透過您的圖書館登入
IP:18.224.33.107
  • 學位論文

兩對數常態分配平均數差與平均數比例之區間估計

The interval estimates for the difference and the ratio of two lognormal means

指導教授 : 蘇秀媛

摘要


對數常態分配(lognormal distribution)為一種單尾機率分配,並且具有取對數後即呈現常態分配之特性。在醫療研究當中,得到的資料往往呈現對數常態分配,由於平均數能夠直接反映出整個群體的變化,所以當研究過程需要比較兩獨立樣本從中獲得資訊時,研究人員對於估計兩樣本母體平均數的差或比例較感興趣,例如:在藥物試驗中,測試新藥對於兩個不同試驗單位所產生之效應,又如醫療成本方面,診斷是否罹患憂鬱症之診斷支出。因此,研究者利用區間估計法分別計算出針對對數常態分配的平均數差以及平均數比例建立100(1-α)%之信賴區間。 本論文分別使用拔靴法(bootstrap method)、廣義樞紐量法(generalized pivotal quantity method)及無母數法(nonparametric method)三種方法來估計兩對數常態母體平均數差及平均數比例之信賴區間。結果顯示在計算平均數差之信賴區間時,無母數法在樣本數小的時候,有較準確的覆蓋機率且信賴區間長度較短,且計算上則較為精簡快速。廣義樞紐量法在樣本數大的時候,所呈現出之覆蓋機率較為準確且具有最短之信賴區間長度。計算平均數比例之信賴區間時,廣義樞紐量法具有較準確之覆蓋機率。當樣本數小的時候,拔靴法呈現出與廣義樞紐量法一樣的結果,且有最短之信賴區間長度。同時,因無母數兩樣本平均數比例之信賴區間目前無相關文獻,因此本論文中我們提出二種無母數平均數比例之信賴區間估計法,此方法可快速計算估算值。

並列摘要


The log-normal distribution is a one-tailed probability distribution of any variable, logarithm is normally distributed. Health research often gives rise to data that are positive and highly skewed. In many situations, the data follow lognormal distributions. This thesis investigates the interval estimates for the difference and the ratio of two lognormal means by using there different methods: bootstrap, generalized pivotal quantity and nonparametric method. The result showed that the generalized pivotal quantity method has the good coverage probability and the shortest interval length in computation confidence interval. There is no research in literature about the confidence interval for the ratio of two lognormal means in nonparametric method. Therefore, in this thesis, we propose two confidence interval estimations for the ratio of means in nonparametric case. The computations of the two estimations are much faster then those obtained from other methods.

參考文獻


1.Barrett J., Goldsmith L.(1976). When is n sufficiently large? American Statistician ; 30: 67-70.
2.Chen Y-H, Zhou X-H.(2006). Interval estimates for the ratio and difference of two lognormal means. Statistics in Medicine; 25:4099-4113.
3.Hall P.(1992a). The Bootstrap and Edgeworth Expansion . New York: Springer.
6.Krishnamoorthy K., Mathew T.(2006). Generalized p-values and confidence intervals: A novel approach for analyzing lognormally distributed exposure data. Journal of Occupational and Environmental Hygiene; 3:642-650.
7.Krishnamoorthy K.,Mathew T.(2003).Inferences on the means of lognormal distributions using generalized p-values and generalized confidence intervals. Journal of Statistical Planning and Inference; 115:103-120.

延伸閱讀