基于区间数证据推理方法的用户满意度调查

目前,加权平均法是一种比较常见的满意度调查结果的汇总方法,但是这种方法的前提条件是决策者的偏好结构满足加性独立条件,否则需要采用非线性综合方法。本文旨在考虑决策者偏好不满足加性独立条件下,将用户满意度抽样调查过程中产生的置信度和置信区间与调查问卷中的用户不确定的评价结果统一进行考虑,并采用mass函数值为区间数的证据推理方法分析基于抽样调查得到的以置信区间表示的用户满意度调查的结果综合问题。最后以某网络信息中心用户满意度调查为例展开实证分析。

Customer Satisfaction Survey Based on Evidential Reasoning Approach with Belief Intervals

Customer satisfaction has become an important marketing research topic.Customer satisfaction can be seen as an important indicator for the career success of an employee in the public service sector.Many scholars have conducted research on customer satisfaction and surveyed different subjects.The execution of a customer satisfaction survey generally follows three stages:(1)questionnaire design(indicator system),(2)survey methodology,and(3)analysis of survey results.The survey methodology has the limitations of time and cost.As a result,most studies adopting this methodology often use samples to represent the population.When interpreting the analysis result of a survey study,a scholar needs to be cautious about its limitations,such as feedback uncertainty,and lack of choices in answers,and unclear questions. The analysis of the survey results is mainly based on statistical principles.The statistical analysis result based on samples is only an estimation of the entire population.The accuracy of statistical results depends on sample size.Thus,it is important to consider the degree of representativeness for the adopted sample.In customer satisfaction surveys,the frequency statistics for each option of a survey question is a point estimation based on the sample data.Data reliability is missing from frequency statistics.The interval estimation based on the confidence level can give an estimation of the confidence interval for the entire population. During a survey,it is common to see that some customers leave no choice or have more than one choice for some unclear questions.Therefore,a customer satisfaction survey contains two types of information uncertainty:confidence intervals in the sample statistics,and question uncertainty(e.g.no choice or multiple choices)for some customers,resulting in no choice or multiple choices.Decision makers need to consider these two information uncertainties when interpreting information on service level,and service quality,and when proposing future development strategies. The Weighted Average Method(WAM)is commonly used to analyze the result of a customer satisfaction survey.Customer satisfaction index(CSI)compiles and converts all scores into a number or percentage.For example,the index can be an average of fifty entries that a customer scores,or it can be the weighted average with each entry given a weight according to its importance.WAM requires that the preference structure of decision-makers(making decision based on overall satisfaction)meet a series of conditions,such as ’additive independence’ etc.If the preference structure of decision-makers does not meet the "additive independence" condition,a new nonlinear approach for summarizing customer satisfaction survey results need to be proposed.This paper uses the evidential reasoning approach with the Interval Mass function to address problems related to customer satisfaction survey.This paper further discusses approaches to minimizing information uncertainty caused by the sampling approach. This paper discusses the analysis result based on a survey with the Network Information Center.The survey result is expressed as intervals for each grade to precisely reflect the degree of customer satisfaction.Decision-makers can conduct an in-depth analysis based on the survey result,and adjust their strategic development and resource allocation programs.Doing so not only can help improve service levels and service capabilities,but also enhance customer satisfaction.