Open Access. Powered by Scholars. Published by Universities.®
Physical Sciences and Mathematics Commons™
Open Access. Powered by Scholars. Published by Universities.®
Articles 1 - 2 of 2
Full-Text Articles in Physical Sciences and Mathematics
A Single, Powerful, Nonparametric Statistic For Continuous-Data Telecommunications Parity Testing, J. D. Opdyke
A Single, Powerful, Nonparametric Statistic For Continuous-Data Telecommunications Parity Testing, J. D. Opdyke
Journal of Modern Applied Statistical Methods
Since the enactment of the Telecommunications Act of 1996, extensive expert testimony has justified use of the modified t statistic (Brownie et al., 1990) for performing two-sample hypothesis tests comparing Bell companies’ CLEC and ILEC performance measurement data (known as parity testing). However, Opdyke (Telecommunications Policy, 2004) demonstrated this statistic to be potentially manipulable and to have literally zero power to detect inferior CLEC service provision under a wide range of relevant data conditions. This article develops a single, nonparametric statistic that is easily implemented (i.e., not computationally intensive) and typically provides dramatic power gains over the modified t while …
Maximum Tests Are Adaptive Permutation Tests, Markus Neuhäeuser, Ludwig A. Hothorn
Maximum Tests Are Adaptive Permutation Tests, Markus Neuhäeuser, Ludwig A. Hothorn
Journal of Modern Applied Statistical Methods
In some areas, e.g., statistical genetics, it is common to apply a maximum test, where the maximum of several competing test statistics is used as a new statistic, and the permutation distribution of the maximum is used for inference. Here, it is shown that maximum tests are special cases of adaptive permutation tests. The 30-year old idea of adaptive statistical tests is more flexible than previously thought when permutation tests are used, and the selector statistic is calculated for every permutation. Because the independence between the selector and the test statistics is no longer needed, the test statistics themselves can …