A test for heteroscedasticity in the form of the size of random error increasing proportionally to changes in one or more exogenous variables. The test is performed by regressing the absolute values of ordinary least squares residuals from the main regression equation on the variables in question. Under the null hypothesis of homoscedasticity the test statistic, NR², has asymptotically chi-square distribution with h degrees of freedom, where h is the number of variables in question, N is the sample size, and R² is the coefficient of determination from the test regression. The Glejser test is only valid when the random error is symmetrically distributed; a number of modifications exist for skewed errors.