Ecological Validity of Indicator Data as Predictors of Survey Findings

In a series of sh1die.r the validity of preexisting indicator data, based on child and family variables, was tested by means of an extensive sample survey with individual case interviews in two counties in New York State. The counties included one of the highest and one of the lowest in the state, as measured on a specially developed index. On comparable variables a close fit was found between the indicator and the survey data. When the more extensive field data were subjected to multiple regression analyses to determine how broadly predictive the indicators are, however, it was found that prediction was much better for subcounty areas than for counties as a whole. These findings have implications for planning at both national and local levels.