Help choosing right statistics

The experiment consisted of testing the effect of several diets on the weight of fish. I want to check what groups differ from each other. There is no homogeneity of variance between groups. Is it correct if i use the Kruskall Wallis test (I have read that the kruskall-wallis should not be used when the variances are different)? I know this test only answer the question " is there a difference" and not " where is the difference". What test can i do after to check where are those difference?
I'm confused with the amount of information i have found.. It sometimes seem it is more a matter of preference than of being right or wrong..
If someone could help me i would be greatly appreciated. I can give more details of the experiment if necessary.
Thanks in advance,
André Barreto


TS Contributor
How many groups are there, how large is your total sample size, are sample sizes different between groups? How large is the smallest standard deviation and how large is the largest standard deviation within groups?

With kind regards



Omega Contributor
Did you randomize the diet or can you rule out needing to control for confounders?

Why not use linear regression analysis?