Modeling the Evolution in Learning Strategies: A Multi-Indicator Latent Growth Analysis

• In Event: On the Dynamics and Relatedness of Students’ Approaches to Learning in Higher Education

Abstract

In recent years, whether and how learning strategies evolve during higher education has been increasingly the focus of attention in research (Phan, 2008). Applying the technique of latent growth analysis (Duncan et al., 2001), this research study has the purpose to address following research questions:
1. How do students’ processing and regulation strategies evolve throughout the three years of higher education?
2. Do students vary in this evolution?
3. How well does the latent growth model fit the data?

One cohort of students entering a University College was followed up during the three years of higher education. 245 students participated at three measurement waves by filling out the Inventory of Learning Strategies – Short Version (ILS-SV; Donche & Van Petegem, 2008). This questionnaire maps four processing strategies (memorizing, analysing, critical processing and relating & structuring) and three regulation strategies (external regulation, self-regulation and lack of regulation), using a five-point likert scale. To assess changes in learning strategy scales, a multi-indicator latent growth model was calculated using Mplus 6.1.

Results indicate an increase in critical processing, relating and structuring and self-regulation throughout higher education. On average, the scores on the memorizing strategy, external regulation and lack of regulation scale decrease. For the analysing scale, no significant changes were found. Of these trends, the largest change over the 26 months under study amounted up to half a standard deviation. In sum, students generally do seem to evolve towards high-quality learning but at a slow pace.

Concerning the differential evolution in learning strategy scales, students were found to vary in their initial level for each scale. Future research could probe into the predictors for this variation at the start of higher education. None of the scales showed significant variability in the slopes. Concerning the research question on the predictive strength of the growth model, results indicate that between 50.6% and 92.9% of the variance in latent factors of processing and regulation strategy scales was explained by the linear growth trajectory. At the same time, for each of the learning strategy scales, at least two of the three errors variances proved significant. This result pleads for predictors, aside of the growth trajectory. Moreover, to better trace the data curve and to detect opportunities for stimulating learning strategy development, future research should preferably include more than three measure moments across time.

Authors

• Liesje Coertjens, University of Antwerp

• Vincent Donche, University of Antwerp

• Sven De Maeyer, University of Antwerp

• Gert Vanthournout, University of Antwerp

• Peter Van Petegem, University of Antwerp