Comparer des méthodes

	Oméga de McDonald robuste	Alpha de Cronbach robuste
Domaine	Psychométrie	Psychométrie
Famille	Latent structure	Latent structure
Année d'origine	1999 (omega); robust variant formalized in 2000s–2010s	2002–2016
Auteur d'origine	Roderick P. McDonald (omega); robust extension via robust SEM estimators (MLR, DWLS)	Derived from Lee J. Cronbach (1951); robust variants formalized by Yuan & Bentler (2002) and Zhang & Yuan (2016)
Type	Reliability coefficient	Robust reliability coefficient
Source fondatrice	McDonald, R. P. (1999). Test theory: A unified treatment. Lawrence Erlbaum Associates. ISBN: 978-0805830408	Yuan, K.-H., & Bentler, P. M. (2002). On robustness of the normal-theory based asymptotic distributions of three reliability coefficient estimates. Psychometrika, 67(2), 251–268.
Alias	robust omega, omega total (robust), robust omega-total, robust composite reliability	robust alpha, outlier-resistant Cronbach's alpha, robust internal consistency, robust coefficient alpha
Apparentées	4	3
Résumé	Robust McDonald's omega estimates the internal consistency reliability of a composite scale using factor-analytic loadings obtained through robust estimation methods (such as MLR or DWLS). Unlike standard omega or Cronbach's alpha, it remains accurate when item distributions are non-normal, skewed, or when the sample contains influential outliers — conditions common in applied psychological and educational measurement.	Robust Cronbach's alpha adapts the classical internal consistency coefficient to data that violate the assumption of multivariate normality or contain influential outliers. By replacing the conventional sample covariance matrix with a robust counterpart, it yields a reliability estimate that is resistant to distortion by non-normal response distributions, contaminated observations, or small violations of model assumptions common in applied psychometric work.