中文摘要：

目的为二分类敏感性问题提供科学的较复杂抽样调查方法及其统计量的计算公式。方法采用Co—chran W．G．的抽样理论及研究结果、随机应答技术的双无关问题模型、分层两阶段整群抽样调查方法、全慨率公式、方差的基本性质、效度评价等理论和方法。结果推导出二分类敏感问题双无关问题模型在分层二阶段整群抽样下总体比例的估计量及其方差的计算公式，并成功应用该调查方法及相关公式调查估计得苏州大学新校区学生婚前性行为发生比例为16．77％，将其与效度准则指标作假设检验得P=0．1696。结论本文提供的敏感问题双无关问题模型的分层二阶段整群抽样调查方法及相应的统计量计算公式科学有效。

英文摘要：

Objective To explore scientific sampling methods and corresponding formulas for sensitive questions survey with the sample selected by stratified two-stage cluster sampling. Methods Cochrart W. G. ＇ s classic sampling theories, Greenberg model, stratified two-stage cluster sampling, total probability formulas, properties of its variance and the as- sessment methods of validity were used. Results Formulas for the estimation of the population proportion and its variance for dichotomous sensi- tive questions on Greenberg model in stratified two-stage cluster sampling were deduced. The survey methods and formulas were employed successful- ly in a survey of premarital sexual behavior among students at Soochow U- niversity. The estimated populated proportion of premarital sexual behavior was 16.77%. Comparing with standard method,high validity was proved （P = 0. 16967 ）. Conclusion The methods and corresponding formulas on Greenberg model with the sample selected by stratified two-stage cluster sampling are feasible.