China
Africa
Explore chapters and articles related to this topic
Hypotheses Testing versus Confidence Interval
Published in Shein-Chung Chow, Innovative Statistics in Regulatory Science, 2019
However, the size of TW is max(α1,α2), which is usually smaller than α since α1 + α2 = α. Thus, TW (or CIW) is not recommended if the desired size is α. It should be noted that the size of a test is not the only measure of its successfulness. The TOST procedure is of size α yet biased because the probability of rejection of the null hypothesis when it is false (power) may be lower than α. Similarly, tests TE and TL are biased. Berger and Hsu (1996) proposed a nearly unbiased and uniformly more powerful test than the TOST procedure. Brown et al. (1997) provided an improved test that is unbiased and uniformly more powerful than the TOST procedure. These improved tests are, however, more complicated than the TOST procedure.
The generalized inference on the sign testing problem about the normal variances
Published in Journal of Applied Statistics, 2018
Wei-Ya Wu, Wei-Hwa Wu, Hsin-Neng Hsieh, Meng-Chih Lee
As a whole, the proposed testing procedure is intuitive and is uniformly more powerful test than LRT, Gutmann's test, L-S test and L-C test. Through this study, the performance of GPV-based method and alternative GPV are better than all the other four methods in view of the empirical sizes and the empirical powers. Therefore, based on these study results, the GPV-based method and alternative GPV are suitable choice on the sign testing problem about the normal variances. Consequently, this is the reason why the concept of GPV have been successfully applied in many situations including this study.
A method for combining p-values in meta-analysis by gamma distributions
Published in Journal of Applied Statistics, 2019
We propose a new meta-analysis method by combining p-values from independent studies. There is no uniformly most powerful test available for pooling p-values across individual studies [3,5,7,20]. The new meta-analysis method, p-value combination method in meta-analysis. The proposed test