Lmst
Login

#MultipleImputation

Jan R. Boehnkejrboehnke
2024-09-10

A on reporting in observational studies that use for causal effect estimation:
bmcmedresmethodol.biomedcentra

Informative read beyond the particular application.

As a I am still working more on the problem that most studies do not even report their sampling and take it for granted that complete-data only analyses are fine.

As noted before 😉
mastodon.social/deck/@jrboehnk

Flow chart across stages of sample selection. The first stages are labelled as "not reported in paper" and comprise n=5234 approached for participation n=1489 agreed to screening interview n=981 fulfilled criteria n=823 provided consent n=580 returned survey and of these latter, n=218 are excluded due to missing data. The paper reports N=362 as analysed.
Nathaniel D. Porterndporter@sciences.social
2023-04-19

Just used #multipleimputation in #SPSS for the first time (used Stata and R before). It's not bad (I'd give it a B) and the reporting pooled estimates works well. That said (1) I had a terrible time imputing categorical variables and (2) skip patterns don't work well because I can't specify constraints to only impute certain variables for given values of another variable.

Client Info

Server: https://mastodon.social
Version: 2025.04
Repository: https://github.com/cyevgeniy/lmst