- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
I am trying to obtain pooled p value from ANOVA statistics of 50 imputed datasets created by proc mi. Would appreciate if anyone know the code.
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
The code for what?
- You have done PROC MI with nimpute=50 (for 50 imputed datasets), correct?
- Then you need to do the analysis you are interested in (like proc reg data=; by _Imputation_ ;).
- Then with PROC MIANALYZE you can combine the 50 sets of (in this case regression) coefficients.
Enter this in the Google search bar and you will find (blogs with) example code:
MI MIANALYZE site:blogs.sas.com
Kind regards,
Koen
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
F statistics for 50 imputed dataset. What SAS procedure I need to obtain
the 1 pooled p value. Thanks,
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
There isn't a way to combine F-statistics directly in SAS, although the process of doing so would be rather easy to program yourself basd on the rules in the paper below.
https://www.sciencedirect.com/science/article/abs/pii/S2452306221001076
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
Thank you for the link. From my understanding is that for Levene test, we can use Fisher Z transformation for F statitsics to eventually obtain the pooled p value. My follow-up question is that given calculation of pooled v (denominator degree of freedom) is
ν1= (m − 1)(1 + g −1 ) 2,
ν2= 0.5(p + 1)v1 (this is the final v)
and g = (1 + m−1 ) trace(B U¯ −1 )/p
I assume B is a matrix but I can't find anywhere in the paper the mention this matrix. Do you happen to know how to obtain the B matrix?
Thank you for your help.
- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content
I found another paper explains with B is. It is the covariance matrix of the estimates Qˆ m. Thank you so much for your guidance.