Hello, I am using the survey package in R to analyze my complex survey data. We have all the marginal distributions for each strata of interest in the population (age, sex, region, type). We have a complex weighting system including:
-design weights oversampling certain strata
-nonresponse adjustment based on logistic regression propensity score
-post stratification using postStratify() command
What would be the reason for total poststratified sums not matching perfectly with the actual population strata totals? In some strata they're fairly close, but in others they're off by several hundred. I assume it's something not working right in the postStratify() command but I can't find anything about it. Thank you.
Suggest you start with the attached screenshot which hopefully will be of some assistance Best wishes David Booth
Hi David, thank you I have been through the package instructions and was not able to solve it.
I am suggesting that you expand your search for solutions. First look at the Lumley paper that is showing in the screen shot. Next contact the package coordinator which you can find by looking at the material in this new screen shot. Then go from there. Best wishes David Booth
Do you have any missing values in your population? Maybe Poststratify h've dropped them.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey