交通大学、清华大学 统计学研究所 专题演讲 题　目：Omitting Meaningless Digits in Point Estimates : The Probability Guarantee of Leading-Digit Rules 主讲人：桑慧敏教授(清华大学工业工程与工程管理学系) 时　间：98年9月25日(星期五)上午10:40-11:30 (上午10:20-10:40茶会於交大统计所429室举行) 地　点：交大综合一馆427室 Abstract Motivated by the question of how many point-estimator digits to report in a statistical experiment, we investigate the probabilistic behavior of the digits of the point estimator as a function of the standard error. We lay the foundation of defining meaningful and meaningless digits of the Point. We define meaningful and meaningless digits of the point estimator by comparing to the discrete uniform distribution of a meaningless digit. We show that reporting the point estimate through the leading digit of the standard error implies that the first unreported digit has no more than 0.117 probability of being correct, which is not much better than the one-in-ten chance of a uniformly distributed random digit. When the standard error is estimated, the left-most unreported point-estimate digit contains even less information. 敬 请 公 布 欢 迎 参 加 --

※ 发信站: 批踢踢实业坊(ptt.cc) ◆ From: 140.113.252.20