10-06-2016 12:17 PM
I would be pleased if anyone familiar with this can help explain if there is something wrong:
I am analyzing my dataset, which has multiple factors. After running the model, I got a Type 3 result for which one of the interaction (fert*char) was 0.1713, indicating a non-significant interaction. On the other hand, if you check the lsmeans, you will realize that the p values are all significant (I have shown the values in red).
My question is, should the p value shown in the Type 3 test be the same as p value shown in the lsmean table?
10-06-2016 10:23 PM
Type 3 tests are about effects. The fert*char effect doesn't change the predicted mean in a statistically significant way.
LSMEANS tests are about means. The predicted means for every level of fert*char are significantly different from zero.