). This means that student re- (ELA) classes. They found that emotional engagement mediated the
ports of their teachers' practices and their individual engagement relationship between CLASS Emotional Support and student
may be related based on student reporters as common data sources achievement. However, they did not find any association between
rather than teachers affecting students' outcomes. Furthermore, the other two CLASS teaching domains (i.e., Instructional Support
there is some evidence that student perceptions of the classroom and Classroom Organization) and student engagement or achieve-
are shaped to a certain extent by their experiences in prior course ment. They focused on a more general construct of engagement
or other concurrent courses (e.g., Friedel, Cortina, Turner, & using Furrer and Skinner’s (2003) Engagement vs. Disaffection
Midgley, 2010). Scale (e.g., “I feel good when I’m in my ELA class”). While longitu-
Apart from these methodological aspects, there are several ad- dinal in design, this study did not assess how teaching practices
vantages of trained observers of teaching practices in comparison predicted changes in student engagement over the course of the
to student reports. Trained observers can be aware of teaching as- year and did not assess student behavioral engagement.
pects that students are not aware. Because the trained observer Our second identified study, Ruzek et al. (2016), examined
focuses on concrete behavior of teachers and students seen in class, behavioral engagement instead of emotional engagement, focusing
the results lend themselves to be used for teacher training purposes on only one domain of the CLASS, Emotional Support, as a predictor.
in particular if combined with observations and video footage of This study was centered around Emotional Support because it was
the class period assessed. A limitation of third-person observations designed to assess whether the three facets of self-determination
apart from the higher costs of the classroom is that they only theory (students’ sense of autonomy support, competence, and
capture one (or a few) class periods and cannot assess students’ peer belongingness) mediated observed Emotional Support and
actual cognitive processes unless they are verbalized (e.g., a self-reported behavioral engagement. They found Emotional Sup-
misconception in a science class). Accounting for these weaknesses, port predicted behavioral engagement. They also found students
observing teaching practice can offer an effective alternative lens to sense of autonomy support and peer belongingness mediated the
understanding teaching practices. This is especially true if ob- relationship between Emotional Support and behavioral engage-
servers are trained to use an accepted rubric and adhere closely to ment. These mediators closely align with the CLASS dimension of
this training. Regard for Student Perspective, which is one component of the
Over the past few decades, there has been increasing attention Emotional Support domain. For example, Regard for Student
and progress in developing standardized observation protocols to Perspective includes indicators such as “relaxed structure for
better understand e and fairly asses e teaching practices movement about the classroom” and “peer sharing and group
(Danielson, 2007; 2011; Pianta et al., 2012). Calls for standardized work” paralleling autonomy support and peer relatedness respec-
protocols come out of efforts to better identify which classroom tively (Ruzek et al., 2016, p. 98).
teaching processes (e.g., management, formats, interactions) are Our third identified study, Rimm-Kaufman et al. (2015), exam-
most closely linked with student achievement outcomes; referred ined how several types of engagement were related to all three
to as the process-product paradigm (Brophy, 2010; Brophy & Good, domains of the CLASS. They focused on a sample of 5th grade
1986; Cochran-Smith & Lytle, 1990). The process-product approach students and their math classrooms, using student engagement at
4 S.E. McKellar et al. / Teaching and Teacher Education 89 (2020) 102936

