How Students Create Motivationally Supportive Learning Environments For Themselves: The Concept of Agentic Engagement

Journal of Educational Psychology © 2013 American Psychological Association
2013, Vol. 105, No. 3, 579 –595 0022-0663/13/$12.00 DOI: 10.1037/a0032690
How Students Create Motivationally Supportive Learning Environments for

Themselves: The Concept of Agentic Engagement
Johnmarshall Reeve
Korea University
The present study introduced “agentic engagement” as a newly proposed student-initiated pathway to
greater achievement and greater motivational support. Study 1 developed the brief, construct-congruent,
and psychometrically strong Agentic Engagement Scale. Study 2 provided evidence for the scale’s
construct and predictive validity, as scores correlated with measures of agentic motivation and explained
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
independent variance in course-specific achievement not otherwise attributable to students’ behavioral,

This document is copyrighted by the American Psychological Association or one of its allied publishers.
emotional, and cognitive engagement. Study 3 showed how agentically engaged students create moti-
vationally supportive learning environments for themselves. Measures of agentic engagement and
teacher-provided autonomy support were collected from 302 middle-school students in a 3-wave
longitudinal research design. Multilevel structural equation modeling showed that (a) initial levels of
students’ agentic engagement predicted longitudinal changes in midsemester perceived autonomy sup-
port and (b) early-semester changes in agentic engagement predicted longitudinal changes in late-
semester autonomy support. Overall, these studies show how agentic engagement functions as a
proactive, intentional, collaborative, and constructive student-initiated pathway to greater achievement
(Study 2) and motivational support (Study 3).
Keywords: achievement, agency, agentic engagement, autonomy support, student engagement
Engagement refers to a student’s active involvement in a learn- ally, emotionally, and cognitively involved in the learning activi-
ing activity (Christenson, Reschly, & Wylie, 2012). It functions as ties their teachers provide (e.g., write an essay, solve a math
a student-initiated pathway to highly valued educational outcomes, problem), and their extent of effort, enjoyment, and strategic
such as academic progress and achievement (Jang, Kim, & Reeve, thinking does predict important outcomes, such as achievement.
2012; Ladd & Dinella, 2009; Skinner, Kindermann, Connell, & But students also do more than this. Students also, more or less,
Wellborn, 2009; Skinner, Zimmer-Gembeck, & Connell, 1998). It proactively contribute into the flow of instruction they receive as
is a multidimensional construct consisting of three distinct, yet they attempt not only to learn but also to create a more motiva-
intercorrelated and mutually supportive, pathways to academic tionally supportive learning environment for themselves (Bandura,
progress—namely, its behavioral, emotional, and cognitive aspects 2006).
(Christenson et al., 2012; Fredricks, Blumenfeld, & Paris, 2004;
Skinner, Kindermann, Connell, & Wellborn, 2009). Behavioral
engagement refers to how involved the student is in the learning Agentic Engagement
activity in terms of attention, effort, and persistence; emotional Reeve and Tseng (2011) initially proposed the concept of agen-
involvement refers to the presence of positive emotions during task tic engagement. They defined it as “students’ constructive contri-
involvement such as interest and to the absence of negative emo- bution into the flow of the instruction they receive” (p. 258). To
tions such as anxiety; and cognitive engagement refers to how characterize students’ agentic engagement, these researchers of-
strategically the student attempts to learn in terms of employing fered classroom-based examples such as “offer input, express a
sophisticated rather than superficial learning strategies, such as preference, offer a suggestion or contribution, ask a question,
using elaboration rather than memorization. This three- communicate what they are thinking and needing, recommend a
dimensional portrayal of what actively involved students do is goal or objective to be pursued, communicate their level of inter-
accurate, but it is also incomplete. Students do become behavior- est, solicit resources or learning opportunities, seek ways to add
personal relevance to the lesson, ask for a say in how problems are
to be solved, seek clarification, generate options, communicate
likes and dislikes, . . . ” (Reeve & Tseng, 2011, p. 258). These
This article was published Online First April 29, 2013. examples were extracted from extensive field notes in which
This research was supported by the WCU (World Class University) trained raters used the Hit-Steer Observation System (Fiedler,
Program funded by the Korean Ministry of Education, Science and Tech-
1975; Koenigs, Fiedler, & deCharms, 1977) to typify how students
nology, consigned to the Korea Science and Engineering Foundation
(Grant R32-2008-000-20023-0).
proactively attempt to learn and contribute into the flow of instruc-
Correspondence concerning this article should be addressed to Johnmar- tion their teachers provide. From this database and from the
shall Reeve, 633 Uncho-Useon Hall, Department of Education, Anam- writings of motivation theorists who described what agentically
Dong Seongbuk-Gu, Korea University, Seoul 136-701, Korea. E-mail: motivated students do— e.g., deCharms’s (1976) origins in the
reeve@korea.ac.kr classroom, Bandura’s (1997) efficacious learners, and Ryan and
579
580 REEVE
Deci’s (2000) autonomously motivated students—Reeve and students’ underlying motivation and (b) students attain learning-
Tseng (2011) operationally defined the agentic engagement con- related outcomes in proportion to which they exert effort, experi-
struct with the following five items: ence enthusiasm, think strategically, and contribute constructively.
• During class, I ask questions. What all four horizontal lines have in common is that they repre-
• I tell my teacher what I like and what I don’t like. sent naturally occurring expressions of students’ underlying aca-
• I let my teacher know what I’m interested in. demic motivation that functionally translate students’ academic
• During class, I express my preferences and opinions. motivation into positive educational outcomes. Of particular im-
• I offer suggestions about how to make the class better. portance is the addition of the fourth horizontal line to communi-
What these five acts of agentic engagement have in common is cate that agentic engagement explains unique variance in students’
that each is a unilateral contribution into the flow of instruction. learning and achievement. The curved line in the lower portion of
Communicating one’s preferences or asking a question, however, the figure communicates a unique property of agentic engage-
may or may not advance one’s learning or improve one’s learning ment—namely, that, through their acts of agentic engagement,
conditions. What would be more likely to do so would be trans- students—more or less—attempt to join forces with the provider of
actional (Sameroff, 2009) or dialectical (Reeve, Deci, & Ryan, the learning environment (i.e., the teacher) to create for themselves
2004) classroom activity. With transactional activity, positive stu- a more motivationally supportive learning environment. These acts
dent outcomes are not a function of student activity (agentic of agentic engagement are qualitatively distinct from the other
engagement) but, rather, are the result of the unfolding of recip- three aspects of engagement in that they are intentional, proactive,
rocal processes between student and teacher. What students do and teacher-collaborative ways of engaging in learning activities.
(display engagement) affects and transforms what teachers do If these agentic contributions do transform how motivationally
(provide instruction) and vice versa, and it is these emerging supportive the learning environment becomes (e.g., greater auton-
transactions that lead to greater or lesser positive student out- omy support, greater access to interesting and personally valued
comes. With dialectical activity, student-initiated questions and learning activities), then the learning environment becomes in-
communications affect change in and transform the teacher’s in- creasingly conducive to the types of student motivation (i.e., origin
structional behavior, just as the teacher’s instructional behavior in motivation, autonomous motivation, academic efficacy) capable of
turn affects change in and transforms the quality and quantity of energizing, directing, and sustaining students’ classroom engage-
the student’s engagement. Hence, agentic engagement can be ment.
viewed not just as a student’s contributions into the flow of
instruction but also as an ongoing series of dialectical transactions
Agentic Engagement as a New Educational Construct
between student and teacher.
How agentically engaged students contribute transactionally and Engagement represents the range of action students take to
dialectically into the instructional flow is illustrated graphically in advance from not knowing, not understanding, not having skill,
Figure 1. The four horizontal lines communicate that (a) student and not achieving to knowing, understanding, having skill, and
engagement emerges out of and publically expresses the quality of achieving. It is what students do to make academic progress. To
Figure 1. Four interrelated aspects of student engagement that explain students’ positive outcomes (four
horizontal lines) plus agentic engagement’s unique contribution to constructive changes in the learning envi-
ronment (curved line at the base of the figure). The six curved lines with double-sided arrows communicate the
positive intercorrelations among the four aspects of engagement.
AGENTIC ENGAGEMENT 581
make progress in learning a foreign language, for instance, stu- classroom concept that best captures student-initiated, proactive,
dents can pay close attention to sources of information, invest intentional, collaborative, and constructive action.
effort, and persist in the face of setbacks, which is behavioral Agentic engagement is not only a pathway to academic prog-
engagement. Or, they can enhance their curiosity and minimize ress, but it is also a student-initiated pathway to a more motiva-
their anxiety and frustration, which is emotional engagement. Or, tionally supportive learning environment. There is some evidence
students can apply sophisticated learning strategies and carry out that students’ behavioral engagement also pulls a more supportive
mental simulations to diagnose and solve problems, which is style out of teachers and that students’ behavioral disengagement
cognitive engagement. These are three empirically validated path- pulls out a more conflictual or controlling style (Pelletier, Seguin-
ways to academic progress (Christenson et al., 2012). But agentic Levesque, & Legault, 2002; Skinner & Belmont, 1993), although
engagement is a fourth pathway. Proactively, students can contrib- we can find no evidence that emotional or cognitive engagement
ute into the flow of instruction both to enhance their learning and pulls a more supportive style out of teachers. This behavioral
to negotiate for the interpersonal support they need to energize engagement effect on teachers’ motivating styles is, however, an
their task-related motivation. To do so, they can express their indirect or inadvertent (albeit fortuitous) effect. In their reasoning
preferences, ask questions, and let the teacher know what they as to why teachers respond to students’ behavioral engagement,
need, want, and are interested in. The acknowledgment of this Skinner and Belmont (1993) argued that teachers find students’
breadth of engagement activity expands the current three-aspect behavioral disengagement to be aversive and that this leads teach-
conceptualization of student engagement into a four-aspect con- ers to feel incompetent or unliked by their students. As a result,
ceptualization, as represented in Figure 1. teachers tend to offer less support to these students. Behavioral
Agentic engagement is similar to the other three aspects of disengagement also signals poor student motivation, and teachers
engagement, as it too is a constructive student-initiated pathway to sometimes seek to remediate such low motivation by applying
academic progress; but it is also meaningfully different. Concep- pressure and coercion to motivate the unmotivated. Thus, behav-
tually, agentic engagement is a uniquely proactive and transac- iorally engaged students may not intentionally recruit their teach-
tional type of engagement. Proactively, agentically engaged stu- ers’ autonomy support any more than behaviorally disengaged
dents take action before the learning activity begins (e.g., students intentionally recruit teachers’ control and coercion. In
contrast, agentic engagement is intentional, purposive student-
“Teacher, can we do x?”); transactionally, they negotiate for a
initiated action to render the learning environment to become more
more motivationally supportive learning environment (e.g., how
motivationally supportive.
challenging, personally relevant, need-satisfying, or goal-
congruent the learning activity is). The other three types of en-
gagement largely take the teacher’s instruction as it is given, as Purpose of the Investigation and Overview of the
students use their behavior, emotion, and cognition as ways of Three Studies
translating that teacher-provided instruction into student-acquired
knowledge, understanding, and skill. Empirically, agentic engage- The purpose of the present investigation was to introduce agen-
tic engagement as a newly proposed student-initiated pathway to
ment has been shown (a) to correlate only modestly with the other
academic success, defined in the present study as the two-fold
three aspects of engagement and (b) to explain unique variance in
student outcome of higher achievement and greater motivational
students’ positive outcomes that the other three aspects cannot
support. Before we could investigate the utility of this pathway, we
explain (Reeve & Tseng, 2011). That is, agentically engaged
found it necessary to expand the instrument designed to assess
students are taking achievement-fostering action that is something
agentic engagement—the Agentic Engagement Scale (AES)1—to
more than just their behavioral, emotional, and cognitive engage-
consider not only unilateral contributions into instruction but also
ments.
more transactional and dialectical contributions. While Study 1
Agentic engagement is also meaningfully different from other
sought to refine the AES into a brief, construct-congruent, and
proactive, collaborative, and constructive classroom events. For
psychometrically strong self-report questionnaire, Study 2 sought
instance, formative assessments and personal response systems
to validate that refined scale. Using the criteria introduced by
(“clickers”) are collaborative, constructive, and sometimes proac- Reeve and Tseng (2011), a valid agentic engagement scale should
tive approaches to instruction that facilitate learning, but they do five things: (a) correlate positively with agency-rich types of
represent teacher-initiated, rather than student-initiated, action. A students’ classroom motivation; (b) correlate negatively with
teacher’s instructional effort to design and implement a construc- agency-impoverished types of motivation; (c) be distinct from the
tivist learning environment also fits within this category of other three aspects of engagement; (d) predict independent (i.e.,
teacher-initiated action (Brown & Campione, 1996). Instrumental unique) variance in course-related achievement (after controlling
(or “adaptive”) help-seeking is student-initiated and collaborative, for the variance in achievement otherwise attributable to behav-
but it is not necessarily either proactive or constructive (i.e., it ioral, emotional, and cognitive engagement); and (e) predict the
generally does not correlate with student achievement; Pajares, extent to which students create a more motivationally supportive
Cheong, & Oberman, 2004). Self-regulated learning involves learning environment for themselves. The fourth and fifth criteria
student-initiated, proactive, intentional, and constructive regula-
tory processes and actions (Zimmerman & Schunk, 2011), but
1
existing theoretical frameworks do not yet incorporate the concept The name of the original scale was changed from the “Agentic En-
gagement Questionnaire” to the “Agentic Engagement Scale” to avoid
of agentic engagement into their conceptualization of how students potential confusion that the AEQ acronym might have with the widely-
actively involve themselves in learning activities (as will be elab- used “Achievement Emotions Questionnaire” (Pekrun, Goetz, Frenzel,
orated on in the General Discussion). Agentic engagement is the Barchfeld, & Perry, 2011).
582 REEVE
were considered centrally important to the present study because Likert-type scale with the following response options: 1 ⫽
their confirmation would validate agentic engagement as an addi- strongly disagree; 2 ⫽ disagree; 3 ⫽ slightly disagree; 4 ⫽ neither
tionally important student-initiated pathway to positive outcomes agree nor disagree; 5 ⫽ slightly agree; 6 ⫽ agree; and 7 ⫽
(i.e., greater achievement, greater motivational support). strongly agree.
Motivation. Two questionnaires assessed students’ agentic-
centric motivational status. To assess psychological need satisfac-
Study 1
tion, we used the Activity-Feelings States (AFS; Reeve & Sick-
In Study 1 we sought to operationally define the agentic en- enius, 1994). The AFS offered the stem, “During this class, I feel,”
gagement construct through the refinement of its measurement and listed four items to assess perceived autonomy (e.g., “free”),
instrument. The methodological strategy was to start with Reeve three items to assess perceived competence (e.g., “capable”), and
and Tseng’s (2011) original five items and test the merits of adding three items to assess perceived relatedness (e.g., “I belong and the
both new and revised construct-consistent candidate items that people here care about me”). The overall 10-item assessment
reflected not just students’ unilateral contributions (as with the showed acceptable reliability (␣ ⫽ .83). We used this particular
original items) but transactional and dialectical ones as well. As measure of psychological need satisfaction because it was used in
before, the source of material to create the new candidate items the Reeve and Tseng (2011) investigation and because it had been
came from extensive classroom field notes from observations of shown to produce scores capable of predicting students’ classroom
what middle- and high-school students say and do during instruc- engagement and course grades (Jang, Reeve, Ryan, & Kim, 2009;
tion. Two key inclusion criteria were established for the Study 1 Reeve, Nix, & Hamm, 2003). To assess self-efficacy, we used the
scale development process: (a) The candidate item must correlate academic efficacy scale from the Patterns of Adaptive Learning
significantly with students’ agency-centric motivational status, and Scales (Midgley et al., 2000). The academic efficacy scale in-
(b) the candidate item must not be so highly correlated with the cluded five items (e.g., “I’m certain I can master the skills taught
other three aspects of engagement (behavioral, emotional, and in this class this year”) and showed an acceptable reliability (␣ ⫽
cognitive) as to be confounded with them. The first criterion— one .88). We used this particular measure to expand the conceptual-
of construct validity—is important to confirm that the candidate ization of agentic motivation beyond just psychological need sat-
item was closely associated with the types of motivation that the isfaction to include academic efficacy and because it had been
original theoretical work on agentic engagement cited as its mo- shown to produce scores capable of predicting students’ classroom
tivational origins, including psychological need satisfaction (Ryan engagement and course grades (Linnenbrink, 2005; Midgley et al.,
& Deci, 2000) and self-efficacy (Bandura, 1997). These motiva- 2000).
tions are agency-centric in that they both energize proactive and Engagement. We assessed four aspects of student engage-
intentional (i.e., agentic) changes on the learning environment ment—agentic, behavioral, emotional, and cognitive. To assess
(Bandura, 2006; Deci & Ryan, 1985). The second criterion— one agentic engagement, we used the five original items from the
of divergent (or discriminant) validity—is important to confirm Reeve and Tseng (2011) measure and then created five new
that the candidate item assesses a uniquely distinct type of engage- candidate items. These 10 items appear in Table 1. Items 1–5
ment. are the original AES items. Item 1b is a revised version of
original Item 1; this item was revised to test the merit of
limiting the context of question-asking to that which is specific
Method
to learning. Items 6 and 7 represented two new candidate items
Participants and procedure. Participants were 271 (48 fe- designed to assess a proactive and transactional contribution
male, 222 male, 1 unknown) college students from one of six into the learning environment. Items 8 and 9 represent two new
different courses within the College of Engineering at a large candidate items designed to assess a personal contribution to
university in Incheon, South Korea. All students (and their teach- one’s own learning that was not necessarily interpersonal, trans-
ers) were ethnic Korean. These particular students were selected actional, or dialectical but would instead apply during work on
because the School of Engineering was considering a curricular a personal project, homework assignment, or independent work.
program change to move classroom instruction away from tradi- The psychometric properties associated with these items and the
tional lecture to more student-centered approaches, such as per- scale that emerged from these items will be presented in the
sonal projects, which prompted school administrators’ interest in Results.
collecting data on their students’ existing motivation and engage- To assess both behavioral engagement and emotional en-
ment profiles. Participation in the study was voluntary, and par- gagement, we used the behavioral engagement and emotional
ticipants were told by the research assistant who administered the engagement scales from the Engagement Versus Disaffection
questionnaire that their scores were confidential and anonymous. with Learning measure (Skinner, Kindermann, & Furrer, 2009).
Students who consented to participate completed the survey at the The behavioral engagement scale included five items (listed in
beginning of a class period midway through the semester. The Table 2), and it showed acceptable internal consistency (␣ ⫽
survey took 5–10 min to complete, and participants were asked to .86). The emotional engagement scale included five items
rate their experiences of motivation and engagement in reference (listed in Table 2), and it too showed acceptable internal con-
to the specific course in which they completed the questionnaire. sistency (␣ ⫽ .90). We used these particular scales because both
Measures. Each measure (other than the newly created AES have been shown to produce scores that predict important
candidate items) was a Korean-translated version of a widely used student outcomes, including course grades (Skinner & Belmont,
and previously validated questionnaire that had been used success- 1993; Skinner, Kindermann, & Furrer, 2009). To assess cogni-
fully with previous Korean samples. Each questionnaire used a 1–7 tive engagement, we used the learning strategy items from the
Table 1
Correlations for All 10 Candidate Items With Two Measures of Agentic Motivation, Study 1
Item Psychological need satisfaction Academic efficacy
Original items to assess a unilateral contribution to the learning environment

