Comment by b1c1jones

I've found timings in general were lognormal a long time ago. I had to routinely apply transformations to get meaningful ANOVA and DOE results.