Timeline for Combining multiple imputation and survey non-response adjustments (IPW)
Current License: CC BY-SA 4.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| S Jan 10, 2023 at 13:24 | history | suggested | nstjhp | CC BY-SA 4.0 |
Need to use the calculated weights in the modelling function
|
| Jan 10, 2023 at 13:09 | review | Suggested edits | |||
| S Jan 10, 2023 at 13:24 | |||||
| Oct 25, 2022 at 15:29 | vote | accept | nstjhp | ||
| Oct 25, 2022 at 14:44 | history | edited | Noah | CC BY-SA 4.0 |
added 29 characters in body
|
| Oct 25, 2022 at 11:34 | comment | added | nstjhp |
Minor code improvements for future readers (@Noah please correct if wrong): The 1st fn should return the di_nonmissY dataframe. The 2nd needs an extra argument to map_dfr i.e. .id = ".imp". In my case fits should be the result of a binomial glm.
|
|
| Oct 25, 2022 at 11:34 | comment | added | nstjhp | Thank you, helps a lot. Useful comment on Seaman et al too. They confuse me in their Sec4 "[MI/IPW has] the disadvantage of having to specify an imputation model for $X$ with that of losing out on the potential efficiency gains of imputing $Y$". Nevertheless it seems other authors describe it like I asked for and how you show i.e. MI for item non-response and IPW for unit non-response. Did I/others read their paper wrongly? | |
| Oct 24, 2022 at 19:21 | history | answered | Noah | CC BY-SA 4.0 |