I have two replicates (tanks) for 4 treatments (hyperthermia, hypoxia, combined hyperthermia and hypoxia, and control), with 10 randomly assigned organisms for each tank. The response variable is non-normal. I have tested for treatment effect through Kruskal-Wallis H and Mann-Whitney U test (Wilcoxon) as posthoc analysis to evaluate all pairwise comparisons. For these analyses, I considered all individuals for each treatment (aggregated individuals from the two replicate tanks).
I know the conditions between replicates are the same, and individuals were randomly distributed before the experiment. Reviewers posted the question of the potential blocking effect between the two replicates for each condition. Is there a non-parametric test I can use to demonstrate no blocking effect?
