I have $10$ groups - sample size $n>700$: resampled to $710$ for ANOVA - visually these distributions are not normal, slight bimodlity in the sets.
I ran an ANOVA, and got a $P\approx 0.089$. It coincides with what I expected from the histograms, they should roughly have the same mean - how catastrophic is non-normality in this case? Can I assume robustness due to sample size?
Note: This is not for regression - not sure if residual normality still applies...
