Content uploaded by Johnmarshall Reeve
Author content
All content in this area was uploaded by Johnmarshall Reeve
Content may be subject to copyright.
Engaging Students in Learning Activities: It Is Not Autonomy Support or
Structure but Autonomy Support and Structure
Hyungshim Jang
University of Wisconsin—Milwaukee
Johnmarshall Reeve
Korea University
Edward L. Deci
University of Rochester
We investigated 2 engagement-fostering aspects of teachers’ instructional styles—autonomy support
and structure—and hypothesized that students’ engagement would be highest when teachers pro-
vided high levels of both. Trained observers rated teachers’ instructional styles and students’
behavioral engagement in 133 public high school classrooms in the Midwest, and 1,584 students in
Grades 9–11 reported their subjective engagement. Correlational and hierarchical linear modeling
analyses showed 3 results: (a) Autonomy support and structure were positively correlated, (b)
autonomy support and structure both predicted students’ behavioral engagement, and (c) only
autonomy support was a unique predictor of students’ self-reported engagement. We discuss, first,
how these findings help illuminate the relations between autonomy support and structure as 2
complementary, rather than antagonistic or curvilinear, engagement-fostering aspects of teachers’
instructional styles and, second, the somewhat different results obtained for the behavioral versus
self-report measures of students’ classroom engagement.
Keywords: autonomy support, structure, engagement, self-determination theory, teacher behavior
Engagement expresses the behavioral intensity and emotional
quality of a student’s active involvement during a learning
activity (Fredricks, Blumenfeld, & Paris, 2004; Skinner, Furrer,
Marchand, & Kindermann, 2008; Wellborn, 1991). In class-
room settings, engagement is particularly important because it
functions as a behavioral pathway through which students’
motivational processes contribute to their subsequent learning
and development (Connell & Wellborn, 1991), including the
skills they develop (Skinner & Belmont, 1993) and the grades
they make (Finn & Rock, 1997). In contrast, disengaged stu-
dents are distracted, passive, do not try hard, give up easily in
the face of challenge or difficulty, express negative emotions,
fail to plan or monitor their work, and generally withdraw (e.g.,
“When I am in class, I usually think about other things”;
Skinner & Belmont, 1993).
When students engage in classroom learning, there is almost
always some aspect of the teacher’s behavior that plays a role in
the initiation and regulation of the engagement. To better under-
stand students’ academic engagement, many researchers have in-
vestigated supportive sociocontextual factors (Skinner et al.,
2008), such as teachers’ instructional style, which is generally con-
ceptualized as a stable pattern in a teacher’s methods of instruction,
classroom management, and interpersonal style with students
(Schultz, 1982). In the present study, we examined two particular
engagement-promoting aspects of teachers’ instructional style: provi-
sion of autonomy support (vs. being controlling) and provision of
structure (vs. chaos). According to the existing literature, when teach-
ers focus on supporting students’ autonomous motives (e.g., interests,
needs, preferences, personal goals) to guide their learning and activity,
these instructional acts support students’ engagement by presenting
interesting and relevant learning activities, providing optimal chal-
lenges, highlighting meaningful learning goals, and supporting stu-
dents’ volitional endorsement of classroom behaviors (Assor, Kaplan,
& Roth, 2002; Reeve & Jang, 2006; Reeve, Jang, Carrell, Jeon, &
Barch, 2004). Further, when teachers provide high structure by com-
municating clear expectations and framing students’ learning activity
with explicit directions and guidance, these instructional acts support
students’ engagement by keeping students on task, managing their
behavior, and avoiding chaos during transitions (Skinner & Belmont,
1993; Tucker et al., 2002).
While both autonomy support and structure make important
contributions to supporting students’ classroom engagement, the
nature of the relation between them has been portrayed rather
confusingly in the literature in at least three different ways—as
being antagonistic, curvilinear, and independent. These differing
portrayals of the relation between provision of autonomy support
and provision of structure call for a deeper understanding of their
interrelation as the multiple portrayals mislead and confuse teach-
ers and researchers alike. To shed light on the relation between
Hyungshim Jang, Department of Educational Psychology, University of
Wisconsin—Milwaukee; Johnmarshall Reeve, World Class University
Project Group, Department of Education, Korea University, Seoul, Korea;
Edward L. Deci, Department of Psychology, University of Rochester.
This research was supported by the Grant for Excellence in Urban
Education awarded to Hyungshim Jang.
Correspondence concerning this article should be addressed to
Hyungshim Jang, who is now at the Department of Education, #325
West Building, Inha University, Incheon 402-751, South Korea. E-mail:
hjang@inha.ac.kr
Journal of Educational Psychology © 2010 American Psychological Association
2010, Vol. 102, No. 3, 588 – 600 0022-0663/10/$12.00 DOI: 10.1037/a0019682
588
teacher-provided autonomy support and teacher-provided struc-
ture, we pursued two research questions. First, how do these two
engagement-fostering aspects of a teacher’s instructional style
relate to one another in teachers’ naturally occurring instruction?
Second, do both autonomy support and structure contribute posi-
tively and uniquely to the support of students’ engagement during
learning activities? Before presenting our hypotheses, we, in the
following sections, discuss autonomy support and structure as two
different engagement-fostering aspects of teachers’ instructional
styles and then discuss their potential interrelation.
Autonomy Support
Self-determination theory (Ryan & Deci, 2000, 2002) proposes
that a teacher’s instructional style can be conceptualized along a
continuum that ranges from highly controlling to highly autonomy
supportive (Deci, Schwartz, Sheinman, & Ryan, 1981). In general,
teachers who adopt an autonomy-supportive style engage students
by facilitating an on-going congruence between students’ autono-
mous sources of motivation and their moment-to-moment class-
room activity. Autonomy-supportive teachers facilitate students’
personal autonomy by taking the students’ perspective; identifying
and nurturing the students’ needs, interests, and preferences; pro-
viding optimal challenges; highlighting meaningful learning goals;
and presenting interesting, relevant, and enriched activities.
More specifically, what autonomy-supportive teachers say
and do to engage students during learning activities can be
characterized by three categories of instructional behavior: (a)
nurture inner motivational resources, (b) rely on noncontrolling
informational language, and (c) acknowledge the students’ per-
spective and feelings (Deci, Eghrari, Patrick, & Leone, 1994;
Mageau & Vallerand, 2003; Reeve & Jang, 2006; Reeve, Jang,
et al., 2004; Ryan & La Guardia, 1999). When autonomy-
supportive teachers nurture students’ inner motivational re-
sources, they create opportunities for students to take the ini-
tiative during learning activities by building instruction around
students’ interests, preferences, personal goals, choice making,
and sense of challenge and curiosity, rather than relying on
external sources of motivation such as incentives, conse-
quences, directives, and deadlines. When autonomy-supportive
teachers rely on noncontrolling informational language, they
provide explanatory rationales for requested tasks and commu-
nicate through messages that are informative, flexible, and rich
in competence-related information, rather than neglecting ra-
tionales and by communicating through messages that are eval-
uative, controlling, pressuring, or even rigidly coercive. When
autonomy-supportive teachers acknowledge the students’ per-
spectives and feelings, they consider and communicate a valu-
ing of the students’ perspectives during learning activities,
inquire about and acknowledge students’ feelings, and accept
students’ expressions of negative affect as a potentially valid
reaction to classroom demands, imposed structures, and the
presentation of uninteresting or devalued activities.
Each of these aspects of what autonomy-supportive teachers say
and do during instruction is important because empirical research
shows that students with autonomy-supportive teachers, compared
to students’ with controlling teachers, display an impressive and
wide range of positive educational outcomes (Reeve, 2009; Reeve,
Deci, & Ryan, 2004), including enhanced classroom engagement
(Jang, 2008; Reeve, Jang, et al., 2004). For instance, Reeve, Jang,
and colleagues (2004) showed that raters’ scoring of high school
students’ engagement increased markedly after their teachers par-
ticipated in a training program on how to nurture inner motiva-
tional resources, rely on informational language, provide explan-
atory rationales, and acknowledge and accept negative feelings. A
series of laboratory studies with college students showed that both
students’ self-reported engagement and rater-scored engagement
were greater when tutors or teachers relied on informational lan-
guage, provided explanatory rationales, and acknowledged nega-
tive feelings (Deci et al., 1994; Jang, 2008; Reeve, Jang, Hardre, &
Omura, 2002). The reason students benefit so widely and so
substantially (e.g., engagement, preference for optimal challenge,
conceptual understanding, grades, psychological well-being) from
their exposure to teachers with an autonomy-supportive motivating
style is because such a style supports students’ internal perceived
locus of causality, experience of volition, and sense of choice
during learning activities (i.e., it supports students’ autonomous
motivation; Reeve, 2009).
Structure
Another aspect of a teacher’s instructional style that has been
used to promote students’ engagement is structure (Connell &
Wellborn, 1991; Skinner & Belmont, 1993; Skinner et al., 2008;
Skinner, Zimmer-Gembeck, & Connell, 1998). Structure refers to
the amount and clarity of information that teachers provide to
students about expectations and ways of effectively achieving
desired educational outcomes (Skinner & Belmont, 1993; Skinner
et al., 1998). Its opposite is chaos in which teachers are confusing
or contradictory, fail to communicate clear expectations and di-
rections, and ask for outcomes without articulating the means to
attain them.
