摘要
设x1,…,xn;y1,…,ym为独立随机样本,x1,…,xn同分布,x1~F(x),F未知,y1,…;ym同分布,y1~Gθ(y),Gθ(y)的形式已知,θ为未知参数.本文结合非参数似然思想和参数似然方法讨论F和Gθ的分位数差异的检验问题,在一定的条件下得到了半经验似然比统计量的渐近分布.
Let x1,..., xn, y1,... , ym be independent and suppose that the {xi} are identically distributed as unknown F(. ) and the {yi} are identically distributed as Gθ(' ), whereGθ(' ) is of known form with unknown parameter θ. In this paper, we combine nonparametriclikelihood thoughts and parametric methods to test the difference of the two sample quantiles. Under some common conditions, the asymptotic distributions of the semi-empiricallikelihood ratio statistic are obtained.
出处
《应用数学学报》
CSCD
北大核心
1998年第1期103-112,共10页
Acta Mathematicae Applicatae Sinica
基金
国家自然科学基金
国家教委博士点基金
中国科学院特别支持经费资助
关键词
经验似然
假设检验
半经验似然
分位数
检验
Empirical likelihood, hypotheses tests, semi-empirical likelihood