In this paper we focused on the two-sample scale problem. We consider the following tests: the F test, the modified Fligner-Killeen (1976) test, the W50 test, the M50 (Pan, 1999) test, the L50 (Pan, 1999) test, the O’Brien R (1979) test, and two tests of Shoemaker (1995 and 1999). Bootstrap and permutation versions of W50, L50 and R tests have been also considered. Type-one error rate and power of the tests are investigated. The computationally simplest robust test is the W50. Higher power, while preserving robustness, is achieved by considering the permutation R test (recommended for normal- and light-tailed distributions) and the bootstrap L50 test (recommended for heavy-tailed and asymmetric distributions).
A Comparative Study of Tests for Scale Equality with Application in Industrial Statistics
MAROZZI, Marco
2008
Abstract
In this paper we focused on the two-sample scale problem. We consider the following tests: the F test, the modified Fligner-Killeen (1976) test, the W50 test, the M50 (Pan, 1999) test, the L50 (Pan, 1999) test, the O’Brien R (1979) test, and two tests of Shoemaker (1995 and 1999). Bootstrap and permutation versions of W50, L50 and R tests have been also considered. Type-one error rate and power of the tests are investigated. The computationally simplest robust test is the W50. Higher power, while preserving robustness, is achieved by considering the permutation R test (recommended for normal- and light-tailed distributions) and the bootstrap L50 test (recommended for heavy-tailed and asymmetric distributions).I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
