Permutaatiotestaus

Permutaatiotestaus, also known as permutation testing or randomization testing, is a non-parametric statistical method used to assess the statistical significance of observed data. It is particularly useful when the assumptions of traditional parametric tests, such as normality, are violated or when dealing with small sample sizes. The core idea of permutation testing is to generate a null distribution by randomly reordering the observed data points.

The process typically begins with a specific hypothesis, usually a null hypothesis stating there is no effect or difference. The observed data is analyzed, and a test statistic is calculated. Then, the labels or group assignments of the data are randomly shuffled many times, creating numerous permuted datasets. For each permuted dataset, the same test statistic is recalculated. This process generates an empirical distribution of the test statistic under the null hypothesis.

Finally, the observed test statistic is compared to this empirical distribution. The p-value is determined by the proportion of permuted test statistics that are as extreme or more extreme than the observed test statistic. If this proportion is small (typically below a chosen significance level, like 0.05), the null hypothesis is rejected, suggesting that the observed effect is statistically significant. Permutation testing is computationally intensive but does not rely on distributional assumptions, making it a robust alternative in many scenarios.