What Works, for Whom? Motivation Moderates Self-Regulated Learning Intervention Efficacy in a Postsecondary Biology Course

• In Event: Cognitive, Metacognitive, and Motivational Interventions Supporting Student Success in STEM: What Works, When, for Whom?

Abstract

Background/Aims
Long-standing postsecondary STEM attrition challenges have prompted educators to implement self-regulated learning (SRL; Greene, 2018) interventions to support students (Bernacki et al., 2020; Rosenzweig et al., 2021; Theobald, 2021). However, the efficacy of such interventions may depend upon how motivated students are to internalize them and strive for success in the course (Robinson et al., 2022). We explore this prediction in this study.

Method
Our study involved 443 consenting college students enrolled in an introductory, high-structure biology course at a large United States postsecondary institution. Participants completed motivation instruments at the beginning of the semester (i.e., achievement goals, Elliot & Murayama, 2008; self-efficacy, Midgley et al., 2000; values, Perez et al., 2014; cost, Eccles & Wigfield, 2020; mindsets, Dweck & Leggett, 1986). A learning analytics prediction model classified students as "in need of support" or not based on digital traces of their behaviors over the first two weeks of the course (Greene et al., 2019).

Participants in need of support were randomly assigned to one of three conditions: viewing a biology multimedia presentation (control), viewing a multimedia science of learning to learn treatment (SoL2L), or attending in-person coaching treatment (coaching). Participants not identified as needing support were also assigned to the control condition (not-needing-support).

Results
Prediction Modeling. Using a prediction model developed in a previous semester, we classified 243 students as in-need-of-support and 207 as not.

Motivation. Confirmatory factor analyses (CFA) and exploratory structural equation modeling (ESEM), with and without bifactors (see Figure 1; Part et al., 2020) showed that achievement goal factors were best modeled using an ESEM model, self-efficacy with a CFA model, costs and values with a bifactor-ESEM model, and mindsets with a CFA model. Factor scores from models were retained and submitted to latent profile analysis, resulting in a four-profile solution (see Table 1).

Moderation Analysis. Analyses involving the full sample showed students in the not-needing-support condition outperformed students in the other conditions, who did not statistically significantly differ from one another. Next, we split the sample into the four motivation profiles and conducted ANOVAs within each subsample to examine profile-specific condition differences on outcomes. The full-sample results were replicated in three of the four subsamples. Descriptively, the SoL2L condition performed similarly to the not-needing-support condition and better than the control and coaching conditions in the "low achievement goals, value, and mindsets with high-cost" profile (see Figure 2). However, likely due to being underpowered, there was no statistically significant overall condition effect in this profile, even though the partial eta-squared was practically significant (.073). That trend was mirrored for the final course grade (condition effect stat ns; partial eta-squared = .062), except in addition the coaching group performed similarly to the not-needing-support condition (see Table 2).

Significance
Our findings suggest understanding the effect of an SRL intervention may require accounting for participants’ motivational profiles. Thus, SRL theory may need to be modified to better account for motivational profile differences in SRL functioning, and SRL interventions may need to be tailored similarly.

Authors

• Jeff A. Greene, University of North Carolina - Chapel Hill
  Presenting Author

• Robert D Plumley, University of North Carolina - Chapel Hill
  Non-Presenting Author

• Matthew L. Bernacki, University of North Carolina - Chapel Hill
  Non-Presenting Author

• Shelbi Laura Kuhlmann, University of North Carolina - Chapel Hill
  Non-Presenting Author

• Michael Berro, University of North Carolina - Chapel Hill
  Non-Presenting Author

• Alaina Garland, University of North Carolina - Chapel Hill
  Non-Presenting Author

• Laura Ott, University of North Carolina - Chapel Hill
  Non-Presenting Author

• Kelly Hogan, University of North Carolina - Chapel Hill
  Non-Presenting Author

• Marc Alan Howlett, University of North Carolina - Chapel Hill
  Non-Presenting Author

• Kimberly Abels, University of North Carolina - Chapel Hill
  Non-Presenting Author