透過您的圖書館登入
IP:3.133.109.30
  • 學位論文

個數資料之過離散性的強韌推論

Inference for overdispersion in count data without making distributional assumptions

指導教授 : 鄒宗山
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


本文之目的在於利用,當估計模型假設錯誤時,Bartlett的第二等式不正確的性質,來提出一個估計具有過離散性的個數資料之過離散係數的方法。再根據Presnell與Boos(2004)在附錄所提出的方法來估計過離散係數估計量的變異數,並探討估計方法的有效性。 論文中提出一個不需知道正確模型下估計過離散係數之方法,適用於對數迴歸模型或其他合理的迴歸模型。

並列摘要


This thesis provides a method for estimating the over-dispersion count data. And this method adopts the poisson distribution as the working model. The violation of the Bartlett’s second identity is then made use of to give rise to a useful formula for the estimation of the over-dispersion. This new means is applicable for any sensible link function that relates the response probabilities to the variates.

參考文獻


1.Bartlett, M. S. (1953). Approximate confidence intervals. Biometrika. 40, 12-19.
2.Bissell, A. F. (1972). A Negative Binomial Model with Varying Element Sizes. Biometrika. 59, 435-441.
3.Brett, P and Dennis, D. B. (2004). The IOS Test for Model Misspecification. Journal of the American Statistical Association. 99, 216-227.
4.Casella, G. and Berger, R. L. (2002). Statistical Inference. (2nd ed.) Pacific Grove, CA: Thompson Learning.
5.Harville, D. A. (1997). Matrix Algebra From a Statistician’s Perspective. New York:Springer-Verlag.

被引用紀錄


陳冠甫(2012)。利用 Bartlett 第二等式來推論模型假設錯誤下的變異數函數〔碩士論文,國立中央大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0031-1903201314435785

延伸閱讀