| 摘要: |
| 针对分层情况下的疾病流行率问题，考虑了金标准存在时，在各层的敏感度和特异度不同时二项比例（疾病流行率）的齐性检验；提出了基于渐近检验过程的7种统计量和基于Bootstrap重抽样检验过程的4种统计量,并通过蒙特卡罗模拟研究来比较了各种检验的犯第一类错误的概率和检验功效；研究表明：score统计量、似然比统计量和4种基于bootstrap重抽样的检验统计量具有良好的统计性质，推荐用于实际中；最后，实际数据进一步验证了方法的有效性。 |
| 关键词: 威布尔分布 山区风速 极大似然 风功率密度 |
| DOI： |
| 分类号: |
| 基金项目: |
|
| Homogeneity Test for Binomial Proportions Based on Stratified Partially Validated Data |
|
LI Tian-jiao
|
|
School of Science, Chongqing University of Technology, Chongqing 400054,China
|
| Abstract: |
| Aiming at the problem of disease prevalence in stratified situation, the homogeneity test of binomial proportion (disease prevalence) is considered when the sensitivity and specificity of each strata are different in the presence of gold standard. Seven kinds of statistics based on the asymptotic test process and four kinds of statistics based on the bootstrap resampling test process are proposed, and the empirical type I error and the empirical power in various tests are compared by Monte Carlo simulation. The research shows that score statistics, likelihood ratio statistics and four kinds of test statistics based on bootstrap resampling have good statistical properties and are recommended to be used in the practice. Finally, a real data is further used to verify the proposed methods. |
| Key words: gold standard homogeneity test Bootstrap resampling method stratified partially validated data |