1. During class, I ask questions. .39ⴱⴱ .39ⴱⴱ
1b. Revised version: During class, I ask questions to help me learn. .48ⴱⴱ .63ⴱⴱ
2. I tell the teacher what I like and what I don’t like. .27ⴱⴱ .22ⴱⴱ
3. I let my teacher know what I’m interested in. .36ⴱⴱ .36ⴱⴱ
4. I offer suggestions about how to make the class better. .29ⴱⴱ .27ⴱⴱ
5. During this class, I express my preferences and opinions. .45ⴱⴱ .40ⴱⴱ
New candidate items to assess a transactional contribution to the learning environment

6. I let my teacher know what I need and want. .36ⴱⴱ .32ⴱⴱ
7. When I need something in this class, I’ll ask the teacher for it. .36ⴱⴱ .44ⴱⴱ
New candidate items to assess a personal contribution to one’s own learning

8. I try to make whatever we are learning as interesting as possible. .55ⴱⴱ .57ⴱⴱ
9. I adjust whatever we are learning so I can learn as much as possible. .51ⴱⴱ .57ⴱⴱ
Note. N ⫽ 271. Items deleted from further consideration appear in italicized type.
ⴱⴱ
p ⬍ .01.
Metacognitive Strategies Questionnaire (Wolters, 2004). The candidate items that were distinct from the other three aspects
scale featured four items (listed in Table 2), and it showed of engagement, these two items were considered to be unduly
acceptable internal consistency (␣ ⫽ .84). We used this partic- confounded with cognitive engagement and were therefore re-
ular scale because it conceptualized cognitive engagement as moved.
strategic learning (e.g., using elaboration-based learning strat- The five items retained for the refined AES appear in Table 3.
egies) and because it had been shown to predict course grades The table provides the descriptive statistics associated with each
(Reeve & Tseng, 2011; Wolters, 2004). individual item and with the five-item scale as a whole. Table 3
provides the statistics associated with the samples utilized in
Results Studies 2 and 3 as well.
For each of the 10 candidate agentic engagement items, Table

