= Latent factor
{title2=$F_j$}
A <latent factor> is an unobserved coordinate introduced to explain shared variation among measurements. In an <orthogonal factor model>, <latent factors> have <variance> one and are mutually uncorrelated, with coefficients specified by the <factor loadings>. They differ from a <regression factor>, which is a categorical predictor. A fitted <covariance> alone generally leaves the <latent factor> orientation undetermined; <factor rotation> gives equivalent representations.
Back to article page