Teacher-provided structure has been studied extensively within
the classroom management literature as establishing order (Doyle,
1986), introducing procedures (Emmer, Evertson, & Anderson,
1980), communicating policies about how to get things done (e.g.,
how to give completed work to the teacher; Carter & Doyle, 2006),
and minimizing misbehavior while encouraging engagement and
achievement (Brophy, 2006). It has also been examined in the
literature on promoting students’ social and cognitive development
(Huston-Stein, Friedrich-Cofer, & Susman, 1977) and on imple-
menting “structured conversations” within peer learning
(O’Donnell, 2006). For those who study structure from a motiva-
tional point of view, teacher-provided structure further helps stu-
dents to develop a sense of perceived control over school out-
comes—that is, to develop perceived competence, an internal
locus of control, mastery motivation rather than helplessness,
self-efficacy, and an optimistic attributional style (Skinner, 1995;
Skinner et al., 2008). Hence, when taken as a whole, teachers
provide structure by clearly communicating expectations and di-
rections, taking the lead during some instructional activities, pro-
viding strong guidance during the lesson, providing step-by-step
directions when needed, scheduling student activities, marking the
boundaries of activities and orchestrating the transitions between
them, offering task-focused and personal control-enhancing feed-
589
AUTONOMY SUPPORT AND STRUCTURE
back, and providing consistency in the lesson (Brophy, 2006;
Doyle, 2006; Huston-Stein et al., 1977; Skinner & Belmont, 1993).
More specifically, what structured teachers say and do can be
characterized by three categories of instructional behavior: (a)
present clear, understandable, explicit, and detailed directions; (b)
offer a program of action to guide students’ ongoing activity; and
(c) offer constructive feedback on how students can gain control
over valued outcomes (Brophy, 1986; Skinner, 1995; Skinner &
Belmont, 1993; Skinner et al., 1998). When teachers establish
clear and understandable directions, they establish clear expecta-
tions with respect to students’ future behavior and prescribe ways
for students to manage their moment-to-moment activity during a
forthcoming learning activity. When teachers offer strong guid-
ance, they provide students with the leadership and the scaffolding
needed for students to instigate and maintain effort toward achiev-
ing their plans, goals, and learning objectives. When teachers offer
constructive feedback, they help students diagnose and build on
their skills and sense of competence. In these ways, teachers can
prescribe to students what is expected of them and help them come
to understand “what it takes to do well in school and whether I’ve
got it” (Skinner, Wellborn, & Connell, 1990, p. 22).
These aspects of what structured teachers say and do during
instruction are important because empirical research shows that
students with structured teachers, compared to chaotic or laissez-
faire ones, display positive educational outcomes (Brooks, 1985;
Brophy, 1986, 2006; Brophy & Good, 1986; Evertson & Wein-
stein, 2006; Huston-Stein et al., 1977), including enhanced class-
room engagement (Skinner & Belmont, 1993; Tucker et al., 2002).
For instance, Skinner and Belmont (1999) conducted time-lagged
analyses to show that elementary school students (Grades 3–5)
showed enhanced engagement in the fall when their teachers
provided highly structured learning environments in the spring.
Skinner and her colleagues (2008) showed a similar result with
older children (Grades 4 –7). Sierens, Vansteenkiste, Goossens,
Soenens, and Dochy (2009) showed that teacher-provided struc-
ture was associated with high levels of high school students’
self-regulating management of their classroom engagement. The
reason students benefit from their exposure to teachers with a
structured instructional style is because such a style supports
students’ perceptions of competence, perceived control over val-
ued outcomes, and self-regulated learning strategies (Skinner et al.,
1998; Sierens et al., 2009).
The term structure is often used in discussions of education
generally and of education of students with special needs in
particular. However, in many cases, there is a confluence between
the concepts of structure and control. That is, people use the term
structure to refer to demands, insistences, sanctions, and rigid
rules. We emphasize, in contrast, that whereas structure can be
used in controlling ways and often is, control is by no means
essential to structure. Indeed, we hypothesized that when structure
is used in controlling ways, it will be detrimental to, rather than
facilitative of, student engagement; whereas when it is used in
autonomy-supportive ways, it will be facilitative of engagement.
As an illustration of the confluence between structure and con-
trol—and of the more general point that structure can be provided
in either controlling or autonomy-supportive ways—Koestner,
Ryan, Bernieri, and Holt (1984) introduced rules to structure
children’s painting activity. When the rules were imposed in a
controlling way (without explanatory rationale, without acknowl-
edging the children’s perspective and feelings), children’s engage-
ment in the painting activity was lower than it was for children
randomly assigned to a control group without the rules. When
these same rules were provided to the children in an autonomy-
supportive way (with explanatory rationales, with the acknowledg-
ment of the children’s perceptive and feelings), however, children
showed no decrease in engagement compared to the children in the
control group and greater engagement than children who received
the rules presented in a controlling way. Beyond rules, other
research has shown essentially the same effects for other elements
of structure—including communications, goals, expectations, re-
wards, and feedback (Deci, Koestner, & Ryan, 1999; Grolnick &
Ryan, 1987; Jang, 2009; Ryan, Mims, & Koestner, 1983; Schuh,
2004)—namely, that any element of structure can be presented in
either an autonomy-supportive or controlling way and, also, that
student outcomes are enhanced when the element of structure is
presented in an autonomy-supportive way, yet diminished when
the element of structure is presented in a controlling way (Sierens
et al., 2009).
Relation Between Autonomy Support and Structure
Teacher-provided autonomy support and structure both make
important contributions to supporting students’ classroom engage-
ment. The nature of their relation with each other, however, has
been portrayed in the literature in at least three different ways.
First, autonomy support and structure have been conceptualized
as being both antagonistic to each other and opposites in their
effects on students’ engagement. Some argue that the various
elements of classroom structure, such as rules, interfere with
teachers’ provision of choice, spontaneity, and the cultivation of
personal responsibility (Daniels & Bizar, 1998). This view of the
relation between structure and autonomy support proposes that a
greater implementation of one aspect of a teacher’s style neces-
sarily leads to a lesser implementation of the other. Our view,
however, is that this supposed antagonism between autonomy
support and structure is not inherent in the concepts, but rather
results from the implicit inclusion of control in some approaches to
using structure.
Second, some have proposed that there is a curvilinear relation
between structure and autonomy support, with teachers moderate
in structure being highest in autonomy support, and also that high
autonomy support with moderate structure yields optimal engage-
ment (deCharms, 1984). According to deCharms (1984), when
teachers provide too little structure, students fail to develop the
prerequisite skills they need to experience engagement-fostering
personal causation (i.e., high perceived autonomy, high perceived
competence). When teachers provide too much structure, students
may learn task-relevant skills but come to hate the experience if it
is overly scripted and hence void of a sense of personal causation.
It is only with moderate structure—some supervision but not
totalitarian supervision—that students learn both prerequisite skills
and the experience of personal causation that promotes engage-
ment.
Here too we believe there may be some confusion between
structure and control. Specifically, to provide structure when it is
not needed, to be continually reiterating instructions and guidelines
590
JANG, REEVE, AND DECI
that are already understood, is not to be high in structure but is,
instead, to be controlling and intrusive.
1
Third, autonomy support and structure have been conceptual-
ized as two independent aspects of teachers’ instructional styles,
each of which can contribute support to students’ motivation and
engagement (Connell & Wellborn, 1991; Skinner & Belmont,
1993; Tucker et al., 2002). Although this has not been empirically
tested, the authors speculated that teachers’ provision of high
structure (i.e., regulations and guidelines) can be combined with
either freedom and encouragement (i.e., high autonomy support) or
with pressure and coercion (i.e., low autonomy support). Thus,
autonomy support and structure can be viewed as two conceptually
distinct aspects of teachers’ styles such that some teachers will
display high levels of both styles, other teachers will display low
levels of both styles, while still others will display a high level of
one style with a low level of the other.
As mentioned earlier, structure and control have often been
confused, but we herein suggest that, although they are separate
concepts, the optimal learning environment for classroom engage-
ment involves structure provided in an autonomy supportive way.
Thus, we predicted that engagement would be highest when teach-
ers provide high levels of both autonomy support and structure.
Purpose of the Study and Hypotheses
Although it is interesting to understand how different theoretical
perspectives approach the integration of autonomy support and
structure in teachers’ classroom practices, it is important to use an
empirical approach to examine how they relate both to one another
and to students’ classroom engagement. Teachers’ instructional
styles differ, with some teachers being high on both autonomy
support and structure, some being low on both, and some being
high on one and low on the other. In the present study, we
examined the relation between these two aspects of teachers’
instructional styles and the relations of each to students’ classroom
engagement. Specifically, we investigated (a) whether these two
aspects of teachers’ instructional styles are either positively or
negatively correlated, relate to one another in a linear or curvilin-
ear way, or are independent of one another, and (b) how each of
these aspects of teachers’ styles predicts students’ engagement. In
line with Skinner and Belmont (1993), we hypothesized that both
aspects of teachers’ instructional styles would function as unique,
positive supports for students’ engagement. To understand how
teacher-provided autonomy support and structure relate to stu-
dents’ classroom engagement, we assessed engagement with both
an objective behavioral measure (scored by trained raters) and a
subjective self-report measure (self-reported by students). By im-
plementing both measures, we consciously focused on testing the
extent to which both autonomy support and structure might
uniquely predict various indicators of students’ classroom engage-
ment. Further, acknowledging that there is some controversy in the
literature as to which is the better and more meaningful way to
assess students’ engagement (e.g., Rosenholtz & Simpson, 1984),
we also hoped our findings would shed light on this issue, as none
of the studies reviewed above assessed both self-report and rater-
scored engagement.
Method
Participants
Participants were 133 teachers and their 2,523 students from
nine public high schools in the Midwest. Of the 133 observed
classrooms, 38% were ninth grade, 32% were 10th grade, and 30%
were 11th grade. Observed teachers were teaching either English,
math, science, or social studies. Teachers included 48% female and
52% male; they were 75% Caucasian, 13% Hispanic, 8% African
American, and 4% Asian American; this was comparable to the
gender and ethnic makeup of teachers at the nine schools and in the
school district. On average, teachers had 13.1 years of teaching
experience, and they taught to a class size of 19.0 students who
attended class on the day of our classroom visit. In each school, at
least 95% of the teachers were certified. The nine schools had an
average student body of 1,272 students (range ⫽ 579 –1,971), a
teacher–student ratio of 1:17.2 (range ⫽ 1:15–1:19), and 20% of
students who received free lunch (range ⫽ 1%– 44%). Students
included 54% female and 46% male; they were 47% Caucasian,
45% African American, 6% Hispanic, and 2% Asian American.
No teacher was rated on more than one occasion, and any student
who completed the engagement questionnaire in a previous class
did not complete the questionnaire on a second occasion.
Procedure
After gaining the school principal’s permission to observe class-
rooms and also the teachers’ permission to visit one of their
classrooms, we preannounced to teachers only that graduate stu-
dents (i.e., raters) associated with the local university would come
to one of their classes to observe the classroom dynamics. While
teachers gave a priori permission to visit their classrooms, they did
not know which regularly scheduled class would be visited. This
feature of the study was included so that we could maximize the
raters’ opportunity to observe each teacher’s naturally occurring
instructional style. Which teachers were visited and during which
class periods they were visited were determined at the beginning of
the day by a schedule prepared by an assistant to the school
principal. When the raters arrived minutes before the beginning of
a class session, they asked teachers to reserve, if possible, the last
2 min of class time so the raters could administer to students a
questionnaire with four briefly worded items. Each class lasted
between 52 and 55 min. Eighty-four of the 133 teachers (63%)
were able to allocate the last 2 min of class time to the question-
naire, while the rest of the teachers ran out of available class time
due to events such as extended question-and-answer periods,
homework assignments, or achieving adequate closure on the
learning session. In the 84 classrooms that were able to allocate the
time, the questionnaires were introduced, administered, and col-
lected by the raters. Students were instructed to remain anonymous
1
We recognize that there may be some circumstances in which educa
-
tional settings are overstructured, leading to less engagement. In a setting
where all students are highly autonomously motivated and interesting
activities are available to them, for instance, low structure may be optimal
to facilitate engagement and creativity. Still, as a general principle in most
classroom settings, we expected high levels of structure to be associated
with greater engagement.
591
AUTONOMY SUPPORT AND STRUCTURE
(i.e., not to write their name anywhere on the questionnaire), and
they were informed that their responses were confidential. On the
questionnaire, students were asked to rate their class experience
during the just-completed lesson.
Raters
Five raters who were trained with classroom observational skills
and who were blind to the study’s purpose and hypotheses rated
teachers’ provision of autonomy support, teachers’ provision of
structure, and students’ collective engagement (and a few addi-
tional items to help disguise the purpose of the rating sheet). For
56 (42%) of the classroom visits, raters worked in pairs. In doing
so, they sat nonintrusively in the back of the classroom and made
independent ratings. We included a pair of raters within 56 class-
room visits to estimate interrater reliabilities for each measure. For
77 (58%) of the classroom visits, only one rater scored the ratings.
Rating Sheet
The rating sheet featured three clusters of items to assess the
measures of teacher’s autonomy support, teacher’s structure, and
students’ behavioral engagement. Each item was scored using a
1–7 Likert scale. Items were taken and slightly modified from an
existing measure that was used and validated previously in high
school classrooms by Reeve, Jang, and colleagues (2004).
Autonomy support. As shown in Figure 1, raters scored three
autonomy-supportive instructional behaviors—nurtures inner mo-
tivational resources, relies on informational language, and ac-
knowledges and accepts students’ negative affect (range of inter-
rater rs ⫽ .72–.88). The bipolar descriptors for these three
behaviors were as follows: relies on extrinsic sources of motiva-
tion versus nurtures inner motivational resources, controlling lan-
guage versus informational language, and counters and tries to
change students’ negative affect versus acknowledges and accepts
students’ negative affect. Collectively, these behaviors reflect self-
determination theory’s conceptualization of autonomy support
(Ryan & Deci, 2000), and each instructional behavior has been
validated as an act of autonomy support in prior classroom use
(Reeve, Jang, et al., 2004). As found in previous research, the three
autonomy-supportive acts of instruction were positively intercor-
related, and we therefore averaged them into a single overall
teacher-provided autonomy support score (␣⫽.81).
Structure. As shown in Figure 2, raters scored three instruc-
tional behaviors to represent teacher-provided structure— clear
and explicit directions, strong guidance during the lesson, and
constructive feedback (range of interrater rs ⫽ .84 –.88). The
bipolar descriptors for these three behaviors were as follows:
during the introduction, absent, unclear, ambiguous, confusing
directions versus clear, understandable, explicit, detailed direc-
tions; during the lesson, weak guidance versus strong guidance;
and during feedback, no feedback or ambiguous feedback versus
skill-building and instructive feedback. Collectively, these behav-
iors reflect Skinner’s motivation-based conceptualization of struc-
ture, and each instructional behavior has been validated as an act
of structure in prior classroom use (Skinner & Belmont, 1993). As
has been found in previous research, the three structure-providing
acts of instruction were positively intercorrelated, and we therefore
averaged them into a single overall teacher-provided structure
score (␣⫽.85).
Differentiating autonomy support from structure. This
study was based on the premise that the construct of teacher-
provided autonomy support was distinct from the construct of
teacher-provided structure. To test the validity of this assumption,
we used structural equation modeling (using LISREL 8.51; Jöres-
kog & Sörbom, 1993) to conduct a confirmatory factor analysis
that tested the extent to which the measured indicators adequately
and uniquely related to their associated latent variables. To eval-
uate model fit, we relied on the chi-square test statistic and the
standardized root-mean-square residual (SRMR). A nonsignificant
chi-square serves as the basic test of whether a hypothesized model
adequately describes the data (Bollen & Long, 1993), though Hu
and Bentler (1999) recommend priority be given to the SRMR
Teacher’s Autonomy Support
Relies on
Extrinsic Sources of Motivation 1 2 3 4
5 6 7 Nurtures Inner Motivational Resources
* Offers Incentives, Consequences, Directives * Interest, Enjoyment, Sense of Challenge
* Makes Assignments, Seeks Compliance * Creates Opportunities for Initiative
Controlling Language 1 2 3 4 5 6 7 Informational Language
* Pressuring, Ego-Involving * Informational, Flexible
* Should, Must, Have to, Got to * Provides Choices, Options
* Neglects Value, Meaning, Use, * Identifies Value, Meaning, Use,
Benefit, Importance of Requests Benefit, Importance of Requests
Counters & Tries to Change 1 2 3 4 5 6 7 Acknowledges & Accepts
Students’ Negative Affect Students’ Negative Affect
* Blocks/Counters Expressions of Negative Affect * Listens Carefully, Openly, Understandingly
* Negative Affect is Not OK, Is Unacceptable, * Accepts Negative Affect,
Is Something to be Changed/Fixed Complaints are OK
Note for Each Rating: Use the bold, underlined 4
as your starting/anchor point.
Figure 1. Rating sheet to score the three teacher-provided autonomy support items.
592
JANG, REEVE, AND DECI
when evaluating the fit of a measurement model (i.e., a confirma-
tory factor analysis), with a value of .08 or lower for the SRMR
considered indicative of a good model fit. The hypothesized two-
factor model fit the data reasonably well, as the chi-square statistic
was significant,
2
(8) ⫽ 20.19, p ⫽ .01, while the SRMR fit index
suggested a good model fit of .04 (also comparative fit index ⫽
.98, nonnormed fit index ⫽ .97). We also compared our hypoth-
esized two-factor model to a one-factor model, and the two-factor
model fit the data significantly better than did the one-factor
model, ⌬
2
(1) ⫽ 173.67, p ⬍ .01. For the two-factor model, each
of the six individual indicators loaded positively and significantly
( p ⬍ .01) on its associated latent factor, including the three
indicators for autonomy support (informational language, ⫽.94;
nurture inner resources, ⫽.89; and acknowledge feelings, ⫽
.75) and the three indicators for structure (strong guidance, ⫽
.96; clear directions, ⫽.95; and instructive feedback, ⫽.78).
