Non-parametric statistics

Non-parametric statistics are used with categorical/ordinal outcomes and statistical assumption violations

Non-parametric statistics are used with categorical and ordinal outcomes. Non-parametric statistics are also used when the statistical assumptions of parametric statistics are violated. The statistical assumptions of parametric statistics include normality, linearity, homogeneity of variance (homoscedasticity), and model fit (residual analysis). Non-parametric statistics should also be used when examining small sample sizes (n < 20). 

Types of non-parametric statistics

Click on a button below to access the methods for conducting and interpreting non-parametric statistics.
There are many different kinds of non-parametric statistics: Chi-square Goodness-of-fit, one-sample median test, chi-square, Mann-Whitney U, Kruskal-Wallis, McNemar's test, Wilcoxon, Cochran's Q, Friedman's ANOVA, phi-coefficient, point biserial, rank biserial, biserial, Spearman's rho, logistic regression, multinomial logistic regression, proportional odds regression, Kaplan-Meier, Cochran-Mantel-Haenszel, and Cox regression.