This paper presents an analysis of the importance of incorporating bot
h individual and neighborhood risk factors into predictive mental heal
th needs assessment models. The analyses are based on data from the me
tropolitan portion of Wave I of the Epidemiologic Catchment Area Surve
y, a large community survey, and the 1980 decennial census. Logistic r
egression procedures were used to estimate the odds ratios for the cla
ssic social area dimensions (neighborhood social rank, life style/urba
nization and race/ethnicity) as well as the standard individual (age,
gender, race/ethnicity, marital status, and high school education). Th
e results suggest that, with the exception of neighborhood social rank
, neighborhoods social area dimension rarely make substantively import
ant contributions to main effects needs assessment models. Thus, the i
nclusion of neighborhood risk factors, with the exception of neighborh
ood social rank, in main effects needs assessment models, while useful
, may not always be necessary, particularly if their inclusion is cost
ly and the substantively important individual risk factors are easily
available. (C) 1998 Elsevier Science Ltd. All rights reserved.