Statistics Glossary → McDonald's ω (omega) reliability coefficient
McDonald's ω (omega) reliability coefficientconcept
An internal-consistency reliability coefficient that estimates, from factor loadings, how much of the variance in a scale's total score is due to the common factor(s).
In a one-factor model, ω = (Σλᵢ)² / [(Σλᵢ)² + Σψᵢ], where λᵢ are the item loadings and ψᵢ the residual (unique) variances. It ranges from 0 to 1 and is read like Cronbach's alpha. Alpha, however, assumes that all items load equally on the factor (tau-equivalence), and when loadings differ it usually understates reliability. Omega does not need this assumption and equals alpha when the loadings are equal.
For multidimensional scales two kinds of omega are distinguished. ω_t (total) is the proportion of variance explained by the general and group factors together. ω_h (hierarchical) is the proportion explained by the general factor alone, showing how far the total score reflects a single construct. ω_h is usually estimated through a Schmid-Leiman transformation or a bifactor model. Omega rests on the assumption that the factor model is correctly specified; correlated errors or the wrong number of factors can make it misleading.
An everyday example
In a 10-item anxiety scale, some items reflect anxiety very well (loading 0.8) and others weakly (loading 0.3). Alpha comes out at 0.74 and omega at 0.78. The gap shows that the items do not contribute equally and that alpha slightly underestimates reliability in this case.
In MerQur
In Cronbach's Alpha and Reliability Analysis, McDonald's omega_t is always computed. If m ≥ 2 group factors are entered in 'ω_h number of group factors' and there are at least 2m items, Schmid-Leiman-based omega_h and omega_t are reported as well.
Analyses where this concept appears
Cronbach's Alpha and Reliability Analysis
Related concepts
Split-half reliability and Guttman's λFactor (component) loadings