Bootstrapping the Measures of Association

• Main Authors: Lai, Yuan-Hsin, 賴永興
• Other Authors: Lee Tai-Ming
• Format: Others
• Language: zh-TW
• Published: 1996
• Online Access: http://ndltd.ncl.edu.tw/handle/55059147566421828416

碩士 === 輔仁大學 === 數學系 === 84 === Measures of association is a statistical concept which has drawn remarkable attention since Pearson (1904,1913) and Yule (1900, 1912). A great amount of measures have been proposed and modified, for example, odds ratio, Matel-Haenszel log-odds- ratio, relative risk, concentration coefficient, uncertainly coefficient, gamma, tau-b, somers' d, kappa ..etc. These measures were proposed under the consideration of better interpretation for both nominal and ordinal variables with different sampling schemes. However, all these measures are just a number, and it's difficult to access its bias, variability and its p-value. Although the Delta method gives the asymptotic variance of these measures, the accuracy has never been justified, and the confidence intervals based on the asymptotic variance are questionable, especially when the sample size is small. I proposed the Sampling-Based approach to solve all these problems with both of classical and Bayesian's points of view.