Analyzing coding behaviour of novice programmers in different instructional settings: Creating vs. Debugging

Abstract

Many methods exist for teaching programming in a physical computing context. However, it is not clear what the advantages and disadvantages of these methods are in practice. Our research compares two methods for integrating programming in a primary robotics workshop. Both methods use a custom-designed programming environment based on Google Blockly. Moreover, one method lets learners create their programs from scratch while the second method requires the learners to fix faulty programs. We compared the differences between the integration methods by analyzing programming environment logging data and linked it to the results on a programming knowledge test. Our results show that the learners in the create group spend more time solving the programming assignment and are more often distracted by code blocks they don’t need. The learners in the debug group require less time. However, they perform a disproportionate amount of code changes and apply the "tinkering" strategy more often. Nevertheless, the drawbacks of the create method, the learners in this group did show significantly higher scores on our programming knowledge test.