Using PROC GLM to calculate an ANOVA for unbalanced data

Chloe1 — Wed, 26 Apr 2017 11:35:13 GMT

Hello, I'm currently running a model very similar to this example:

https://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_glm_sect053.htm

But I'm struggling to understand how the numbers 2.137, 2.2408 and 2.3162 are calculated in the Output 39.8.3 Expected Values of Type III Mean Squares. When using balanced data they should be all the same - the number of observations of each factor combination. But when we have unbalanced data the number of observations of each factor combination is different. Does anyone know how to calculate them or have any idea where they come from?

Thanks in advance!

Chloe

Re: Using PROC GLM to calculate an ANOVA for unbalanced data

Rick_SAS — Wed, 26 Apr 2017 17:41:02 GMT

The mathematical formulas are in the section "Computation of Expected Mean Squares for Random Effects"

topic Using PROC GLM to calculate an ANOVA for unbalanced data in SAS Procedures

Using PROC GLM to calculate an ANOVA for unbalanced data

Re: Using PROC GLM to calculate an ANOVA for unbalanced data