Imputing missing data in surveys
Problem description
National survey lots of missing values. In particular the response has 82% missingness and explanatory variables range from almost complete to around the same level of missingness. How can we impute the missing values?
Consulting response
Questions
- How was the data gathered? Simple random sample? Stratified?
- Why are you interested in doing imputation?
Comments
- 82% seems like a lot of missingness, especially in the response
- The client will likely need to get comfortable with missing data approaches.
Relevant links
- Statistical Analysis with Missing Data
- https://stats.stackexchange.com/questions/149140/how-much-missing-data-is-too-much-multiple-imputation-mice-r
- https://www.theanalysisfactor.com/mar-and-mcar-missing-data/
- https://en.wikipedia.org/wiki/Missing_data
- http://www.stat.columbia.edu/~gelman/arm/missing.pdf
blog comments powered by Disqus