The robust estimation of examinee ability based on the four-parameter logistic model when guessing and carelessness...

Area of Science:

• Psychometrics
• Educational Measurement
• Statistical Modeling

Background:

• Traditional logistic models like the three-parameter Logistic Model (3PLM) and four-parameter Logistic Model (4PLM) aim to mitigate biases from guessing and carelessness in test responses.
• However, these models may inadvertently affect examinees who do not exhibit guessing or careless behaviors, introducing potential inaccuracies.
• Addressing this limitation is crucial for accurate assessment of true ability.

Purpose of the Study:

• To introduce a novel robust estimation approach, the four-parameter Logistic Model-Robust (4PLM-Robust), designed to overcome the limitations of existing models.
• To incorporate a critical-probability guessing parameter and a carelessness parameter within the 4PLM-Robust framework.
• To evaluate the effectiveness of the 4PLM-Robust against established methods in reducing estimation bias.

Main Methods:

• The proposed 4PLM-Robust method was developed incorporating specific parameters for guessing probability and carelessness.
• Comparative analysis was conducted against the two-parameter Logistic Model with Maximum Likelihood Estimation (2PLM-MLE), 3PLM-MLE, 4PLM-MLE, Biweight estimation, and Huber estimation.
• Bias was assessed through an illustrative example and three distinct simulation studies.

Main Results:

• The 4PLM-Robust demonstrated effectiveness as a robust estimation method, significantly reducing bias in the analyzed data.
• Performance comparisons indicated that the 4PLM-Robust offers a more accurate estimation compared to standard MLE-based logistic models when response disturbances are present.
• Computational efficiency was observed, with the 4PLM-Robust proving simpler to calculate than the Biweight and Huber estimation methods.

Conclusions:

• The 4PLM-Robust presents a valuable advancement in psychometric modeling for handling response disturbances.
• This method offers a more accurate and robust estimation of item and ability parameters, particularly in the presence of guessing and carelessness.
• The simplicity of calculation further enhances its practical applicability in educational and psychological assessments.