Identifying Disengaged Responding in Multiple-Choice Items: Extending a Latent Class Item Response Model With Novel...

Area of Science:

• Educational Measurement
• Psychometrics
• Cognitive Psychology

Background:

• Disengaged responding threatens the validity of educational assessments.
• Current methods using response times risk misclassification.
• Process data offers richer insights into test-taking behavior.

Purpose of the Study:

• To investigate the utility of process data (text reread, item revisit, answer change) in identifying disengaged responding.
• To develop an extended latent class item response model incorporating these indicators.
• To compare the extended model with a baseline model using only response time.

Main Methods:

• Developed an extended latent class item response model.
• Included text reread, item revisit, and answer change as predictors of engagement.
• Applied the model to a sample of 1,932 German university students.

Main Results:

• The extended model showed a better fit than the baseline model.
• Item response time and text reread were significant predictors of engagement.
• No systematic differences were found in parameter estimation or classification.

Conclusions:

• Process data, specifically text rereading, offers a marginal improvement in identifying disengaged responding.
• Further research is needed to fully leverage process data for detecting unmotivated test-takers.