![]() |
Dr. Tsung-Shan Tsou 鄒 宗 山 |
研究方向 本人主要之研究領域包含二方面:1.統計原理,2.醫學統計。
統計原理之研究目的為探討現在廣為使用的統計推論工具
( 如決策理論、貝氏統計 ) 的正確性及尋求正確的統計推論方法。舉例來說一般在做統計檢定時所用的Neyman-Pearson檢定法,可以用簡單的例子來說明其謬誤之處。事實上Neyman-Pearson檢定法所解決的是決策理論而非統計推論的問題。同樣的因為貝氏統計必須有先驗資料,因此亦非正確的推論工具。而與此研究有關的著作包含
Liang and Tsou (1992), Tsou and Royall (1995) 和
Tsou (1996)。同時目前尚有二篇有關論文已投稿發表。
在有關醫學統計方面,曾參與美國行為科學家合作有關於工作壓力對心臟血管疾病間相關性之研究計劃。主要探討的因素包含心理壓力、體力負荷、工作環境等可能危險因子對心血管疾病的影響,成果則可參考Hall,
Johnson and Tsou (1993)。
著重於有關統計在牙醫學應用之研究。例如有關牙科材料學方面之問題及有關牙周病與生活習慣中危險因子間關係的探討。已發表之著作者有
Wang, Lu, Shiau and Tsou (1995).而有關牙周病方面之研究。已完成的部份包含抽煙對牙周病的影響。其中應用了目前廣為使用的廣義估計方程式來分析具“群”結構的牙齒資料。目前已有二篇相關論文分爺寄出發表,並都在修正當中。而目前正在進行的則是有關檳榔與牙周病關係的研究。 本人近年來致力於強韌統計推論方法的研究上。研究之方向是探討如何對一些實作概似(working likelihood)函數做修正,而修正後的概似函數,不論資料真正的分配為何,在大樣本的情形下,皆為正確的概似函數。如,在廣義線性模型的架構之下,針對迴歸係數的推論方面,提出了強韌迴歸(robust regression)方法。針對珈瑪(gamma)及常態(normal)迴歸模型分別提出了強韌珈瑪迴歸與強韌常態迴歸模型。而針對個數資料(count data)也提出了強韌的迴歸模型。相關的研究還針對二個母體的變異數比較問題上,提出了較目前常用的統計方法,更為有效(efficient)的有母數(parametric)強韌方法。
|
Foundations
of Statistical and Biostatistics Foundations of statistics emphasizes
the difference between three main areas of Statistics, namely, Statistical
Inference, Statistical Decision and Bayesian Statistics. Statistical
Inference answers question such as “What the data say?”, Statistical
Decision answers question like “What should I do, now that I see the
data?”, and Bayesian Statistics answers question like “What Do I
believe, now that I see the data?”.
Different questions need different statistical tools for answers.
Unfortunately, the distinction is not well understood by most.
Therefore, many study results are misinterpreted, and policies are
misguided. For instance,
simple examples can show that the Neyman-Pearson Paradigm is wrong for
making a statistical inference. It
is in fact a statistical decision tool.
However, the N-P paradigm is used for inferential purpose by every
statistical practitioner. Related
publications in this area can be found in the references. Biostatistics focuses on the application of statistics to biomedical data. Disease-Exposure relation is the main focus, for example, the impact of life risk factors on the development of cancers. Recently I have spent time on the association between periodontal disease and risk factors, such as age, sex and smoking. More risk factors will be considered in the future, such as Betdnut chewing. Also the idea of surrogates to account for covariates lacking measurement accuracy will be a further research topic. |
Selected
Recent Publications: 1.
Liang, K.Y. and Tsou, T.S.
(1992) Empirical Bayes and Conditional Inference with Many Nuisance
Parameters. Biometrika 79, pp 261-270. 2.
Hall, E.M., Johnson, J.V. and Tsou, T.S. (1993)
Women, occupation, and Risk of Cardiovascular Morbidity and Mortality.
Occupational Medicine 8, pp 709-719. 3.
Lo, S.K, Li, I.T., Tsou, T.S. and
See L. (1995) Non-significant in Univariate but Significant in
Multivariate Analysis: A Discussion with Examples. The Chang Gung
Medical Journal 18, No. 2. 4.
Wang, H-Y., Lu, Y-C., Shiau, Y-Y. and Tsou, D.
(1995)
Vertical
Distortion in Distal Extension Ridges and Palatal Area of Casts Made by
Different Techniques. Jour. of Prosthetic Dentistry, 75, pp 302-308. 5.
Tsou,
T.S and Royall, R.M. (1995) Robust Likelihoods. J.
