摘要
简述了偏最小二乘用于多元回归建模的方法。对顾客购后行为意向及其成因指标的调查数据采用偏最小二乘法,根据变量投影重要性指标和因子载荷分析,从10个成因指标中筛选出8个组成自变量指标集合,然后以重复购买、交叉购买、正面推荐等三种顾客购后行为意向指标作为因变量集合,建立了多元回归模型,取得了比较满意的拟合与预测效果。主要结论如下:对于电视机产品,(1)期望和愿望自身对于顾客购后行为意向没有显著影响;(2)期望和愿望的满足程度影响顾客的购后行为意向;(3)企业形象和发布信息的真实性影响顾客的购后行为意向;(4)质量因素比价格因素的影响强烈;(5)在三种顾客购后行为中,顾客满意度对于正面推荐的影响系数最大。
The modeling method using partial least squares (PLS) is briefed and applied to the sample of customers' postpurchase behaviors and its determinants. According to the criteria of VIP (variable importance in the projection) and factor loadings analysis, eight independent variables are selected out of ten variables, based on which, a multiple regression model is set up with three postpurchase behavior variables as dependent variables. Satisfying predicting capability, and four conclusions are made as follows for durable product such as TV: (1) customer expectation and desire have no effect on postpurchase behavior; (2) the extents to which that customer perceived performance match their expectations and desires have notable influence on postpurchase behavior; (3) company image and true advertisement information impact on postpurchase behavior; (4) quality has stronger impact than price does; (5) in the last three postpurchase behaviors, customer satisfaction has the strongest influence over intention to recommend.
出处
《工业工程与管理》
2003年第3期14-19,共6页
Industrial Engineering and Management
基金
国家自然科学基金资助项目(70271040)
