Type III sum of squares for all predictors using regstats

Hi All, I'm running regstats with an 18 column x predictor and would like to obtain the type III sum of squares for each predictor in the model. Regstats just gives the overall value for the model, and when I tried using anovan Matlab keeps running out of memory. Is there a simpler way?

Answer 1

[Tom Lane]

Aloha. You've run into the issue that regstats doesn't want to do anova, and anovan wants to treat its x variables as categorical. But you can tell it not do do that:

x = randn(100,5);
y = 10 + x*(1./(1:5)') + 5*randn(100,1);
anovan(y,x,'continuous',1:5)

[Laura Kloepper]

Ah, Tom. I can always count on you. Thank you!