The two latent factors intercorrelated positively and significantly
(⫽.59, p ⬍ .01).
Students’ collective behavioral engagement. Raters scored
six engagement-related aspects of students’ behaviors at the class-
room level—attention, effort, verbal participation, persistence,
positive emotion, and voice (interrater rs ranged from .63 to .92).
The bipolar descriptors for the first five of these behaviors were as
follows: dispersed attention versus focused attention; passive,
slow, or minimal effort versus active, quick, and intense effort;
students don’t talk, ask questions, or discuss versus students talk,
ask questions, and discuss; during challenge, failure, or confusion
students give up easily and decrease effort over time versus main-
tain or increase effort over time; bored, disinterested, and flat
emotional tone versus enjoyment, interested, and fun. These five
indicators reflect Skinner’s conceptualization of engagement (e.g.,
Skinner & Belmont, 1993; Skinner et al., 2008). The sixth engage-
ment measure, voice, was based on deCharms conceptualization of
engagement and was calculated as a ratio to index students’
attempts to influence the flow of instruction in a constructive way
(voice ⫽ frequency of students’ influence attempts, divided by
frequency of students’ influence attempts ⫹ frequency of the
teacher’s influence attempts; Koenigs, Fiedler, & deCharms,
1977). Each of these six expressions of students’ collective behav-
ioral engagement has been validated in prior classroom research
(Reeve, Jang, et al., 2004), as each rating has been shown to
produce high interrater reliability, to intercorrelate highly and
positively with the other five engagement ratings (i.e., high reli-
ability), and to be sensitive to teaching variables known to affect
students’ engagement (i.e., high validity). As has been found in
these previous studies, the six indicators of classroom engagement
were positively intercorrelated, so we converted the classroom
rating on each indicator into a z score and averaged them (equally
weighted) to form a single overall collective engagement score for
the group of students in each classroom (␣⫽.92).
Students’ Individual Self-Report Engagement
(Student Questionnaire)
The student questionnaire included four briefly worded items to
assess the extent to which students were engaged during the class
period. We developed the measure to reflect Fredricks et al.’s
(2004) three-component conceptualization of engagement, which
features behavioral, cognitive, and emotional aspects. The ques-
tionnaire began with the stem, “During this class. . .” and included
the following four items: “I paid attention” “I worked very hard”
“I tried to learn as much as I could” and “I enjoyed today’s class.”
The first two items were designed to reflect the behavioral aspects
of engagement, the third item was designed to reflect the cognitive
aspect of engagement, and the fourth item was designed to assess
the emotional aspect of engagement. Each item was scored on a
unipolar 7-point Likert scale that ranged from 1 (not at all true)to
7(extremely true). To create a single score for each student, we
averaged the four items into an individual self-report engagement
score (␣⫽.88).
Teacher’s Structure
During Introduction:
Absent, Unclear, Ambiguous, 1 2 3 4 5 6 7 Clear, Understandable,
Confusing Directions Explicit, Detailed Directions
* “What to do” is Absent, Confusing * “What to do” is Clear
* Poorly Organized * Well Organized
* No Clear Frame for Upcoming Lesson * Frames Upcoming Lesson Well
During Lesson
:
Weak Guidance 1 2 3 4 5 6 7 Strong Guidance
* Little Guidance, Leadership * Much Guidance, Leadership
* No Action Plan, No Goal * Clear Action Plan, Clear Goal
* Few Control-Establishing Hints * Many Control-Establishing Hints
During Feedback
:
Feedback: 1 2 3 4 5 6 7 Feedback:
None, Ambiguous
Skill-Building, Instructive
* None, Off-Task, Rambling * Constructive, Informative
* Competence-Irrelevant Information * Competence-Relevant Information
Note for Each Rating: Use the bold, underlined 4
as your starting/anchor point.
Figure 2. Rating sheet to score the three teacher-provided structure items.
593
AUTONOMY SUPPORT AND STRUCTURE
Statistical Analyses
We used two engagement measures—students’ collective be-
havioral engagement and students’ individual self-report engage-
ment. For students’ collective behavioral engagement, raters
scored the average engagement of groups of students (class size
averaged 19.0 students) sitting in the classrooms of 133 different
teachers. These 133 classrooms were nested within nine schools.
Multilevel modeling (using hierarchical linear modeling [HLM])
was therefore ideally suited to our research question. For some
analyses, on the first level of the multilevel model (between-
teachers level), regression equations were modeled for character-
istics that differed between teachers (autonomy support, structure).
Level 1 predictor variables were centered around their means. At
the second level (between-schools level), regression equations
were modeled for the different schools in which the teachers
taught.
For students’ individual self-report engagement, 1,584 students
individually completed the self-report engagement questionnaire.
These 1,584 responses were nested within 84 classrooms (recall
that students did not complete the questionnaire in the remaining
49 classrooms). Further, the 84 classrooms were nested within the
nine schools. As a check, we compared the rater-scored instruc-
tional styles of the 84 teachers who allocated 2 min for the student
questionnaires at the end of the class time to the 49 teachers who
did not, and these two groups of teachers did not differ in terms of
their provision of either autonomy support (t ⬍ 1) or structure (t ⫽
1.15). For analyses of these data, on the first level (between-
students level), regression equations were modeled to detect en-
gagement differences among students sitting in the same class-
room. At the second level (between-teachers level), regression
equations were modeled for characteristics that differed between
teachers (autonomy support, structure) and these Level 2 predictor
variables were centered around their means. At the third level
(between-schools level), regression equations were modeled for
the different schools in which the teachers taught. A detailed
presentation of multilevel modeling can be found in Raudenbush
and Bryk (2002), and all analyses were conducted with HLM 6
(Raudenbush, Bryk, Cheong, & Congdon, 2004).
Results
Preliminary analyses were conducted to investigate school-level
differences in all the variables. Overall means for the pair of
engagement measures appear in Table 1. School-based means
ranged from – 0.48 to 0.32 for students’ collective behavioral
engagement (scored with z scores) and from 5.38 to 5.88 for
students’ individual self-report engagement (scored on a 1–7
scale). HLM analyses were used to investigate the between-
schools effects on the engagement measures. Interclass correlation
coefficients indicated that between-school differences accounted
for 7% of the total variance in students’ collective behavioral
engagement and for less than 1% of the total variance in students’
individual self-report engagement. These figures suggest that most
of the variance in both engagement measures was not due to
systematic school differences. We therefore turned our attention to
investigating between-teachers (i.e., between-classrooms) differ-
ences in the engagement measures. Interclass correlation coeffi-
cients indicated that between-teachers differences accounted for
93% of the total variance in students’ collective behavioral en-
gagement. For students’ individual self-report engagement, inter-
class correlation coefficients indicated that between-teachers dif-
ferences accounted for 14% of the total variance, while within-
classroom variance accounted for the remaining (unexamined)
86% of the variance.
Table 1 shows the descriptive statistics and correlation matrix
for the four measures expressed at the classroom level. The two
teacher ratings (autonomy support, structure) intercorrelated pos-
itively and significantly with one another, as did the two student
engagement measures. Further, teacher-provided autonomy sup-
port and teacher-provided structure both individually and posi-
tively correlated with both students’ collective behavioral and
students’ individual self-report engagement (all ps ⬍ .01).
Table 1
Descriptive Statistics and Correlation Matrix for the Measures of Teachers’ Instructional Styles
and Students’ Classroom Engagement
Measure 1 2 3 4
1. Teacher-provided autonomy support — .60
ⴱⴱ
.70
ⴱⴱ
.36
ⴱⴱ
2. Teacher-provided structure — .76
ⴱⴱ
.30
ⴱⴱ
3. Students’ collective behavioral engagement — .41
ⴱⴱ
4. Students’ individual self-report engagement —
M 4.59 5.04 0.00 5.41
SD 1.17 1.06 0.84 0.59
Range of scores 1.33 to 6.83 1.40 to 7.00 ⫺2.86 to 1.78 3.88 to 6.78
Skewness ⫺0.38 ⫺0.61 ⫺0.39 ⫺0.24
Kurtosis ⫺0.40 0.62 0.42 0.06
Sample size 133 133 133 84
a
a
Our primary unit of analysis was at the teacher level. Hence, the statistics for students’ self-report engagement
are expressed at the classroom (teacher) level. That is, the data points above for students’ individual self-report
engagement represent the class mean of the students in that class. When students’ self-report engagement
responses are expressed at the individual level, sample size increases from 84 to 1,582, range of scores increases
to 1.00 to 7.00, and standard deviation increases to 1.33, while the remaining statistics (mean, skewness,
kurtosis) remain largely unchanged.
ⴱⴱ
p ⬍ .01.
