本篇論文回顧Zhu et al.(2013)的逐次檢定方法檢測兩類別變數之間的關聯程度,並使用他們所提出的指標tau來衡量兩類別變數的關聯強度。本研究進一步利用Zhu et al.(2013)的發現建構一些新的指標以修正原有的指標,並且針對這些指標與Cramer''s V係數、列聯係數,進行有限樣本的模擬研究,以比較這些指標的表現。在模擬研究當中,第一部分驗證了Zhu et al.(2013)的逐次檢定統計量的抽樣分配會近似卡方分配。第二部分透過大量模擬發現以新指標來衡量兩類別變數的關聯程度表現得最好,其中指標tau4為指標tau乘上((L+1)/L)再開根號,L為行數或列數取較小的再減1。為了示範新指標tau4的執行,針對台北榮民總醫院所提供的燒燙傷研究的資料,本文以新指標tau4評估對於感興趣的兩類別變數間的關聯程度。 This paper reviews the sequential test method proposed by Zhu et al.(2013) to test the degree of association of two categorical variables. Zhu et al.(2013) also proposed an index tau to measure the strength of association of two categorical variables. In this study, some another similar indexes are constructed and their performance are then compared with the Cramer''s V and contingency coefficient. According to the result of the extensive simulation studies, the index tau4 has the best performance, where tau4 is the square root of the index tau multiplied by ((L+1)/L) and L+1 is the number of rows or the number of columns, whichever is less. Another simulation studies confirm that the sampling distribution of the sequential test statistic proposed by Zhu et al.(2013) approximates the chi-square distribution. An acute burn injury research from the Veterans General Hospital database has been used to illustrate the proposed method as well.