Amer. Stat. Ass. 90, pp 316-320. 6. Tsou,
T.S (1996) The Quantification of
Information Contained in Margins in a 2 by 2 Table. J. Chinese Stat.
Ass., 34, pp 347-355. 7. Tsou, T-S (2000) Exchangeable cluster binary data correlation coefficient estimation with generalized estimating equations. Stat. & Prob. Letters, 50, pp 179-186. 8. Tsou,T-S (2000) A brief discussion on differences among Bayesian analysis, Decision theory and Statistical inference. Formosan J. of Medicine 4, pp 446-450. 9. See, Lai-Chu, Tsou, T.S and Lin, Lih-Hwa (2000) Program management and statistical comparison of two dependent proportions in multiple choice questions. Formosan J. of Medicine 4, pp 75-79. 10. Sing Kai Lo & Tsou, T-S (2002). On the Use of General Linear Models in Medical Research. Changhua Jour. of Medicine, 7, 193-200. 11. Chang, M-C, Chang, Y-C, Chiou, J-F, Tsou, T-S and Lin, C-C (2002) Overcoming Patient-related barriers to cancer pain management for home care patients. Cancer Nursing, 25, 470-476. 12. Lin, C-C, Tasi, H-F, Chiou, J-F, Lai, Y-H, Kao, C-C and Tsou, T-S (2003) Changes in levels of hope after diagnostic disclosure among Taiwanese patients with cancer. Cancer Nursing, 26,155-160. 13. Hsu, T-H, Lu, M-S, Tsou, T-S and Lin, C-C. (2003) The relationship of pain, uncertainly, and hope in Taiwanese lung cancer patients. J. Pain Symptom Manage, 26, 835-842. 14. Royall, R.M. and Tsou, T-S (2003). Interpreting statistical evidence using imperfect models: Robust adjusted likelihood functions. JRSS-B, 65, 391-404. 15. Tsou, T-S (2003). Comparing two population means and variances – a parametric robust way. Comm. Stat.- Theor, Meth,32, 10, 2013-2019. 16. J.T. Horng, H.D. Huang, K.Y. Tseng, T-S. Tsou, and B.J. Liu.(2003). Mining Correlations of Tissue Gene Expression from Digital Expression Profiles of Human Tissues. Special Issue on Bioinformatics, Journal of Information Science and Engineering, Vol. 19, No. 6, 909-921 17. H.D. Huang, H.L. Chang, T-S. Tsou, B.J. Liu, and J.T. Horng. (2003). A Data Mining Method to Predict Transcriptional Regulatory Sites Based on Differentially Expressed Genes in Human Genome. Journal of Information Science and Engineering, Vol. 19, No. 6, 923-942 18. H.D. Huang, J.T. Horng, C. H. Chang, T-S. Tsou, J.Y. Hong and B.J. Liu. (2004) A computational approach to discover differential cooperation of regulatory sites in functionally related genes in Yeast genome. Journal of Information Science and Engineering, Vol. 20, No. 6, 1139-1157. 19. Tsou, T-S and K-F Cheng (2004) Parametric robust regression analysis of contaminated data. Comm. Stat.- Theor, Meth,33, 1887-1898. 20. Tsou, T-S (2005). Robust inferences for the correlation coefficient – a parametric robust way. (Comm. Stat.- Theor, Meth, 34, 147-162.) 21. Tsou, T-S and Chien, L-C (2005). Parametric robust tests for multiple regression parameters under generalized linear models (Advances and Applications in Statistics, 5, 51-86.) 22. Tsou, T-S (2004) Robust Poisson regression. (To appear in Journal of Statistical Planning and Inference) 23. Tsou, T-S (2001) Measuring strength of statistical evidence using imperfect models. 2001第三屆海峽兩岸統計學術研討會 24. Tsou, T-S. (2002) Robust Poisson regression. 2002中華機率統計學術研討會 25. Tsou, T-S and Liu, Ming-In (2002). Robust inference for large data sets. 2002台北 資料採礦研討會 26.Chien, L-C and Tsou, T-S (2003). Regression Diagnostic under Model Misspecification. (Submitted) 27.Tsou, T-S (2004). Parametric robust inferences for regression parameters under generalized linear models. (Submitted) 28. Tsou, T-S (2004). Robust inverse Gaussian models for regression parameters under generalized linear models. (Submitted) 29. Tsou, T-S (2004). Parametric robust test for several variances with unknown underlying distributions. (Submitted) 30. Tsou, T-S (2004). A parametric robust approach to the determination of sample size for testing regression coefficients in generalized linear models. (Submitted)
|