肖维勒准则和格拉布斯准则的比较

Comparison Between Chauvenet's Criterion and Grubbs Criterion

目前大多物理实验工作者剔除实验异常数据时,多选择肖维勒准则作为判断工具。但肖维勒准则具有概率分布不均匀的缺点,对异常数据的剔除存在误判的可能性。以《测直流电源的电动势和内阻》为例,利用Origin软件进行实验数据的线性拟合以及调用程序编辑功能,实现肖维勒准则和格拉布斯准则对异常数据的判断和剔除。根据两个剔除结果的差异,结合实验数据的特点,从实验数据的测量次数和剔除原则的显著水平两个角度来比较两个剔除准则。提出应根据实验次数和实验要求来选择异常数据剔除准则。

Nowadays,lots of physics experimenters use Chauvenet＇s Criterion as a tool to reject bad data points.But it has the disadvantages of uneven probability distribution,which will cause an error in rejection.This article aim to take the experiment——Measuring Eletriomotive Force and Internal Resistance in Direct Circuit as an example,and apply Origin as a linear fitting tool and editor program to achieve rejection in two ways.Discussion based on the difference between results,is comparing Chauvenet＇s Criterion to Grubbs＇s Criterion in two aspects——the number of measurements and significant level.Conclusion,the criteria should be selected based on these two.