Assessing students' understanding of object structures
Andreas Mühling, Carsten Schulte, Jens Bennedsen, Lea Budde, Gregor Große-Bölting · 2019
We present a theoretically derived and empirically tested competence model related to the concepts of "object state" and "references" that both form an important part of object-oriented programming. Our model characterizes different levels of programming capability with a focus on possible learning stages of beginning learners. It is based on the notion of understanding objects and their interaction with each other during the runtime of a program. Based on a hierarchical description of our theory, we derive a two-dimensional structure that separates the hierarchy into two facets "structure" (how are objects structured/stored) and "behaviour" (how do objects interact and access each other). Based on this, we have developed a set of items and collected data in a CS1 course (N = 195) to validate the item-set. We analyzed the data using a Rasch model to check item difficulty and the presence of different difficulty levels, and factor analysis to check the dimensions of the model. Furthermore, we argue for the validity of the items with the help of additional data collected from the students. The results indicate that our theoretical assumptions are correct and that the items will be usable with some minor modifications.