Assumptions and nonparametric tests

For linear models the following assumptions should be met (see chapter 7)

• Linearity
• Equal variance/ Homoscedasticity
• Normal distribution of residuals
• Independence (except for linear mixed models)

If the assumptions are not met, you have two options

• Transformation of data
  • Logarithmic transformation
  • Square roots transformation
• Nonparametric tests
  • For the ordinary linear model (see chapter 8)
    • Spearman´s rho
    • Kendall´s Tau
    • Kruskal-Wallis test
    • 
      
  • For the linear mixed model (see chapter 14)
    • Friedman´s test
    • Wilcoxon´s signed ranks test