three time points over the course of the year. They confirmed that dimensions actually fall under three domains or whether a
students had higher cognitive, social, and emotional engagement in different factor or domain structure is best, such as one or two
classrooms with greater Emotional Support and Classroom Organi- domains (Allen et al., 2013; Bihler et al., 2018; Hafen et al., 2015;
zation. They also found only one of the three domains of the CLASS, Virtanen et al., 2018). A dimensional approach instead of a domain
Classroom Organization, predicted observer reports of student approach to predict engagement may help address some of these
behavioral engagement. Rimm-Kaufmann and colleagues were recent questions. Pianta et al. (2012) also recommended that future
surprised that Instructional Support did not predict behavioral studies examine teaching practices at a “granular level” in relation
engagement and hypothesized a few possible reasons, calling to student engagement (p. 380). Researchers and teachers will
future studies to examine the role of differences in student ability easily agree that Emotional Support, Instructional Support, and
levels and the role of specific dimensions within the domain of Classroom Organization will all be positively associated with desir-
Instructional Support. able student outcomes, but the present study built upon prior work
Overall, more research is needed to understand the relation- by more closely examining the extent to which the 11 CLASS
ships between the CLASS domains of Instructional Support and teaching dimensions predict two types of student engagement,
Classroom Organization and different types of engagement. None- behavioral and emotional engagement, in early adolescence.
theless, when taken together, these three studies provide evidence
that the Emotional Support domain of the CLASS predicts student 1.5. Purpose and hypotheses
emotional and behavioral engagement in secondary ELA and math
classrooms. While these studies used the three broad domains of In classrooms, teachers may take different approaches to
the CLASS to assess the relationships between teaching practices engaging students and tap into different facets of student engage-
and student engagement, they ultimately used the CLASS’s 11 ment. For example, one might imagine a classroom where students
teaching dimensions to explain the implications of their findings. are behaviorally engaged due to the high expectations and aca-
For example, when discussing findings related to the Emotional demic rigor rather than emotional support. Conversely, one can
Support domain of the CLASS, Ruzek et al. (2016) and Rimm- imagine a classroom where students are emotionally engaged as a
Kaufman et al. (2015) suggested ways in which student engage- result of teacher warmth and positive relational climate rather than
ment may be shaped by practices that demonstrate Regard for instructional supports. While teachers use a variety of practices to
Student Perspective (e.g., providing students with autonomy and engage students, perhaps specific practices promote specific types
opportunities to connect with classmates). Similarly, when dis- of student engagement. The purpose of this study was to system-
cussing findings about the Classroom Organization domain of the atically examine the ways in which specific observed teaching
CLASS, Rimm-Kaufman and colleagues explained how classrooms practices predict specific forms of engagement.
that have “interesting learning formats, clear statement of expec- The present study used the 11 teaching dimensions of the CLASS
tations, high productivity” enhance students’ behavioral engage- to understand behavioral and emotional engagement. While this
ment (p. 182). study was largely exploratory, especially related to understanding
Despite these strides in understanding how domains of which dimensions of the CLASS predict changes in behavioral
observed teaching practices are related to student engagement, the engagement, we had two broad hypotheses based on Pianta and
strength of the CLASS’s ability to record 11 specific teaching prac- colleagues' (2012) theoretical review and the three studies linking
tices has not fully been utilized. Few studies have parsed out these the CLASS to engagement during early adolescence (Reyes et al.,
11 dimensions of the CLASS to outline how different teacher- 2012; Rimm-Kaufman at al., 2015; Ruzek et al., 2016). Based on
student interactions may differentially predict distinct types of Reyes et al.’s (2012) and Rimm-Kaufman at al. ’s (2015) findings, our
student engagement, despite the fact that from the practical first hypothesis was that one or more dimensions within the
perspective of teacher training it is particularly beneficial to be as domain of Emotional Support would be the strongest predictor of
specific as possible. Teachers may be able to improve their emotional engagement. Second, we hypothesized that one of the
instructional approaches with greater insights on the extent to domains of Classroom Organization would also strongly predict
which student emotional and behavioral engagement are related to emotional engagement. There was insufficient prior research to
each of the 11 teaching dimensions of the CLASS (e.g., Teacher make dimensional-based hypotheses related to behavioral
Sensitivity, Content Understanding, and Productivity). Deemer and engagement, though Rimm-Kaufman et al.’s and Ruzek et al.’s work
Hanich (2005) stressed the benefits of this kind of specificity, discussed Productivity and Regard for Student Perspective as
arguing that a framework including detailed dimensions of teach- possible dimensional predictors. Lastly, in light of Rimm-Kaufman
ing (i.e., Epstein’s (1988) and Ames’ (1992) TARGET framework) can at al. ’s (2015) findings that the relationship between Classroom
help instructors improve student motivation and engagement. For Organization and behavioral engagement was stronger for 5th
example, they describe how understanding teacher recognition of grade boys than girls, we also explored gender differences among
student effort and appropriate evaluation of student lear- students.
ningdcomponents in the CLASS’s dimension Quality of Feed-
backdcan support teachers in increasing student engagement, 2. Method
especially when used as part of a multidimensional framework.
Moreover, the developers of the CLASS and their colleagues, Hafen 2.1. Procedure and participants
et al. (2015), created an intervention program, called MyTeach-
ingPartner, to improve teaching practices based on the Teaching Data were collected as part of a larger longitudinal study
Through Interactions framework and the CLASS. While MyTeach- examining early adolescent social and academic adjustment in
ingPartner uses all 11 teaching dimensions of the CLASS to offer school. Schools were recruited from three school districts located in
feedback, this specificity has not been used to understand sec- small urban communities in the Midwest. To provide a common
ondary student reports of their engagement in prior empirical reference point across the different school settings, we focused on
studies. Although a domain approach to understanding engage- the classroom context in the domains of math and science. Prior
ment is useful, studying the specific dimensions of the CLASS may work (Allen et al., 2013) found that the CLASS was related to gains
better support teachers in their efforts to improve their practices. in math and science achievement similarly, providing support for
There has also been some debate about whether these 11 including both subjects. All math and science teachers in the 6th
S.E. McKellar et al. / Teaching and Teacher Education 89 (2020) 102936 5

grade at the middle schools agreed to participate, and we chose one 7% Hispanic and 9% other ethnic groups). For all of the variables in
of their classes to observe and administer surveys. For the teachers our study, no differences were found between students who left
from the feeder elementary schools, we aimed to focus on math or after the fall (i.e., missing data in the spring) compared with stu-
science in equal proportions. dents who participated in the fall and the spring.
Observations were conducted in October and November. As the Instructions and items were read aloud while students read
school year commenced in mid-August in these school districts, along and responded. Students were assured that the information
this was about two or three months into the school year. Research in the survey would be kept confidential. In addition, students were
assistants scheduled classroom observations on days that the told that filling out the survey was voluntary and that they could
teachers deemed “typical” days of math or science instruction. stop at any time. We visited the schools one additional day to
About a week or two following observations, two trained research administer make-ups for students who were absent for survey
assistants administered surveys to students in their classrooms. administration. Survey administration was repeated about six
Four classrooms did not complete the student survey aspect of the months later in the spring of the school year.
project due to scheduling conflicts and were not included in our
study. 2.2. Measures
The Upper Elementary Version of the CLASS certification (grades
4e6; Pianta, Hamre, & Mintz, 2010) was obtained by our six Classroom Assessment Scoring System (CLASS). The CLASS is a
classroom observers; the Collaborative Institutional Training well-established observational measure of teaching practices in the
Initiative (CITI) Human Subject Protection trained university re- classroom (Pianta, La Paro, & Hamre, 2008). The CLASS is comprised
searchers. Certification on the CLASS observation protocol meant of three domains each with three or four dimensions, so 11 di-
that all six coders had to achieve at least 80% correct on the test at mensions total related to teaching, each with a set of criteria to
the end of the CLASS course. To obtain this, score codes must be score (see Table 1). The first domain of Emotional Supports
within one point (on a seven point continuum) from what the (a ¼ 0.83) contains three dimensions of Positive Climate (relation-
developers consider the correct code when coding segments of ships, affect, respect, communication), Negative Climate (punitive-
videos of classrooms. Furthermore, 20% of the classroom observa- ness, sarcasm/disrespect, negativity), Sensitivity (awareness,
tions in our sample were conducted in pairs. When we employed responsiveness, action to address problems, comfort), and Regard
the same criteria in the field as the CLASS certification test, 94% of for Student Perspectives (flexibility, support for autonomy, connec-
the time our observation pairs were coded within one point of each tions to current life, and meaningful peer interactions). The second
other. domain of Instructional Supports (a ¼ 0.77) is comprised of four
For the present study, students came from 54 classrooms (27 5th dimensions: Content Understanding (teachers help students un-
grade classrooms, 18 math and 9 science; and 27 6th grade class- derstand both the broad framework and key ideas in an academic
rooms, 16 math and 11 science). The 5th grade teachers were pre- discipline), Analysis and Problem Solving (promotion of higher-
dominantly female (N ¼ 25, 93%) and White (N ¼ 23, 85%). The 6th order thinking skills, opportunities for application), Quality of
grade teachers were predominantly female (N ¼ 21, 78%) and White Feedback (feedback loops, encouragement of responses, explana-
(N ¼ 25, 93%). Only students who returned signed permission from tion of performance), and Instructional Dialogue (content-driven
parents were allowed to participate. The percentage of students exchanges, distributed talk and facilitation strategies). The third
with parental consent averaged 84% and was similar at both grade domain of Classroom Organization (a ¼ .78) includes three di-
levels. The total sample (N ¼ 860) was comprised of students who mensions: Behavior Management (clear expectations, proactiveness
had data at both waves. Roughly half of our student sample was and redirection), Productivity (maximized time use, efficient rou-
female (51.1%) and there was heterogeneity of student ethnicity tines and transitions), and Instructional Learning Formats (engaging
across our sample (36% African American, 47% European American, approach that maximizes learning opportunities). The CLASS

