Related Experiment Videos
The problem of missing clinical data for research in psychopathology: some solution guidelines
1Department of Psychiatry and Behavioral Sciences, University Health Sciences/The Chicago Medical School, Illinois 60064, USA.
The Journal of Nervous and Mental Disease
|April 1, 1994
Summary
Psychiatric researchers can use an intermediate value imputation method for missing data, which is effective even with nonrandomly missing values. This practical approach is more informative than dropping subjects from analyses.
Area of Science:
- Psychiatry
- Biostatistics
- Clinical Research
Background:
- Statistical adjustment for missing data is a challenge in psychiatric research.
- Missing data in clinical assessments can arise nonrandomly, particularly in conditions like schizophrenia.
Purpose of the Study:
- To address the lack of guidelines for handling missing data in psychiatric research.
- To evaluate the effectiveness of different methods for statistically adjusting missing data.
Main Methods:
- Utilized structured clinical instruments to collect data from 241 patients.
- Compared a simple intermediate value imputation method with a more complex vectoring method.
- Analyzed nonrandomly missing data, specifically unrated items for hallucinations and delusions.
Main Results:
- Nonrated items were more common in chronic schizophrenics and those with high psychopathology scores.
- The intermediate value imputation method effectively handled nonrandom missing data.
- This simple method demonstrated comparable patient discrimination to the vectoring method.
Conclusions:
- Recommends the intermediate value method for handling missing data in psychiatric research, especially with linear relationships.
- Emphasizes that missing values do not equate to missing information.
- Advocates against subject attrition (dropping subjects) as the least informative approach to missing data.