中文摘要：

网络调研已广泛应用于领导满意度评价等敏感性问题.网络调研时不存在令用户充分信任的随机数生成机制,从而导致传统敏感性问题调研的随机作答方法（RRT）失效.提出分组调研方法,通过合并分组降低问题敏感性、提高如实作答可能.同时依据样本量最小的原则,给出了最优分组方式,并在合理的参数范围内通过数值求解得到分组人数与总人数的经验公式k=（n＋1）/2.方法也可以应用于随机数生成机制不可信任条件下的其它敏感性调研问题.

英文摘要：

Online survey has been widely applied to sensitive question surveys such as satisfaction evaluation for superior colleagues.Traditional survey methods for sensitive questions（RRT） could not be adopted for online surveys,because none of the random number generators is trustworthy enough to be used online.A grouping survey method is therefore introduced to reduce survey sensitivity and to increase faithful response rate by grouping sensitive questions.In addition,the paper provides the optimal grouping method in accordance with the principle of minimal sample size.An empirical formula k =（n ＋ 1）/2 was obtained by numerical solution in reasonable parameter range.This method is also applicable in other sensitive question surveys where random number generator fails.