Table 1
Classroom assessment scoring system domains and (class) dimensions (upper elementary version, grades 4e6; Pianta, Hamre & Mintz, 2010).

Domain Dimension Indications/Description

Emotional Positive Climate Relationships, Positive affect, Positive communications, and Respect
Support Negative Climatea Negative affect, Punitive control, and Disrespect
Teacher Sensitivity Awareness, Responsiveness, Effectiveness in addressing problems, and Student comfort
Regard for Student Flexibility and student focus, Connections to current life, Support for student autonomy and leadership, and Meaningful peer
Perspectives interactions

Instructional Content Understanding Depth of understanding, Communication of concepts and procedures, Background knowledge and misconceptions,
Support Transmission of content knowledge and procedures, and Opportunity for practice of procedures and skills
Analysis and Problem Inquiry & Analysis, Opportunities for novel application, and Metacognition
Solvingb
Quality of Feedback Feedback loops, Scaffolding, Building on student responses, and Encouragement and affirmation
Instructional Dialogue Cumulative, content-driven exchanges, Distributed talk, and Facilitation strategies

Classroom Behavior Management Clear expectations, Proactive, Effective redirection of misbehavior, and Student behavior
Organization Productivity Maximizing learning time, Routines, Transitions, and Preparation
Instructional Learning Learning targets/organization, Variety of modalities, strategies, and materials, Active facilitation, and Effective engagement
Formats*

Note 1. The domain and dimension changes outlined above have been made to the CLASS since Hafen and colleagues' (2015) study. This table represents the CLASS teaching
dimensions during our data collection, and we used Pianta, Hamre & Mintz's (2010) conceptualization of the CLASS dimensions and domains.
Note 2. In addition to the 11 dimensions outlined above, there is one CLASS dimension, Student Engagement, that does not fall under any other domains and is not considered
an indicator of teaching practices.
a
Domain Changes: Negative Climate became part of Classroom Organization and Instructional Learning Formats became part of Instructional Supports.
b
Dimension Change: Analysis and Problem Solving has been renamed to Analysis and Inquiry.
6 S.E. McKellar et al. / Teaching and Teacher Education 89 (2020) 102936

Table 2
Mean, Standard Deviation, and Correlations between Student Gender, Race, and Student Behavioral and Emotional Engagement.

Emotional D Emotional Behavioral D Behavioral


Engagement Engagement Engagement Engagement
Correlations Correlations Correlations Correlations

M SD Fall Spring Fall Spring

Gender (male 0.49 0.50 -.01 .00 .05 -.09** -.08* .02
students ¼ 1)
Black students 0.38 0.49 .06y .00 -.06y -.02 -.03 -.01
White students 0.49 0.50 -.08* -.03 .05 .01 .02 .00
Latino students 0.08 0.27 .00 -.01 -.02 .02 -.03 -.03
Asian students 0.06 0.24 .02 .05 .03 .00 .04 .05
D Emotional Engagement 0.21 1.02 -.38*** .50*** -.17*** .31*** .51***
D Behavioral Engagement 0.15 0.77 -.11** .34*** -.36*** .58***

Note 1. D represents the difference between fall and spring engagement, M represents mean, and SD represents standard deviation.
Note 2. “D Emotional Engagement Correlation” and “D Behavioral Engagement Correlation” show the correlations between demographic variables and the difference between
mean fall and spring emotional engagement.
Note 3. Correlations between emotional and behavioral engagement and changes in engagement ranged from .40 to .65, and are all p<.001.
yp<.1, *p<.05, **p<.01, ***p<.001

