Problemlöseprozesse sichtbar machen: Analyse von Logdaten aus einer computerbasierten Bürosimulation

Problem-solving competence is of great relevance both in general education as well as vocational education. Although problem solving itself is a process, the analysis of problem-solving competence usually focuses on products but ignores the quality of problem-solving processes. Computer-based testing environments provide new possibilities for analysing problem-solving processes on the basis of big data from interactions with the test environment (log data). The theoretical section includes a distinction between types and characteristics of problems and various strategies – domain-general heuristics, knowledge-based (domain-specific) heuristics and domain-specific algorithms. Analyses of log data which are used to identify individual problem-solving strategies can ideally be divided into data-driven and model-driven approaches. This paper exemplarily discusses approaches to log file analysis based on data from domain-specific problem scenarios that have been processed in a computer-based office simulation by 780 students in commercial vocational education and training. In a first step of data preparation, the particular interactions are consolidated into related activities. Second, the 30-minute scenarios are divided into six intervals of five minutes each. Finally, the selected activities are analysed for each of five performance groups (grouped according to their problem-solving competence). The number of activities, the use of notepad and the handling of relevant documents are examined for two problem-solving scenarios that differ in terms of their approach. All analyses show that the group of more successful problem solvers has more of the above mentioned activities. In the course of time, there are some interesting differences between both, the problem scenarios and the performance groups. These differences indicate that more successful problem solvers align their problem-solving processes with the problems’ inherent approaches.