Discussion
1 displays its correlations with psychological need satisfaction
and academic efficacy. Item 1 correlated well with both indi- The result of applying the construct and discriminant (or
cators of agentic motivation, but its revised version correlated factorial) validity criteria to the 10 candidate items was the
noticeably better with academic efficacy (rs ⫽ .63 vs. .39, z ⫽ five-item AES listed in Table 3. Overall, the refined scale
3.82, p ⬍ .01). Revised Item 1b was therefore preferred over showed strong internal consistency and produced a normal
original Item 1, because it was more closely associated with distribution of scores (see Table 3). An important question to
agentic motivation. Items 2 and 4 failed to correlate with either answer from Study 1 was whether the revised scale was superior
psychological need satisfaction or academic efficacy above a to the original scale. The revised scale represents a conceptually
threshold of 10% shared variance (r ⫽ .34; or R2 ⬎ .10), and psychometrically stronger instrument in four ways. First,
whereas all six remaining candidate items did. From these Item 1 was strengthened to restrict question-asking only to
correlations and from the need to test candidate items from all questions specifically related to learning. Second, two new
three categories of agentic engagement (i.e., unilateral, trans- items were added to represent students’ transactional (as well as
actional, and personal contributions), we retained Items 1b, 3, proactive) contributions into the learning environment. The
5, 6, 7, 8, and 9 for further analyses while we removed Items 1, addition of these two items (Items 1 and 3) allowed the AES to
2, and 4 from further consideration. better align with the conceptual definition of the construct (that
We next performed an exploratory factor analysis (EFA) on emphasized transactional contributions). Third, the two original
the seven retained agentic engagement candidate items, the five items that correlated only weakly with agentic-centric motiva-
behavioral engagement items, the five emotional engagement tions were removed. Fourth, the two candidate items designed
items, and the four cognitive engagement items. The four-factor to represent a personal contribution were not retained. While
solution (based on eigenvalues ⱖ 1) using principal axis fac- both items strongly correlated with agentic-centric sources of
toring with oblique rotation appears in Table 2. Of central motivation, their high factor loadings on the cognitive engage-
importance was whether the seven candidate items would load ment factor suggested that they expressed cognitive engage-
on their own separate (i.e., distinct) factor. Five candidate items ment at least as much as they expressed agentic engagement,
did load separately (on Factor 2), but the two new candidate and actually more so.
items designed to assess a personal contribution unexpectedly From the data obtained in Study 1, we conclude that the AES
loaded on the factor defined by the cognitive engagement items. is a psychometrically sound scale—a measure that is internally
Because the purpose of the factor analysis was to identify only consistent, closely aligned with agentic-centric sources of mo-
584 REEVE
Table 2
Factor Loadings From an Exploratory Factor Analysis Using Principal Axis Factoring With Oblique Rotation for the 21 Items
Assessing the Four Aspects of Student Engagement, Study 1
Factor 1 Factor 2 Factor 3 Factor 4

Questionnaire item/factor intercorrelations (47.6%) (10.3%) (8.3%) (4.6%)
Behavioral Engagement items

When I’m in this class, I listen very carefully. .86
I pay attention in this class. .82
I try hard to do well in this class. .59
In this class, I work as hard as I can. .57
When I’m in this class, I participate in class discussions. .35 .36
Agentic Engagement (candidate) items
I let my teacher know what I need and want. .79
I let my teacher know what I am interested in. .78
During this class, I express my preferences and opinions. .78

During class, I ask questions to help me learn. .56

When I need something in this class, I’ll ask the teacher for it. .51
I adjust whatever we are learning so I can learn as much as possible. ⫺.67
I try to make whatever we are learning as interesting as possible. ⫺.55
Cognitive Engagement items
When I study for this class, I try to connect what I am learning with my own experiences. ⫺.92
I try to make all the different ideas fit together and make sense when I study for this class. ⫺.86
When doing work for this class, I try to relate what I’m learning to what I already know. ⫺.65
I make up my own examples to help me understand the important concept I study for this class. ⫺.59
Emotional Engagement items
When we work on something in this class, I feel interested. ⫺.86
This class is fun. ⫺.81
I enjoy learning new things in this class. ⫺.75
When I’m in this class, I feel good. ⫺.68
When we work on something in this class, I get involved. .52
Factor intercorrelations
Factor 1: Behavioral Engagement — .32 ⫺.45 ⫺.67
Factor 2: Agentic Engagement — ⫺.44 ⫺.44
Factor 3: Cognitive Engagement — .56
Factor 4: Emotional Engagement —
Note. Factor loadings ⬍ .30 are not shown.
tivation, and meaningfully distinct from the other three aspects external regulation (motivated reactively by environmental con-
of engagement. Still, it is not yet clear if scores produced by the tingencies). By focusing on this distinction, the goal was to
AES will correlate negatively with agency-impoverished as- establish discriminant validity in that we expected agentic en-
pects of motivation, will explain independent variance in stu- gagement scores to correlate positively with autonomous mo-
dent achievement (i.e., horizontal line in Figure 1), and will tivation but negatively with controlled motivation.
explain student-initiated constructive changes in the learning Second, in a test of the scale’s discriminant validity, we
environment (i.e., curved line in Figure 1)— questions to be followed-up Study 1’s EFA with a confirmatory factor analysis
addressed in Studies 2 and 3. (CFA) in Study 2. In the CFA, we included items to represent
behavioral, emotional, cognitive, and agentic engagement and
Study 2 hypothesized that a corresponding four-factor model would fit
While the purpose of Study 1 was to refine the AES at the the data well (with no cross-loadings and no correlated errors).
item level, Study 2 sought to validate the scale as a whole. To Third, in a test of the scale’s predictive validity, we predicted
do so, we employed a three-part strategy. First, in a test of the that scores produced by the AES would explain independent
scale’s construct validity, we correlated participants’ scores on variance in students’ course achievement even after controlling for
the AES with their reports of autonomous and controlled mo- students’ behavioral, emotional, and cognitive engagements. This
tivation. In their self-determination theory, Ryan and Deci test of the scale’s unique predictive validity is important because
(2000, 2002) conceptualized autonomous motivation as an there are two essential criterion-related rationales to justify the
agentic-laden type of motivation that consists of both intrinsic need to add agentic engagement as a new fourth aspect of engage-
motivation (e.g., motivated proactively by interest) and identi- ment. First, agentic engagement can explain unique variance in
fied regulation (motivated proactively by value) while they student achievement. This is a first test of the unique predictive
conceptualized controlled motivation as an agentic- validity of the AES, and it constituted the primary purpose of
impoverished type of motivation that consists of both intro- Study 2. Second, agentic engagement can predict constructive
jected regulation (motivated reactively by internal pressure) and changes in the learning environment. This is a second test of the
Table 3
Descriptive Statistics for Each Individual AES Item and for the AES Scale as a Whole Across Studies 1, 2, and 3
Study 1 Study 2 Study 3

Variable M SD M SD M SD
Individual AES item

1. I let my teacher know what I need and want. 2.83 1.29 2.80 1.00 3.96 1.53
2. During this class, I express my preferences and opinions. 3.37 1.47 2.61 0.98 4.65 1.29
3. When I need something in this class, I’ll ask the teacher for it. 3.75 1.49 3.52 1.03 4.61 1.22
4. During class, I ask questions to help me learn. 4.03 1.42 2.91 1.05 4.11 1.46
5. I let my teacher know what I am interested in. 2.79 1.39 3.54 0.86 4.01 1.44
Overall 5-item AES score
Range 1–7 1–5 1–7
M 3.35 3.08 4.17
SD 1.13 0.71 1.09

