有序列联表线性趋势X^2检验

Linear Tendency Chi-square Test for Ordered Contingency Table

为避免常规x^2检验对有序分类表数据分析的误用所导致的信息丢失,提出线性趋势x^2检验。该方法首先对有序分类变量的各水平按其边际频数分布配置标准化秩得分,然后按线性秩统计量的分布特征构造线性趋势x^2统计量。并用2个临床实例说明其过程。结果显示该检验比常规x^2检验的灵敏度高,而且具有简单、易行的优点。

In order to avoid information loss caused by misapplication of ordinary Chi-square test in data analysis for the ordered contingency table, a method of linear tendency Chi-square test was proposed. The first step of this method was to allocate standard rank scores to the levels of ordered class variables according to the marginal frequency distribution of the ordered variables, then to form the linear tendency Chi-square statistic referring to the characteristics of linear rank statistic distribution. To demostrate the test process, two clinical examples were given. The results showed that linear tendency Chi-square test was more sensitive and more practicable than ordinary Chi-square test.