Grenier B
Pathol Biol (Paris). 1985 Nov;33(9):934-8.
The respective effectiveness of two qualitative tests is usually compared by their results when applied to a same specimen or subject. The results reported in a fourfold table show coincident numbers of double positive and double negative results. That straight concordance is fallacious, including the chance coincidences. Test Kappa measures the true agreement, without the chance coincidences. Kappa value may be compared to a maximum value of agreement depending on the marginal results of the fourfold table. As a matter of fact, the test Kappa is a non parametric correlation function comparable to the non parametric correlation coefficient Phi computed from the value of the chi square test.
通常通过将两种定性测试应用于同一标本或受试者时的结果来比较它们各自的有效性。四格表中报告的结果显示了双阳性和双阴性结果的一致数量。这种直接一致性是错误的，包括偶然的巧合。检验kappa测量的是真实一致性，不包括偶然巧合。kappa值可以与取决于四格表边缘结果的最大一致值进行比较。事实上，检验kappa是一种非参数相关函数，类似于根据卡方检验值计算出的非参数相关系数Phi。
