Trends Analysis

#1
Hello all.

I am using SPSS to analyze associations between genetic polymorphisms and certain variables with abnormal distribution. I would like to add a trend analysis in kruskal-Wallis test. Is this possible at all or should I use another test to calculate p for trends for differences between variables with abnormal distribution?

Thanks in advance
 
#2
Hi, can you tell a little bit more about your outcome and predictor variables, e.g., is this genetic polymorphism variable a continuous variable? And what nature are the predictors, and what exactly is not normally distributed? The predictor variables itself do not have to be normally distributed for a classical ANOVA, but only the response variable e.g. with respect to different levels of predictor variables. And where do the trend appear in your model? If you want to test for a trend and assumptions of the classical ANOVA are violated, you should use the Jonckheere-Terpstra-Test, which is like the Kruskal-Wallis-Test a non-parametric ANOVA, but also allows the testing for trends.