配对四格表资料的SPSS软件实现及相关结果解读  

The SPSS software implementation of paired 2×2 tables and interpretation of related results

作　　者：金英良 赵华硕 黄水平 孙桂香 王可 朴雪 曾平 

机构地区：[1]徐州医科大学公共卫生学院流行病与卫生统计教研室,江苏徐州221004 [2]徐州医科大学医学信息学院医学智能教研室,江苏徐州221004

出　　处：《预防医学论坛》2017年第3期171-173,179,共4页Preventive Medicine Tribune

基　　金：江苏省教育厅高校哲学社会科学研究基金资助项目(2016SJD910001);徐州市科技计划项目(KC16SW169)

摘　　要：目的通过验证SPSS软件分析实现配对四格表资料的统计分析两种方法的一致性和相关主要结果的解读。方法SPSS软件提供两种方法可进行配对四格表资料的统计分析,一是列联表资料分析方法,定义行列变量后点击主菜单的Statistics按钮,选择McNemar选项;二是两相关样本的非参数检验方法,定义配对变量后选择主菜单的McNemar选项。结果 b+c>25时,列联表分析法和非参数检验法获得结果几乎相同,分别采用二项分布累积计概率和四格表资料校正公式法,P值分别为0.652和0.651;,当b+c≤25时,列联表分析法和非参数检验法获得结果相同,均采用二项分布累积概率法,P值均为0.057。采用列联表资料分析方法,结果是采用二项分布原理计算的双侧尾部累积概率值;采用非参数检验方法,当b+c>25时,结果采用配对四格表资料的校正公式计算的统计量和相应的概率P值,当b+c≤25时,结果采用基于二项分布原理计算的双侧尾部累积概率值。结论配对四格表资料采用列联表资料分析法与非参数检验法的计算结果一致。SPSS软件的配对四格表资料的分析结果主要是基于二项分布原理进行的精确检验和采用配对四格表资料的校正公式计算的结果。Objective To validaterealize the analysisconsistency of paired 2 × 2 tables through SPSS software and two methodsexplain the main results. Methods To analyze paired 2 × 2 tables, two methods were used by SPSS soft ware. The first was erosstabs,clicked the statistics button on the main menu and selected the McNemar option;the second was 2 related samples test of nonparametric test and selected the McNemar option from the main menu. Results Based on example 1 and contingency table analysiscrosstabs, the two-sidedbinomial distribution principle of cumulative probability result showed that there was no significance （ P =0. 652）area of binomial distribution was calculated ; the second method was based on nonparametric test. When b＋c was greater than 25 ,the corrected chi square test result also showed that there was no significance （ P =0. 651）statistic and related P value of paired 2 × 2 tables were calculated through the nonparametric test. When b＋c was less than or equal to 25 from example 2,the results of two-sided cumulative probability value was calculated by binomial distributionshowed that contingency table analysis and nonparametric test was the same based on binomial distribution cumulative probability method and P value was 0. 057. Conclusion The analysis results of paired 2 × 2 tables through SPSS software are consistent based on nonparametric test and contingency table analysisbinomial distribution with exact test and the corrected chi-square of paired 2 × 2 tables.

关 键 词：配对设计 SPSS软件 二项分布 MCNEMAR检验 

分 类 号：C81[社会学—统计学]