Comparing concurrent versus fixed parameter equating with common items: using the dichotomous and partial credit...

Area of Science:

• Psychometrics
• Educational Measurement
• Statistical Modeling

Background:

• Equating is crucial for large-scale assessments to ensure score comparability across different test forms.
• Calibration methods, including anchoring, are debated for their impact on equating accuracy.
• Limited research exists comparing fixed calibration with concurrent calibration, particularly in psychometric practice.

Purpose of the Study:

• To compare the accuracy of fixed calibration versus concurrent calibration in test equating.
• To investigate the influence of sample size and common items on calibration method performance.
• To evaluate these methods using dichotomous Rasch and Rasch partial credit models.

Main Methods:

• Simulation study comparing fixed calibration and concurrent calibration.
• Utilized dichotomous Rasch and Rasch partial credit models.
• Employed the WINSTEPS computer program for calibration and equating.

Main Results:

• Contrary to expectations, concurrent calibration did not consistently yield greater accuracy in item parameter estimation.
• The accuracy of calibration methods was confounded by sample size and the number of common items.
• No definitive benefit was found for using one method over the other for parallel test forms.

Conclusions:

• The choice between fixed and concurrent calibration methods for equating parallel test forms offers no inherent advantage.
• Equating accuracy is influenced by design factors (sample size, common items) rather than the calibration process alone.
• Psychometricians should consider these confounding factors when selecting calibration strategies in large-scale assessments.