Reliability generalization of the Inventory of Learning patterns of Students in Higher Education: Consistency according to territories, age and versions

The theoretical model of learning patterns and the Inventory of Learning patterns of Students (ILS) have had a relevant impact, prompting international research on students’ learning in different territories. Using a meta-analysis design, this study aims at summarizing the available evidence in Web of Science, Scopus, PsycINFO, ERIC and Google Scholar, regarding the reliability of the ILS scores in Higher Education (HE) students. The analyses focused on synthesizing alpha coefficients through random-effects models and analyzing the influence of moderating variables with weighted regression and meta-ANOVA. Inclusion criteria were different type of primary studies using students of HE and upper, non-compulsory secondary education. Literature or empirical reviews, qualitative studies were excluded. Results of 46 applications of the ILS were included, which provided 698 alpha coefficients. The combined reliability coefficients ranged from α=.55 (“personal interest”) to α=.83 (“deep processing”). Moderation analyses revealed that age of the participants, territory and ILS version affected to a greater extent the reliability of learning conceptions and orientation, rather than processing and regulation strategies. Findings support the use of ILS in different territories, indicate some areas of improvement for specific subscales and contexts of application, and highlight the possibility of adapting it to other contexts. The relevance of identifying the components analyzed by the ILS for their usefulness in improving learning and teaching strategies in HE is discussed. Nonetheless, incomplete reporting of key study details -like sample traits, age ranges, version differences, and translation practices- reduces the accuracy and clarity of moderation analyses.