Nonparametriske

Nonparametriske (nonparametric) methods are statistical techniques that do not assume a specific parametric form for the population distribution. This makes them suitable when data do not meet the assumptions of parametric tests, such as normality, or when data are ordinal or contain outliers. Nonparametriske approaches focus on the ranks, signs, or resampling of data rather than on estimating parameters of a distribution.

Common techniques include rank-based tests such as the Mann-Whitney U test (also called the Wilcoxon rank-sum test), the Wilcoxon signed-rank test, and the Kruskal-Wallis test. For association or correlation, nonparametric measures such as Spearman's rho and Kendall's tau are used. For comparing more than two groups across repeated measures, the Friedman test is used. Permutation tests and bootstrap methods are widely used nonparametric approaches that rely on resampling to derive inference without strong distributional assumptions.

Advantages of nonparametric methods include robustness to deviations from normality and applicability to small samples or ordinal data. Limitations include typically lower statistical power relative to parametric methods when the parametric assumptions are satisfied, and interpretation that often centers on medians or ranks rather than means.

In modern practice, nonparametric techniques are common across disciplines and are often combined with resampling to produce p-values and confidence intervals that do not rely on distributional assumptions.