Paper Summary

Direct link:

The Many Faces of Self-Regulation in Educational Contexts: Painting the Bigger Picture

Mon, April 8, 8:00 to 10:00am, Sheraton Centre Toronto Hotel, Floor: Mezzanine, Chestnut East


Research on self-regulation (SR) in educational contexts is based on various constructs from different “theory families” that have been individually linked to learning outcomes but remain largely disintegrated. In the current study we investigate the relations between four core facets of college students’ self-regulated learning in school and lab contexts, covering five different content domains ranging from math to art.

Theoretical framework
To overcome the fragmented research on different facets of SR in education, we propose a comprehensive framework for SR in educational settings (see Figure 2). At its core, the framework posits that self-regulated learning processes are constituted by learning activities (LA; e.g., (meta-)cognitive strategies), driving forces (DF) underlying these activities (e.g., motivation), and limited resources (LR) required by them (e.g., working memory). Moreover, personal predispositions (PP - e.g., grit, conscientiousness) influence how these processes unfold, finally leading to learning outcomes and achievement. Single facets of the proposed model have been linked to learning and achievement, such as personal characteristics (e.g., conscientiousness, grit; Duckworth, Peterson, Matthews & Kelly, 2007; Poropat, 2009), cognitive and metacognitive strategies (e.g., Schunk & Greene, 2018), and executive control functions (e.g., working memory, Cowan, 2014). In the current study we explore the relative contribution of each facet in our framework to learning outcomes in different domains and in different contexts.

We assessed 102 undergraduate students in a laboratory study (M=23.49 years, SD=2.89 years, 70.59% female). Participants completed questionnaires assessing various personal characteristics and cognitive strategies, and completed cognitive tasks assessing executive functioning and learning in several academic domains (see Table 2 for a summary of measures). Z-scores were computed for all measures for use in the data analyses. To evaluate the proposed models’ structure, we computed three domain-general facets of SR (personal predispositions, limited resources, learning activities) by mapping all measures into a cluster structure based on their covariance matrix, using state of the art machine learning algorithms (Kluger, Basri, Chang & Gerstein, 2003). Second, we used the resulting clusters together with domain-specific motivations (i.e., driving forces) in regression models to predict different learning outcomes (self-reported grades, learning task outcomes) in each of the five domains (math, biology, physics, art, and history; see Table 3). Third, Bayes factors were computed to assess how well the models represent the data.

The cluster solution supported the three domain-general facets of SR proposed in the framework (PP, LR, LA, see Figure 3). The regression models better predicted school grades than laboratory learning outcomes. Regarding the relative importance of different SR facets, we found that LR and domain-specific DF were predictive for all domains whereas the clusters representing LA and PP were only predictive for some domains. Moreover, LR best predicted lab-learning outcomes, whereas DF best predicted school grades.

Our results demonstrate the importance of painting a bigger picture of SR than usual by considering how various facets of SR simultaneously predict outcomes in various learning settings. Our approach is particularly important for obtaining a comprehensive understanding of performance in different domains and learning contexts.


©2019 All Academic, Inc.   |   Privacy Policy