observational measure can be used for any subject matter and model with spring engagement as the outcome to calculate the
across elementary school subjects. Each dimension of the CLASS is intraclass correlation (ICC) as a measure of the amount of variance
rated on a scale from 1 to 7 by certified observers. The scoring does that existed between classrooms. Next, in Model 1 (see Table 4), we
not factor in aspects of the classroom outside of the teaching introduced fall engagement in order to model change, and student
practices in relation to their students (e.g., there are no scores for demographics (gender and race). In subsequent models, we added
instructional materials, physical environment, or curriculum class-level predictors using a stepwise approach in order to avoid
adoption; Pianta & Hamre, 2009). instability due to multicollinearity as reported in prior research
For the current study, observers visited classrooms on two (Rimm-Kaufman et al., 2015). We included in the model the
different days (but at the same time). During these observations the dimension of CLASS that showed the highest, significant effect if
entire lesson was observed and between four to six rating cycles added to the model as a single additional predictor, hence reflecting
were completed. Each rating cycle consisted of a 15 minute the strongest predictor of change. In a subsequent step, we
observation of classroom practices followed by a 5 to 10 minute inspected the potential of further CLASS dimensions to significantly
time period where the observers rated the 11 CLASS dimensions. add to the prediction of change after the first predictor was
The length of the lessons varied from 48 min to an hour. The eleven included. This process of adding predictors continued until addi-
dimensions were averaged to produce scores on the three domains tional predictors did not significantly predict engagement and the
as outlined above. model fit did not improve (e.g., the Chi-square change between the
Student Engagement. We adapted an established 10-item self- prior and current model was non-significant), emulating standard
report measure of student engagement to assess behavioral and “stepwise” procedures in regression analysis in software packages
emotional engagement in the classroom (Skinner et al., 2009). For (e.g., SPSS). This analytic approach has the advantage that it doc-
behavioral engagement, items assessed the extent to which stu- uments at every step the potential instability of regression co-
dents try hard, participate in classroom activities, exert effort, pay efficients for predictors already in the model. We ran these
attention, and persist. For example, items included “I try hard to do stepwise models separately for behavioral engagement and
well in math/science class” and “I pay attention in my math/science emotional engagement.
class.” For emotional engagement, items assessed the extent to
which students enjoyed and had positive feelings related to their
experiences in their class. For example, items included, “When we
work on something in math/science class, I feel interested” and “I
3. Results
enjoy learning new things in my math/science class.”
All items were answered on a 5-point Likert response scale
3.1. Descriptive statistics and correlations
(1 ¼ not at all true of me, 3 ¼ somewhat true, and 5 ¼ very true of
me). The scores ranged from 1 to 5 with a 5 indicating more or
The means, standard deviations, and correlations are displayed
positive behavioral or emotional engagement. The validity of this
in Table 2 (student level variables) and Table 3 (classroom level
measure has been demonstrated in research showing that students
variables). For student level variables, gender and race were not
who rate themselves high on engagement are more likely to be
correlated with changes in either type of engagement from fall to
identified by their teacher as engaged (Skinner et al., 2009). Our
spring. For classroom level variables, eight of the 11 dimensions of
measure of behavioral engagement was reliable in the current
the CLASS across all three broader domains were significantly
sample (six items, Cronbach’s alpha ¼ .86 at Wave 1 and 0.89 at
correlated with classroom aggregates of student behavioral
Wave 2) as well as the measure of emotional engagement (four
engagement in the spring. Quality of Feedback and Regard for Stu-
items, Cronbach’s alpha ¼ .89 at Wave 1 and 0.92 at Wave 2).
dent Perspectives were significantly correlated with the change in
classroom aggregates of student behavioral engagement from fall to
2.3. Analytical procedure spring. For emotional engagement, three dimensions of the CLASS
observations (Positive Climate, Teacher Sensitivity, Regard for Stu-
Given that the students were nested within classrooms, and dent Perspectives) were correlated with aggregates of student
therefore dependent observations, we used multilevel modeling, emotional engagement in the spring. Regard for Student Perspec-
specifically two-level Hierarchical Linear Modeling (HLM; tives was significantly correlated with the change in classroom ag-
Raudenbush & Bryk, 2002). First, we ran a fully unconditional gregates of student emotional engagement from fall to spring.
S.E. McKellar et al. / Teaching and Teacher Education 89 (2020) 102936 7

Table 3
Mean, Standard Deviation and Correlations between the 11 CLASS Dimensions of Teaching Practices and Classroom Aggregates of Student Engagement.

Emotional D Emotional Behavioral D Behavioral


Engagement Engagement Engagement Engagement
Correlations Correlations Correlations Correlations

M SD Fall Spring Fall Spring

Emotional Support
Positive Climate 4.57 1.09 .22 .32* -.11 .15 .34* -.27y
Negative Climate 1.94 0.95 -.13 -.21 .09 -.20 -.38** .26y
Teacher Sensitivity 3.88 1.07 .19 .29* -.11 .15 .34** -.27
Regard for Student Perspectives 2.76 1.13 .13 .42** -.34* .26y .45** -.28*
Classroom Organization
Behavior Management 4.88 1.10 .15 .21 -.07 .31* .41** -.18
Productivity 4.53 0.96 .16 .27y -.12 .22 .40** -.27y
Instructional Learning Formats 3.76 0.75 .01 .08 -.08 -.02 .13 -.19
Classroom Organization
Content Understanding 3.52 1.04 .06 .06 .01 .05 .24y -.25y
Analysis and Problem Solving 1.98 0.93 .16 .20 -.04 .20 .171 .01
Quality of Feedback 3.11 0.84 .08 .25y -.21 .12 .38** -.36**
Instructional Dialogue 2.79 0.99 .23y .27y -.03 .27* .38** -.19

Note. D represents the difference between fall and spring engagement, M represents mean, and SD represents standard deviation.
yp < .1, * p < .05, **p < .01, ***p < .001.

