【作者】 熊晶晶；

【导师】 何桢；

【作者基本信息】 天津大学 ， 管理科学与工程， 2004， 硕士

【摘要】 本论文主要研究均值分析（ANOM）方法及其应用。均值分析是一种有效的质量工具，也是休哈特控制图技术的延伸，从本质上讲，它是一种将一组个体均值与总平均(个体均值的均值)在图上作比较，判断是否有个别均值显著地偏离总平均的技术。均值分析有着和方差分析（ANOVA）相同的实验目的，都是用来检验实验因素对输出的影响是否显著，但均值分析只适用于因子水平固定的实验类型。它的突出优点是，计算过程简单，容易掌握，分析结果容易理解，检验均值异常的能力强，能同时反映数据在统计和工程上的显著性，而且可以用ANOM图来描述分析结果，简单、清楚。本论文首先介绍了均值分析的基本分析过程，它可以应用于单因子、双因子以及更复杂的实验设计，无论是计量值数据，还是计数值数据，均值分析检验异常值的能力都很强。文中给出许多实际应用例子，并介绍了均值分析检出力曲线的使用方法。本论文重点研究了均值分析的应用，包括：均值分析和方差分析在检验异常值灵敏度和所需样本含量大小的比较研究；均值分析在估计正态总体参数 上的应用；均值分析和极差分析（ANOR）在田口正交实验中的应用；均值分析在方差一致性（HOV）检验中的应用；均值分析在营销管理和公司员工绩效考评中的应用。更多还原

【Abstract】 The thesis studies a kind of effective quality method - the analysis of means (ANOM) and its applications. It is thought as an extension of Shewhart control chart techniques. In essence, ANOM is a technique for comparing a group of means to see if any of them are significantly different from the overall mean. As such, it can be thought as an alternative to the analysis of variance for experimental designs with factors at fixed levels. They both provide a study whether the effects of the factors are significant. It is especially recommended for ease of interpretation, the simple computation steps, an assessment of practical significance as well as statistical significance and graphical presentation of results. The thesis gives an introduction to the use of analysis of means step by step. The procedure is applied to one-way, two-way and more complicated factorial designs. For variable data and attribute data, the ability to detecting different means of ANOM is both strong. Examples illustrating its usefulness for analyzing variable data, proportions and counts are given. How to use the power curve of ANOM is also introduced.The thesis makes great efforts to study the application of ANOM, including: the comparison of ANOM with ANOVA on the sensitivity for detecting different means and sample sizes, the introduction of a procedure for pooling data by ANOM, the application of ANOM and ANOR (Analysis of Ranges) for signal and noise in Taguchi design, an analysis-of-means-type test for the homogeneity of variance (HOV), the application of ANOM on the assessment of the company employee performance and ANOM in marketing research.更多还原