Including Empirical Prior Information in the Reliable Change Index.

Area of Science:

• Psychometrics
• Statistical Modeling

Background:

• The reliable change index (RCI) is a standard metric for assessing individual change over time.
• Existing RCI methods have undergone numerous improvements since their inception.
• Item response theory (IRT) offers a sophisticated framework for psychometric analysis.

Purpose of the Study:

• To extend an existing IRT-based RCI by incorporating empirical prior information.
• To evaluate the impact of empirical priors on RCI computations, particularly with quantifiable group-level differences.
• To enhance the accuracy and power of the RCI in detecting genuine changes in latent traits.

Main Methods:

• Reanalysis and extension of a prior simulation study.
• Implementation of empirical priors within an IRT version of the RCI.
• Comparison of the proposed RCI with competing estimators.

Main Results:

• Inclusion of empirical priors introduces minimal bias when no true change exists.
• Empirical priors generally improve the Type I error rate of the model-based RCI.
• Bias and sampling variability are reduced when non-zero changes are present, increasing detection power.

Conclusions:

• The proposed IRT-based RCI with empirical priors offers improved performance.
• This enhanced RCI provides a more powerful and reliable tool for detecting individual changes.
• The method is particularly beneficial when group-level differences can be quantified from post-test data.