A Test for Heteroscedasticity Based on Ordinary Least Squares Residuals
Authors:
- 1 June 1979
- journal article
- research article
- Published by Taylor & Francis in Journal of the American Statistical Association
- Vol. 74 (366a) , 494-499
Abstract
Two variants of a test for heteroscedasticity based on ordinary least squares residuals are proposed: a bounds test and an “exact” test. Both variants are parametric and make use of tables of the F distribution. Under a variety of circumstances, the power of the bounds test compares favorably with the powers of the tests of Goldfeld and Quandt, Theil, and Harvey and Phillips. The computational simplicity of the bounds test compared with other tests makes it a particularly attractive practical procedure. The power of the “exact” test generally exceeds that of the other tests, but the “exact” test is rather difficult to apply in practice.Keywords
This publication has 14 references indexed in Scilit:
- A Class of Parametric Tests for Heteroscedasticity in Linear Econometric ModelsEconometrica, 1978
- The Robustness of Some Standard Tests for Autocorrelation and Heteroskedasticity when Both Problems Are PresentEconometrica, 1977
- A comparison of the power of some tests for heteroskedasticity in the general linear modelJournal of Econometrics, 1974
- Independent Stepwise Residuals for Testing HomoscedasticityJournal of the American Statistical Association, 1970
- Estimation in a Heteroscedastic Regression ModelJournal of the American Statistical Association, 1968
- Testing Single-Equation Least Squares Regression Models for Autocorrelated DisturbancesEconometrica, 1966
- Some Tests for HomoscedasticityJournal of the American Statistical Association, 1965
- Computing the distribution of quadratic forms in normal variablesBiometrika, 1961
- TESTING FOR SERIAL CORRELATION IN LEAST SQUARES REGRESSION. IIBiometrika, 1951
- A General Expression for the Moments of Certain Symmetrical Functions of Normal SamplesBiometrika, 1933