Non-Equivalent Group Statistical Analyses - is ANOVA appropriate?

#1
Hi there,

I have pre/post data on the same tool for two different populations (Group A: Higher needs, Group B: Lower needs) and two different services (Intensive counseling for group A, Brief counseling for group B). Other variables include age, gender and length of service.

Would it be inappropriate to conduct a mixed 2x2 anova to look at the variance across the groups at start of their treatment and at the end of their treatment?
I am getting confused with statistical literature that seems to suggest this is inappropriate.

Also, I was thinking that it would be appropriate to do a one-way anova of the post-treatment scores using the IV as the two treatments, the DV as the post score, and the Covariate. -- Should I be using the post score as DV or should I compute a change score (pre minus post).

Thanks for any feedback!
 

Karabiner

TS Contributor
#2
So you have 2 groups, one with higher needs and intensive counseling,
one with lower needs and brief counseling. And you have pre-post data.
I suppose the dependent variable is measured on an interval scale. So,
what exactely do you mean with
to conduct a mixed 2x2 anova to look at the variance across the groups at start of their treatment and at the end of their treatment?
I am getting confused with statistical literature that seems to suggest this is inappropriate.
Why inappropriate, for what reason? (by the way, you would ultimately
look at the group means, not at variances).


With kind regards

K.
 
#3
Hi Karabiner,

Thank you for your response!

I realize now that I was probably using the wrong method. I spent a lot of time going back and forth, largely with trying to focus my research question(s). I had wanted to test for significant differences (and effect sizes) between the two treatment groups; however I am not looking for an interaction effect.

My initial concern was about internal validity when there are nonequivalent groups being analyzed in a mixed anova based on articles, such as this: http://www.socialresearchmethods.net/kb/statnegd.php

Thank you for your time!