Multiple Measurements of U.S. Income Inequality
利用1980年美国州级人口普查数据，将基尼系数、泰尔指数等八种不平等测度分别与十九个社会经济变量进行回归，发现教育程度标准差是收入不平等的最强预测因子，而以往研究中占主导的家庭平均收入仅居次位。
Using 1980 state Census data, the Gini ratio, Theil index, coefficient of variation, Atkinson measure (four values), and Nelson index as separate measurements of income inequality are regressed with nineteen SES variables traditionally found significant with Gini in past research. The standard deviation of educational attainment is dominant among all independent variables as the strongest predictor of inequality. Mean family income, dominant in prior research, is a distant second in predictiveness. More diversity than commonality is evident among the nineteen SES variables as they predict income inequality measured by the eight different techniques. Copyright 1988 by MIT Press.