Hi all,
I am beginning to run some tests looking at the effect of an intervention on emergent reading ability. Two groups, control and experimental, tested on a series of subskills (i.e., hearing first sound in words, hearing last sounds in words, etc) both before and after intervention. One-way ANOVA on starting scores showed the experimental group performed significantly lower on 4/6 subskills assessed compared to control group. All 6 subskills were assessed post-intervention - my question is: is a one-way ANCOVA appropriate, using pre-intervention scores as the covariate, or is MANCOVA more appropriate given that there are 6-subskills being measured post-intervention (i.e., dependent variables)? (these subskills are related skills)

Also, post-intervention we also looked at non-word reading (not assessed pre-intervention) between the groups - would a one-way ANOVA be most appropriate for analyzing between group differences on that dependent variable? My understanding was that ANOVA requires three or more means to compare however I have come across a few sites saying that this can be done with just two means????

Any advice would be appreciated :)
For each subskill i you need to create variable X_i = Measurement After - Measurement Before. All analysis should be run on these differences only. If the sample size is substantial, it is best to run 6 univariate ANOVA and apply the Bonferroni correction. If the sample size is modest (like 30-50), it is best to run MANOVA since the Bonferroni correction would be nuclear in this case.