Evaluation of the tests
In this chapter we compare powers of the tests against positive autocorrelation and against heterovariance, as described in Section 2.5, , in order to answer the question of which disturbance estimator should be applied to which test. The six disturbance estimators, identified by û, hyp, sel, mod. BLUS, BLUS,and z, are described below. The powers are calculated for ℋA : ρ = 0.8 in test(Q), for ℋA : η = 0.9 in test(S), and for ℋA : γ = 0.83 in test(V),at significance levels γ = 0.05 and α = 0.10 (we also carried out computations for JℋA : ρ = 0.3, or η = 0.75, or γ = 0.5, and α = 0.025, the results of which are not presented in this study: we found the same relative performances of the disturbance estimators, so that the results of the power comparisons are not restricted to the rather arbitrary choice of ℋ A and α). The X-matrices used in this chapter are also described below. The section is concluded by a comparison between the test statistics using mod. BLUS and those using BLUS.
KeywordsNull Distribution Significance Point Selection Device Beta Approximation Mixed Test
Unable to display preview. Download preview PDF.