BOOTSTRAP TESTS FOR GENERALIZED LEAST-SQUARES REGRESSION-MODELS

This paper shows that certain statistics in the normal linear regression model may be bootstrapped to obtain test procedures with faster rates of convergence than the usual normal approximation. Specifically, bootstrapping a standardized estimator of the regression coefficients is shown to produce errors with smaller order than n-1, and bootstrapping a variance-adjusted statistic gives errors of smaller order than n-2. An advantage of the bootstrap is that it provides an additional step in asymptotic accuracy without the necessity of explicitly calculating correction terms. © 1990.