BackHome
Refresh: OffView Personal Schedule
Paper Summary
Share...

Direct link:

How Do Motivation Profiles Moderate the Efficacy of a Targeted Self-Regulated Learning Intervention?

Thu, April 21, 11:30am to 1:00pm PDT (11:30am to 1:00pm PDT), Marriott Marquis San Diego Marina, Floor: North Building, Lobby Level, Marriott Grand Ballroom 12

Abstract

Background and Aim. The United States’ leaky STEM pipeline necessitates reforming undergraduate education (PCAST, 2012) and supporting students’ refinement of their self-regulated learning (SRL; Greene, 2018) skills. Bernacki and colleagues (2020) utilized trace data from students’ interactions with digital course resources to identify and intervene with students in need of SRL support; responsive students improved exam performance. This study extends that approach to another undergraduate context where we replicated the identification and intervention models, examined effects, and conducted exploratory analyses to determine whether students’ motivation profile (i.e., achievement goal theory [Harackiewicz et al., 2002] and expectancy-value theory scores [Eccles & Wigfield, 2020]) and prior knowledge moderated those effects.

Method.
Sample, Context, and Data Collection
Participants included 399 undergraduate students from two sections of an introductory biology course, utilizing the same syllabus and pedagogy, at a large, southeast US university. Motivation and prior knowledge were assessed in Week 2. Our prediction model used observations of participants’ interactions with digital course resources through week 3 to differentiate students who would benefit from support (i.e., flagged students, projected as likely to earn a final grade requiring re-enrollment in the course) from their peers (i.e., non-flagged). Flagged students were randomly assigned to treatment (i.e., viewed an SRL strategy advice page, added to an instructor-delivered check-in survey) or control (i.e., survey without advice page).

Data Analysis
Prediction Modeling. We classified students as flagged or non-flagged using an elastic-net logistic regression algorithm composed of features reflecting students’ digital interactions.
Motivation. Confirmatory factor analysis was used to test a model with specific motivation factors (e.g., mastery approach, attainment value) versus one with the addition of general bifactors (e.g., achievement goals, value, cost; Reise, 2012). Then, we used motivation factor scores as indicators in a latent profile analysis.
Moderation Analysis. We conducted exploratory covariance pattern growth mixture modeling (McNeish & Harring, 2020) of exam scores using motivation profiles as known classes, and condition status (i.e., treatment, control, non-flagged), prior knowledge score, and their interaction were used as predictors of exam trajectory, within each class.

Results. Treatment effects were statistically non-significant across all exams (ps > .050; Appendix A). Confirmatory factor analyses revealed the specific-factor-plus-general-bifactor-model, with a performance-item-only achievement goal theory bifactor, had the best data-model fit (Appendices B and C). Latent profile analyses using motivation factor scores supported a 5-class model (Appendix D, Figure 1).
Class 1 had a productive motivational profile, with moderate achievement goal, high value, and low cost means. Growth mixture modeling revealed more productive motivational profiles had higher average exam scores (Appendix E, Figure 2). In Class 1, the interaction of treatment-versus-control with pretest was a statistically significant predictor of the intercept latent factor. Specifically, Class 1 treatment participants with low pretest scores performed better than their non-flagged peers, and similarly to their control peers (Figures 3-5).

Significance. Our findings support further investigation of how motivation and prior knowledge moderate the effect of SRL interventions upon STEM students in need of support, and if replicated, suggest tailoring those interventions to motivational profiles.

Authors