Longitudinal Complexity of Affect and Mathematics Self-Efficacy in Middle School

• In Event: 61.015 - Embracing and Modeling the Complex Dynamics of Motivation and Engagement: Contextual, Temporal, Dynamic, and Systematic

Abstract

Background & Theoretical Framework
Within-person approaches to studying psychological experiences are important for modeling processes as they unfold in context (Hamaker, 2012). Accordingly, researchers have noted the contextualized, dynamic nature of motivational constructs and begun work to employ methods that capture them in authentic learning settings (e.g., Kaplan & Garner, 2017). Findings from a time-intensive study of student motivation and engagement with high school students indicated that these constructs are sensitive to daily occurrences in class, and the authors called for investigations of variability in student experience associated with lesson content (Martin et al., 2015). We extend this literature by utilizing data collected during the normal course of middle school mathematics instruction to examine how student self-perceptions of efficacy and emotion vary with mathematics tasks and content.
Perceived self-efficacy, or one’s beliefs and judgements about their capability to complete a task (Bandura, 2006), is typically considered a malleable and domain-specific construct. Theoretical perspectives grounded in social cognitive theory have linked perceived self-efficacy with effort and persistence in learning as well as emotional experience (Zimmerman, 2000). However, the extent to which these experiences are coupled within contextualized lessons over time is not well understood.
Method
Participants were 125 7th and 8th grade students enrolled in a pre-Algebra course recruited from a single middle school (details in Bernacki et al., 2013) who spent two math class periods per week solving units of problems in the Cognitive Tutor (CT) intelligent tutoring system. Students responded to questionnaires embedded at the end of each CT unit. This adaptive technology is self-paced, so students varied in the number of units completed (M = 26.23; SD = 10.75).
Data
This study utilized unit-level self-efficacy scores and single-item measures of engagement and boredom for each CT unit students completed. Data presented here were collected in every other unit through unit 24 (i.e., units completed by majority of participants).
Results and Significance
Preliminary analyses using univariate descriptive statistics suggest that boredom, engagement, and self-efficacy are moderate and relatively stable across the academic year (see table 1). Graphic depictions of the same data generated in SAS (9.4) make clear that this summary is insufficient to explain variability in trajectories across the year (see figures 1-3). Because self-efficacy is theorized as malleable in nature and influenced by contextual factors, analyses limited to fixed, linear functions would insufficiently explain the data over all units. Random variance components are expected over time, and the coupling of within-unit constructs are expected to vary depending on the nature of the unit. Additional analyses to be presented further demonstrate contextual sensitivity of self-efficacy after accounting for within-person stability across units. Unit-specific factors that further explain sensitivity of efficacy can include factors fixed across learners per unit (i.e., difficulty of objectives, presence of visual aids) as well as student-specific metrics (e.g., time spent, hints requested, performance). Modeling these contextual factors as random effects and as time-varying covariates address significant gaps in the literature created by a focus on cross-sectional motivational research and linear modeling.

Authors

• Chanler Hilley, Arizona State University

• Gwen C. Marchand, University of Nevada - Las Vegas

• Matthew L. Bernacki, University of North Carolina Chapel Hill