摘要
满意度调查越来越为各个行业所重视,学术界也越来越重视对调查结果处理方法的研究.文中就满意度调查结果建立了数学模型,提出了相关的假设检验问题,并利用指数族的性质给出了显著性水平为α的最优势检验,然后由假设检验的接受域与置信限(区间)之间的关系得到了调查结果的置信水平为1 -α的最精确置信限(区间).针对具体的问题给出了拒绝域与置信限(区间)的确定方法, 最后应用所提出的方法进行了实例分析.
In this paper,the authours investigate the satisfaction survey result, and show that the possibility of result follows the multinomial distribution. Because multinomial distribution belongs to exponential family, we suppose a method of hypotheses verification. Following the features of exponential family, these verification is uniformly most powerful (or unbiased) level-α tests. The critical regions of the tests for practice are obtained. Meanwhile, the uniformly most accurate (or unbiased) level-(1-α) conference limit(or interval) by the relationship of the acceptance region and the conference limit(or interval) are given. Finally, we apply our method in an example and illustrate that our method is convenient for application.
出处
《沈阳理工大学学报》
CAS
2005年第1期91-94,共4页
Journal of Shenyang Ligong University
关键词
满意度调查
假设检验
置信限和置信区间
指数族
检验的水平与势
satisfaction survey
hypotheses verification
confidence limitand interval
exponential family
power and level of verification
