Technology-based assessments can provide many advantages, for example, by allowing the collection of data on the course of item processing (so-called process data) in addition to the recording of "traditional" outcome data such as item responses. Such process data includes log data (i.e., events, event-related attributes, and timestamps) that are accumulated in log files during computer-based testing. While log data was still considered a "by-product" in the last decade, its strategic collection and deliberate analysis is steadily increasing. The possibilities of analyzing log data are manifold, but can be quite challenging due to the amount and structure of data, lack of conceptual considerations or insufficient documentation. The workshop will provide an introduction on how to work with log data.
Participants will gain insight into the preparation and analysis of log data using the log data from a previous study. They will construct and analyze log data indicators themselves in hands-on exercises using the R package LogFSM. Finally, different strategies of analysis (i.e., log data indicators as dependent or independent variable in regression analyses, methods of machine learning, latent variable models) are briefly reviewed in an overview.
The all-day workshop is given by Carolin Hahnel (DIPF). Workshop language is English. For more information see the "Call for Presentations".
Leibniz Institute for Educational Trajectories (LIfBi)
Wilhelmsplatz 3 96047 Bamberg Germany