where
xijℓ	=	the individual's score on the jth variable or attribute predicted by the model,
gik	=	an individual's weight on the Kth pure type (or group),
λkjℓ	=	a dimension's score on the jth variable or attribute,
K	=	number of dimensions, and
j	=	number of variables (and ℓ is the number of different types of responses to the variable).