03-27-2015 03:49 PM
Without more detail to your query, it is had to do much more than point you to the proper procedure and tell you to read the documentation.
ANCOVA is a special case of the General Linear Model. If straightforward, I might first approach it using PROC GLM. If you do a Google search for
you will find a number of relevant hits that your data might fit into.
If your transaction data has repeats within customer then you might be in the area of mixed models. The Google search I mentioned also includes some references to those.
03-28-2015 03:38 AM
I read the ANCOVA procedure and was looking to implement with my available data. I searched for the documentation and couldn't figure out the complete one.
May I request you to provide me the link to utilize the example which you want me to look for?
Thanks in advance.
04-02-2015 02:28 PM
The covariates are selected based on the science of the problem. The covariate is expected to be continuous* and it's relationship to the dependent variable is assumed to be linear.
*by the Central Limit Theorem, a binary (0/1 coded) covariate meets the criteria if the sample size is sufficient. For 'how big is big enough', you need to reference some of the linear models textbooks.