Abstract
Under a robust Bayesian framework, we study the problem of power estimation in binomial hypothesis testing problems. For testing the binomial mean, the observed powers are maximum likelihood estimates of power functions for one-sided and two-sided problems. With respect to these functions, we show that the observed powers are too large when the data is significant for some beta families of priors. A practical implication about the problem of sample size calculation is discussed.
Citation
Chih-Chien Tsai. C. Andy Tsao. "POST-DATA ESTIMATION OF BINOMIAL TESTING POWER." Taiwanese J. Math. 13 (6B) 1867 - 1883, 2009. https://doi.org/10.11650/twjm/1500405646
Information