對多重假設檢定，傳統上係在控制整體誤差率之下來選取顯著性因子。惟傳統的多重假設檢定，如 Bonferroni 法，雖然程序簡易，但隨著待選因子個數的上升，不易選取到具顯著性的因子，且檢定力低。Benjamini and Hochberg (1995) 提出控制偽陽率 (false positive) 以逐次 P 值法來選取顯著因子，除了改善傳統多重假設檢定的缺點，其執行程序亦相當簡單。惟Benjamini and Hochberg (1995) 在其選取顯著性因子的過程中未考慮潛在的次要因子，此為其中美中不足之處。為了更貼近真實情況，本研究將次要因子加入模型中，用以修正逐次 P 值法。模擬結果證明修正後的逐次 P 值法比其他多重假設檢定方法有較高的檢定力。我們以成大醫院的二組資料說明所建議方法之應用。

　　Peoele usually need to use the technique of multiple hypotheses testing to search for significant factors under the control of familywise error rate.Traditionally, the procedure of multiple hypotheses testing, like Bonferroni procedure,is very simple and easy to perform, but it tends to select fewer statistically significant facotrs and has smaller power.Benjamini and Hochberg (1995) proposed the sequential p-value approach by controlling the false discovery rate to select statistically significant ones.
This approach of Benjamini and Hochberg (1995) is also simple, and has larger power as compare to traditional method.In the thesis, we consider the sequential p-value approach with covariates adjusted.Through simulation we found that the performances of the adjusted sequential p-value approach is superior to sequential p-value method and Bonferroni approach. We use two real expamles to illustrate the application of the suggested method.

1. Benjamini, Y. and Hochberg, Y. (1995) Controlling the False Dis-
covery Rate: A Practical and Powerful Approach to Multiple
Testing. J. Roy. Statist. Soc. Ser. B 57, 289-300.
2. Benjamini, Y. and Yekutieli, D. (2001) The Control of the False
Discovery Rate in the Multiple Testing Under Dependency. Ann.
Statist., 29, 1165-1188.
3. Hommel, G. (1988) A Stagewise Rejective Multiple Test Pro-
cedure Based on a Modi ed Bonferroni Test. Biometrika, 75,
383-386.
4. John D. Storey, (2003) The Postive False Discovery Rate: A
Bayesian Interpretation and the q-value. Ann. Statist., 0, 1-23.
5. Simes, R. J. (1986) An Improved Bonferroni for Multiple Tests
of Signi cance. Biometrika, 73, 751-754.