Reference
Xiao, J., & Bulut, O. (2020). Evaluating the Performances of Missing Data Handling Methods in Ability Estimation From Sparse Data. Educational and Psychological Measurement, 80(5), 932-954. https://doi.org/10.1177/0013164420911136
Review
In the article "Evaluating the Performances of Missing Data Handling Methods in Ability Estimation From Sparse Data" (2020), Xiao and Bulut conducted two Monte Carlo simulation studies to evaluate the performance of four methods in handling missing data when estimating ability parameters. These methods include full-information maximum likelihood (FIML), zero replacement, and multiple imputations with chain equations utilizing classification and regression trees (MICE-CART) and random forest imputation (MICE-RFI). The authors assessed the accuracy of ability estimates for each method using bias, root mean square error, and the correlation between true ability parameters and estimated ability parameters.
The results of the study showed that FIML outperformed the other methods under most conditions. Interestingly, zero replacement provided accurate ability estimates when the missing proportions were very high. MICE-CART and MICE-RFI demonstrated similar performances, but their effectiveness appeared to vary depending on the missing data mechanism. As the number of items increased and missing proportions decreased, all methods performed better.
The authors also found that incorporating information on missing data could improve the performance of MICE-RFI and MICE-CART when the dataset is sparse and the missing data mechanism is missing at random. This research is valuable for educational assessments, where large amounts of missing data can distort item parameter estimation and lead to biased ability estimates.
