Toward an empirically derived typology of obese persons: derivation in a nonclinical sample.

The purpose of this research was to empirically derive a typology of obese persons and validate a typology derived in earlier research (Allison & Heshka [1991] International Journal of Obesity). Biological, behavioral, and psychological variables were assessed through survey among 719 (641 females, 78 males) obese members of The National Association to Advance Fat Acceptance (NAAFA). All variables were subjected to principal components analyses which extracted 12 biological and 12 psychological components. A two-cluster solution from a k-means clustering on biological components was replicated via Ward's method. Agreement between the solutions was significant (Phi = .33, Kappa = .19, p < .05). The solution was validated through entering psychological component scores into discriminant analysis. One significant function (p < .001) substantially separated the clusters. A component measuring early onset/familial history powerfully discriminated between the clusters. Early onset obese were more obese, more active, and restricted caloric intake to a greater degree. Late onset obese were more likely to be "settled down," engage in substance abuse, eat at night, and have diabetes (p < .05). Results were substantially consistent with those of prior research.