Metaphor types as strategies for teaching regression to novice learners

Abstract

Metaphors are well-known tools for teaching statistics to novices. However, educators might overlook metaphor theoretical developments that offer nuanced and testable perspectives on their pedagogical applications. This article introduces the notion of metaphor types—“correspondence” (CO) and “class inclusion” (CI)—as different strategic ways of presenting metaphors and reports an experimental study on their effectiveness in teaching basic regression to language and communication majors. Briefly, CO emphasizes systematic links while CI emphasizes holistic perceptions of similarity between the source and target of a metaphor. Both competency and attitudinal measures were compared in view of the latter’s importance as intended outcomes of the typical introductory course. The results show that while CO outperformed CI in assessments of manual calculations (e.g., SST/SSR/SSE/R2), CI outperformed CO in essay assessments requiring elaboration of general conceptual understanding. CI was also linked to more positive perceptions of the practical utility of regression analysis and its contribution to personal growth. Correlations between performance and attitudes were stronger in CO than CI, which further suggests CO’s greater perceived resemblance to a “rote learning” approach. The attendant implications are discussed in the growing context of general statistics education for nonstatistics majors. Directions for further research are suggested.