Intermediate efficiency of some max-type statistics |
| |
| Affiliation: | 1. Lappeenranta University of Technology, LUT Chemistry, Laboratory of Green Chemistry, Sammonkatu 12, FI-50130 Mikkeli, Finland;2. Colleges of Chemistry & Chemical Engineering, Hunan Normal University, China;3. University of Jyväskylä, Department of Chemistry, Laboratories of Inorganic and Analytical Chemistry, P.O. Box 35, FI-40014 Jyväskylä, Finland;4. Department of Environmental Science, University of Eastern Finland, P.O. Box 1627, FI-70211 Kuopio, Finland;1. Department of Economics, University of Perugia, Via A. Pascoli, 20, 06123 Perugia, Italy;2. Department of Economics and Social Sciences, Università Politecnica delle Marche and MoFiR, P.le Martelli, 8, 60121 Ancona, Italy |
| |
| Abstract: | Asymptotic comparison of two recent tests for constant regression via intermediate efficiency approach is developed here. The following work shows that the constructions proposed by Eubank and Hart (Ann. Statist. 20 (1992) 1412) and Fan and Huang (J. Amer. Statist. Assoc. 96 (2001) 640) are efficient for one type of deviation only, which is the same for both tests. It is also inferred that, for other directions, the second solution outperforms the first one.The approach elaborated in this paper also allows one to calculate the intermediate efficiency of the classic Kolmogorov–Smirnov test, thus supplementing several earlier developments on this statistic. |
| |
| Keywords: | |
| 本文献已被 ScienceDirect 等数据库收录！ |
|
