Literaturnachweis - Detailanzeige
Autor/in | Rule, David L. |
---|---|
Titel | A Simulation-Based Comparison of Several Stochastic Linear Regression Methods in the Presence of Outliers. |
Quelle | (1993), (34 Seiten)
PDF als Volltext |
Sprache | englisch |
Dokumenttyp | gedruckt; online; Monographie |
Schlagwörter | Analysis of Covariance; Bayesian Statistics; Comparative Analysis; Computer Simulation; Estimation (Mathematics); Goodness of Fit; Least Squares Statistics; Mathematical Models; Matrices; Regression (Statistics); Research Methodology; Sample Size; Scores |
Abstract | Several regression methods were examined within the framework of weighted structural regression (WSR), comparing their regression weight stability and score estimation accuracy in the presence of outlier contamination. The methods compared are: (1) ordinary least squares; (2) WSR ridge regression; (3) minimum risk regression; (4) minimum risk 2; (5) goodness of fit index (GFI); and (6) WSR reduced rank regression. Three population covariance matrices were used that were drawn from applied behavioral science literature as the basis for generating samples, some of which were contaminated. A bootstrap method was used to compare the regression methods. Analysis resulted in 4 sets of bootstrap samples for each of the population systems, 12 in all. Results support the notion of increased efficacy of the adaptive forms of WSR in small sample applications where outlier contamination exists. The improvement over conventional least squares is not always substantial, but it is notable that adaptive forms of WSR based on the concept of empirical Bayes covariance estimation can provide consistent and sometimes substantial improvement over conventional methods. Five tables present analysis data. Appendix A provides the basis for the WSR class of methods. Appendices B and C each contain three tables of analysis information. (SLD) |
Erfasst von | ERIC (Education Resources Information Center), Washington, DC |