Tests of H0: ρ1 = 0 for Autocorrelation Estimators rF1 and rF2

Abstract
Two new tests of the hypothesis of no lag-1 autocorrelation in a time-series process, i.e., H0: ρ1 = 0, are presented for two reduced-bias autocorrelation estimators that were introduced by Huitema and McKean in 1994. The performance of the new tests, ZF1 and tF2, was evaluated using Monte Carlo methods. Both new tests were superior to the conventional Bartlett asymptotic test in the case of small and intermediate N; tF2 outperforms ZF1 substantially at small and intermediate values of N. The true probability of Type I error associated with test tF2 is exceedingly close to the nominal value for all values of α (.01, .05, and .10) and N (6—500) investigated. The tF2 test is also more powerful against positive values of ρ1 than are the Bartlett and ZF1 tests.

This publication has 4 references indexed in Scilit: