Agentic Engagement Explains Independent Variance in Students' Achievement and Perceived Autonomy Support

• In Event: Agentic Engagement: A New, Important, and Progress-Enabling Educational Construct

Abstract

Purpose
To demonstrate that agentic engagement predicts students’ course achievement (Study 1) and perceived autonomy support (Study 2) above and beyond the predictive effects of the other three components of engagement (behavior, emotion, cognition).

Theoretical Framework
Engagement is widely viewed as a multidimensional construct that consist of the three components of behavior, emotion, and cognition (Christenson, Reschly, & Wylie, 2012; Fredricks, Blumenfeld, & Paris, 2004) that predicts valued educational outcomes (Jang, Kim, & Reeve, 2016; Ladd & Dinella, 2009). Students do become behaviorally, emotionally, and cognitively involved in the learning activities their teachers provide but they further, more or less, offer their input, communicate their preferences, and ask questions to help them learn and make progress (i.e., agentic engagement).

To the extent that any one engagement component can predict unique variance in important educational outcomes, then such a finding would provide supportive evidence to warrant its inclusion in the larger engagement construct. We hypothesized that agentic engagement would correlate with the other three engagement components yet predict unique variance in course achievement (Study 1) and perceived autonomy support (Study 2), even after controlling for the predictive effects of behavioral, cognitive, and emotional engagement.

Methods
Study 1. Participants were 406 ethnic Korean middle school students (217 females, 189 males) enrolled in 11 classes with 4 subject matters. Students completed the study questionnaire twice—early and late in the semester. For academic achievement, course grades were taken from the actual objective school record (range, 0 to 100). We assessed behavioral and emotional engagement (1—5 response scale) with the Engagement versus Disaffection with Learning measure (Skinner, Kindermann, & Furrer, 2009), cognitive engagement with the Deep Learning Strategies measure (Senko & Miles, 2008), and agentic engagement with the Agentic Engagement Scale (Reeve, 2013). All alphas > .80.

Study 2. Participants were 483 ethnic Korean middle school students (190 females, 293 males) enrolled in the classes of nine physical education teachers. Students completed the same questionnaires as in Study 1 (all ’s > .80; 1-7 scale) at the beginning, middle, and end of the semester—plus the Learning Climate Questionnaire (Williams & Deci, 1996) to assess perceived autonomy support.

Results
Study 1. To predict academic achievement, the four T1 and T2 engagement components, along with the five statistical controls (gender, grade level, and the subject matters), were entered as fixed effects predictors into a multilevel regression. As shown in Table 1, T2 agentic engagement individually predicted course achievement, along with T1 behavioral engagement, T2 behavioral engagement.

Study 2. To predict end-of-semester (T3) perceived autonomy support, the four T1 and T2 engagement components, along with the three statistical controls (T1 perceived autonomy support, gender, and grade level), were entered as fixed effects predictors into a multilevel regression. As shown in Table 2, only T2 agentic engagement predicted end-of-semester perceived autonomy support (controlling for T1 perceived autonomy support).

Discussion
Agentic engagement needs to be added to the larger engagement construct. Doing so increases the capacity of the engagement construct to explain key educational outcomes.

Authors

• Johnmarshall Reeve, Australian Catholic University

• Hyungshim Jang, Hanyang University