Related Experiment Videos
Practical methods for dealing with 'not applicable' item responses in the AMC Linear Disability Score project
Rebecca Holman1, Cees A W Glas, Robert Lindeboom
1Department of Clinical Epidemiology and Biostatistics, Academic Medical Center, Amsterdam, The Netherlands. r.holman@amc.uva.nl
Health and Quality of Life Outcomes
|June 18, 2004
Summary
Three imputation methods effectively handle missing data in the AMC Linear Disability Score (ALDS) item bank. Cold deck imputation proved unsuitable, while hot deck imputation and other models showed similar results for functional status and quality of life assessments.
Area of Science:
- Psychometrics
- Health Measurement
- Statistical Modeling
Background:
- Questionnaires assessing functional status and quality of life often have non-response items.
- This study addresses handling 'not applicable' responses within the AMC Linear Disability Score (ALDS) item bank.
Purpose of the Study:
- To evaluate different strategies for managing non-response data in the ALDS item bank.
- To identify suitable methods for imputation in the context of item response theory.
Main Methods:
- Analysis of data from 392 respondents to 32 items using the one-parameter logistic item response theory model.
- Comparison of four imputation strategies: cold deck, hot deck, item omission, and a response tendency model.
Main Results:
- Hot deck imputation, item omission, and the response tendency model yielded similar parameter estimates.
- Cold deck imputation produced substantially different estimates, indicating unsuitability.
Conclusions:
- Cold deck imputation is not recommended for the ALDS item bank.
- Hot deck imputation, item omission, and the response tendency model are viable options for the ALDS item bank and similar datasets using item response theory.