Breakthrough measuring neighborhoods

Abstract

An empirical strategy is presented for transforming ordinal counts and percentages to interval scale measures by recoding them as ordered categories and estimating Rasch model rating scale parameters. This strategy is demonstrated for a neighborhood construct socioeconomic disadvantage operationally defined by eight characteristics of Chicago neighborhoods (N = 77). Results show surprisingly sound model fit and satisfactory scale invariance between 1980 and 1990 census. A striking finding obscured by traditional methods is many Chicago neighborhoods are four times more disadvantaged than official U.S. poverty threshold. Intramodel construct validation confirms this scale structure is consistent with sociological expectations about property values, income, and race. A general benefit of this approach over conventional categorical socioeconomic indices is neighborhood measurement on a linear scale.