Two-way ANOVA assumptions

lauramb — Tue, 30 Jan 2018 01:21:49 GMT

Hello everyone,

I'm new to SAS and I have a few doubts about how to check for the assumption in a two-way ANOVA.

For what I understand proc univariate looks at the distribution of the variable, not the residuals.

So i have this general script that I put together looking at different posts in SAS cummunities.

Is for a two way anova with interactions with two independent variables (A and B) and a response variable (Y).

PROC glm DATA=data1;
class A B;
model Y= A B A*B;
output out = resid p=predicted rstudent=rstd residual=res;
run;
quit;

Proc gplot data = resid; /*Homocedasticity*/
plot rstd* (predicted);
run;

proc univariate data = resid normal; /*Normality*/

var res;

histogram/normal;

run;

I don't know if this is the most parsimonious way to do this or there is a statement inside MIXED or GLM that I'm missing?

How can I include a QQplot of the sample vs theoretical quantiles?

Thanks for all your help!

Laura

Re: Two-way ANOVA assumptions

PGStats — Tue, 30 Jan 2018 05:17:36 GMT

Consider plots=diagnostics, as in:

proc glm data=sashelp.cars plots=diagnostics;
class origin type;
model msrp = origin | type;
run;

lauramb — Tue, 30 Jan 2018 19:28:17 GMT

There was an easier way...