Analysis of data from 13 planned new communities, 13 control communities, and two retirement communities indicates that a highly effective model for predicting service preferences can be constructed using per capita income, age, sex and community type. This model limits itself to data available to all planned organizations and suggests that, although the elderly are a diverse population, there may be basic needs. Furthermore, the findings suggest a reexamination of the traditional "problem" focus of services. (Author)