Kolb learning style inventory: a meta-analysis

This study addressed the effect of the Kolb Learning Style Inventory (LSI) in learning environments and especially those which involve computers. The goal was accomplished by conducting a meta-analysis of the findings of 101 randomly selected studies that compared the LSI to other variables. After a thorough electronic and traditional bibliographic search, and rigorous screening, 101 randomly selected studies were analyzed. The meta-analysis utilized Glass (1981) techniques, and used F, r, t and chi square values to calculate effect sizes. Of 59 studies utilizing correlations, only one exhibited what Kerlinger (1973) called very high standards of correlation. Meta-analysis produced far more meaningful results than traditional vote counting techniques of literature review, summarization, and synthesis (Cooper and Rosenthal, 1980). Results: (1) The average effect sizes for the 10 LSI variables were found to be weak overall, yet stronger in sub-populations. (2) There were major differences between studies which involved learning and using computers, and those studies which did not. (3) Older studies had somewhat higher effect sizes than more recent studies. (4) Unpublished studies had considerably higher effect sizes, which was polar to the "file drawer problem" which Rosenthal (1979) described. (5) Studies using undergraduates had considerably different effect sizes than the remainder of the population. This finding re-affirmed the "sophomore" effect described by Gordon, Slade, and Schmidt (1987). (6) Varying populations produced very little difference in effect sizes. In conclusion, although the LSI has intuitive value in many studies, when quantitatively combined with other variables, despite reported significance levels, the LSI did not contribute heavily to effect size.