Speaker: Philipp Doebler

Authors: Hamdollah Ravand, Susanne Frick, Farshad Effatpanah Hesari, Olga Kunina-Habenicht, Purya Baghaei, Philipp Doebler

Abstract

The full transition to digital assessment in TIMSS 2023 opens new possibilities beyond traditional score reports. The log files students generate as they navigate through the assessment contain rich information about how they approach test items—whether they rush, persist when items are difficult, or revisit their answers. We introduce a five-level classification of test-taking behavior. Drawing on the distinction between motivation and efficiency proposed by Lundgren and Eklöf (2020), we argue that rapid guessing on incorrect items signals motivational disengagement—the student gives up when faced with difficulty—whereas rapid guessing on correct items may simply reflect efficient responding. Our taxonomy—Severely Disengaged, Moderately Disengaged, Error-Revising Engaged, Correct-Response Reviewing, and Direct Completion Engaged—provides a more nuanced description of test-taking behavior than the conventional engaged/disengaged distinction, offering a useful framework for process data analysis in large-scale assessments.