594
JANG, REEVE, AND DECI
Predicting Students’ Collective Behavioral
Engagement
HLM was used to examine whether students’ collective behav-
ioral engagement could be predicted by the two teacher-level
variables of autonomy support and structure. The HLM equations
were constructed to simultaneously test the teacher-level variables
as joint predictors of students’ collective behavioral engagement.
Hypothesis testing for fixed effects revealed significant t statistics
for both teacher variables as individually significant predictors of
students’ collective behavioral engagement (see Table 2). Hence,
the more teachers were rated by observers as providing autonomy
support and the more teachers were rated by observers as provid-
ing structure, the greater collective engagement the observers
reported seeing in their students.
Predicting Students’ Individual Self-Report
Engagement
HLM was also used to examine whether students’ individual
self-report engagement could be predicted by the two teacher-level
variables of autonomy support and structure. The HLM equations
were again constructed to simultaneously test the teacher-level
variables as joint predictors of students’ individual self-report
engagement. Hypothesis testing for fixed effects revealed a sig-
nificant t statistic for teacher-provided autonomy support as an
individually significant predictor of students’ individual self-
reported engagement but a nonsignificant t statistic for teacher-
provided structure (see Table 2). Hence, the more teachers were
rated by observes as providing autonomy support (but not neces-
sarily structure), the greater individual engagement their students
reported experiencing.
Relation Between Autonomy Support and Structure
The zero-order correlation between autonomy support and struc-
ture reported in Table 1 was significant and positive, r(133) ⫽ .60,
p ⬍ .01.
As to the curvilinear hypothesis, we used raters’ scores to divide
the 133 teachers into three equal groups of providing low (n ⫽ 46),
moderate (n ⫽ 43), or high (n ⫽ 44) levels of structure so that we
could perform a one-way analysis of variance with tests for both
linear and quadratic trend effects on teacher-provided autonomy
support. In the prediction of autonomy support, the quadratic trend
was not significant, F(1, 130) ⬍ 1, while the linear trend was
significant, F(1, 130) ⫽ 60.92, p ⬍ .01, based on the following
means: low ⫽ 3.81; moderate ⫽ 4.57; high ⫽ 5.42. Hence, the
effect that higher levels of structure had on teachers’ provision of
autonomy support was linear, not curvilinear (i.e., quadratic). In
addition, we tested whether the effect of structure on students’
collective behavioral engagement was linear or quadratic. Again,
the quadratic trend was not significant, F(1, 130) ⬍ 1, while the
linear trend was significant, F(1, 130) ⫽ 117.03, p ⬍ .01 (low,
M ⫽ – 0.70; moderate, M ⫽ – 0.04; and high, M ⫽ 0.70). Finally,
we tested whether the effect of structure on students’ self-report
engagement was linear or quadratic. Again, the quadratic trend
was not significant, F(1, 81) ⬍ 1, while the linear trend was
significant, F(1, 81) ⫽ 13.20, p ⬍ .01 (low, M ⫽ 5.24; moderate,
M ⫽ 5.49; high, M ⫽ 5.81). Overall, these data show strong
support for the idea that the relation between structure and auton-
omy support (and the relation between structure and both types of
engagement) are linear; there was no support for the idea that the
relations are curvilinear.
Discussion
In the present study, we focused on two aspects of teachers’
instructional styles—autonomy support and structure—to assess
their association with each other and with students’ classroom
engagement. Specifically, we investigated the interrelation be-
tween autonomy support and structure (Was it antagonistic, cur-
vilinear, or independent?) and the relation of each to student
engagement. We found that teacher-provided autonomy support
and structure positively covaried rather than being antagonistic,
curvilinear, or independent. Further, correlational analyses showed
that students’ classroom engagement was quite strongly and pos-
itively associated with both aspects of teachers’ instructional
styles. Finally, the regression-based HLM analysis (capable of
controlling for the effect of the other predictor variable) showed
that autonomy support uniquely predicted both measures of stu-
dents’ engagement, while structure uniquely predicted students’
Table 2
Results From Hierarchical Analyses Predicting Students’ Collective Behavioral Engagement and Students’ Individual
Self-Report Engagement
Students’ collective behavioral engagement
(133 classrooms)
Students’ individual self-report engagement
(1,582 students in 84 classrooms)
Variable Coefficient SE t(130) Variance SD Coefficient SE t(81) Variance SD
Fixed effects
Intercept
a
0.02 .08 0.23 5.50 .06 87.34
ⴱ
Teacher autonomy support 0.36
ⴱ
.05 7.31
ⴱ
0.19
ⴱ
.07 2.54
ⴱ
Teacher structure 0.38
ⴱ
.05 7.78
ⴱ
0.06 .08 0.80
Random effects
Level 1 (R) .21 .46 .23 .48
Intercept (U
0
)
.04
ⴱ
.20 .00 .01
a
Raw scores for students’ collective behavioral engagement are expressed in z scores; raw scores for students’ individual self-reported engagement are
expressed on a 1–7 Likert scale.
ⴱ
p ⬍ .05.
595
AUTONOMY SUPPORT AND STRUCTURE
collective behavioral engagement but not students’ individual self-
report engagement. When taken as a whole, these findings have
implications that speak to the nature of the relation between
autonomy support and structure, to explaining students’ classroom
engagement, to classroom climate and teacher behavior, and to
classroom practices.
Relation Between Autonomy Support and Structure
From our review of the literature, we identified three possible
ways that teacher-provided autonomy support and structure might
relate both to each other and to students’ high classroom engage-
ment. We return to these possible relations here to examine
whether the current findings support these possible relations.
Are autonomy support and structure antagonists? One
view of the relation between autonomy support and structure is
that they are antagonists, such that the provision of one interferes
with the provision of the other. While both anecdotal and empirical
examples can be offered to illustrate how a specific element of
structure might suppress a specific element of autonomy support
(e.g., Perlmuter & Monty, 1979), the correlations reported in Table
1 indicated that autonomy support and structure were positively
correlated. In fact, autonomy support and structure shared a com-
mon variance of 36%, based on the r ⫽ .60 reported in Table 1.
That is, the high school teachers in our study who nurtured
students’ inner motivational resources, used informational lan-
guage, and accepted expression of students negative affect were
more likely, not less likely, to communicate clear and explicit
directions, display strong guidance, and offer constructive feed-
back. Further, as shown in the HLM regression-based analyses, the
variance in students’ engagement accounted for by autonomy
support and structure together was greater than was the variance
accounted for by either autonomy support alone or structure alone.
This suggests that autonomy support and structure functioned in a
complementary, not in an antagonistic, way.
The curvilinear hypothesis. A second view of the relation
between autonomy support and structure is the curvilinear hypoth-
esis. Providing too much structure conceivably undermines both
teacher-provided autonomy support and students’ engagement be-
cause it tends to create a rigid, compliance-based, and personal
causation-depleting environment. Providing too little structure
does not interfere with a teacher’s provision of autonomy support,
though it does conceivably undermine students’ engagement be-
cause it tends to create a chaos-inviting laissez-faire environment
and neglects the development of students’ personal control beliefs.
Thus, the optimal amount of structure to support teacher-provided
autonomy support and students’ classroom engagement is, con-
ceivably, a moderate amount: “too much is bad—too little is
chaos” (deCharms, 1984, p. 287). Importantly, the curvilinear
hypothesis applies only to teacher-provided structure, as deCharms
(1984) argued that high autonomy support was always better than
low autonomy support. The reason deCharms (1984) cautioned
that high structure was problematic for student engagement was
that it too often undermined teacher-provided autonomy support
(implying that, at high levels, structure could be antagonistic to
autonomy support, as discussed above). The trend analyses for
how structure affected teacher-provided autonomy support and
how structure affected the pair of student engagement measures
showed no support for the curvilinear hypothesis among low,
moderate, and high levels of teacher-provided structure. Instead,
these analyses showed only strong support for the linear hypoth-
esis, as teacher-provided autonomy support was highest with the
provision of a highly structured instructional style and student
engagement was also at its highest with the provision of a highly
structured instructional style.
Two independent aspects of instructional style. A third
view of the relation between autonomy support and structure is
that they are separate and independent aspects of teachers’ instruc-
tional styles, each contributing its own unique role to support
students’ motivation and engagement. Rather than supporting the
conclusion that the two instructional styles were independent and
uniquely predictive of students’ engagement, our data supported
the conceptualization that the two styles are complementary and
uniquely predictive. By complementary, we mean that the two
styles are both positively correlated and contribute to the predic-
tion of students’ engagement. That is, both aspects of teachers’
styles correlated positively and strongly with one another and with
students’ engagement (see Table 1), and both aspects produced a
significant main effect that explained unique variance in students’
collective behavioral engagement (see Table 2). Following the
thinking outlined by Skinner and Belmont (1993), we believe that
both autonomy support and structure support students’ engage-
ment but that they likely do so in somewhat different ways. That
is, the two aspects of teachers’ instructional styles support different
aspects of students’ underlying motivation, as autonomy support
primarily enriches students’ perceived autonomy and sense of
personal causation, while structure primarily enriches students’
perceived competence and perceptions of control over outcomes.
