A modified hidden Markov model for detecting insufficient effort responses in questionnaires.

Area of Science:

• Psychometrics
• Data Quality Assessment
• Behavioral Data Analysis

Background:

• Insufficient effort response (IER) significantly degrades questionnaire data quality and impacts research validity.
• Existing IER detection methods often lack the ability to capture diverse IER types or account for participant state changes.

Purpose of the Study:

• To develop and validate an advanced statistical method for the dynamic detection of insufficient effort response (IER) in questionnaires.
• To enhance the accuracy and comprehensiveness of IER identification by integrating response patterns and response times.

Main Methods:

• Reconstruction of response and response time (RT) models within a hidden Markov model (HMM) framework.
• Development of a modified hidden Markov model (M-HMM) specifically designed to identify IER characteristics.
• Simulation studies to assess the M-HMM's parameter recovery and detection sensitivity under various conditions.

Main Results:

• The M-HMM demonstrated effective parameter recovery in simulation studies.
• Detection sensitivity was influenced by IER prevalence, RT distribution differences, and IER heterogeneity.
• Analysis of empirical data using M-HMM provided deeper insights into IER occurrences.

Conclusions:

• The M-HMM offers a robust approach for dynamically detecting diverse types of insufficient effort response (IER).
• This method enhances the assessment of item quality and questionnaire data integrity.
• Findings provide valuable tools for researchers and practitioners aiming to improve data validity.