␣
.86 .84 .81

Skewness (m3) 0.47 0.15 0.36
Kurtosis (m4) 0.31 ⫺0.18 0.03
t test for gender differences (1 ⫽ female; 2 ⫽ male) 1.78, p ⬍ .08 1.81, p ⬍ .08 0.02, ns
Sample 271 university 248 university 302 middle-school
engineering education majors physical education
majors from 6 from 4 classes students (at Time
classes 2) from 9 classes
Note. AES ⫽ Agentic Engagement Scale.
unique predictive validity of the AES, and it constituted the pur- course-related intrinsic motivation (e.g., “because it’s fun”), and
pose of Study 3. four items assessed identified regulation (e.g., “because it is some-
thing that is personally important to me”). To create a single index
Method of autonomous motivation, we followed the tradition established
within this area of research (e.g., Williams, Grow, Freedman,
Participants and procedure. Participants were 248 (132 fe- Ryan, & Deci, 1996) and combined these eight items into a single
male, 116 male) college students from a large university in Seoul, autonomous motivation score (␣ ⫽ .84). Four items assessed
South Korea who were taking one of four different sections of the
course-related introjected regulation (e.g., “because I’ll feel bad
same course in the Department of Education. All students (and
about myself if I don’t”), and four items assessed external regu-
their teachers) were ethnic Korean. Participation in the study was
lation (e.g., “because I’ll get in trouble if I don’t”). To create a
voluntary, and participants were told by the research assistant who
single index of controlled motivation, we again followed the
administered the questionnaire that their scores were confidential
tradition within this literature and combined these eight items into
and anonymous. Students who consented to participate completed
a single controlled motivation score (␣ ⫽ .76). As expected, the
the survey at the beginning of one class midway through the
two measures were moderately negatively correlated, r(248) ⫽
semester, while the achievement data (course grades) were col-
–.31, p ⬍ .01.
lected at the end of the semester. The survey took 5–10 min to
complete, and participants were asked to rate their experiences of Engagement. To assess agentic engagement, we used the five-
motivation and engagement in reference to the specific course in item scale developed in Study 1 (see Table 3). In Study 2, the AES
which they completed the questionnaire. showed an acceptable level of internal consistency (␣ ⫽ 84). To
Measures. As in Study 1, each measure came from a previ- assess the behavioral, emotional, and cognitive aspects of engage-
ously validated and widely used questionnaire that had been pre- ment, we used the same three questionnaires from Study 1 (see the
viously translated and successfully used in Korean. Each question- items listed in Table 2)—with two exceptions. Because it loaded
naire used a 1–5 Likert-type scale with the following response poorly as a behavioral engagement item in the Study 1 factor
options: 1 ⫽ strongly disagree; 2 ⫽ disagree; 3 ⫽ neither agree analysis (see Table 2), we excluded behavioral engagement Item 5
nor disagree; 4 ⫽ agree; and 5 ⫽ strongly agree. (“When I’m in this class, I participate in class discussions.”) from
Motivation. To report their class-specific motivation, partici- Study 2. And, because it loaded poorly as an emotional engage-
pants completed the Academic Self-Regulation Questionnaire ment item, we excluded emotional engagement Item 5 (“When we
(ASRQ; Ryan & Connell, 1989). The ASRQ features the stem “I work on something in this class, I get involved.”). All three
do my classwork . . . ” followed by 16 items to assess four aspects measures showed acceptable levels of internal consistency in
of student motivation that lie along a continuum from highly Study 2, including behavioral engagement (four-items, ␣ ⫽ .87),
autonomous to highly controlled motivation. Two scales—intrin- emotional engagement (four-items, ␣ ⫽ .91), and cognitive en-
sic motivation and identified regulation—are conceptualized as gagement (four-items, ␣ ⫽ .72).
autonomous and proactive types of motivation, and two scales— Achievement. For course-specific achievement, we collected
introjected regulation and external regulation—are conceptualized each student’s final semester grade from the objective school
as controlled and reactive types of motivation. Four items assessed record for the particular class in which he or she completed the
586 REEVE
questionnaires. Course achievement scores were reported on a The hypothesized 17-item, four-factor model fit the data well,
0 –100 point scale. ␹2 (266) ⫽ 360.61, p ⬍ .01, RMSEA (90% confidence interval
Data analysis. The hypothesized measurement model to test [CI]) ⫽ .076 (.065–.088), SRMR ⫽ .069, CFI ⫽ 0.98, NNFI ⫽
the scale’s discriminant validity assessed behavioral, emotional, 0.98, and all five agentic engagement items loaded positively and
cognitive, and agentic engagement as latent variables. To assess highly (p ⬍ .001) on the latent factor. In the CFA, we included no
behavioral, emotional, and cognitive engagement, we used the four cross-loadings and no correlated error terms. The standardized and
items from their respective scales as observed indicators, and for unstandardized beta weights associated with all 17 indicators in-
agentic engagement we used the five items from the AES as cluded in this analysis appear in Table 4.3
observed indicators. The measurement model therefore included Unique predictive validity. To test the unique predictive
17 indicators for four latent variables. The hypothesized structural validity of the AES, we tested whether each aspect of engagement
model to test the scale’s predictive validity used these same 17 could serve as an independent predictor of student achievement
indicators and four latent variables but also added student gender (i.e., final course grade). The test of the 17-item, four-factor model
as a single-indicator predictor variable and course grade as a in the preceding discriminant validity analysis confirmed the fit of
single-indicator of the achievement outcome. To prepare these two the underlying measurement model. To create the structural (i.e.,
models for statistical testing, we first calculated multilevel analy- hypothesized) model, we simply added gender to serve as a single-
ses using hierarchical linear modeling (HLM, Version 7; Rauden- item predictor (a statistical control) and course grade to serve as
bush, Bryk, Cheong, Congdon, & du Toit, 2011) to determine the single indicator for the achievement outcome. The intercorre-
whether between-teacher differences affected students’ self- lations among the four aspects of engagement (as latent variables)
reports and course grades. The percentage of variance attributable and course achievement appear below the diagonal in Table 5.
to between-teacher differences determined from unconditional Each latent variable correlated significantly with course achieve-
models exceeded 10% for two of the 18 measured variables, and ment (rs ranged from .21 to .42, ps ⬍ .01). Gender did not
the mean intraclass correlation (ICC) across the 18 measured correlate with any of the four engagement latent variables, but it
variables was 5.9%. The ICCs associated with the 17 engagement did correlate with achievement (r ⫽ .16, p ⬍ .01; as females had
indicators appear in the right most column of Table 4 (the ICC for somewhat higher course grades than did males), so we included
the 18th indicator, course grade, was 0.4%). Given the meaningful gender as a fifth predictor to function as a statistical control
between-teacher effects on some of the assessed measures, we (male ⫽ 1, female ⫽ 2). The hypothesized structural model fit the
chose to use multilevel structural equation modeling (LISREL 8.8; data well, ␹2 (327) ⫽ 436.36, p ⬍ .01, RMSEA (90% CI) ⫽ .074
Jöreskog & Sörbom, 1996) in the test of both the hypothesized (.063–.085), SRMR ⫽ .068, CFI ⫽ .98, NNFI ⫽ .98. Collectively,
measurement model (the CFA) and the hypothesized structural the four aspects of engagement (and gender) explained 25% of the
model (to predict achievement). In the conduct of the multilevel variance in course achievement. Only two of the hypothesized
HLM analyses, we used maximum-likelihood estimation. By using paths were individually significant, however, as behavioral en-
multilevel modeling, the results may be interpreted as student-level gagement (beta ⫽ .28, t ⫽ 2.80, p ⬍ .01) and agentic engagement
results (n ⫽ 248) that are statistically independent of any teacher- (beta ⫽ .25, t ⫽ 3.20, p ⬍ .01) individually predicted achievement
level effects within the data. Unfortunately, the low number of while this was not true for either emotional engagement (beta ⫽
teachers/classrooms (k ⫽ 4) prevented us from analyzing these .07, t ⫽ 0.87, ns) or cognitive engagement (beta ⫽ –.06, t ⫽ 0.75,
data at the teacher level. To evaluate model fit we relied on the ns). Gender also significantly predicted achievement. The dia-
chi-square test statistic and multiple indices of fit (as recom- gram showing these standardized parameter estimates appears
mended by Kline, 2011), including the root-mean-square error of in Figure 2.
approximation (RMSEA), the standardized root-mean-square re-
sidual (SRMR), the comparative fit index (CFI), and the non- Discussion
normed fit index (NNFI).2
In this second data set, the overall five-item AES showed strong
internal consistency, produced a normal distribution of scores, was
Results associated positively with autonomous motivation, was associated
negatively with controlled motivation, and explained independent
Construct validity. The five-item AES correlated positively
variance in student achievement that the other three aspects of
and significantly with the measure of autonomous motivation,
engagement were unable to explain. These data suggest that the
r(248) ⫽ .44, p ⬍ .01, and negatively and significantly with
controlled motivation, r(248) ⫽ –.14, p ⬍ .05. These correlations
appear above the diagonal in Table 5 in rows 1 and 2. As can be 2
RMSEA and RMSR values of .08 or less indicate good fit, at least as
seen from these correlations, the positive correlation between long as the upper bound of the RMSEA’s 90% confidence interval is ⱕ .10
autonomous motivation and agentic engagement and the negative (Hu & Bentler, 1999; Kline, 2011); CFI and NNFI values of .95 or greater
correlation between controlled motivation and agentic engagement indicate good fit, at least as long as these values co-occur with a SRMR
value of ⱕ .08 (Hu & Bentler, 1999; Kline, 2011).
were similarly true for the other three aspects of engagement. Thus 3
AES Item 3 had a noticeably low factor loading of ␤ ⫽ .30. This same
engagement in general—rather than agentic engagement in partic- item did not have a low factor loading in the data sets from Studies 1 and
ular—was associated positively with autonomous motivation and 3, however. Item 3 had a low factor loading in the Study 2 data set because
negatively with controlled motivation. this item also had a very high ICC of 38.6%. A close inspection of
students’ responses on this item showed that scores were heavily skewed
Discriminant validity. We calculated a multilevel confirma- by one particular teacher/class, as students in one class reported substan-
tory factor analysis (CFA) to assess how well the 17 items assess- tially higher scores on this item than did students from the other three
ing the different aspects of engagement fit to a four-factor model. classes.
Table 4
Unstandardized and Standardized Beta Weights and Interclass Correlation Coefficients Associated With All 17 Observed Indicators
Within the Measurement Model, Study 2
Observed variable B SE B ␤ ICC (%)
Latent factor including the Behavioral Engagement items

When I’m in this class, I listen very carefully. .98 .06 .87 4.3
I pay attention in this class. 1.00 .89 4.8
I try hard to do well in this class. .79 .06 .70 3.0
In this class, I work as hard as I can. .77 .06 .68 2.1
Latent factor including the Agentic Engagement items
I let my teacher know what I need and want. .92 .07 .80 6.9
I let my teacher know what I am interested in. 1.00 .87 1.5
During this class, I express my preferences and opinions. .34 .07 .30 38.6
During class, I ask questions to help me learn. .65 .07 .57 17.4
When I need something in this class, I’ll ask the teacher for it. .73 .07 .64 8.2
Latent factor including the Cognitive Engagement items

When I study for this class, I try to connect what I am learning . . . .93 .11 .66 3.1
I try to make all the different ideas fit together and make sense . . . 1.00 .71 2.5
When doing work for this class, I try to relate what I’m learning . . . .92 .11 .65 3.3
I make up my own examples to help me understand the concept . . . .67 .11 .48 3.2
Latent factor including the Emotional Engagement items
When we work on something in this class, I feel interested. .96 .05 .88 2.7
This class is fun. .91 .05 .84 1.0
I enjoy learning new things in this class. 1.00 .92 0.6
When I’m in this class, I feel good. .84 .05 .76 2.7
Note. ICC ⫽ interclass correlation coefficient.
AES is a reliable and valid scale, but they do not speak to that early-semester changes in agentic engagement would longitu-
the crucial and as yet unexplored question of whether scores on the dinally predict late-semester changes in perceived teacher
AES can predict student-initiated constructive changes in the autonomy-support (after controlling for both initial and midsemes-
learning environment. Accordingly, Study 3 was designed to test ter perceived autonomy support).
this prediction. In doing so, Study 3 deliberately sampled middle-
school students, because the first two data sets used college stu- Method
dents. By sampling middle-school students, we returned to the type
of sample from which the AES item content originated (i.e., field Participants and procedure. Participants who consented to
notes from middle- and high-school classrooms using the Hit-Steer participate were 315 middle-school students (146 female, 169
Observation System). We also wanted assurance that the findings male) from nine physical education (PE) classes situated in nine
in Studies 1 and 2 were not limited to the engagement dynamics of different schools in the Seoul, Korea metropolitan area. All stu-
college classrooms but could also be extended to the engagement dents (and their teachers) were ethnic Korean. Class sizes averaged
dynamics of middle-school classrooms. 35.0 students per class (range ⫽ 31 to 38). Students attended their
PE class each day, and each class lasted 55 min.
Participants completed a brief questionnaire three times during
Study 3
the semester—2 weeks into the semester (T1), a week after the
If agentic engagement allows students to create more motiva- midterm exam (T2), and the next-to-last week of the semester
tionally supportive learning environments for themselves, then (T3). The T1 questionnaire assessed students’ demographic infor-
students who begin a class with high levels of agentic engagement mation, class-specific agentic engagement, and perceptions of
should experience a corresponding constructive change in their teacher-provided autonomy support; the T2 questionnaire assessed
classroom learning environment over the course of the semester. students’ agentic engagement and perceived autonomy support;
This change should occur as students act agentically during in- and the T3 questionnaire assessed only perceived autonomy sup-
struction—as they express their preferences and interests, ask port. The research assistant who administered the questionnaire
questions, and communicate what they want and need. To test this told participants that their responses would be confidential, anon-
hypothesis, we assessed students’ perceptions of teacher-provided ymous, and used only for purposes of the research study. Three
autonomy support at the beginning, middle, and end of an 18-week hundred fifteen (315) students agreed to complete the question-
semester. The first prediction was that early-semester agentic naire at T1, while only 308 of these same students agreed to
engagement would longitudinally predict midsemester changes in complete the questionnaire at T2. The loss of seven students at T2
perceived teacher autonomy-support (after controlling for early- represented a dropout rate of 2.2%. T2 persisters did not differ
semester perceived autonomy support). In addition, we expected from dropouts on either T1 agentic engagement or T1 perceived
that students who experience increased agentic engagement during autonomy support (ts ⬍ 1). Three hundred two (302) students,
the course would experience a corresponding constructive change including 139 females and 163 males, agreed to complete the
in their learning environment. The second prediction was therefore questionnaire at all three time points. The loss of an additional 6
588 REEVE
Table 5
Descriptive Statistics and Intercorrelations Among the Observed Variables (Above the Diagonal) and Latent Variables (Below the
Diagonal), Study 2
Variable M SD 1 2 3 4 5 6 7 8
1. Autonomous Motivation 3.55 0.62 — ⫺.31ⴱⴱ .44ⴱⴱ .52ⴱⴱ .69ⴱⴱ .53ⴱⴱ .26ⴱⴱ ⫺.01
2. Controlled Motivation 2.53 0.80 — ⫺.14ⴱ ⫺.12 ⫺.30ⴱⴱ ⫺.17ⴱⴱ ⫺.03 .05
3. Agentic Engagement 3.08 0.71 — .49ⴱⴱ .41ⴱⴱ .47ⴱⴱ .32ⴱⴱ ⫺.12
4. Behavioral Engagement 3.58 0.75 .53 — .58ⴱⴱ .30ⴱⴱ .40ⴱⴱ .01
5. Emotional Engagement 3.86 0.75 .40 .65 — .40ⴱⴱ .30ⴱⴱ .01
6. Cognitive Engagement 3.72 0.66 .45 .55 .42 — .21ⴱⴱ ⫺.05
7. Academic Achievement 76.9 15.3 .37 .42 .32 .21 — .17ⴱⴱ
8. Gender (Male ⫽ 1, Female ⫽ 2) 1.53 0.50 ⫺.12 .00 .01 ⫺.12 .16 —
Note. N ⫽ 248. Possible range for all variables ⫽ 1–5, except for academic achievement, which ranged from 0 to 100. Descriptive statistics and the
correlations above the diagonal are for observed (measured) variables; correlations below the diagonal listed in italics are for latent variables.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01.
students at T3 represented an overall study-wide dropout rate of percentage of variance attributable to between-teacher differences
4.1% (13/315), or a retention rate of 95.9% (302/315). The six T3 determined from unconditional models exceeded 10% in six of the
dropouts did differ from the 302 T3 persisters on T1 and T2 28 measured variables with a mean intraclass correlation (across
perceived autonomy support (ps ⬍ .01) and on T1 agentic engage- all 28 measured variables) of 5.5%. Given these between-teacher
ment (p ⬍ .01), but not on T2 agentic engagement. These data effects, we chose to use multilevel structural equation modeling
suggest that while the sample loss at T2 did not bias the analyzed with maximum-likelihood estimation in the tests of the measure-
sample, the sample loss at T3 did in that it somewhat underrepre- ment and structural models. To evaluate model fit, we again relied
sented initially agentically disengaged students who reported low on the chi-square test statistic and multiple indices of fit.
autonomy support.
Measures. The questionnaires to assess students’ agentic en-
Results
gagement and perceived autonomy support used a 1–7 Likert-type
scale with the following response options: 1 ⫽ strongly disagree; The descriptive statistics for and the intercorrelations among the
2 ⫽ disagree; 3 ⫽ slightly disagree; 4 ⫽ neither agree nor five latent variables included in Study 3 appear in Table 6.
disagree; 5 ⫽ slightly agree; 6 ⫽ agree; and 7 ⫽ strongly agree. Main analyses. In testing both the measurement and structural
Agentic engagement. To assess agentic engagement, we used models, we allowed (a) the between-wave error term of each
the five-item scale AES shown in Table 3. In Study 3, the AES observed indicator to correlate with itself from T1 to T2, from T2
showed an acceptable level of internal consistency at both T1 (␣ ⫽ to T3, and from T1 to T3 and (b) the errors between the within-
.77) and T2 (␣ ⫽ .81). wave T2 variables to correlate in the structural model (as depicted
Perceived autonomy support. To assess students’ perceptions in the curved line between T2 perceived autonomy support and T2
of teacher-provided autonomy support, participants completed the agentic engagement in Figure 3). The measurement model fit the
six-item short version of the Learning Climate Questionnaire data well, ␹2 (723) ⫽ 857.14, p ⬍ .01, RMSEA (90% CI) ⫽ .059
(LCQ; Williams & Deci, 1996). The short-version of the LCQ has (.052–.066), SRMR ⫽ .047, CFI ⫽ .99, NNFI ⫽ .99, and each
been widely used in classroom-based investigations of autonomy item designed to assess its corresponding latent construct loaded as
support (Black & Deci, 2000; Jang et al., 2009), and includes the expected. The structural model testing the hypothesized model also
following six items: “I feel that my teacher provides me with fit the data well, ␹2 (725) ⫽ 863.71, p ⬍ .01, RMSEA (90% CI) ⫽
choices and options”; “I feel understood by my teacher”; “My .060 (.053–.067), SRMR ⫽ .049, CFI ⫽ .99, NNFI ⫽ .99. Both
teacher encourages me to ask questions”; “My teacher listens to hypothesized paths from agentic engagement to perceived auton-
how I would like to do things”; “My teacher conveys confidence omy support were individually significant, as early-semester (i.e.,
in my ability to do well in the course”; and “My teacher tries to initial) agentic engagement predicted midsemester perceived au-
understand how I see things before suggesting a new way to do tonomy support controlling for early-semester perceived autonomy
things.” The LCQ showed strong reliability across all three waves support (beta ⫽ .30, t ⫽ 4.15, p ⬍ .01), and early-semester
of data collection (␣s of .87, .88, and .92 at T1, T2, and T3). changes in agentic engagement predicted late-semester changes in
Data analysis. Both agentic engagement and perceived auton- perceived autonomy support controlling for early and midsemester
omy support were assessed and entered into the analyzed model as perceived autonomy support (beta ⫽ .23, t ⫽ 3.09, p ⬍ .01). The
latent variables. For agentic engagement, we used the five items path diagram showing the standardized estimates for each path in
from the AES as observed indicators. For perceived autonomy the model appears in Figure 3.
support, we used the six items from the LCQ as observed indica- Supplemental analyses. In follow-up analyses, we checked
tors. Hence, the data analysis involved 28 indicators for five latent for three possible results. First, we tested whether the two excluded
variables. To prepare the hypothesized measurement and structural (nonhypothesized) paths from T1 perceived autonomy support to
models for statistical testing, we first calculated multilevel analy- T2 agentic engagement and from T1 agentic engagement to T3
ses using hierarchical linear modeling to determine whether perceived autonomy support might be significant. Adding the path
between-teacher differences affected students’ self-reports. The from T1 perceived autonomy support to T2 agentic engagement
Figure 2. Standardized parameter estimates for the test of the predictive validity model, Study 2. Solid lines
represent significant paths; dashed lines represent nonsignificant paths. Disturbances with gender are not shown,
but they were as follows: agentic engagement, –.12; behavioral engagement, –.01; emotional engagement, .01;
cognitive engagement, –.12.
did not result in an improved model fit, ⌬␹2 (1) ⫽ 1.78, ns, and agentic engagement had on T3 perceived autonomy support (i.e.,
that added path was not individually significant (beta ⫽ .09, t ⫽ r ⫽ .46, p ⬍ .01, from Table 6). T2 agentic engagement did
1.35, ns). The reason why we did not expect this path to be significantly mediate this direct effect (z ⫽ 2.91, p ⬍ .01). Thus,
individually significant was because teacher-provided autonomy changes in agentic engagement (T2)—not initial agentic engage-
support facilitates student motivation, and it is student motivation, ment (T1)— explained late-semester changes in (T3) perceived
rather than teacher-provided autonomy support, that facilitates and autonomy support.
explains changes in students’ engagement, including students’ Third, we tested for a possible (but not hypothesized) inter-
agentic engagement (see Jang et al., 2012). Adding the direct path action effect between the two predictors at both T1 and T2. To
from T1 agentic engagement to T3 perceived autonomy support do so, we created a first latent interaction variable involving the
did not result in an improved model fit, ⌬␹2 (1) ⫽ 1.18, ns, and two T1 latent variables to test for their interactive relation to T2
that added path was not individually significant (beta ⫽ –.09, t ⫽ perceived autonomy support, and we created a second latent
1.07, ns). The reason why we did not expect this path to be interaction variable involving the two T2 latent variables to test
individually significant was because we expected T2 agentic en- for their interactive relation to T3 perceived autonomy support.
gagement, rather than T1 agentic engagement, to predict and To do so, we used the single product indicator approach intro-
explain changes in students’ T3 perceived autonomy support. duced by Kenny and Judd (1984), refined by Jöreskog and Yang
Second, we used a Sobel test to evaluate if T2 agentic engage- (1996), and validated by Li et al. (1998). This seven latent
ment did actually fully mediated the otherwise direct effect that T1 variable model (the five latent variables included in Figure 3
590 REEVE
Table 6
Descriptive Statistics and Intercorrelations Among the Variables in Study 3
Variable M SD 1 2 3 4 5
1. Agentic Engagement, T1 4.03 1.08 — .65 .53 .48 .46

