【摘要】 为了选取适合于对细胞毒性试验数据的异常值进行检验的方法、提高检测数据的可靠性,对常用的5种异常值检验方法的计算方法及判定原则进行了比较研究,并运用这5种异常值检验方法对同一组细胞毒性试验数据进行了分析。结果发现,拉依达准则计算简便,适合样本量较大时对异常值进行初筛;样本量较小时适合选取狄克逊准则、罗马诺夫斯基准则对数据中的异常值进行判定;从计算方法来看,肖维勒准则、格拉布斯准则、罗马诺夫斯基准则相似;对数据中异常值进行检出时,罗马诺夫斯基准则最为严格,肖维勒准则次之;狄克逊准则较适合于对多个异常值进行逐个检出。细胞毒性试验多孔平行数据属于小样本量数据,待检出的异常值不仅限于1个,选用狄克逊准则对细胞毒性试验多孔平行数据进行检验较为适宜。更多还原

【Abstract】 In order to improve the reliability of the cytotoxicity test′s data,select a suitable abnormal value detection method,comparative study on five kinds of abnormal value statistical methods is carried out.And a group data from a cytotoxicity test is analyzed using these statistical methods.The results show that Pauta test is suitable for screening abnormal value in large sample size,Dxion test and t-test are applicable to detect abnormal value when sample size is small,the calculation models of Chauvenet test,Grubbs test and t-test are similar.t-Test is the most stringent criterion method.Dixon test is more suitable for repeated use.In view of the data of cytotoxicity test usually belongs to small sample size and abnormal value often more than one,Dxion test as the abnormal value detection method is more appropriate.更多还原