Change within groups and differences between groups

Hi All,

I have a question that perhaps is basic for most of you. I am hoping I can rely on your expertise to answer this question.

I am trying to study the effectiveness of an intervention. All participants receive the intervention. Upon analyzing the demographics of the participants I realize there are about four clearly distinct groups of participants receiving the interventions (all have different sample size). What I want to do is to:

1. Find if for all of these groups of participants there is a significant change in a number of outcome measure. In other words, is there significant change for each group - independent of what happens in the other groups.

2. If there are significant differences between the groups in terms of the change. In other words, did one group benefited more from the intervention than another or all others

I am collecting pre- and post-treatment measures of my outcomes. I have about 5 different outcomes all measured at both time. I realize that the methodology is not the strongest to study change but this is, in a way, a "pilot" study that could evolve into a study with three/four measurement points.

My question is: how would I ago about analyzing my data with these goals in mind (i.e., what procedures would you use). Also, I want to point out that I'll be using SPSS of the analyses.

I appreciate any and all comments you might have.

I wanted to add some thoughts about how I am thinking of analyzing the data.

For the within-group comparison (i.e., investigate whether there is significant change from pre to post for each) I thought I would created gain scores (post - pre = gain scores) and then I would run a one-sample t-test against a Ho = 0.

For my second question where I want to know there are differences between groups (i.e., how change for each group compares with change of the other groups) I thought I would run an ANCOVA where I would have my pre-scores as covariates and the post-scores as DV. I would ask for post-hoc comparisons for each pair of groups.

My hesitation with this plan of analysis is that (1) I am not sure if there is a more efficient way of doing these analyses. By this I mean that I am not sure if a mixed ANOVA (for instance) would yield the same results I am looking for. (2) I also am unsure if because I have unequal sample sizes in my groups (largest having an N=~1400 participants while smaller group an N=~44) would be problematic for assumption checking for the analyses - especially the ANCOVA.

I would appreciate any comments you have.