Dealing with missing data in a multi-question depression scale: a comparison of imputation methods
2006

Comparing Methods for Handling Missing Data in Depression Scale

Sample size: 1580 publication Evidence: high

Author Information

Author(s): Fiona M Shrive, Heather Stuart, Hude Quan, William A Ghali

Primary Institution: University of Calgary

Hypothesis

This study aims to compare six different imputation techniques for dealing with missing data in the Zung Self-reported Depression scale (SDS).

Conclusion

Multiple imputation is the most accurate method for dealing with missing data in most scenarios assessed for the SDS.

Supporting Evidence

  • Multiple imputation produced the highest Kappa statistic (0.89) indicating near perfect agreement.
  • Individual mean imputation performed comparably well in many scenarios.
  • Random selection method yielded poor correlations as missing data increased.

Takeaway

When researchers have missing answers in surveys, using a method called multiple imputation helps fill in those gaps more accurately.

Methodology

The study compared six imputation methods using simulated missing data scenarios on responses from 1580 participants who completed the SDS.

Limitations

The random simulations may not reflect real missing data patterns, and the study excluded participants with truly incomplete questionnaires.

Participant Demographics

Participants were surgical patients, with a total of 1931 surveyed, of which 1580 completed all items.

Digital Object Identifier (DOI)

10.1186/1471-2288-6-57

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication