摘要
建立二项分布参数的置信检验方法 ,包括强检验、弱检验和弱强检验 ,可以满足工程中成败型试验结果以及多质量特征产品的统计检验需求。传统假设检验依据的是“小概率原则” ,而置信检验依据的是“大概率原则” ,因此 ,置信检验能够以高概率确认二项分布参数是否满足规定的指标 ,克服了显著性假设检验在接受原假设时仅仅意味着不拒绝原假设的缺点 ,能够大大提高工程检验的精度。并且强检验和弱强检验都具有序贯抽样特点 ,在样本较小时就能进行高置信水平的判断。
The confidence test methods for binomial distribution are presented, which include the strong test, the weak test and the weak-strong test. The success/failure data and the multi-attribute products can be tested by the methods. The traditional hypothesis tests are based on Law of Small Probability. In contrast, the confidence tests are based on Law of Large Probability. The strong test can overcome the disadvantage that to accept a null hypothesis may simply mean that it is not rejected by the hypothesis tests, thus it can be used to decide with high probability whether the binomial distribution parameter reaches the specification. In addition, the strong test and the weak-strong test have the property of sequential sampling and can make conclusion at a high confidence level under small sample conditions.
出处
《机械强度》
CAS
CSCD
北大核心
2003年第5期513-518,共6页
Journal of Mechanical Strength
基金
国防科技预研项目(413280203)资助。
关键词
二项分布
置信检验
强检验
弱检验
弱强检验
成败型试验
抽样检验
质量控制
Binomial distribution
Confidence test
Strong test
Weak test
Weak-strong test
Success/failure test
Sampling inspection
Quality control
