STATS MAKE ME CRY

I just completed a study with colleagues where we used two different types of social skills interventions with students in three different categories. Students in all three categories were places in one of the two interventions. Pre- and Post-test data was taken using a behavior rating scale and a social skills rating scale. There are 29 variables on which we have pre- and post-test data. Our n=23 so we have a small sample size. Intervention groups are just about even with 11 in one group and 12 in the other. When you break it down to student categories our groups become 5, 6, and 12 in size.

I'm using SPSS for data analysis. I ran within-sample t-test for the two different intervention groups and found significance in some scales using p <.10. So my thought was to run a more complex analysis to determine if there is change between intervention groups.

So I'm trying now to figure out with running a Repeated Measures ANOVA or MANOVA would be better for this? If I run a MANOVA using the pre-test scores as covariates then which line in the SPSS output am I really looking at b/c it would compare every since post-test against every single pre-test?

Any clarification would help greatly.
Thanks!

Thanks. I figured with breaking down the sample into three groups breaks them down into much. So even I ran a repeated measures on a two group breakdown with one group with an n=11 and the other group with an n=12 do you think it's still too small?

I was just really hoping to be able to get more out of the data. :)

thanks again!