Non-parametric equivalent of an independent factor two way ANOVA?

[Rachyxxx]

Hi there,
I have been running ANOVAs with two IVs, the assumption of normality has a been violated on some of the levels and the distribution of residuals as been violated on some of the levels. I have therefore decided to carry out a non-parametric alternative to these ANOVAs. Would Kruskal-Wallis be a suitable alternative, or will it not work because I have two independent variables? Is there any alternatives? Any help is much appreciated

 

[fed2]

SAS gives the 'Van Elteren' test, this is with no interaction int he 2-way, which i guess is what independant means
https://support.sas.com/kb/25/022.html

with an interaction, you got me, I think it is the case that non-parametric tests and interactions do not get along, so you don't see it really that much.

 

[GretaGarbo]

Skip the normal distribution and go for a generalized linear model (glm)