Thomas Lumley • 7/22/2021

Subsets and subpopulations in survey inference

This technical article addresses when it is valid to take a subset of a probability sample for survey inference. Using R and the survey package, it demonstrates correct and incorrect approaches to estimating totals for subpopulations (e.g., high school students), highlighting how improper subsetting leads to inaccurate standard errors due to misrepresenting the sample design.

0 comments

#R Programming #Statistical Inference #Sampling Design