topic Re: within group p-value in proc glm in Statistical Procedures

within group p-value in proc glm

Melk — Mon, 30 Oct 2017 17:53:27 GMT

I am running an ANOVA (on change score controlling for pre) on some pre-post data for a 2 group randomized trial. I have 1 pre and 1 post measurement. I want a p-value for the within group change from pre to post for each group, separately. Any way I can output this in my GLM model?

Re: within group p-value in proc glm

Reeza — Mon, 30 Oct 2017 18:05:33 GMT

Do you mean you want the ANOVA run for each group separately? If so, try using a BY statement.

Re: within group p-value in proc glm

Melk — Mon, 30 Oct 2017 18:10:09 GMT

What p-value would give me the significance of the difference between pre and post then?

Re: within group p-value in proc glm

sld — Mon, 30 Oct 2017 20:51:00 GMT

If your response variable is the difference = (pre - post) such that you have a model statement that looks something like

class group;
model difference = group;
lsmeans group / cl;

then could you test whether the estimate of each group mean is different from zero? You could even put a confidence interval around the estimate.

Re: within group p-value in proc glm

Melk — Mon, 30 Oct 2017 22:10:51 GMT

Don't know why I didn't think of that - thanks!