The Relationship Between Programming Quality and Different Measures of Computational Thinking

Abstract

Computational thinking (CT) is often promoted as the literacy of the 21st century and as the foundation of many concepts in computer science and related fields. Although most research has shown that including CT in computer science education has positive effects on programming, there are conflicting results. These inconsistencies may occur because of different frameworks, which lead to different ways of measuring that kind of thinking. This raises questions about the role of CT in programming, in particular, how is CT applied when solving a programming task (RQ1) and whether different measures of CT can be relevant predictors for programming quality (RQ2). Based on a literature review from two fields, computer science and psychology, a conceptual framework of CT is developed in this thesis. This conceptual framework builds the foundation for an instrument to observe CT behaviour. In order to answer the research questions, participants worked in pairs (n = 27) to solve a programming task in Scratch. The solving process were videotaped and analysed based on CT activities. In addition, participants’ Scratch projects were analyses based on programming quality criteria. A set of adjusted Bebras tasks were used as unplugged measure of CT. To control for confounding effects, a measure for nonverbal intelligence was completed by the participants as well. Results showed that not all CT associated behaviour was equally often apparent while participants were working on the programming task. Participants engaged only infrequently in decomposition or abstract thought about a problem. Instead, of thinking about the problem, they tried to create solutions from the beginning. Correlations and regression analysis also revealed that CT measures differ in their suitability for revealing the relationship with programming quality. Only the behaviour based measure of CT revealed that relationship. On this basis, it is recommended that educators should focus on different parts of CT in order to enhance that kind of thinking. In order to analyse the unique impact that CT might have on programming, instruments must be chosen with care.