While both aspects of teachers’ instructional styles appear to
contribute positive support to students’ engagement, the two as-
pects may not contribute equally, as autonomy support accounted
for unique variance in both students’ collective behavioral engage-
ment and students’ individual self-report engagement, while struc-
ture accounted for unique variance only in students’ collective
behavioral engagement. This pattern of findings suggests that
teacher-provided autonomy support may be associated with the
full range of students’ engagement, while teacher-provided struc-
ture may be associated more narrowly with the on-task behavioral
aspects of engagement (e.g., attention, effort, persistence).
Explaining Students’ Classroom Engagement
For the purposes of the present study, we distinguished between
students’ behavioral (objective) and self-reported (subjective) en-
gagement. Behavioral/objective engagement is what raters (and
teachers) publicly see during the lesson—students’ on-task atten-
tion, effort investment, persistence in the face of difficulty, and so
forth. Self-reported/subjective engagement, on the other hand, is
what students privately experience during the lesson—intentional
learning, positive feelings, deep information processing, and gen-
eral proactivity to contribute their sense of voice to the ongoing
flow of instruction. Teacher-provided autonomy support was as-
sociated with both of these aspects of students’ engagement, while
teacher-provided structure was more narrowly associated only
with the behavioral/objective aspect.
In practice, this pattern of findings suggests that structure-based
instructional strategies might be insufficient to support the full
range of students’ engagement. When provided with a highly
596
JANG, REEVE, AND DECI
structured learning environment, students do generally display
high levels of attention, effort, and persistence (i.e., behavioral
engagement). Our findings suggest that to further support students’
subjective engagement in the lesson, however, teachers need to
find ways to administer elements of classroom structure that not
only structure the lessons but also support students’ autonomy
while doing so, as it was teacher-provided autonomy support, not
teacher-provided structure, that was uniquely associated with stu-
dents’ subjective engagement. Hence, elements of structure might
guide students’ behavioral engagement, but these elements (e.g.,
communications, goals, feedback) need to be offered in autonomy-
supportive ways if they are to support both overt behavioral
displays of engagement and private subjective experiences of
engagement.
This study was designed to examine between-classes rather than
within-class effects on students’ engagement. That is, our focus
was on how different aspects of teachers’ instructional styles
affected students’ engagement. This focus on between-classes
effects, however, is a different focus from the majority of existing
research investigating students’ motivation and engagement,
which tends to focus on within-class effects (Anderman & Maehr,
1994; Church, Elliot, & Gable, 2001; Kaplan & Midgley, 1999;
Roeser, Midgley, & Urdan, 1996). The logic in an emphasis on
within-class effects is that students’ perceptions of the classroom
climate are the key factors predicting students’ motivation and
engagement. We agree that students’ perceptions of their teachers’
behaviors and their perceptions of their own engagement are impor-
tant variables. In fact, our HLM analyses of students’ individual
self-report engagement showed that within-class effects constituted
86% of the variance, while between-classes effects constituted only
14% of the variance (with between-schools effects constituting ⬍
1%). Clearly, student individual differences such as perceived
competence, goal orientation, and various motivational beliefs
account for substantial within-class differences in students’ moti-
vation (Wolters, 2004) and engagement (Skinner & Belmont,
1993). Still, we focused on objective differences between teachers
rather than on subjective differences between students. In doing so,
it is our hope to confirm the important sociocontextual contribu-
tion added by examining between-classes (between-teachers) ef-
fects to what researchers already know are important within-class
(between-students) effects. This between-classes and between-
teachers focus allows us rather uniquely to speak to research on
classroom climate and teacher behavior.
Classroom Climate and Teacher Behavior Research
The approach taken in the present study parallels research on
classroom climate (Fraser, 1994) and teacher behavior (see Wub-
bels, Brekelmans, den Brok, & van Tartwijk, 2006). While differ-
ent researchers tend to use somewhat different terminology, most
studies of teaching behavior emphasizes the two teacher dimen-
sions of influence (submission to dominance) and proximity (op-
position to cooperation) (Wubbels, Creton, Levy, & Hooymayers,
1993). These dimensions represent two orthogonal axes within a
coordinate system that features eight categories of teacher behav-
ior: leadership, friendliness, understanding, student responsibility,
uncertain, dissatisfied, admonishing, and strict (Wubbels et al.,
2006; Wubbels & Levy, 1991). The category of teacher behavior
that corresponds most closely to our conceptualization of structure
is leadership (“lead, organize, give orders, set tasks, determine
procedure, structure the classroom situation”); the opposite of
which is uncertain (“keep a low profile, apologize, wait and see
how the wind blows”). The category of teacher behavior that
corresponds most closely to our conceptualization of autonomy
support is understanding (“listen with interest, empathize, show
confidence and understanding, accept apologies, look for ways to
settle differences, be patient, be open”); the opposite of which is
admonishing (“get angry, take pupils to task, express irritation and
anger, forbid, correct, punish”; quotes from Wubbels et al., 2006,
p. 1166). In research based on this model, students’ classroom
effort and enjoyment (i.e., engagement) is highest when teachers
show high leadership and high understanding and lowest when
teachers show high uncertainty and high admonishing (Brekel-
mans, 1989; Goh & Fraser, 1996; den Brok, Brekelmans, &
Wubbels, 2004).
2
Prior research on classroom climate and teacher behavior offers
a potential insight as to how autonomy support and structure might
combine during the ongoing flow of instruction to enhance stu-
dents’ engagement. By analyzing teacher behavior from one mo-
ment to the next, researchers found that teaching at central mo-
ments in the lesson (e.g., when the teacher was in front of the class
and introducing a new learning activity) was crucial to predicting
students’ subsequent classroom engagement (Brekelmans, Slee-
gers, & Fraser, 2000; van Tartwijk, Brekelmans, Wubbels, Fisher,
& Fraser, 1998). To set the conditions under which students could
later regulate their own learning in an autonomous and responsible
way— especially during less supervised group and individual
work—it was helpful for a teacher to first display a strong sense of
leadership (i.e., high structure) during these central lesson seg-
ments. Notice that our rating sheet to score structure (see Figure 2)
was partitioned into three lesson segments— during introduction,
during lesson, and during feedback. Perhaps student engagement
thrives when teachers provide consistently high autonomy support
(high understanding, low admonishing) and especially well-timed
high structure (high leadership, low uncertainty).
Implications for Classroom Practice
The current findings have implications for teachers wrestling
with the daily goal of supporting students’ engagement during
learning activities. We found that teacher-provided autonomy sup-
port and teacher-provided structure functioned as rather important
predictors of students’ collective classroom engagement. The im-
plication that comes from testing the antagonistic, curvilinear,
independent, and complementary hypotheses on the relation that
autonomy support and structure have on each other and on stu-
dents’ engagement is that teachers seeking engagement-fostering
instructional strategies need not choose between providing auton-
omy support or structure but, instead, can focus their instructional
energies on providing autonomy support and structure.
2
Some of these research studies also merge multiple categories of
teacher behavior into distinct teacher profiles. The profiles most strongly
associated with students’ classroom engagement are directive (moderately
high influence, moderately high proximity), authoritative (moderately high
influence, very high proximity), and tolerant (moderate influence, very
high proximity; Brekelmans, 1989).
597
AUTONOMY SUPPORT AND STRUCTURE
As to the quality of just what teachers would say and do to
provide students with an autonomy-supportive, structured learning
environment, our goal was to conceptualize and operationally
define autonomy support and structure in a comprehensive way.
That is, for the provision of autonomy support, we suggest that
teachers might want to initiate learning activities by involving
students’ inner motivational resources, communicating in noncon-
trolling and informational ways and acknowledging students’ per-
spectives and negative feelings when motivational (e.g., listless-
ness) and behavioral (e.g., disrespectful language) problems arise.
For the provision of structure, we suggest that teachers might want
to initiate learning activities by offering clear and detailed expec-
tations and instructions, offering helpful guidance and scaffolding
as students try to profit from the lesson, and providing feedback to
enhance perceptions of competence and perceived personal control
during a reflective postperformance period.
We view our conceptualization of teacher-provided autonomy
support and structure as a comprehensive framework, and this
belief in the comprehensiveness of our conceptual and operational
definitions is based on the conceptual models, instructional rec-
ommendations, and empirical assessments of past research (Skin-
ner & Belmont, 1993; Reeve, Jang, et al., 2004). That said, we
recognize that our conceptual and operational definitions might
capture only the essential, rather than the comprehensive, elements
of teacher-provided autonomy support and structure. We recognize
that others—especially those with behavioral or sociocultural
(rather than motivational) perspectives—may emphasize aspects
of autonomy support or structure that we did not include in this
investigation. Hence, future researchers might want to make it a
priority to answer the question of whether our conceptualization of
these two characteristics of instructional style includes the com-
prehensive, versus the merely essential, elements of teacher-
provided autonomy support and teacher-provided structure.
