GEE procedure

[tan1721]

I am trying to fit GEE model and using "Proc GEE" method in SAS. This method gives "GEE parameter estimate" table. But "Proc GENMOD" also gives the "GEE parameter estimate". One difference I found is GENMOD procedure has the option to provide Wald statistics, but GEE procedure does not. Also, most of the journal papers in my filed (transportation) reported Wald statistics with their parameter estimate. Now, I am confused about two things: 1) should I use GENMOD procedure or GEE procedure? 2) should I report Wald statistics or Z statistic for GEE model?. 

I have been searching in internet, but was not able to find out the answer. Help from SAS community will be highly appreciated.

[PaigeMiller]

---

@tan1721 wrote:

I am trying to fit GEE model and using "Proc GEE" method in SAS. This method gives "GEE parameter estimate" table. But "Proc GENMOD" also gives the "GEE parameter estimate". One difference I found is GENMOD procedure has the option to provide Wald statistics, but GEE procedure does not. Also, most of the journal papers in my filed (transportation) reported Wald statistics with their parameter estimate. Now, I am confused about two things: 1) should I use GENMOD procedure or GEE procedure? 2) should I report Wald statistics or Z statistic for GEE model?. 

I have been searching in internet, but was not able to find out the answer. Help from SAS community will be highly appreciated.

---

If GENMOD gives what you want, that seems to be a reason to use it. Regarding whether or not you should use Wald statistics or Z statistics, that is really up to you and your particular analysis needs.

[tan1721]

Thank you!

[StatDave · Accepted solution]

The Wald and t-tests are the same thing and give the same p-value. The Wald is just the square of the 1 df t-test. So, you can use either equally.