topic Re: Trying to find a family p-value from proc mianalyze in Statistical Procedures

Trying to find a family p-value from proc mianalyze

RAGC — Fri, 07 May 2021 18:11:19 GMT

Hi everyone,

I hope you are doing well! I feel like there is a simple solution to my question, and I have just missed it in the documentation so hopefully you can help!

I am conducting a regression analysis using imputed data. I have included my interaction term for the proc mixed analysis. As part of proc mixed, I have specified type3 in the model statement to give me the score statistics for type 3 GEE analysis as pictured here:

However, when I use proc mianalyze, I only get the individual p-values which I would typically see as part of the analysis of GEE parameters estimates.

Essentially, my question is is there a way to get the overall p-values as pictured above from proc mianalyze?

Re: Trying to find a family p-value from proc mianalyze

ballardw — Fri, 07 May 2021 20:20:13 GMT

Since the type of output that may be generated is dependent on the options used you should provide, as a minimum, the code used to generate your output.

You are comparing output from one proc with another, so maybe you need to provide the code from both procs.

Re: Trying to find a family p-value from proc mianalyze

RAGC — Fri, 07 May 2021 20:43:14 GMT

The code I used to generate the first statement is:

proc genmod data = long;
class id timeptn(ref='0') sexn;
model weight = age timeptn|sexn BMI TT/type3;
repeated subject=id /type=exch MODELSE;
by _imputation_;
ods output GEEModPEst=Out1;
run;
quit;

The code that I have been using for proc mianalyze that is missing the output I am looking for is:

proc mianalyze parms(classvar=level)=Out1 edf=226;
class timeptn sexn;
modeleffects intercept timeptn sexn timeptn*sexn age BMI TT;
run;