topic Re: Test for continuous-by-class interaction in Statistical Procedures

Test for continuous-by-class interaction

issac — Fri, 17 Aug 2012 20:55:10 GMT

Hi folks;

Am working with 9.3 version. I wonder what test is the best for testing possible interactions with one continuous variable and one class variable. Let's say, I have Length-Of-Stay as continuous and Race as class and would like to test whether the interaction is significant or not.

Thanks!

Issac

Re: Test for continuous-by-class interaction

art297 — Fri, 17 Aug 2012 22:01:54 GMT

It's late on a Friday, but I'm either missing something or you are looking for something that can't exist. Isn't an interaction, by definition, differences in a dependent variable based on the levels of two or more independent variables.

If you just want to identify whether differences are significant, and your data meet the test's assumptions, couldn't you get what you want by simply using a one way anova?

Take a look at: http://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_anova_sect003.htm

Re: Test for continuous-by-class interaction

PGStats — Fri, 17 Aug 2012 22:14:02 GMT

Search for "Homogeneity-of-slopes" in the documentation. One should do this test before any ANCOVA.

Re: Test for continuous-by-class interaction

issac — Sat, 18 Aug 2012 14:40:17 GMT

Thank Arthur and PG

I think the way to go is to ANCOVA since there is one class and one continuous variables and check for slope homogeneity as well.

Issac