3.2. Multilevel analyses results Student Perspective.


Additional findings. Across all models, we controlled for gender
Null and Covariate Model: For spring engagement, 5% of the and race as fixed effect. After controlling for prior engagement,
variance of behavioral engagement and 9% of the variance of neither effect was significant.
emotional engagement was attributable to between-classroom
differences. The covariate model (see Models 1 in Table 4) refers 4. Discussion
to models with spring engagement predicted by fall engagement,
gender, and race. In the covariate model, we established that fall The purpose of this study was to use standardized classroom
engagement predicted spring engagement for both types (behav- observations of specific teaching practices to predict changes in
ioral engagement g ¼ 0.70, p < .001 and emotional engagement early adolescent emotional and behavioral engagement from fall to
g ¼ 0.64, p < .001). The proportion of classroom variance was spring. Overall, the Quality of Feedback was the strongest predictor
accounted for by covariates was 62%. Gender and race did not of student behavioral engagement and Regard for Student
significantly predict spring engagement when controlling for fall Perspective was the strongest predictor of student emotional
levels; however, we kept them in subsequent models as controls. engagement.
Behavioral engagement. Our HLM exploratory analysis run for While several practices were positively associated with change
Model 1 of behavioral engagement the CLASS dimension with the in behavioral engagement, Quality of Feedback and Productivity
highest t-value was for Quality of Feedback (b ¼ 0.04, SE ¼ 0.01, explained nearly all of the predictive power of the CLASS di-
t ¼ 3.581; see Table S1 in the online Supporting Information). As mensions for behavioral engagement. Given our exploratory
shown in Table 4 Model 2, teachers’ Quality of Feedback was hence approach and Brophy’s (2010) work on motivation to learn, it is not
included in the next step and is a strong predictor of behavioral surprising that teachers’ Quality of Feedback, a dimension of
engagement, b ¼ 0.12, SE ¼ 0.04, t ¼ 3.43, p < .001. After including Instructional Support, was the strongest predictor of behavioral
Quality of Feedback, Productivity, the second highest original po- engagement. Our findings align well with Pianta and Hamre’s
tential predictor with b ¼ 0.01, SE ¼ 0.00, t ¼ 1.56,1 was not a sig- (2009) framework that outlines Quality of Feedback as a teacher’s
nificant predictor of behavioral engagement above and beyond ability to scaffold content and build on student responses for stu-
teachers’ Quality of Feedback; however, there was a marginally dent learning. It is also not surprising that teachers’ Productivity, a
significant trend, b ¼ 0.05, SE ¼ 0.03, t ¼ 1.87, p ¼ .068. dimension of Classroom Organization, was the second strongest
Emotional engagement. For emotional engagement, the Model predictors of behavioral engagement. This parallels Rimm-
1 HLM exploratory analysis revealed Regard for Student Perspective Kaufmann et al. s’ (2015) finding that Classroom Organization pre-
as the CLASS dimension with the highest t-value (b ¼ 0.06, dicts observer reports of student behavioral engagement.
SE ¼ 0.02, t ¼ 3.321; see Table S2 in the online Supporting Infor- The strong relationship between Quality of Feedback and
mation). As shown in Table 4 Model 2, Regard for Student behavioral engagement does not mean that other strengths of the
Perspective significantly predicts emotional engagement in the instruction are irrelevant. However, for increasing behavioral
spring when controlling for fall engagement. Instructional Learning engagement, our results suggest that when observers see high
Formats, which had the second highest t-value as potential addi- quality in the way teachers give feedback, the instruction tends to
tional predictor in Model 1 was not a significant predictor of be strong in other observed dimensions as well. Overall, our find-
emotional engagement above and beyond Regard for Student ings for behavioral engagement suggest that teachers’ ability to
Perspective. Consistently, as seen in Model 3, adding Instructional scaffold lessons and offer constructive feedback is most likely to
Learning Formats did not improve the model fit. We conclude that increase students’ participation, effort, and focus in classroom
there were no additional dimensions of the CLASS that significantly learning tasks from fall to spring. Along with these findings, Pro-
predict emotional engagement above and beyond Regard for ductivity and Regard for Student Perspective are also likely to
predict behavioral engagement based on our exploratory analyses,
and they are likely to be working in conjunction with Quality of
1
p values are not generated in HLM exploratory analyses. Feedback.
8
Table 4
Multilevel Models for the CLASS Teaching Practices Predicting Changes in Emotional and Behavioral Engagement from Fall to Spring.

S.E. McKellar et al. / Teaching and Teacher Education 89 (2020) 102936


S.E. McKellar et al. / Teaching and Teacher Education 89 (2020) 102936 9

Change in student emotional engagement was largely predicted examination of the dimensions of the CLASS may offer insights that
by teachers’ Regard for Student Perspective. Different from can support educators in targeting their professional development
behavioral engagement, other dimensions of the CLASS did not goals towards bolstering behavioral and/or emotional engagement.
predict emotional engagement above and beyond this critical Our findings support current intervention efforts to increase stu-
predictor. Our results relate to Rimm-Kaufmann and colleagues' dent engagement. For example, the developers of the CLASS and
(2015) findings that Emotional Support predicted engagement, their colleagues’ intervention, MyTeachingPartner, could use these
since teachers’ Regard for Student Perspectiveda dimension of findings as they evaluate teachers across all 11 dimensions of the
Emotional Supportdwas the strongest and only significant predic- CLASS and suggests targeted areas for improvement (Hafen et al.,
tor of emotional engagement when looking at multiple dimensions. 2015).
According to Pianta et al. (2012), this association means that
teachers’ consideration of students’ autonomy needs and promo- 4.1. Strength and limitations
tion of meaningful peer interactions supports students’ enjoyment
in classroom learning tasks. These findings for emotional engage- Among the strengths of this study was our focus on under-
ment were expected, given the motivational needs outlined by standing changes in student engagement over the course of a
Ryan and Deci’s (2000) self-determination theorydautonomy, school year. This accounts for the individual student characteristics
competence, and belongingness dparallel the CLASS’s dimension that predict student engagement prior to them entering the
of Regard for Student Perspective. However, we were surprised that classroom. Another strength of the study was our racially diverse
we did not find links between dimensions within the Classroom sample of students from a range of socio-economic backgrounds,
Organization domain, since Rimm-Kaufmann et al. (2015) found predominantly middle to lower income students.
emotional engagement was related to both Emotional Support and Despite these strengths, the study was limited by our inability to
Classroom Organization. assess student engagement more than twice in a given school year.
When interpreting these findings, it is important to consider the We also did not collect data to assess the stability of the observed
extent to which different types of engagement are amenable to teaching practices, and this study was limited by only having fall
teacher influence; to what extent can teachers shape students’ observations of teaching practices. Nonetheless, our findings par-
behaviors in their math class versus their feelings about their math allel Rimm-Kaufmann et al. s’ (2015) work looking at three time
class? Our preliminary analyses provided evidence that classroom points. Further research should assess engagement and teaching
differences in behavioral engagement were better explained by the practices several times throughout the year in order to test Finn’s
CLASS than classroom difference in emotional engagement. An (1989) participation-identification model because this would
estimated 17% of the between-classroom variance for emotional offer evidence for the reciprocal effects between behavioral and
engagement was explained by Regard for Student Perspective and emotional engagement in relation to teacher practices. We were
Instructional Delivery Formats, while 30% of the between- also limited in only assessing two types of engagement. Future
classroom variance for behavioral engagement was explained by work may also consider additional types of engagement, including
Quality of Feedback and Productivity. Moreover, the correlations of academic, agentic, affective, social, and cognitive engagement
CLASS dimensions with emotional engagement are systematically (Reschly & Christenson, 2012). Prior work has noted the impor-
lower than the correlations with behavioral engagement. From tance of peers for both behavioral and emotional engagement
these findings, our study suggests that observed teaching practices (Kindermann, 2016). Thus, additional research may also consider
are better designed to explain changes in behavioral engagement the role of peer dynamics alongside teaching practices. This is
than changes in emotional engagement. Students’ behavioral especially important during adolescence given the importance of
engagement is part of the social interactions of the classroom itself peer influence for student engagement (Shin & Ryan, 2014).
and therefore logically related to the visible teaching practices or Perhaps the most important lesson to be learned from this study
interactions with the students, while emotional engagement is a was that distinct practices can be leveraged by teachers to support
more private attitude towards the learning process that might not distinct types of student engagement. Quality of Feedback and
be immediately visible and hence less reliably observed. Productivity can be leveraged to support student behavioral
Overall, our study reaffirms the importance of specifically engagement, and Regard for Student Perspective can be leveraged
examining what teachers are doing to support student effort and to support student emotional engagement. Depending upon
enjoyment during learning activities. Most importantly, this sup- teaching goals and learning objectives, a teacher may want to pri-
ports the need to have observational tools to more clearly assess oritize facilitation of student enjoyment in learning activities
practices, as many student-report measures have a tendency to instead of prioritizing facilitation of student effort and focus on
have a halo effect, where students will rate a teacher highly across learning tasks. While these objectives are reciprocal according to
all teaching practices neglecting to differentiate across aspects of Finn’s (1989) participation-identification model, our study can
teaching, whereas third-party observations delineate between support teachers in targeting practices to areas that need the most
teaching practices that support distinct types of student engage- attention as well as catering instruction to their engagement pri-
ment. For example, if a teacher has students who appear to enjoy orities for students. With this in mind, observed teaching practices
time in math but are failing to participate in more cognitively are better able to measure behavioral engagement than emotional
challenging ways, the teacher may want to consider focusing their engagement. This work paves the way for intervention work to
attention on the types of feedback offered to students. Yet a teacher equip teachers with the tools that best align with their needs and
whose students are working hard but failing to enjoy learning tasks explore less common ways of assessing teaching-practices (e.g.,
may want to focus primarily on greater attunement with students’ third party observations) as they relate to student engagement.
needs and taking students’ voices into consideration.
While the three dimensions of the CLASS allow for parsimony in Author note
research studies, examining the 11 teaching dimensions of the
CLASS allowed for a more nuanced understanding of what predicts Thank you to all of the teachers and students who participated
behavioral and emotional engagement. Although our study does in the Classroom and Peer Ecologies (CAPE) project. We would like
not address concerns over the validity of the three larger domains to acknowledge the original funding source provided by the
of the Teaching Through Interactions framework, a closer Spender Foundation. Correspondence concerning this article
10 S.E. McKellar et al. / Teaching and Teacher Education 89 (2020) 102936

