基于偏最小二乘路径模型的分位效应测度

Measuring Quantile Effects Based on Partial Least Squares Path Model

对于偏最小二乘路径模型的效应分析,为了测度路径模型的分位效应,文章首先给出了偏最小二乘路径模型建模的具体过程.其次,基于潜变量得分与分位回归提出估计平滑分位效应的方法,给出了平滑分位效应的Bootstrap置信带的算法.最后,考虑顾客满意度的分位异质性,对满意度模型的分位效应进行分析.结论表明,该方法是对传统偏最小二乘路径模型的一种补充且可获得更有深度的决策信息.

To measure quantile effects in the path model for the path effects of Partial Least Squares(PLS)path model,this article firstly gives the specific procedure of PLS path modeling.Furthermore,the estimators of smoothing quantile effects are derived based on latent variable score and quantile regression,as well as the algorithm of bootstrap confidence corridor for smoothing quantile effects are given.Finally,we use the proposed method to analyze the case-study regarding quantile effects of consumer satisfaction index,in which data quantile heterogeneity regarding consumers1 satisfaction is considered.The results also illustrate the proposed approach complements and provides insights over and above a traditional PLS path model.