ABSTRACT
Combining theory-driven and data-driven approaches, this study used both self-reported and observational measures to examine: (1) the joint contributions of students' self-reported undergraduates' motivation and emotion in their self-regulated learning, their observed online learning interactions, and their academic success in blended course designs; and (2) the extent to which the self-reported and observational measures were consistent with each other. The participants in the study were 54 social sciences undergraduates in the Czech Republic. The participants' self-reported self-efficacy, intrinsic goals, and anxiety were assessed using a Czech version of three scales from the Motivated Strategies for Learning Questionnaire. Their online engagement was represented by students' observed frequency of interactions with the six online learning activities recorded in the learning management system. The results of a hierarchical regression analysis showed that the self-reported and observational measures together could explain 71% of variance in academic success, significantly improving explanatory power over using self-reported measures alone. Departing from the theory-driven approach, students were clustered as better and poorer self-regulated learners by their self-reports, and one-way ANOVAs showed that better self-regulated learners had significantly more frequent online interactions with four out of six online learning activities and better final exam results. Departing from the data-driven approach, students were clustered as higher and lower online-engaged learners by the observed frequency of their interaction with online learning activities. One-way ANOVAs showed that higher online-engaged learners also reported having higher self-efficacy and lower anxiety. Furthermore, the strong association between the students' profiles in both self-reported measures and observational measures in cross-tabulation analyses showed that the majority of better self-regulated learners by self-reporting also had higher online engagement by observation, whereas the majority of poorer self-regulated learners by self-reporting were lower online-engaged learners, demonstrating consistency between theory-driven and data-driven approaches.
ABSTRACT
Introduction: The present investigation examined the stability of the twofold multidimensional structure of academic self-concepts (ASCs) in three domains, namely Chinese, math, and general school using four-wave data collected over 2 years among 552 Chinese secondary school students. Method: Adopting both a within-network and a between-network approach, confirmatory factor analyses (CFAs) and factor correlations were performed in Mplus 8.2. Results: The within-network results showed that CFA models wherein competence and affect dimensions were conflated generated unacceptable fit. In contrast, the CFAs in which competence and affect were modeled as separate latent factors consistently produced superior fit to the data. The between-network results demonstrated that in the Chinese and math domains and across the four-time waves, the competence components were more strongly related to the achievements in matching domains than the affect components were. Furthermore, both the competence and affect components of ASCs and achievements were positively correlated in the non-matching domains, which were somewhat contradictory to the internal/external frame of reference model predicting zero or negative relations. Discussion: Such results seem to suggest more involvement in social comparison than in dimensional comparison of Chinese students, which might be attributed to the collectivistic Chinese culture and the common phenomenon of academic social comparisons among Chinese adolescents in schools.