should be addressed to Sarah E. McKellar, Combined Program in Fisher, R. J. (1993). Social desirability bias and the validity of indirect questioning.
Journal of Consumer Research, 20(2), 303e315. https://doi.org/10.1086/209351.
Education and Psychology, 610 E. University Ave, Ann Arbor, MI
Fredricks, J. A., Reschly, A. L., & Christenson, S. L. (2019). Interventions for student
48109; Email: smckella@umich.edu. engagement: Overview and state of the field. In J. A. Fredricks, A. L. Reschly, &
S. L. Christenson (Eds.), Handbook of student engagement interventions (pp.
1e11). https://doi.org/10.1016/B978-0-12-813413-9.00001-2.
Appendix A. Supplementary data Friedel, J. M., Cortina, K. S., Turner, J. C., & Midgley, C. (2010). Changes in efficacy
beliefs in mathematics across the transition to middle school: Examining the
effects of perceived teacher and parent goal emphases. Journal of Educational
Supplementary data to this article can be found online at
Psychology, 102(1), 102e114. https://doi.org/10.1037/a0017590.
https://doi.org/10.1016/j.tate.2019.102936. Furrer, C., & Skinner, E. (2003). Sense of relatedness as a factor in children’s aca-
demic engagement and performance. Journal of Educational Psychology, 95(1),
148e162. https://doi.org/10.1037/0022-0663.95.1.148.
References Gonida, E. N., Voulala, K., & Kiosseoglou, G. (2009). Students’ achievement goal
orientations and their behavioral and emotional engagement: Co-examining
Abbott-Chapman, J., Martin, K., Ollington, N., Venn, A., Dwyer, T., & Gall, S. (2014). the role of perceived school goal structures and parent goals during adoles-
The longitudinal association of childhood school engagement with adult cence. Learning and Individual Differences, 19(1), 53e60. https://doi.org/10.1016/
educational and occupational achievement: Findings from an Australian na- j.lindif.2008.04.002.
tional study. British Educational Research Journal, 40(1), 102e120. https:// Hafen, C. A., Hamre, B. K., Allen, J. P., Bell, C. A., Gitomer, D. H., & Pianta, R. C. (2015).
doi.org/10.1002/berj.3031. Teaching through interactions in secondary school classrooms: Revisiting the
Allen, J., Gregory, A., Mikami, A., Lun, J., Hamre, B. K., & Pianta, R. (2013). Obser- factor structure and practical application of the classroom assessment scoring
vations of effective teacher-student interactions in secondary school class- systemesecondary. The Journal of Early Adolescence, 35(5e6), 651e680. https://
rooms: Predicting student achievement with the classroom assessment scoring doi.org/10.1177/0272431614537117.
system–secondary. School Psychology Review, 42(1), 76e97. Hamre, B. K., Pianta, R. C., Downer, J. T., DeCoster, J., Mashburn, A. J., Jones, S. M.,
Ames, C. (1992). Classrooms: Goals, structures, and student motivation. Journal of et al. (2013). Teaching through interactions. The Elementary School Journal,
Educational Psychology, 84(3), 261e271. https://doi.org/10.1037/0022-0663.84. 113(4), 461e487.
3.261. Jimerson, S. R., Campos, E., & Greif, J. L. (2003). Toward an understanding of defi-
Bihler, L.-M., Agache, A., Kohl, K., Willard, J. A., & Leyendecker, B. (2018). Factor nitions and measures of school engagement and related terms. California School
analysis of the classroom assessment scoring system replicates the three Psychologist, 8(1), 7e27. https://doi.org/10.1007/BF03340893.
domain structure and reveals no support for the bifactor model in German Kindermann, T. A. (2016). Peer group influences on students’ academic motivation.
preschools. Frontiers in Psychology, 9. https://doi.org/10.3389/fpsyg.2018.01232. In K. R. Wentzel, & G. B. Ramani (Eds.), Handbook of social influences in school
Brophy, J. E. (2010). Motivating students to learn (3rd ed.). New York, NY: Routledge contexts: Social-emotional, motivation, and cognitive outcomes (pp. 31e47).
https://doi-org.proxy.lib.umich.edu/10.4324/9780203858318. NewYork, NY: Routledge.
Brophy, J., & Good, T. (1986). Teacher behavior and student achievement. In Lawson, M. A., & Lawson, H. A. (2013). New conceptual frameworks for student
M. C. Wittrock (Ed.), Handbook of research on teaching (3rd ed., pp. 376e391) engagement research, policy, and practice. Review of Educational Research,
(New York: McMillan). 83(3), 432e479.
Burniske, J., & Meibaum, D. (2012). The use of student perceptual data as a measure of Lei, H., Cui, Y., & Zhou, W. (2018). Relationships between student engagement and
teaching effectiveness (Briefing paper from the Texas Comprehensive Center). academic achievement: A meta-analysis. Social Behavior and Personality: Inter-
Austin, TX: SEDL. Retrieved from http://www.sedl.org/txcc/resources/briefs/ national Journal, 46(3), 517e528. https://doi.org/10.2224/sbp.7054.
number_8/index.php. Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports
Chase, P. A., Warren, D. J. A., & Lerner, R. M. (2015). School engagement, academic on mental processes. Psychological Review, 84(3), 231e259. https://doi.org/10.
achievement, and positive youth development. In P. E. Bowers, J. G. Geldhof, 1037/0033-295X.84.3.231.
K. S. Johnson, J. L. Hilliard, M. R. Hershberg, V. J. Lerner, et al. (Eds.), Promoting Pianta, R. C., Belsky, J., Vandergrift, N., Houts, R., & Morrison, F. J. (2008). Classroom
positive youth development: Lessons from the 4-H study (pp. 57e70). Cham, effects on children’s achievement trajectories in elementary school. American
Switzerland: Springer. Educational Research Journal, 45(2), 365e397. https://doi.org/10.3102/
