Robust sample size formulae for testing the regression parameter when binary data are correlated

• Main Authors: Ruei-Ai Wang, 王睿愛
• Other Authors: Tsung-Shan Tsou
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/rbv35n

碩士 === 國立中央大學 === 統計研究所 === 96 === Abstract The aim of this research is to make use of the robust likelihood method proposed by Royall and Tsou (2003) to establish sample size formulae for testing parameter in logistic regression when binary data are correlated. We adopted the binomial distribution as the working model and robustified the naïve likelihood using the robust technique by Royall and Tsou (2003). Robust sample size formulae for testing the regression parameter associated with the cluster-specific covariate is provided. The two versions of the sample size required to achieve a predetermined power, namely, the naïve and the robust formulae, are compared through simulations and analyses of several real data sets.