为提高塑料弯曲性能检测实验室的能力水平，山东非金属材料研究所开展了“塑料弯曲性能的检测”能力验证计划。利用经典法、稳健统计算法A和稳健统计四分位距法3种不同统计方法对塑料弯曲性能检测能力验证统计结果进行比较，评价了塑料弯曲性能检测实验室的整体水平，对可能导致结果偏离的原因进行了分析。结果表明：在能力验证数据统计中，为避免极端数据对统计结果的影响，应尽量采用稳健统计的方法；约有12%的实验室的塑料弯曲能力验证结果存在问题；检测结果的偏差可能跟设备、环境和人员有关。

In order to improve the proficiency level of the laboratories for plastics flexural property testing, Shandong nonmetallic materials institute firstly organized "test for plastics flexural property" proficiency testing program. The statistical results of the proficiency testing for plastics flexural property were analyzed respectively by classic statistical method, iteration robust method and quartile robust method. Also, the overall level of the plastics flexural property was evaluated and the possible causes of deviation were pointed out. The results show that in the data statistics of laboratory proficiency testing, the robust statistical method was the best to avoid the influence of extreme data. Test results for plastics flexural property of about 12% laboratories had problems and the results deviation may be related to the equipment, the environment and the personnel.