MerQuR+Statistics Glossary

Statistics Glossary → Communality

Communalityconcept

The proportion of a variable's variance that is explained by the common factors extracted in factor analysis (or by the components retained in principal component analysis); written h².

Factor analysis splits each variable's variance into two parts: the share held in common with other variables and explained by the common factors (the communality, h²), and the part unique to that variable (the uniqueness, 1 - h²). In an orthogonal solution, a variable's h² is the sum of its squared loadings on all the factors. h² = 0.70 means that 70% of the variable's variance is explained by the factors.

The initial communality is the estimate before any factor is extracted: in principal axis factoring it is usually the squared multiple correlation (SMC) obtained by regressing each variable on all the others, while in principal component analysis it is 1, because all of the variance enters the analysis. A variable with a low extraction communality (e.g. below 0.30) relates weakly to the factor structure and may be considered for removal from the scale. A communality above 1 (a Heywood case) signals an impossible solution and is usually linked to too many factors, a small sample or a misspecified model.

An everyday example

In a job-satisfaction survey, the item 'I am satisfied with my pay' has a communality of 0.65: it overlaps strongly with the general satisfaction factors the survey measures. The item 'My office has a window' has a communality of 0.08: it shares little with the others and largely measures something of its own.

In MerQur

It is given through the 'Communalities (h²) + variable MSA' output in Principal Component Analysis and the 'Initial/extraction communalities and Total Variance Explained' extra output in Exploratory Factor Analysis.

Analyses where this concept appears

Principal Component Analysis (PCA)Exploratory Factor Analysis (EFA)

Related concepts

Factor (component) loadingsProportion of explained variance