Sekce 1 – Teorie a metoda v pedagogice
Autoři: Petr Soukup
Abstrakt: Cílem příspěvku je ukázat na příliš časté používání neparametrických testů pro data z pedagogických výzkumů.
Klíčová slova: Statistické testy, neparametrické testy
Title: Too much ado about nonparametric tests?
Abstract: In pedagogy, the empirical analysis is often pointed out to the fact that data is not normally distributed and therefore it is necessary to use nonparametric tests instead of parametric tests. If we use a non-parametric test instead of a parametric test, the hypotheses we test change (e.g. we switch from testing equality of averages to testing equality of of medians). Furthermore, it should be noted that non-parametric tests have a lower power. An empirical illustration in ESCI (Cummings, Jageman, 2017), showing that the sampling distribution of averages is close to normal for skewed data. At the end of the presentation attention is drawn to the implications for teaching, where it is not necessary to include non-parametric tests in basic courses of statistics or quantitative methods.
Keywords: Statistical tests, nonparametric tests
Download the article in English [PDF]: Not available