ANOCOVA or Gain score?

mrsi

New Member
#1
Hi everyone,

I'm a newbie Ph.D. student and encountering an issue regarding which analysis to use for my study. Please help me out if you have some time.

My study design is basically two treatment groups and one control group with pre and posttest. However, the pretest scores for one of the variables have significant differences across the three conditions which I need to control for that. I tried ANCOVA and the homogeneity of variance was violated. Unfortunately, the group sizes were different (N=56/35/21). I know there has being controversies about using gain score analysis. I wonder what other analysis I can use. Thank you for your time. :D
 

Karabiner

TS Contributor
#2
The third (and straightfoward, IMHO) option would be repeated-measures ANOVA with an
additional grouping factor.

With kind regards

K.
 

mrsi

New Member
#3
Thank you, sir, for the reply. Would you please explain more? What would be the additional grouping factor? Thank you so much for your time.
 

mrsi

New Member
#5
Thank you again for your kind reply. I apologize for my confusion. One more questions. I used a 2 x 3 mixed ANOVA in my original manuscript. But the reviewer commented that I should control for the differences in pretest by using other analysis. Is there any way I can address this issue?
 

Karabiner

TS Contributor
#6
So you already had used the mixed ANOVA approach which I later recommended?

Unfortunately, I do not know an analysis which can really control for differences in baseline values.
Maybe your reviewer has the idea that using baseline as covariate can eliminate baseline differences.

Some discussion about ANCOVA versus gain scores (the latter produce results equivalent to
mixed ANOVA in case of 2 time points) can be found here http://wiki.math.yorku.ca/index.php/Statistics:_Gain_scores_vs._residualized_gain_scores

With kind regards

K.