One way ANOVA or repeated measures? Does anyone know which type of ANOVA is appropriate to test the hypothesis: 'There is no difference in flexibility (measured in cm) due to type of prior warm-up'? There are 6 different warm-up types. The data are normally distributed and have homogeneity of variance.
 A: The question of whether to use one way ANOVA or one way repeated measures ANOVA has to do with the dependence or independence of the samples in each group (group is warm up type in your case).
Dependent samples would be those measured, for example, in the same individual (i.e. each individual is measured for flexibility each time after performing each warm up routine). A common term for this kind of dependency is blocked samples (which are quite like the paired samples, of the paired t test). In your case, if your samples were dependent, your outcome measures (flexibility) would be blocked by individuals.The appropriate analysis here would be one way repeated measures ANOVA.
Independent samples would be those measured, for example, in different individuals (i.e. here's a group of people measured for flexibility after warm up type 1, here's a separate group of people measured for flexibility after warm up type 2, etc.). Such samples are sometimes terms unblocked, though often they're just called independent. The appropriate analysis here would be one way ANOVA.
Of course the dependence or independence of data also depends on what one is doing with it. So while samples used for one way ANOVA are independent, the post hoc pairwise comparisons one makes with the same data induces a dependency among the comparisons, since, for example, the group 1 data used to measure group 1 versus group 2 differences, are reused to measure group 1 versus group 3 differences, etc. (as opposed to separate samples of group 1 being obtained for each pairwise comparison).