Limitations
In interpreting the current findings, we recognize four potential
methodological limitations. One was that the sample included only
high school students and their teachers. In high school, teachers
expect greater personal responsibility and self-regulation from
their students, at least compared to what is expected from elemen-
tary school students (Martinez-Pons, 2002). Also, student motiva-
tion and engagement are generally lower in secondary school
classrooms than they are in elementary school classrooms (Got-
tfried, Fleming, & Gottfried, 2001; Tucker et al., 2002). These
grade-related baseline differences in teachers’ expectations and
demands on students might have important implications for our
findings. Hence, it seems important for future research to test the
generalizability of the findings to elementary and middle-school
classrooms.
A second limitation is that we do not know to what extent
observed teachers might have altered their instructional styles
upon seeing the raters enter their classroom. While our efforts to
provide teachers with minimal forewarning helped minimize this
effect (by preventing nontypical preparation), it is unknown to
what extent such on-the-spot alteration of a teacher’s instructional
style might have occurred. In support of the idea that teachers
probably did not alter their instructional styles is research on
teacher behavior that shows strong consistency in any one teach-
er’s instructional style from one assessment period to the next
(Brekelmans, 1989; Deci et al., 1981). Brekelmans (1989) found
that any one assessment of a teacher’s instructional style during the
year is a valid representation of that teacher’s year-long instruc-
tional style. In addition, because our observed teachers taught
English, math, science, and social studies, we do not know the
extent to which our 133 teachers are representative of the popu-
lation of teachers in other subject areas (e.g., art, music, physical
education).
A third limitation is the possibility of observer bias in the
behavior ratings. While raters were blind to the experimental
hypotheses and while raters’ scores of teachers’ instructional style
did predict students’ self-reported engagement, observer bias may
still have occurred when raters scored both the teacher’s instruc-
tional style and the students’ collective behavioral engagement. It
is possible that, when raters gave teachers generally good scores on
their instructional style, they were also inclined to see the students
as collectively engaged (and vice versa). To the extent that such
observer bias occurred, this shared method variance confound
could be expected to inflate the positive correlation between teach-
ers’ instructional behaviors and students’ behavioral (but not sub-
jective) engagement. The way that past researchers have dealt with
this concern is to use teacher ratings from one rater to predict the
student ratings of a second rater (and also to use teacher ratings
from the second rater to predict the student ratings of the first
rater). When this methodological correction has been applied (see
Reeve, Jang, et al., 2004, p. 163), the cross-rater analyses repli-
cated the findings from the single-rater analyses. We performed
these same cross-rater analyses with our data and found that the
cross-rater analyses replicated the single-rater analyses, though the
sample size was reduce from 133 classroom observations (classes
with at least one rater) to only 56 classroom observations (classes
with two raters). While potential rating artifacts are always a
concern, we conclude that our findings reflect a real association
between teachers’ instructional styles and students’ engagement.
A fourth limitation is that the data are correlational and cross-
sectional. While we generally framed the research questions
around asking how teachers influence students, we recognize that
engagement serves as an important social signal from students to
elicit supportive reactions from teachers (Furrer & Skinner, 2003).
For instance, when students show signs of engagement, teachers
are more likely to provide instructional support and hence to
display greater autonomy support and structure (Skinner & Bel-
mont, 1993); similarly, when students show signs of disengage-
ment, teachers are less likely to provide this same sort of instruc-
tional support (Skinner et al., 2008). To disentangle the effect that
teachers have on students from the effect that students have on
teachers and also to substantiate an interpretation that teachers’
instructional styles have a directional influence on students’ en-
gagement, experimental and longitudinal designs are needed. For-
tunately, past research allows us a measure of confidence in our
hypothesized directional effect (that teachers’ styles affect stu-
dents’ engagement) because the beneficial effects of teacher-
provided autonomy support on students’ engagement have been
shown both in longitudinal (deCharms, 1976) and in longitudinal/
experimental (Reeve, Jang, et al., 2004) studies, just as the bene-
ficial effect of teacher-provided structure on students’ engagement
has been shown in longitudinal/experimental research (Evertson &
Emmer, 1982).
598
JANG, REEVE, AND DECI
References
Anderman, E. M., & Maehr, M. L. (1994). Motivation and schooling in the
middle grades. Review of Educational Research, 64, 287–309.
Assor, A., Kaplan, H., & Roth, G. (2002). Choice is good, but relevance is
excellent: Autonomy-enhancing and suppressing teaching behaviors pre-
dicting students’ engagement in schoolwork. British Journal of Educa-
tional Psychology, 27, 261–278.
Bollen, K. A., & Long, J. S. (Eds.). (1993). Testing structural equation
models. Newbury Park, CA: Sage.
Brekelmans, M. (1989). Interpersonal teacher behavior in the classroom.
Utrecht, the Netherlands: W. C. C.
Brekelmans, M., Sleegers, P., & Fraser, B. J. (2000). Teaching for active
learning. In R. J. Simons, J. van der Linden, & T. Duffy (Eds.), New
learning (pp. 227–242). Dordrecht, the Netherlands: Kluwer Academic.
Brooks, D. (1985). The teacher’s communicative competence: The first day
of school. Theory into Practice, 24, 63–70.
Brophy, J. (1986). Teacher influences on student achievement. American
Psychologist, 41, 1069 –1077.
Brophy, J. (2006). Observational research on generic aspects of classroom
teaching. In P. A. Alexander & P. H. Winne (Eds.), Handbook of
educational psychology (2nd ed., pp. 755–780). Mahwah, NJ: Erlbaum.
Brophy, J., & Good, T. (1986). Teacher behavior and student achievement.
In M. C. Wittrock (Ed.), Handbook of research on teaching (3rd ed., pp.
328 –375). New York, NY: Macmillan.
Carter, K., & Doyle, W. (2006). Classroom management in early childhood
and elementary classrooms. In C. M. Evertson & C. S. Weinstein (Eds.),
Handbook of classroom management: Research, practice, and contem-
porary issues (pp. 373– 406). Mahwah, NJ: Erlbaum.
Church, M. A., Elliot, A. J., & Gable, S. L. (2001). Perceptions of
classroom environment, achievement goals, and achievement outcomes.
Journal of Educational Psychology, 93, 43–54.
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.), Self processes in development: Minnesota
symposium on child psychology (Vol. 23, pp. 167–216). Chicago, IL:
University of Chicago Press.
Daniels, H., & Bizar, M. (1998). Methods that matter: Six structures for
best practice classrooms. Portland, ME: Stenhouse.
deCharms, R. (1976). Enhancing motivation: Change in the classroom.
New York, NY: Irvington.
deCharms, R. (1984). Motivation enhancement in educational settings. In
R. Ames & C. Ames (Eds.), Research on motivation in education:
Student motivation (Vol. 1, pp. 275–310). Orlando, FL: Academic Press.
Deci, E. L., Eghrari, H., Patrick, B. C., & Leone, D. R. (1994). Facilitating
internalization: The self-determination theory perspective. Journal of
Personality, 62, 119 –142.
Deci, E. L., Koestner, R., & Ryan, R. M. (1999). A meta-analytic review
of experiments examining the effects of extrinsic rewards on intrinsic
motivation. Psychological Bulletin, 125, 627– 668.
Deci, E. L., Schwartz, A., Sheinman, L., & Ryan, R. M. (1981). An
instrument to assess adult’s orientations toward control versus autonomy
in children: Reflections on intrinsic motivation and perceived compe-
tence. Journal of Educational Psychology, 73, 642– 650.
den Brok, P., Brekelmans, M., & Wubbels, T. (2004). Interpersonal teacher
behavior and student outcomes. School Effectiveness and School Im-
provement: An International Journal of Research, Policy and Practice,
15, 407– 442.
Doyle, W. (1986). Classroom organization and management. In M. C.
Witrock (Ed.), Handbook of research on teaching (3rd ed., pp. 432–
463). New York, NY: Macmillan.
Doyle, W. (2006). Ecological approaches to classroom management. In
C. M. Evertson & C. S. Weinstein (Eds.), Handbook of classroom
management: Research, practice, and contemporary issues (pp. 97–
125). Mahwah, NJ: Erlbaum.
Emmer, E., Evertson, C., & Anderson, L. (1980). Effective classroom
management at the beginning of the school year. Elementary School
Journal, 80, 219 –231.
Evertson, C. M., & Emmer, E. T. (1982). Effective management at the
beginning of the year in junior high school classes. Journal of Educa-
tional Psychology, 74, 485– 498.
Evertson, C. M., & Weinstein C. S. (Eds.). (2006). Handbook of classroom
management: Research, practice, and contemporary issues. Mahwah,
NJ: Erlbaum.
Finn, J. D., & Rock, D. A. (1997). Academic success among students at
risk for school failure. Journal of Applied Psychology, 82, 221–234.
Fraser, B. J. (1994). Research on classroom and school climate. In D.
Gabel (Ed.), Handbook of research on science teaching and learning
(pp. 493–541). New York, NY: Macmillan.
Fredricks, J. A., Blumenfeld, P. C., & Paris, A. H. (2004). School engage-
ment: Potential of the concept, state of the evidence. Review of Educa-
tional Research, 74, 59 –109.
