Poisson及负二项分布对微核试验数据拟合效果
Poisson distributon and negative binomial distribution fitting for data of cytokinesis-block micronucleus test
-
摘要: 目的 探讨微核试验数据(每1000个双核淋巴细胞中具有微核的淋巴细胞数)的统计分布,为微核试验数据的统计分析提供依据.方法 利用东北某大型钢铁公司158名焦炉工和66名非焦炉工微核检测的实例数据,进行矩法估计Poisson分布和二项分布的参数,同时拟合Poisson分布和负二项分布,比较其拟合效果.结果 拟合优度检验显示,焦炉工和非焦炉工微核数据均不服从Poisson分布(P<0.001);焦炉工微核数据服从负二项分布(P=0.545),非焦炉工微核数据虽不服从负二项分布(P<0.001),但该数据似合负二项分布的效果要好于拟合Poisson分布(拟合负二项分布比值3.1<拟合Poisson分布比值20.3).结论 对于具有聚集性趋向(即方差与均数比值较大)的微核数据,相对于拟合Poisson分布,拟合负二项分布是一较好选择.Abstract: Objective To investigate the statistical distribution of the data of cytok inesis-block micronucleus test to provide basis of statistical analysis.Methods The moment estimate parameters of the Poisson distribution,the negative binomial distribution,and Pearson goodness of fit test were used to evaluate the goodness of fit of the two different distributions.Results Neither the data of cytokines is-block mucronuc leustest for coke-oven workers and non-coke-oven workers presen ted Poisson distribution(P<0.001).The data of coke-oven workers showed a negative binomial distribution(P=0.545)and the data of non-coke-oven worke s did not(P<0.001).Conclusion For the data of cytokinesis-block micro-nucleus test with cluste red characte ristics,the negative binomial distribution fitting was a better choice compared to the Poissond is tribution fitting.
下载: