Missing Data Approaches in eHealth Research
Author Information
Author(s): Eysenbach Gunther, Smit Filip, Streiner David, Blankers Matthijs, Koeter Maarten W J, Schippers Gerard M
Primary Institution: Amsterdam Institute for Addiction Research (AIAR), Academic Medical Center, University of Amsterdam
Hypothesis
What are the best statistical approaches to handle missing data in eHealth research?
Conclusion
Using multiple imputation improves the validity of results when analyzing datasets with missing observations.
Supporting Evidence
- Multiple imputation techniques led to accurate results in the simulation study.
- Amelia II outperformed other methods with the smallest deviation from the reference value.
- Basic approaches like complete case analysis and mean imputation did not perform well.
Takeaway
Sometimes, researchers don't get all the answers they need, and this study shows how to fill in the gaps in a smart way so the results are still good.
Methodology
The study used a simulation approach to test various missing data techniques on a dataset from an online self-help program for problem drinkers.
Potential Biases
Some methods may lead to biased results if the missing data patterns are not properly addressed.
Limitations
The study focused on nonnormally distributed count data, which may not generalize to all types of data.
Participant Demographics
Participants were individuals in an online self-help program for problem drinkers.
Statistical Information
P-Value
0.001
Confidence Interval
95%
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website