Towards a Common Instrument for Measuring Prior Programming Knowledge

Abstract

Computing education researchers and educators use a wide range of approaches for measuring students' prior knowledge in programming. Such measurement can help adapt the learning goals and assessment tools for groups of learners at different skills levels and backgrounds. There seems to be no consensus on if and how prior programming knowledge should be measured. Traditional background surveys are often ad-hoc or non-standard, which do not allow comparison of results between different course contexts, levels, and learner groups. Moreover, surveys may yield inaccurate information and may not be useful due to lack of detail. In contrast, tests can provide much higher detail and accuracy than surveys about student knowledge or skills, but large-scale tests are typically very time-consuming or impractical to arrange. To bridge the gap between ad-hoc surveys and standardized tests, we propose and evaluate a novel self-evaluation instrument for measuring prior programming knowledge in introductory programming courses. This instrument investigates in higher detail typical course concepts in programming education considering the different levels of proficiency. Based on a sample of two thousand introductory programming course students, our analysis shows that the instrument is internally consistent, correlates with traditional background information metrics and identifies students of varying programming backgrounds.