以顾客为导向，追求顾客满意和减少顾客投诉是服务管理理论研究的重要问题。目前，顾客满意度调查被广泛应用，以期定量评价服务质量，但大多数顾客满意度调查与分析忽略了各测评项离散分布对顾客满意程度整体评价不确定性的影响，难以提供全面、准确的信息。同时，大学生公寓服务质量会影响校园稳定和社会稳定，其满意指标的研究具有重要意义。本研究总结了在顾客满意、服务质量、顾客投诉等方面的研究成果，提出了存在的问题，并在已有顾客满意度及服务质量理论的基础上，引入了标准差概念，构建了顾客满意度二维模型，第一维是顾客满意度（CS），第二维是感知质量标准差（SD），并提出应用“四区间”图对顾客满意情况进行全面评价。通过专家访谈和数据统计分析，本研究建立了大学生公寓的顾客满意度二维模型，2005年12月至2008年10月的近3年时间中，在清华大学大学生公寓开展了10次大样本的测评（有效问卷共9638份），同时本研究提出了大样本问卷数据筛选原则与数学模型，编制了数据处理程序，并对实际调查的问卷数据进行了筛选处理。采用结构方程模型，本研究论证了大学生公寓顾客满意度二维模型的有效性，并对2006-2007与2007-2008学年度大学生公寓顾客满意度、质量标准差与学生投诉数量进行了相关性分析。本研究的主要结论：1．用大学生公寓的实证模型验证顾客满意度二维模型是有效的。2．在较高满意度水平下，质量标准差与投诉数量存在显著正相关关系（相关系数大于0.6），顾客满意度与投诉数量之间没有显著的负相关关系。3．与大学生公寓相关的感知质量概念模型具有四个维度，分别是公共卫生、楼长服务、维修服务、安全管理；建立的顾客满意度模型通过了结构模型检验，公共卫生、楼长服务、维修服务、安全管理等维度的感知质量与顾客满意度正相关。4．数据筛除原则有效，数据筛除率最高达12%。本研究的部分理论成果已在清华大学得到应用，有效提升了大学生公寓服务质量。

In the customer-oriented times, improving the Customer Satisfaction and decreasing the Customer Complaints is becoming the main problem of the developments of modern service finance and management theory. At present, surveys of Customer Satisfaction are used extensively in service industry and others as the unique criterion for the quantificational evaluation of service quality. But since surveying results come from a number of customers who usually have different opinions and difference of service quality supplied by enterprise, uncertainty of surveys would affect the result of Customer Satisfaction. And because of the particularity of public industry in service and management of undergraduate dormitory that affect campus and society stability, surveys of Customer Satisfaction might well make an important roll for conjuncture indicator. The study of Customer Satisfaction has become one of hot topics for both Public Industry managers and scholars. This paper proposed the questions of Customer Satisfaction, by which summarized the results about former theory of Customer Satisfaction, service quality and Customer Complaint. In this paper, based on the existed theories of Customer Satisfaction and service quality, a Customer Satisfaction 2-order Model applied in public enterprise along the related industry and characteristic of customer by appending standard deviation was constructed which is composed of the customer satisfaction index (defined as CS) and the service quality standard deviation (defined as SD), and used to evaluate the service quality totally through 4-section by different combination of CS and SD. Large sampling surveys of Customer Satisfaction 2-order Model were implemented in Undergraduate Dormitory of Tsinghua University for 10 times and 3 years (since Dec, 2005 to Oct, 2008, effective numbers of questionnaire are totally up to 9638). Based on the Stat. and Delphi Method, the undergraduate dormitory Customer Satisfaction 2-order Model was presented, and indexes of CS and SD were designed in details. Data ridding algorithm and related filtering model were set up by which the surveys were filtered, and the undergraduate dormitory Customer Satisfaction 2-order conceptual model was proved effectively through the Structural Equation Algorithm. The relativity of CS, SD and number of customer complaints was analyzed in this paper. Throuth using the software AMOS, SPSS and Matlab to check all assumption, the significance of conclusions could be acquired: Firstly, the customer satisfaction 2-order conceptual model was proved effectively by empirical studies of undergraduate dormitory services in Tsinghua University. Secondly, with a high Customer Satisfaction level, perceptive quality standard deviations are exactly related to number of Customer Complaints (relative coefficient is more than 0.6.) and Customer Satisfaction has no remarkable negative relationship to number of Customer Complaints. Thirdly, the conceptual model which related to perceptive quality of undergraduate dormitory service was composed of public cleaning, manager, maintenance, security services. Through theoretical study, the paper build Customer Satisfaction model and established every own indexes. Empirical studies validated reliability and validity of these conceptual models, and perceptive quality of public cleaning, services, maintenance and security services were positively related to customer satisfaction. Fourthly, the data ridding algorithm was effective for applying. During survey processing, the ineffective data through the ridding algorithms was abandoned up to 12%. The partial conclusions about this article have been successfully applied in practical management and service of undergraduate dormitory in Tsinghua University to examine the performance of different branches. Within 3 years, the service quality of undergraduate dormitory has been improved slowly.