摘要
目的针对传统非参数方法在等级数据分析中忽略非线性特征与极化趋势的局限,探索优势分析与ANY极化指数在量化分布差异及不平等程度中的应用价值。方法结合李克特量表模拟数据(不同中位数分布)与美国国家健康与营养调查(National Health and Nutrition Examination Survey,NHANES)真实数据,采用优势分析与ANY指数,对比传统秩次检验,评估方法在揭示等级数据分布特征与极化效应中的效能。结果优势分析通过累积分布函数(cumulative distribution function,CDF)有效区分群体间整体水平差异(F-优势)与内部不平等程度(S-优势);ANY指数通过调节参数α(侧重低端分布)与β(侧重高端分布),灵活捕捉极化趋势。NHANES数据分析中,新方法成功识别男性健康评价低端极化更显著的特征,而传统Mann-Whitney U检验无统计学差异。结论优势分析与ANY指数突破了传统方法对等级数据的简化假设,为卫生政策制定提供了更精准的分布差异与极化分析工具,尤其在资源优化配置与弱势群体健康需求识别中具有重要实践意义。
Objective To address the limitations of traditional nonparametric methods in analyzing ordinal data,which often overlook nonlinear characteristics and polarization trends,this study explores the application value of dominance analysis and the ANY polarization index in quantifying distributional differences and inequality.Methods Simulated Likert-scale datasets(with varying medians)and real-world data from the National Health and Nutrition Examination Survey(NHANES)were analyzed using dominance analysis and the ANY index.Their performance in revealing distributional heterogeneity and polarization was compared with conventional rank-based tests.Results Dominance analysis effectively distinguished inter-group differences in overall levels(F-dominance)and internal inequality(S-dominance)through cumulative distribution function(CDF).The ANY index captured polarization trends by adjusting parametersα(emphasizing lower percentiles)andβ(emphasizing upper percentiles).In NHANES data,the new methods identified significantly higher low-end polarization in males′self-rated health,whereas traditional Mann-Whitney U tests showed no statistical significance.Conclusion Dominance analysis and the ANY index overcome the oversimplified assumptions of conventional methods,offering refined tools for analyzing ordinal data in health policy-making,particularly in optimizing resource allocation and addressing health disparities among vulnerable populations.
作者
李超
杨莉莉
齐士勇
高雅杰
纪伟伟
马婷婷
孙萌
刘爽
崔壮
Li Chao;Yang Lili;Qi Shiyong(Tianjin Medical University Cancer Institute&Hospital,Tianjin 300060)
出处
《中国卫生统计》
北大核心
2025年第3期328-333,共6页
Chinese Journal of Health Statistics
基金
国家自然科学基金(72374153)
天津医科大学2024年度医院管理创新研究项目(2024YG19)。
关键词
等级数据
优势分析
极化指数
Ordinal data
Dominance analysis
Polarization index