2. Agentic Engagement, T2 4.17 1.09 — .34 .68 .59
3. Perceived Autonomy Support, T1 4.23 1.05 — .50 .48
4. Perceived Autonomy Support, T2 4.31 1.07 — .69
5. Perceived Autonomy Support, T3 4.40 1.16 —
Note. N ⫽ 302. Possible range for all variables ⫽ 1–7. T1 ⫽ Time (or Wave) 1; T2 ⫽ Time 2; T3 ⫽ Time 3. Descriptive statistics are for observed
(measured) variables; intercorrelations are for latent variables. All correlations are statistically significant, p ⬍ .01.
plus the interaction of the two T1 exogenous variables and the Discussion
interaction of the two T2 endogenous variables) did not fit the

data well, ␹2 (832) ⫽ 2,665.54, p ⬍ .01, RMSEA (90% CI) ⫽ The overall five-item AES again showed strong internal consis-
.083 (.077–.088), SRMR ⫽ .171, CFI ⫽ .90, NNFI ⫽ .89. The tency and produced a normal distribution of scores. It also showed
added paths from the T1 interaction variable to T2 perceived predictive validity in terms of explaining a student-initiated con-
autonomy support was not significant (beta ⫽ .00, t ⫽ 0.03, ns), structive change in the learning environment. Middle-school stu-
and the added path from the T2 interaction variable to T3 dents who reported high levels of agentic engagement at the
perceived autonomy support was not significant (beta ⫽ ⫺0.05, beginning of the course perceived that their teachers became
t ⫽ 1.19, ns). These two nonsignificant interaction terms sug- significantly more autonomy-supportive toward them by midse-
gest that the effects of agentic engagement on changes in mester, and middle-school students who reported early-semester
perceived autonomy support (shown in Figure 3) were robust gains in agentic engagement perceived that their teachers became
rather than being qualified or moderated either by participants’ significantly more autonomy-supportive toward them late in the
levels of agentic engagement or by their perceptions of auton- semester. These data show that agentic engagement functioned as
omy support. a student-initiated pathway to produce a more motivationally sup-
Figure 3. Standardized parameter estimates for the test of the hypothesized model, Study 3. Solid lines
represent significant paths, p ⬍ .01. AS ⫽ autonomy support; AE ⫽ agentic engagement.
portive learning environment. What these data do not show, how- horizontal “agentic engagement” directional arrow. The findings
ever, is that agentic engagement predicted unique variance in from Study 3 showed that students can, through agentic acts of
teacher-provided autonomy support, because Study 3 did not in- engagement, further create a more motivationally supportive learn-
clude measures of the other three aspects of engagement—a key ing environment for themselves, at least in respect to how auton-
limitation of the study that will be discussed in the general dis- omy supportive they perceive their teachers to be. This second
cussion section. benefit was depicted graphically in Figure 1 by the curved “agentic
The longitudinal effect of students’ initial agentic engagement engagement” directional arrow. Together, these two benefits (i.e.,
on T2 autonomy support was large—about the same in magnitude greater achievement, greater autonomy support) typify and effec-
as was the effect of T1 autonomy support on its T2 assessment tively realize what Albert Bandura meant by his pioneering con-
(betas of .30 vs. .34). This suggests that students perceived their cept of agency: “To be an agent is to influence intentionally one’s
teachers as autonomy supportive at midsemester partly because functioning and life circumstances” (Bandura, 2006, p. 164).
teachers were initially autonomy supportive but also because stu-
dents’ own agentic engagement tended to open up teachers’ mo-
What Is Agentic Engagement?
tivational responsiveness. The longitudinal effect of students’ T2

agentic engagement on T3 autonomy support was only half the Because agentic engagement is proposed as a new educational
magnitude as was the effect of T2 autonomy support on its T3 construct, it is important to be clear about its conceptual nature. It
assessment (betas of .23 vs. .46). This suggests that teachers’ is foremost a type of engagement. Like acting behaviorally, emo-
motivating styles were more stabilized in the second half of the tionally, and cognitively are pathways to positive student out-
semester, although students’ acts of agentic engagement did nev- comes, acting agentically is another student-initiated pathway to
ertheless exert a significant positive effect on teachers’ late- positive outcomes. The five items on the AES identify explicitly
semester openness and responsiveness to support students’ auton- what agentically engaged students are doing—namely, expressing
omy. their preferences, asking questions, and letting the teacher know
One possible criticism of these findings might be that agentic what they like, need, and want. In addition to speaking up in class,
engagement predicted perceived autonomy support simply because agentic engagement also likely occurs more subtly, as through
of item overlap between the two questionnaires. One item on the private conversations and e-mails, various forms of student input,
LCQ is, for instance, “My teacher encourages me to ask ques- and acts of cooperation, negotiation, and reciprocation that allow
tions,” while one item on the AES is, “During class, I ask questions students and teachers to be more in synch with one another.
to help me learn.” It is unlikely that a methodological artifact Agentic engagement is also a new, distinct educational con-
explains the findings, however, because the findings reveal more struct. It is different from other aspects of engagement (as per the
than just a cross-sectional correlation between the two measures. factor analyses in Studies 1 and 2), and it is different from
The longitudinal research design showed that after controlling for classroom events such as formative assessment, clickers, construc-
this T1 correlation (i.e., after controlling for the potential method- tivistic learning environments, and adaptive help seeking (as dis-
ological artifact), T1 agentic engagement predicted changes in cussed earlier). It is also different from more encompassing con-
perceived autonomy support, and it did so at both T2 and T3. structs, such as self-regulated learning (SRL; Zimmerman &
Rather than suggesting a methodological artifact, what this longi- Schunk, 2011). Self-regulation theorists view SRL “as proactive
tudinal effect suggests is an emerging student-teacher synchrony processes that students use to acquire academic skill, such as
as teachers became more responsive to students’ initiatives (in- setting goals, selecting and deploying strategies, and self-
cluding question asking). We would add that interpersonal syn- monitoring one’s effectiveness, rather than as a reactive event that
chrony is a defining characteristic of most high-quality relation- happens to students” (Zimmerman, 2008, pp. 166 –167). Those
ships (De Wolff & van IJzendoorn, 1997; Kochanska, 2002). “proactive processes” include forethought (which revolves around
motivation), performance (which revolves around engagement),
and reflection (which revolves around the outcome). The efforts to
General Discussion
distinguish agentic engagement from SRL and to situate this new
Engagement is motivated action that functions as a student- educational construct within the SRL framework raise two key
initiated pathway to positive educational outcomes (Skinner, Kin- questions: (a) Can agentic engagement be explained within the
derman, Connell, & Wellborn, 2009; Skinner, Kindermann, & SRL framework to the point that this new concept is not needed—
Furrer, 2009). Past research had confirmed that students’ behav- because it already exist by another name? and (b) Where would
ioral, emotional, and cognitive engagements help them make ac- agentic engagement fit if it were to be integrated within the SRL
ademic progress, so the present study sought to extend this liter- framework?
ature by investigating agentic engagement as a potential fourth The answer to the first question is that agentic engagement
student-initiated pathway—not only to well-established learning- cannot be subsumed within the SRL framework, at least not within
related outcomes but also to the new student outcome of a more the framework’s current conceptualization (Zimmerman & Sc-
motivationally supportive learning environment. hunk, 2011). The key difference between the two approaches is the
The findings from Study 2 confirmed that acts of agentic en- role and function of the teacher in supporting the learner’s moti-
gagement uniquely predicted course-specific achievement. This vation and academic progress. Agentically engaged students work
suggests that agentically engaged students are taking achievement- transactionally with the teacher to create learning conditions that
facilitating action during learning activities that is above and can vitalize their otherwise latent inner motivational resources
beyond their applications of effort, enthusiasm, and strategic think- (e.g., autonomy-supportive teaching; Ryan & Deci, 2000). That is,
ing. This first benefit was depicted graphically in Figure 1 by the agentically engaged students solicit and rely on their teachers’
592 REEVE
professional insight, teaching skill, and classroom experience to being (Assor, Kaplan, & Roth, 2002; Cheon, Reeve, & Moon,
create motivationally enriching classroom conditions for them, 2012; Vansteenkiste, Simons, Lens, Sheldon, & Deci, 2004). Be-
such as by providing choices to vitalize their otherwise latent cause of these benefits, it becomes important to understand why
autonomy, optimal challenges to vitalize their competence, and teachers are (and are not) autonomy supportive and also how they
social interaction opportunities to vitalize their relatedness. In SRL might learn to become more autonomy supportive toward students.
theory, social agents are also very important, but their role is to Intervention-based research shows that teachers can learn how to
shift a student’s learning from initially social to eventually self become more autonomy supportive, but this same research also
sources of regulation (Schunk & Zimmerman, 1997). That is, shows that teachers need considerable help and expert guidance to
through initial modeling and social guidance to later social emu- learn how to do so (Su & Reeve, 2011).
lation and corrective feedback, students gradually acquire a greater It strikes us as entirely possible that highly agentically engaged
capacity to regulate their own learning in the absence of social students can provide teachers with in-class opportunities to self-
supports (e.g., during homework; Stoeger & Ziegler, 2011). That learn how to become more autonomy supportive. The defining
is, the goal of SRL is to develop independent learners who can characteristics of autonomy support include taking the students’
monitor and regulate their thinking, wanting, and acting, while the perspective, welcoming their thoughts, feelings, and behaviors into
goal of agentic engagement is to recruit the interpersonal support the delivery of instruction, and providing learning activities in
necessary to create a motivationally supportive learning environ- ways that vitalize (rather than neglect or frustrate) students’ inner
ment. motivational resources (Reeve, 2009). Agentically engaged stu-
The answer to the second question is that agentic engagement dents therefore provide teachers with opportunities to learn how to
would fit within SRL’s “Performance” phase. The Performance be more autonomy supportive when they communicate their per-
phase includes behavioral (i.e., attention focusing, effort) and spective, add their thoughts, feelings, and behaviors into the flow
cognitive (i.e., self-instruction, task strategies, metacognitive mon- of instruction, and let teachers know what they like, need, and
itoring) engagement, but it includes only the suggestive hint of want. To the extent that this is true—and the findings reported in
agentic engagement (i.e., environmental restructuring, help seek- Study 3 did provide rather compelling evidence that students’
ing). What the new concept of agentic engagement can do for the agentic engagement contributes to constructive changes in teach-
SRL framework is what it can do for all educational models of
ers’ motivating styles—then it changes the notion of a teacher’s
students’ motivation and engagement—namely, expand what it
classroom motivating style away from something that resides
means for a motivated learner to be actively involved during a
within the teacher and toward something that, at least in part,
learning activity.
unfolds during interaction with students (Sameroff, 2009; Samer-
The two failed candidate AES items from Study 1 to represent
off & Fiese, 2000).
students’ personal contribution to their own learning (Items 8 and
9 in Table 1) help distinguish SRL and what self-regulated learners
do from agentic engagement and what agentically engaged stu- Limitations and Future Research
dents do. These two items were “I try to make whatever we are
learning as interesting as possible” and “I adjust whatever we are Study 3 showed that agentic engagement predicted longitudinal
learning so I can learn as much as possible.” Both items loaded changes in perceived autonomy support. This conclusion is lim-
with the cluster of cognitive engagement items rather than with the ited, however, because the only aspect of engagement included in
cluster of agentic engagement items. In doing so, they better the study was agentic engagement. Failing to include measures of
captured what cognitively engaged SRLs do—try to gain control behavioral, emotional, and cognitive engagement leaves open the
over their own learning. To help student gain better self-control question of whether any of these other aspects of engagement
over their own learning, teachers tutor, coach, mentor, and scaffold might also individually predict unique variance in perceived au-
students’ forethought, performance, and reflection. In doing so, tonomy support. Thus, while the findings in Study 3 supported the
they help students acquire and then refine ever-more sophisticated conclusion that student-initiated acts of agentic engagement help
self-regulatory capacities. This is, however, fundamentally differ- transform teachers’ motivating styles, they did not establish agen-
ent from the instructional effort to vitalize students’ inherent— tic engagement as the unique engagement-based predictor of these
although sometimes latent—inner motivational resources (Reeve, changes. It may be, but its unique status has not yet been tested.
Ryan, Deci, & Jang, 2008).4 Behaviorally engaged and cognitively Our expectation is that agentic engagement would be the only T1
engaged student may attend to, emulate, and internalize their engagement predictor of these changes in the learning environ-
teachers’ tutoring, coaching, mentoring, and scaffolding, but agen- ment, because it is uniquely proactive, intentional, and collabora-
tically engaged students uniquely try to work collaboratively with tive in ways that the other three aspects are not. However, changes
their teachers to create a personally more motivationally support- in behavioral engagement might also uniquely predict changes in
ive learning environment. Appreciating what these agentically perceived autonomy support. Even if they did, however, these
engaged students are doing seems like a profitable future area of effects would likely be only indirect or inadvertent (as explained in
research (and practice) for SRL theorists. the Introduction).
Agentic Engagement as an Antecedent of Autonomy 4

In self-determination theory, what teachers do to cultivate students’
Support self-regulatory capacities is referred to as teacher-provided structure, while
what teachers do in response to students’ acts of agentic engagement is
A teacher’s autonomy support benefits students in a multitude of referred to as teacher-provided autonomy support (Jang, Reeve, & Deci,
ways, including enhancing their motivation, learning, and well- 2010).
One limitation that extended across all three studies was incon- Conclusion
sistent sampling. Study 1 used only a convenience sample of
engineering studies, Studies 1 and 2 used college students, while Across three studies, agentic engagement was associated posi-
Study 3 used middle-school students. These samples differed in tively with agentic-rich motivation and negatively with agentic-
impoverished motivation, contributed uniquely to course-specific
terms of their ages, subject matters, and classroom dynamics, so it
achievement and predicted longitudinal changes in teachers’ class-
was reassuring to see that the distribution of AES scores produced
room motivating style. The general conclusion is that agentic
by these varying samples were fairly similar (see Table 3). Another
engagement is a new and constructive aspect of student engage-
sampling concern was gender, and it is not yet clear if agentic
ment that allows educators to more fully appreciate how students
engagement is different between males and females. All three
actually engage themselves in learning activities, as they not only
studies tested for gender differences, and none found significant
try to learn and develop skill, but they also try to create a more
differences, although there was a trend in Studies 1 and 2 for males
motivationally supportive learning environment for themselves.
to report higher levels than females. One characteristic that did not
vary from study to study was that all three studies sampled only
Korean students. Students may experience and express their class- References
room engagement differently from one culture to the next, and

these different ways of trying to make academic progress may Assor, A., Kaplan, H., & Roth, G. (2002). Choice is good, but relevance is
excellent: Autonomy-enhancing and suppressing teaching behaviors pre-
possess different predictive utility from one culture to the next.
dicting students’ engagement in schoolwork. British Journal of Educa-
Hence, the hypotheses tested in the present study need to be tional Psychology, 72, 261–278. doi:10.1348/000709902158883, doi:
evaluated in non-Korean classroom contexts as well (especially in 10.1348/000709902158883
Western classroom contexts). Bandura, A. (1997). Self-efficacy: The exercise of control. New York, NY:
Study 3 was limited in that it included only self-report data. An Freeman.
objective measure of teachers’ autonomy-supportive motivating Bandura, A. (2006). Toward a psychology of human agency. Perspectives
style (e.g., scores from trained raters) would be a welcomed on Psychological Science, 1, 164 –180. doi:10.1111/j.1745-6916.2006
addition to this methodology (as per Cheon et al., 2012; Reeve, .00011.x
Jang, Carrell, Jeon, & Barch, 2004; Tessier, Sarrazin, & Ntouma- Black, A. E., & Deci, E. L. (2000). The effects of instructors’ autonomy
support and students’ autonomous motivation on learning organic chem-
nis, 2008).
istry: A self-determination theory perspective. Science Education, 84,
Adding agentic engagement as a fourth aspect of engagement 740 –756. doi:10.1002/1098-237X(200011)84:6⬍740::AID-SCE4⬎3.0
introduces new research opportunities. Here, we identify four. .CO;2-3
First, the present study identified greater achievement and greater Brown, A. L., & Campione, J. C. (1996). Psychological theory and the
motivational support as two student outcomes associated with design of innovative learning environments: On procedures, principles,
agentic engagement. Additional outcomes might also be possible. and systems. In L. Schauble & R. Glaser (Eds.), Innovations in learning:
For instance, agentically engaged students likely spend more class- New environments for education (pp. 289 –325). Hillsdale, NJ: Erlbaum.
room time involved with personally interesting, valued, and goal- Cheon, S. H., Reeve, J., & Moon, I. K. (2012). Experimentally based,
relevant learning activities. Second, because agentically engaged longitudinally designed, teacher-focused intervention to help physical
education teachers be more autonomy supportive toward their students.
students create both greater achievement and greater motivational
Journal of Sport & Exercise Psychology, 34, 365–396.
support, these gains should accumulate into an impressive devel-
Christenson, S. L., Reschly, A. L., & Wylie, C. (Eds.). (2012). The
opmental trajectory. It would be informative to track these benefits handbook of research on student engagement. New York, NY: Springer
longitudinally—not just over the course of a semester (as in Study Science.
3) but over the course of schooling (e.g., from enrollment to deCharms, R. (1976). Enhancing motivation: Change in the classroom.
graduation). The prediction would be that agentically engaged New York, NY: Irvington.
students, compared to their non-agentically engaged counterparts, Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-
would be able to create motivationally supportive classroom con- determination in human behavior. New York, NY: Plenum Press.
ditions that set the developmental stage for an ever-increasing De Wolff, M., & van IJzendoorn, M. H. (1997). Sensitivity and attachment:
A meta-analysis of parental antecedents of infant attachment. Child
growth trajectory of positive outcomes. Third, the present study
Development, 68, 571–591.
limited its focus to the teacher-student interaction. But learning
Fiedler, M. L. (1975). Bidirectionality of influence in classroom interac-
environments are also offered by parents, coaches, and tutors. tion. Journal of Educational Psychology, 67, 735–744. doi:10.1037/
Highly agentically engaged students may attempt to contribute to 0022-0663.67.6.735
these learning environments in the same way that they attempt to Fredricks, J. A., Blumenfeld, P. C., & Paris, A. H. (2004). School engage-
contribute to teacher-provided learning environments. Fourth, fu- ment: Potential of the concept, state of the evidence. Review of Educa-
ture research will likely find Figure 1 to be an oversimplified tional Research, 74, 59 –109. doi:10.3102/00346543074001059
depiction of how students benefit from their classroom engage- Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance
ment. For instance, engagement likely contributes positively to structure analysis: Conventional criteria versus new alternatives. Struc-
tural Equation Modeling, 6, 1–55. doi:10.1080/10705519909540118
constructive changes in students’ motivation (Reeve & Lee, 2013).
Jang, H., Kim, E.-J., & Reeve, J. (2012). Longitudinal test of self-
So, instead of providing a comprehensive model of the interrela- determination theory’s motivation mediation model in a naturally-
tions among classroom conditions, motivation, engagement, and occurring classroom context. Journal of Educational Psychology, 104,
outcomes, the figure more modestly communicates the dual ben- 1175–1188. doi:10.1037/a0028089
efits of students’ agentic engagement (greater learning, greater Jang, H., Reeve, J., & Deci, E. L. (2010). Engaging students in learning
motivational support). activities: It is not autonomy support or structure but autonomy support
594 REEVE
and structure. Journal of Educational Psychology, 102, 588 – 600. doi: Reeve, J., Jang, H., Carrell, D., Jeon, S., & Barch, J. (2004). Enhancing
10.1037/a0019682 students’ engagement by increasing teachers’ autonomy support. Moti-
Jang, H., Reeve, J., Ryan, R. M., & Kim, A. (2009). Can self-determination vation and Emotion, 28, 147–169. doi:10.1023/B:MOEM.0000032312
theory explain what underlies the productive, satisfying learning expe- .95499.6f
riences of collectivistically-oriented South Korean adolescents? Journal Reeve, J., & Lee, W. (2013). Students’ classroom engagement produces
of Educational Psychology, 101, 644 – 661. doi:10.1037/a0014241 longitudinal changes in classroom motivation. Manuscript submitted for
Jöreskog, K. G., & Sörbom, D. (1996). LISREL 8: Structural equation publication.
modeling with the SIMPLIS command language. Hillsdale, NJ: Scien- Reeve, J., Nix, G., & Hamm, D. (2003). Testing models of the experience
tific Software International. of self-determination in intrinsic motivation and the conundrum of
Jöreskog, K. G., & Yang, F. (1996). Non-linear structural equation models: choice. Journal of Educational Psychology, 95, 375–392. doi:10.1037/
The Kenny-Judd model with interaction effects. In G. A. Marcoulides & 0022-0663.95.2.375
R. E. Schumacker (Eds.), Advanced structural equation modeling: Is- Reeve, J., Ryan, R. M., Deci, E. L., & Jang, H. (2008). Understanding and
sues and techniques (pp. 57– 88). Mahwah, NJ: Erlbaum. promoting autonomous self-regulation: A self-determination theory per-
Kenny, D., & Judd, C. M. (1984). Estimating the nonlinear and interaction spective. In D. Schunk & B. Zimmerman (Eds.), Motivation and self-
effects of latent variables. Psychological Bulletin, 96, 201–210. doi: regulated learning: Theory, research, and application (pp. 223–244).
10.1037/0033-2909.96.1.201 Mahwah, NJ: Erlbaum.

