GSE Graduate Student Research
Document Type
Working Paper
Date of this Version
4-2018
Abstract
The Gini coefficient, an indicator that is often used to measure the inequality in the distribution of income within countries, is meaningless when used on standardized test scores. This is because the value of the Gini coefficient depends on the scale’s mean and standard deviation which are arbitrarily selected by the test developers. Keeping the standard deviation of the scale constant, increasing the mean will decrease the Gini coefficient, while keeping the mean of the scale constant, increasing the standard deviation will increase the Gini coefficient. In addition, when Gini coefficients are estimated with scores on two different scales, not only the values of the Gini coefficients but also the country rankings of the Gini coefficients will change. Therefore, for standardized test scores, the value of the Gini coefficient is meaningless, as is comparing the size of the Gini coefficients estimated from different countries. More generally, all relative measures of dispersion, including the Gini coefficient, are meaningless for interval scales (i.e., a scale in which the distance between any two consecutive points are equal, but the scale does not have an absolute zero), such as standardized test scores.
Date Posted: 18 September 2018
Comments
Working paper presented at the Annual Conference of the American Educational Research Association (AERA), New York, NY, April 16, 2018.