Graphical diagnostic methods for regression models: A Monte Carlo study with application of Taiwanese kindergarten educators' work motivation

摘要

The goal of this study is to reveal the patterns of residuals in several graphical diagnostic methods under various regression assumptions and sample sizes by performing Monte Carlo simulation. Moreover, evidence found in the simulation procedures is applied in an educational research of Taiwanese kindergarten educators' work motivation to demonstrate a practical use.;The traditional graphical diagnostic methods and nonparametric regression smoothing techniques are utilized and compared in this research. The nonparametric regression smoothing techniques do improve some weaknesses of traditional methods when sample sizes are appropriate.;The substantive guidelines for evaluating regression model adequacy in related fields are also provided in the last chapter.