Cochran-Smith, M. (2016). Teaching and teacher education: Absence and presence 0002831207308230.
in AERA presidential addresses. Educational Researcher, 45(2), 92e99. https:// Pianta, R. C., & Hamre, B. K. (2009). Conceptualization, measurement, and
doi.org/10.3102/0013189X16639040. improvement of classroom processes: Standardized observation can leverage
Cochran-Smith, M., & Lytle, S. L. (1990). Research on teaching and teacher research: capacity. Educational Researcher, 38(2), 109e119. https://doi.org/10.3102/
The issues that divide. Educational Researcher, 19(2), 2e11. 0013189X09332374.
Connell, J. P., & Wellborn, J. G. (1991). Competence, autonomy, and relatedness: A Pianta, R. C., Hamre, B. K., & Allen, J. P. (2012). Teacher-student relationships and
motivational analysis of self-system processes. In M. R. Gunnar, & L. A. Sroufe engagement: Conceptualizing, measuring, and improving the capacity of
(Eds.), Self processes and development (pp. 43e77). Hillsdale, NJ: Lawrence Erl- classroom interactions. In S. L. Christenson, A. L. Reschly, & C. Wylie (Eds.),
baum Associates, Inc. Handbook of research on student engagement (pp. 365e386). New York, NY, US:
Covington, M. V. (2000). Goal theory, motivation, and school achievement: An Springer. https://doi.org/10.1007/978-1-4614-2018-7_17.
integrative review. Annual Review of Psychology, 51(1), 171e200. https://doi.org/ Pianta, R. C., Hamre, B. K., & Mintz, S. (2010). Classroom assessment scoring system
10.1146/annurev.psych.51.1.171. (CLASS): Elementary 4e6 manual. Baltimore, MD: Paul H. Brooks.
Danielson, C. (2007). Enhancing professional practice: A framework for teaching (2nd Pianta, R. C., La Paro, K. M., & Hamre, B. K. (2008). Classroom Assessment Scoring
ed.). Alexandria, VA: Association for Supervision and Curriculum Development. System (CLASS). Charlottesville: VA: University of Virginia.
Danielson, C. (2011). Evaluations that help teachers learn. Educational Leadership, Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common
68(4), 35e39. method biases in behavioral research: A critical review of the literature and
Deemer, S., & Hanich, L. B. (2005). Using achievement goal theory to translate recommended remedies. Journal of Applied Psychology, 88(5), 879e903. https://
evidence-based principles into practice in educational psychology. The Clearing doi.org/10.1037/0021-9010.88.5.879.
House, 78(5), 197e201. Pressley, M., Roehrig, A. D., Raphael, L., Dolezal, S., Bohn, C., Mohan, L., & Hogan, K.
Doyle, W. (1977). Paradigms for research on teacher effectiveness. Review of (2003). Teaching processes in elementary and secondary education. In
Research in Education, 5(1), 163e198. https://doi.org/10.3102/ W. M. Reynolds, & G. E. Miller (Eds.), Handbook of psychology: Educational
0091732X005001163. psychology (pp. 153e175). Hoboken, NJ, US: John Wiley & Sons Inc.. https://doi.
Eccles, J. S. (2016). Engagement: Where to next? Learning and Instruction, 43, 71e75. org/10.1002/0471264385.wei0708
https://doi.org/10.1016/j.learninstruc.2016.02.003. Quin, D. (2017). Longitudinal and contextual associations between teacherestudent
Eccles, J. S., Midgley, C., Wigfield, A., Buchanan, C. M., Reuman, D., Flanagan, C., et al. relationships and student engagement: A systematic review. Review of Educa-
(1993). Development during adolescence: The impact of stage-environment fit tional Research, 87(2), 345e387. https://doi.org/10.3102/0034654316669434.
on young adolescents’ experiences in schools and in families. American Psy- Quin, D., Hemphill, S. A., & Heerde, J. A. (2017). Associations between teaching
chologist, 48(2), 90e101. https://doi.org/10.1037/0003-066X.48.2.90. quality and secondary students’ behavioral, emotional, and cognitive engage-
Eccles, J. S., & Roeser, R. W. (2011). Schools as developmental contexts during ment in school. Social Psychology of Education: International Journal, 20(4),
adolescence: Schools as developmental contexts. Journal of Research on 807e829. https://doi.org/10.1007/s11218-017-9401-2.
Adolescence, 21(1), 225e241. https://doi.org/10.1111/j.1532-7795.2010.00725.x. Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and
Epstein, J. (1988). Effective schools or effective students? Dealing with diversity. In data analysis methods. SAGE Publications Inc.
R. Haskins, & B. MacRae (Eds.), Policies for America’s public schools: Teacher eq- Reschly, A. L., & Christenson, S. L. (2012). Jingle, jangle, and conceptual haziness:
uity indicators (pp. 89e126). Norwood, NJ: Ablex. Evolution and future directions of the engagement construct. In
Finn, J. D. (1989). Withdrawing from school. Review of Educational Research, 59(2), S. L. Christenson, A. L. Reschly, & C. Wylie (Eds.), Handbook of research on student
117e142. https://doi.org/10.3102/00346543059002117. engagement (pp. 3e19). Retrieved from http://link.springer.com/10.1007/978-1-
Finn, J. D., & Zimmer, K. S. (2012). Student engagement: What is it? Why does it 4614-2018-7.
matter? In S. L. Christenson, A. L. Reschly, & C. Wylie (Eds.), Handbook of research Reyes, M. R., Brackett, M. A., Rivers, S. E., White, M., & Salovey, P. (2012). Classroom
on student engagement (pp. 97e131). New York, NY, US: Springer Science þ emotional climate, student engagement, and academic achievement. Journal of
Business Media. https://doi.org/10.1007/978-1-4614-2018-7_5. Educational Psychology, 104(3), 700e712. https://doi.org/10.1037/a0027268.
S.E. McKellar et al. / Teaching and Teacher Education 89 (2020) 102936 11

