Comparer des méthodes

	ANOVA robuste (moyenne tronquée de Welch)	Inférence par bootstrap	Test par permutation (ou randomisation)
Domaine	Statistique	Statistique	Statistique
Famille	Regression model	Regression model	Regression model
Année d'origine	1951	1979	2005
Auteur d'origine	Welch (1951); robust trimmed-mean approach popularised by Wilcox	Bradley Efron	Good (2005); Edgington & Onghena (2007); resampling tradition
Type	Robust one-way analysis of variance	Resampling-based inference	Nonparametric resampling test
Source fondatrice	Welch, B. L. (1951). On the comparison of several mean values: an alternative approach. Biometrika, 38(3/4), 330-336.	Efron, B. (1979). Bootstrap Methods: Another Look at the Jackknife. Annals of Statistics, 7(1), 1-26.	Good, P. (2005). Permutation, Parametric and Bootstrap Tests of Hypotheses (3rd ed.). Springer. ISBN: 978-0387202792
Alias	Welch ANOVA, trimmed-mean ANOVA, heteroscedastic one-way ANOVA, Robust ANOVA (Welch & Trimmed Mean)	bootstrap, bootstrap resampling, nonparametric bootstrap, Bootstrap Çıkarımı	randomization test, exact permutation test, re-randomization test, Permütasyon Testi
Apparentées	5	5	5
Résumé	Robust ANOVA compares the central tendency of three or more groups when the classical assumptions of normality and equal variances fail. It combines Welch's heteroscedasticity-adjusted statistic, introduced by Welch in 1951, with trimmed-mean tests advanced by Wilcox, giving reliable comparisons in the presence of outliers and unequal group spreads.	Bootstrap inference, introduced by Bradley Efron in 1979, estimates the sampling distribution of a statistic by repeatedly resampling the observed data with replacement. It requires no distributional assumption and produces reliable confidence intervals even in small samples.	The permutation test is a nonparametric resampling procedure that builds the sampling distribution of a test statistic directly from the data by repeatedly shuffling the group labels. Developed in the resampling tradition and treated systematically by Good (2005) and Edgington & Onghena (2007), it requires no parametric distributional assumption and yields an exact p-value.