Modelling programming performance: Beyond the influence of learner characteristics

Abstract

In the 21st century, the ubiquitous nature of technology today is evident and to a large extent, most of us benefit from the modern convenience brought about by technology. Yet to be technology literate, it is argued that learning to program still plays an important role. One area of research in programming concerns the identification of predictors of programming success. Previous studies have identified a number of predictors. This study examined the effect of a combination of predictors (gender, learning styles, mental models, prior composite academic ability, and medium of instruction) on programming performance. Data were collected anonymously through a website from 131 secondary school students in Hong Kong who opted for computer programming in the Computer and Information Technology curriculum. Partial Least Squares (PLS) modelling was used to test a hypothesized theoretical structural model. All of the five aforementioned variables were either direct or indirect predictors of programming performance and the antecedents accounted for 43.6% of the variance in programming performance. While this study shows the influence of learner characteristics such as gender, learning styles, and mental models on programming performance, it highlights the effect that prior composite academic ability and medium of instruction exert on learning outcomes, which is uncommon among studies of similar purpose. These findings have significant implications for policy makers and educators alike. © 2011 Elsevier Ltd. All rights reserved.