臺灣博碩士論文加值系統

這篇碩士論文所研究的主題為針對樣本數大小的不同如何進行顯著水準及P-值校正。研究的開始，我們主要回顧了Royall (1986)的一篇文獻，文中主要提及Lindley (1984)和Peto (1976)對於樣本數大小的不同如何影響P-值在看法上的不同，我們進而從中定義了兩種證據測度以作為之後我們進行顯著水準及P-值校正時的基礎。特別我們提出利用保持Peto證據測度的一致性上進行顯著水準的校正。另外我們也回顧了Berry和Viele (1987)的一篇文章並和我們的結果作比較。

The problem of calibrating significance levels and p-values for sample size is studied. In light of different principles of adjustment suggested in Royall (1986), Lindley (1984) and Peto (1976), two measures of evidence are explicitly defined. These measures are in turn employed as bases to calibrate significance levels and p-values. Particularly, we propose a method of calibrating significane level which maintains the consistency of Peto's measure of evidence.A different method proposed by Berry and Viele (1987) is reviewed and contrasted.