proc genmod SE doesn't match proc ttest SE

[best_means]

What is the difference between how standard error is calculated in proc ttest vs. proc genmod dist=normal.

I ran the following codes:

• proc ttest; class group; var age; run;
• proc genmod; class group; model age=group / dist=normal; lsmeans group; run;

The same means were produced but the standard errors were different.  Anyone know why?

[Ksharp]

I guess the distribution is different . T distribution V.S. Normal distribution.

[PaigeMiller · Accepted solution]

GENMOD uses maximum likelihood to estimate the scale parameter. TTEST uses a pooled estimate of variance. These are not the same.

[best_means]

Thanks for the quick reply.  This is helpful as "The PROC GENMOD scale parameter, in the case of the normal distribution, is the standard deviation."