Furrer, C., & Skinner, E. A. (2003). Sense of relatedness as a factor in
children’s academic engagement and performance. Journal of Educa-
tional Psychology, 95, 148 –162.
Goh, S. C., & Fraser, B. J. (1996). Validation of an elementary school
version of the Questionnaire on Teacher Interaction. Psychological
Reports, 79, 515–521.
Gottfried, A. E., Fleming, J. S., & Gottfried, A. W. (2001). Continuity of
academic intrinsic motivation from childhood through late adolescence:
A longitudinal study. Journal of Educational Psychology, 93, 3–13.
Grolnick, W., & Ryan, R. M. (1987). Autonomy in children’s learning: An
experimental and individual difference investigation. Journal of Person-
ality and Social Psychology, 52, 890 – 898.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance
structure analysis: Conventional criteria versus new alternatives. Struc-
tural Equation Modeling, 6, 1–55.
Huston-Stein, A., Friedrich-Cofer, L., & Susman, E. J. (1977). The relation
of classroom structures to social behavior, imaginative play, and self-
regulation of economically disadvantaged children. Child Development,
48, 908 –916.
Jang, H. (2008). Supporting students’ motivation, engagement, and learn-
ing during an uninteresting activity. Journal of Educational Psychology,
100, 798 –811.
Jang, H. (2009). Preserving and enhancing students’ autonomy by deliv-
ering directed instruction in an autonomy-supportive way. Unpublished
manuscript, University of Wisconsin—Milwaukee.
Jöreskog, K. G., & Sörbom, D. (1993). LISREL 8: Structural equation
modeling with the SIMPLIS command language. Hillsdale, NJ: Scien-
tific Software International.
Kaplan, A., & Midgley, C. (1999). The relationship between perceptions of
the classroom goal structure and early adolescents’ affect in school: The
mediating role of coping strategies. Learning and Individual Differ-
ences, 11, 187–212.
Koenigs, S. S., Fiedler, M. L., & deCharms, R. (1977). Teacher beliefs,
classroom interaction and personal causation. Journal of Applied Social
Psychology, 7, 95–114.
Koestner, R., Ryan, R. M., Bernieri, F., & Holt, K. (1984). Setting limits
on children’s behavior: The differential effects of controlling versus
informational styles on intrinsic motivation and creativity. Journal of
Personality, 52, 233–248.
Mageau, G. A., & Vallerand, R. J. (2003). The coach–athlete relationship:
A motivational model. Journal of Sports Sciences, 21, 883–904.
Martinez-Pons, M. (2002). Parental influences on children’s academic
self-regulatory development. Theory into Action, 41, 126 –131.
O’Donnell, A. M. (2006). The role of peers and group learning. In P. A.
Alexander & P. H. Winne (Eds.), Handbook of educational psychology
(2nd ed., pp. 781– 802). Mahwah, NJ: Erlbaum.
599
AUTONOMY SUPPORT AND STRUCTURE
Perlmuter, L. C., & Monty, R. A. (1979). Choice and perceived control.
Hillsdale, NJ: Erlbaum.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models (2nd
ed.). Thousand Oaks, CA: Sage.
Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., & Congdon, R. T. (2004).
HLM 6: Hierarchical linear and nonlinear modeling. Lincolnwood, IL:
Scientific Software International.
Reeve, J. (2009). Why teachers adopt a controlling motivating style toward
students and how they can become more autonomy supportive. Educa-
tional Psychologist, 44, 159 –175.
Reeve, J., Deci, E. L., & Ryan, R. M. (2004). Self-determination theory: A
dialectical framework for understanding the sociocultural influences on
student motivation. In D. M. McInerney & S. Van Etten (Eds.), Research
on sociocultural influences on motivation and learning: Big theories
revisited (Vol. 4, pp. 31–59). Greenwich, CT: Information Age Press.
Reeve, J., & Jang, H. (2006). What teachers say and do to support students’
autonomy during a learning activity. Journal of Educational Psychology,
98, 209 –218.
Reeve, J., Jang, H., Carrell, D., Jeon, S., & Barch, J. (2004). Enhancing
students’ engagement by increasing teachers’ autonomy support. Moti-
vation and Emotion, 28, 147–169.
Reeve, J., Jang, H., Hardre, P., & Omura, M. (2002). Providing a rationale
in an autonomy-supportive way as a strategy to motivate others during
an uninteresting activity. Motivation and Emotion, 26, 183–207.
Roeser, R. W., Midgley, C., & Urdan, T. (1996). Perceptions of the school
psychological environment and early adolescents’ self-appraisals and
academic engagement: The mediating role of goals and belonging.
Journal of Educational Psychology, 88, 408 – 422.
Rosenholtz, S. J., & Simpson, C. (1984). Classroom organization and
student stratification. Elementary School Journal, 85, 21–37.
Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the
facilitation of intrinsic motivation, social development, and well-being.
American Psychologist, 55, 68 –78.
Ryan, R. M., & Deci, E. L. (2002). An overview of self-determination
theory: An organismic-dialectical perspective. In E. L. Deci & R. M.
Ryan (Eds.), Handbook of self-determination research (pp. 3–33). Roch-
ester, NY: University of Rochester Press.
Ryan, R. M., & La Guardia, J. G. (1999). Achievement motivation within
a pressured society: Intrinsic and extrinsic motivation to learn and the
politics of school reform. In T. Urdan (Ed.), Advances in motivation and
achievement (Vol. 11, pp. 45– 85). Greenwich, CT: JAI Press.
Ryan, R. M., Mims, V., & Koestner, R. (1983). Relation of reward
contingency and interpersonal context to intrinsic motivation: A review
and test using cognitive evaluation theory. Journal of Personality and
Social Psychology, 45, 736 –750.
Schuh, K. L. (2004). Learner-centered principles in teacher-centered prac-
tices? Teaching and Teacher Education, 20, 833– 846.
Schultz, R. A. (1982). Teaching style and sociopsychological climates. The
Alberta Journal of Educational Research, 18, 9 –18.
Sierens, E., Vansteenkiste, M., Goossens, L., Soenens, B., & Dochy, F.
(2009). The synergistic relationship of perceived autonomy support and
structure in the prediction of self-regulated learning. British Journal of
Educational Psychology, 79, 57– 68.
Skinner, E. A. (1995). Perceived control, motivation, and coping. Newbury
Park, CA: Sage.
Skinner, E. A., & Belmont, M. J. (1993). Motivation in the classroom:
Reciprocal effects of teacher behavior and student engagement across
the school year. Journal of Educational Psychology, 85, 571–581.
Skinner, E. A., Furrer, C., Marchand, G., & Kindermann, T. (2008).
Engagement and disaffection in the classroom: Part of a larger motiva-
tional dynamic? Journal of Educational Psychology, 100, 765–781.
Skinner, E. A., Wellborn, J. G., & Connell, J. P. (1990). What it takes to
do well in school and whether I’ve got it: A process model of perceived
control and children’s engagement and achievement in school. Journal
of Educational Psychology, 82, 22–32.
Skinner, E. A., Zimmer-Gembeck, M. J., & Connell, J. P. (1998). Individ-
ual differences and the development of perceived control. Monographs
of the Society for Research in Child Development, 63(2–3, Whole No.
204).
Tucker, C. M., Zayco, R. A., Herman, K. C., Reinke, W. M., Trujillo, M.,
Carraway, K., . . . Ivery, P. D. (2002). Teacher and child variables as
predictors of academic engagement among low-income African Amer-
ican children. Psychology in the Schools, 39, 477–488.
van Tartwijk, J., Brekelmans, M., Wubbels, T., Fisher, D. L., & Fraser,
B. J. (1998). Students perceptions of teacher interpersonal style: The
front of the classroom as the teacher’s stage. Teaching and Teacher
Education, 14, 607– 617.
Wellborn, J. G. (1991). Engaged and disaffected action: The conceptual-
ization and measurement of motivation in the academic domain. Un-
published doctoral dissertation, University of Rochester.
Wolters, C. A. (2004). Advancing achievement goal theory: Using goal
structures and goal orientations to predict students’ motivation, cogni-
tion, and achievement. Journal of Educational Psychology, 96, 236 –
250.
Wubbels, T., Brekelmans, M., den Brok, P., & van Tartwijk, J. (2006). An
interpersonal perspective on classroom management in secondary class-
rooms in the Netherlands. In C. M. Evertson & C. S. Weinstein (Eds.),
Handbook of classroom management: Research, practice, and contem-
porary issues (pp. 1161–1191). Mahwah, NJ: Erlbaum.
Wubbels, T., Creton, H. A., Levy, J., & Hooymayers, H. P. (1993). The
model for interpersonal teacher behavior. In T. Wubbels & J. Levy
(Eds.), Do you know what you look like? Interpersonal relations in
education (pp. 13–28). London, England: The Falmer Press.
Wubbels, T., & Levy, J. (1991). A comparison of interpersonal behavior of
Dutch and American teachers. International Journal of Intercultural
Relations, 15, 1–18.
Received February 12, 2008
Revision received March 30, 2010
Accepted March 31, 2010 䡲
600
JANG, REEVE, AND DECI