Kline, R. B. (2011). Principles and practice of structural equation mod- Reeve, J., & Sickenius, B. (1994). Development and validation of a brief
eling (3rd ed.). New York, NY: Guilford Press. measure of the three psychological needs underlying intrinsic motiva-
Kochanska, G. (2002). Mutually responsive orientation between mothers tion: The AFS scales. Educational and Psychological Measurement, 54,
and their young children: A context for the early development of 506 –515. doi:10.1177/0013164494054002025
conscience. Current Directions in Psychological Science, 11, 191–195. Reeve, J., & Tseng, M. (2011). Agency as a fourth aspect of student
doi:10.1111/1467-8721.00198 engagement during learning activities. Contemporary Educational Psy-
Koenigs, S. S., Fiedler, M. L., & deCharms, R. (1977). Teacher beliefs, chology, 36, 257–267. doi:10.1016/j.cedpsych.2011.05.002
classroom interaction and personal causation. Journal of Applied Social Ryan, R. M., & Connell, J. P. (1989). Perceived locus of causality and
internalization: Examining reasons for acting in two domains. Journal of
Psychology, 7, 95–114. doi:10.1111/j.1559-1816.1977.tb01332.x
Personality and Social Psychology, 57, 749 –761. doi:10.1037/0022-
Ladd, G. W., & Dinella, L. M. (2009). Continuity and change in early
3514.57.5.749
school engagement: Predictive of children’s achievement trajectories
Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the
from first to eighth grade? Journal of Educational Psychology, 101,
facilitation of intrinsic motivation, social development, and well-being.
190 –206. doi:10.1037/a0013153
American Psychologist, 55, 68 –78. doi:10.1037/0003-066X.55.1.68
Li, F., Harmer, P., Duncan, T. E., Duncan, S. C., Acock, A., & Boles, S.
Ryan, R. M., & Deci, E. L. (2002). Overview of self-determination theory:
(1998). Approaches to testing interaction effects using structural equa-
An organismic dialectical perspective. In E. L. Deci & R. M. Ryan’s
tion modeling methodology. Multivariate Behavioral Research, 33,
(Eds.), Handbook of self-determination theory research (pp. 3–33).
1–39. doi:10.1207/s15327906mbr3301_1
Rochester, NY: Rochester University Press.
Linnenbrink, E. A. (2005). The dilemma of performance-approach goals:
Sameroff, A. (Ed.). (2009). The transactional model of development: How
The use of multiple goal contexts to promote students’ motivation and
children and contexts shape each other. Washington, DC: American
learning. Journal of Educational Psychology, 97, 197–213. doi:10.1037/
Psychological Association. doi:10.1037/11877-000
0022-0663.97.2.197
Sameroff, A., & Fiese, B. H. (2000). Transactional regulation: The devel-
Midgley, C., Maehr, M. L., Hruda, L. Z., Anderman, E., Anderman, L.,
opmental ecology of early intervention. In J. P. Shonkoff & S. J. Meisels
Freeman, K. E., . . . Urdan, T. (2000). Manual for the Patterns of (Eds.), Early intervention: A handbook of theory, practice, and analysis
Adaptive Learning Scales. Ann Arbor, MI: University of Michigan. (2nd ed., pp. 135–159). New York, NY: Cambridge University Press.
Pajares, F., Cheong, Y. F., & Oberman, P. (2004). Psychometric analysis doi:10.1017/CBO9780511529320.009
of computer science help-seeking scales. Educational and Psychological Schunk, D. H., & Zimmerman, B. J. (1997). Social origins of self-
Measurement, 64, 496 –513. doi:10.1177/0013164403258447 regulatory competence. Educational Psychologist, 32, 195–208. doi:
Pekrun, R., Goetz, T., Frenzel, A. C., Barchfeld, P., & Perry, R. P. (2011). 10.1207/s15326985ep3204_1
Measuring emotions in students’ learning and performance: The Skinner, E. A., & Belmont, M. J. (1993). Motivation in the classroom:
Achievement Emotions Questionnaire (AEQ). Contemporary Educa- Reciprocal effects of teacher behavior and student engagement across
tional Psychology, 36, 36 – 48. doi:10.1016/j.cedpsych.2010.10.002 the school year. Journal of Educational Psychology, 85, 571–581. doi:
Pelletier, L. G., Seguin-Levesque, C., & Legault, L. (2002). Pressure from 10.1037/0022-0663.85.4.571
above and pressure from below as determinants of teachers’ motivation Skinner, E. A., Kindermann, T. A., Connell, J. P., & Wellborn, J. G.
and teaching behaviors. Journal of Educational Psychology, 94, 186 – (2009). Engagement and disaffection as organizational constructs in the
196. doi:10.1037/0022-0663.94.1.186 dynamics of motivational development. In K. Wentzel & A. Wigfield
Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., Congdon, R. T., Jr., & du (Eds.), Handbook of motivation in school (pp. 223–245). Mahwah, NJ:
Toit, M. (2011). HLM 7: Hierarchical linear and nonlinear modeling Erlbaum.
[Computer software]. Lincolnwood, IL: Scientific Software Interna- Skinner, E. A., Kindermann, T. A., & Furrer, C. J. (2009). A motivational
tional. perspective on engagement and disaffection: Conceptualization and as-
Reeve, J. (2009). Why teachers adopt a controlling motivating style toward sessment of children’s behavioral and emotional participation in aca-
students and how they can become more autonomy supportive. Educa- demic activities in the classroom. Educational and Psychological Mea-
tional Psychologist, 44, 159 –175. doi:10.1080/00461520903028990 surement, 69, 493–525. doi:10.1177/0013164408323233
Reeve, J., Deci, E. L., & Ryan, R. M. (2004). Self-determination theory: A Skinner, E. A., Zimmer-Gembeck, M. J., & Connell, J. P. (1998). Individ-
dialectical framework for understanding the sociocultural influences on ual differences and the development of perceived control. Monographs
student motivation. In D. McInerney & S. Van Etten (Eds.), Research on of the Society for Research in Child Development, 63 (2–3, Whole No.
sociocultural influences on motivation and learning: Big theories revis- 204).
ited (Vol. 4, pp. 31–59). Greenwich, CT: Information Age Press. Stoeger, H., & Ziegler, A. (2011). Self-regulatory training through
elementary-school students’ homework completion. In B. J. Zimmerman Williams, G. C., Grow, V. M., Freedman, Z., Ryan, R. M., & Deci, E. L.
& D. H. Schunk’s (Eds.), Handbook of self-regulation of learning and (1996). Motivational predictors of weight-loss and weight-loss mainte-
performance (pp. 87–101). New York, NY: Routledge. nance. Journal of Personality and Social Psychology, 70, 115–126.
Su, Y.-L., & Reeve, J. (2011). A meta-analysis of the effectiveness of doi:10.1037/0022-3514.70.1.115
intervention programs designed to support autonomy. Educational Psy- Wolters, C. A. (2004). Advancing achievement goal theory: Using goal
chology Review, 23, 159 –188. doi:10.1007/s10648-010-9142-7 structures and goal orientations to predict students’ motivation, cogni-
Tessier, D., Sarrazin, P., & Ntoumanis, N. (2008). The effects of an tion, and achievement. Journal of Educational Psychology, 96, 236 –
experimental programme to support students’ autonomy on the overt 250. doi:10.1037/0022-0663.96.2.236
behaviours of physical education teachers. European Journal of Psy- Zimmerman, B. J. (2008). Investigating self-regulation and motivation:
chology of Education, 23, 239 –253. doi:10.1007/BF03172998 Historical background, methodological developments, and future pros-
Vansteenkiste, M., Simons, J., Lens, W., Sheldon, K. M., & Deci, E. L. pects. American Educational Research Journal, 45, 166 –183. doi:
(2004). Motivated learning, performance, and persistence: The synergis- 10.3102/0002831207312909
tic role of intrinsic goals and autonomy-support. Journal of Personality Zimmerman, B. J., & Schunk, D. H. (Eds.). (2011). Handbook of self-
and Social Psychology, 87, 246 –260. doi:10.1037/0022-3514.87.2.246 regulation of learning and performance. New York, NY: Routledge.
Williams, G. C., & Deci, E. L. (1996). Internalization of biopsychosocial

values by medical students: A test of self-determination theory. Journal Received June 24, 2012
of Personality and Social Psychology, 70, 767–779. doi:10.1037/0022- Revision received February 20, 2013
3514.70.4.767 Accepted March 12, 2013 䡲
New Editors Appointed, 2015–2020

The Publications and Communications Board of the American Psychological Association an-
nounces the appointment of 6 new editors for 6-year terms beginning in 2015. As of January 1,
2014, manuscripts should be directed as follows:
● Behavioral Neuroscience (http://www.apa.org/pubs/journals/bne/), Rebecca Burwell, PhD,

Brown University
● Journal of Applied Psychology (http://www.apa.org/pubs/journals/apl/), Gilad Chen, PhD,
University of Maryland
● Journal of Educational Psychology (http://www.apa.org/pubs/journals/edu/), Steve Graham,
EdD, Arizona State University
● JPSP: Interpersonal Relations and Group Processes (http://www.apa.org/pubs/journals/psp/),
Kerry Kawakami, PhD, York University, Toronto, Ontario, Canada
● Psychological Bulletin (http://www.apa.org/pubs/journals/bul/), Dolores Albarracín, PhD,
University of Pennsylvania
● Psychology of Addictive Behaviors (http://www.apa.org/pubs/journals/adb/), Nancy M. Petry,
PhD, University of Connecticut School of Medicine
Electronic manuscript submission: As of January 1, 2014, manuscripts should be submitted

electronically to the new editors via the journal’s Manuscript Submission Portal (see the website
listed above with each journal title).
Current editors Mark Blumberg, PhD, Steve Kozlowski, PhD, Arthur Graesser, PhD, Jeffry
Simpson, PhD, Stephen Hinshaw, PhD, and Stephen Maisto, PhD, will receive and consider new
manuscripts through December 31, 2013.

How Students Create Motivationally Supportive Learning Environments For Themselves: The Concept of Agentic Engagement

Uploaded by

Copyright:

Available Formats

How Students Create Motivationally Supportive Learning Environments For Themselves: The Concept of Agentic Engagement

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

How Students Create Motivationally Supportive Learning Environments For Themselves: The Concept of Agentic Engagement

Uploaded by

Copyright:

Available Formats

Journal of Educational Psychology © 2013 American Psychological Association

2013, Vol. 105, No. 3, 579 –595 0022-0663/13/$12.00 DOI: 10.1037/a0032690

How Students Create Motivationally Supportive Learning Environments for

independent variance in course-specific achievement not otherwise attributable to students’ behavioral,

Keywords: achievement, agency, agentic engagement, autonomy support, student engagement

Item Psychological need satisfaction Academic efficacy

Original items to assess a unilateral contribution to the learning environment

New candidate items to assess a transactional contribution to the learning environment

New candidate items to assess a personal contribution to one’s own learning

For each of the 10 candidate agentic engagement items, Table

Factor 1 Factor 2 Factor 3 Factor 4

Behavioral Engagement items

During this class, I express my preferences and opinions. .78

During class, I ask questions to help me learn. .56

Study 1 Study 2 Study 3

Individual AES item

SD 1.13 0.71 1.09

.86 .84 .81

Observed variable B SE B ␤ ICC (%)

Latent factor including the Behavioral Engagement items

Latent factor including the Cognitive Engagement items

1. Agentic Engagement, T1 4.03 1.08 — .65 .53 .48 .46

interaction of the two T2 endogenous variables) did not fit the

tivational responsiveness. The longitudinal effect of students’ T2

Agentic Engagement as an Antecedent of Autonomy 4

room engagement differently from one culture to the next, and

10.1037/0033-2909.96.1.201 Mahwah, NJ: Erlbaum.

Williams, G. C., & Deci, E. L. (1996). Internalization of biopsychosocial

New Editors Appointed, 2015–2020

● Behavioral Neuroscience (http://www.apa.org/pubs/journals/bne/), Rebecca Burwell, PhD,

Electronic manuscript submission: As of January 1, 2014, manuscripts should be submitted

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.