Literaturnachweis - Detailanzeige
Autor/inn/en | Millsap, Roger E.; und weitere |
---|---|
Titel | Component Analysis under Linear Equality and Inequality Constraints. |
Quelle | (1986), (8 Seiten)
PDF als Volltext |
Sprache | englisch |
Dokumenttyp | gedruckt; online; Monographie |
Schlagwörter | Cross Sectional Studies; Factor Analysis; Goodness of Fit; Longitudinal Studies; Matrices; Multivariate Analysis; Regression (Statistics) |
Abstract | A constrained component analysis method, which bears a formal resemblance to the confirmatory factor analysis methods developed by K. G. Joreskog (1969) and others, is presented. In confirmatory factor analysis, the constraints allow the testing of formally structural hypotheses within a model that is falsifiable, even in its "just defined" form. In component analysis, the goal is to determine whether a component solution that is restricted in various ways can still account for an adequate share of the variance in the observed data. The consistency and efficiency of the estimates under the constrained analysis are considered; however, no statistical tests of fit of the component are discussed. In this sense, the method is closer to the work in constrained canonical correlation by W. S. DeSarbo and others (1982). Interesting applications of constrained component analysis exist in data that are longitudinal or cross-sectional in nature. In these cases, the natural constraints may involve the property of being stationary or the invariance of the compositing weight, component pattern, or component structure matrices. In data measured at a single occasion in a single population, constraints might be used to impose simple structure or achieve a sensible variable clustering. Boundary constraints might be used to achieve a positive manifold. (TJH) |
Erfasst von | ERIC (Education Resources Information Center), Washington, DC |