Using Learning Analytics in Understanding Students' Behavior in Block-based and Text-based Programming Modality

Abstract

Due to the rapid growth of various programming tools in secondary school computer science classes, the block-based programming environment has replaced or taken precedence over text-based programming languages. Despite their growing use in formal contexts, relatively few empirical works have been done to understand students' programming behavior using block-based and text-based programming environments in secondary school classrooms. In this article, we designed a 5-week, quasi-experimental research comparing isomorphic block-based and text-based programming modalities in an introductory secondary school programming class. Mixed-method approach was used to conduct fine-grained analysis of students in two programming modalities from multi-dimensional perspectives. The findings from this study indicated that students in text-modality tended to write longer lines of codes, make more syntactical errors, and spend much more time between two times of debugging, while students in block-based modality tended to have more attempts in debugging and make fewer syntactical errors. The underlying reasons and the implications of these findings with respect to instructional design were discussed in the final sections.