PERMUTATION TESTS FOR LINEAR MODELS.pdfVIP

Aust. N. Z. J. Stat. 43(1), 2001, 75–88 PERMUTATION TESTS FOR LINEAR MODELS MARTI J. ANDERSON 1 AND JOHN ROBINSON2∗ University of Sydney Summary Several approximate permutation tests have been proposed for tests of partial regression coefficients in a linear model based on sample partial correlations. This paper begins with an explanation and notation for an exact test. It then compares the distributions of the test statistics under the various permutation methods proposed, and shows that the partial correlations under permutation are asymptotically jointly normal with means 0 and vari- ances 1. The method of Freedman Lane (1983) is found to have asymptotic correlation 1 with the exact test, and the other methods are found to have smaller correlations with this test. Under local alternatives the critical values of all the approximate permutation tests converge to the same constant, so they all have the same asymptotic power. Simulations demonstrate these theoretical results. Key words: asymptotics; partial correlations; power; resampling. 1. Introduction The first descriptions of permutation (or randomization) tests for linear statistical models can be traced back to the first half of the 20th century in the work of Fisher (1935) and Pitman (1937a, b, c). However, these tests are computationally intensive and the use of them, as opposed to the traditional normal-theory tests, did not receive much attention in the natural and behavioural sciences until the emergence of widely accessible computer power (Edgington, 1995; Manly, 1997). There is general agreement concerning an appropriate method of permutation for exact test