Rimm-Kaufman, S. E., Baroody, A. E., Larsen, R. A. A., Curby, T. W., & Abry, T. (2015). engagement and disaffection: Conceptualization and assessment of children’s
To what extent do teacherestudent interaction quality and student gender behavioral and emotional participation in academic activities in the classroom.
contribute to fifth graders’ engagement in mathematics learning? Journal of Educational and Psychological Measurement, 69(3), 493e525. https://doi.org/10.
Educational Psychology, 107(1), 170e185. https://doi.org/10.1037/a0037252. 1177/0013164408323233.
Roehrig, G. H., Johnson, C. C., Moore, T. J., & Bryan, L. A. (2015). Integrated STEM Ulmanen, S., Soini, T., Pyha €lto
€, K., & Pietarinen, J. (2014). Strategies for academic
education. In C. C. Johnson, & T. A. Sondergeld (Eds.), STEM road map (pp. engagement perceived by Finnish sixth and eighth graders. Cambridge Journal of
35e50). Routledge. Education, 44(3), 425e443. https://doi.org/10.1080/0305764X.2014.921281.
Ruzek, E. A., Hafen, C. A., Allen, J. P., Gregory, A., Mikami, A. Y., & Pianta, R. C. (2016). Virtanen, T. E., Pakarinen, E., Lerkkanen, M.-K., Poikkeus, A.-M., Siekkinen, M., &
How teacher emotional support motivates students: The mediating roles of Nurmi, J.-E. (2018). A validation study of classroom assessment scoring
perceived peer relatedness, autonomy support, and competence. Learning and systemesecondary in the Finnish school context. The Journal of Early Adoles-
Instruction, 42, 95e103. https://doi.org/10.1016/j.learninstruc.2016.01.004. cence, 38(6), 849e880. https://doi.org/10.1177/0272431617699944.
Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the facilitation of Wang, M. T., & Eccles, J. S. (2012). Social support matters: Longitudinal effects of
intrinsic motivation, social development, and well-being. American Psychologist, social support on three dimensions of school engagement from middle to high
55(1), 68e78. https://doi.org/10.1037/0003-066X.55.1.68. school. Child Development, 83, 877e895. https://doi.org/10.1111/j.1467-8624.
Salmela-Aro, K., & Upadaya, K. (2012). The schoolwork engagement inventory. Eu- 2012.01745.x.
ropean Journal of Psychological Assessment, 28, 60e67. https://doi.org/10.1027/ Westergård, E., Ertesvåg, S. K., & Rafaelsen, F. (2018). A preliminary validity of the
1015-5759/a000091. classroom assessment scoring system in Norwegian lower-secondary schools.
Shin, H., & Ryan, A. M. (2014). Early adolescent friendships and academic adjust- Scandinavian Journal of Educational Research, 1e19. https://doi.org/10.1080/
ment: Examining selection and influence processes with longitudinal social 00313831.2017.1415964.
network analysis. Developmental Psychology, 50(11), 2462e2472. https://doi. Wetzel, C. G., Wilson, T. D., & Kort, J. (1981). The halo effect revisited: Forewarned is
org/10.1037/a0037922. not forearmed. Journal of Experimental Social Psychology, 17(4), 427e439.
Skinner, E. A., & Belmont, M. J. (1993). Motivation in the classroom: Reciprocal ef- https://doi.org/10.1016/0022-1031(81)90049-4.
fects of teacher behavior and student engagement across the school year. Weyns, T., Colpin, H., De Laet, S., Engels, M., & Verschueren, K. (2018). Teacher
Journal of Educational Psychology, 85(4), 571e581. https://doi.org/10.1037/0022- support, peer acceptance, and engagement in the classroom: A three-wave
0663.85.4.571. longitudinal study in late childhood. Journal of Youth and Adolescence, 47(6),
Skinner, E. A., Kindermann, T. A., & Furrer, C. J. (2009). A motivational perspective on 1139e1150. https://doi.org/10.1007/s10964-017-0774-5.

You might also like