Multivariate forms of multiple linear regression, analysis of variance and analysis of co-variance. Multiple discriminant analysis, canonical correlation, and principal-components analysis. Pre: EDEP 603 and EDEP 604, or consent. (Cross-listed as PSY 614 and SW 656)
Modified: August 20, 2006
Pedhazur, E. J., & Schmelkin, L. P. (1991). Measurement, design, and analysis: An integrated approach. Hillsdale, NJ: Elrbaum. [Esp. Chaps. 22, 23, & 24]
Pedhazur, E. J. (1997). Multiple regression in behavioral research: Explanation and prediction (3rd ed.). Orlando, FL: Harcourt Brace. [Esp. Parts 3 and 4]
Stevens, J. (1995). Applied multivariate statistics for the social sciences (3rd ed.). Hillsdale, NJ: Erlbaum.
Wickens, T. D. (1995). The geometry of multivariate statistics. Hillsdale, NJ: Erlbaum.
The student is strongly encouraged to make use of these Internet resouces/links as a way to reinforce one's understanding of basic statistical concepts.
Running SAS in Batch Mode on Unix (Robert A. Yaffee, NYU)
SAS for UNIX (University of Texas Statistical and Mathematical Services)
PROC CALIS (Structural Equation Model procedure)
We will not be considering the textual material in the order in which it is presented in the book. We will begin with a review of basic psychometric principles (reliability and validity) and the logic of quantitative modeling and statistical inference (parameter estimation and hypothesis testing).
We will first consider the "measurement model" which deals with the dimensionalization of the domain of inquiry or the relation between the latent constructs and the observed indicators we use to assess them. Then we will consider the "structural model" which deals with the interrelations of the dimensions (latent constructs) of the domain.
Since the material with which we will be dealing lends itself most reasonably to representation in matrix algebra notation, there will be class time spent on the basic formulations thereof. Tabachnick and Fidell give an introduction in Appendix A. There is also an introduction to matrix algebra in Appendix A of Pedhazur as well at links to pages on matrix algebra listed in the outline below. The discussion of matrix algebra in class will be centered on the representation of linear models in matrix formulation.
A. Quantitative Modeling in the Behavioral Sciences
1. Measurement and Dimensionalization
2. Theoretical/Structural Modeling
B. Matrix Algebra (representation of models in matrix algebra notation)
Introduction to Matrix Algebra (Bertram F. Malle, University of Oregon)
Matrix Algebra for Statistics (Bertram F. Malle, University of Oregon)
C. Measurement: The Dimensionalization of the Domain of Inquiry
1. Principal Components Analysis (PCA)
2. The Common Factor Model
a. Exploratory Factor Analysis (EFA)
b. Confirmatory Factor Analysis (CFA)
Understanding Factor Analysis (R J Rummel, Emeritus, University of Hawaii)
Programs for Number of Components and Factors using Parallel Analysis (Brian P. O'Conner, Lakehead University, Ontario)
Preacher, K. J., & MacCallum, R. C. (2003). Reparing Tom Swift's electric factor analysis machine. Understanding Statistics, 2(1), 13-43.
Eigenvalues: A short introduction (authorship unknown)
Factor Analysis Using SAS PROC FACTOR (Statistical Services, UT-Austin)
Confirmatory factor analysis using SAS (Statistical Services, UT-Austin)
Factor Analysis (G. David Garson, North Carolina State University)
Factor Analysis (Bertram Malle, University of Oregon)
Factor Analytic Models (Leslie F. Marcus, Queens College, CUNY)
Factor Analysis Glossary (John T. Pohlmann, Southern Illinois University)
Review of Exploratory Factor Analysis (Stephen G. Sapp, Iowa State University)
D. Structure: The Canonical Model (estimation of theoretical relation parameters)
1. Estimation based on Ordinary Least Squares (OLS)
a. Analysis of Variance (ANOVA)
b. Analysis of Covariance (ANCOVA)
c. Multiple Regression
Multiple Regression (G. David Garson, North Carolina State University)
d. Discriminant Analysis (dichotomous dependent variable)
Discriminant Function Analysis (G. David Garson, North Carolina State University)
2. Consideration of Multivariate Forms
a. Multivariate Analysis of Variance (MANOVA)
b. Multivariate Analysis of Covariance (MANCOVA)
GLM: MANOVA and MANCOVA (G. David Garson, NCSU)
c. Multivariate Regression
d. Multiple Discriminant Analysis (polychotomous dependent variable)
Discriminant Function Analysis (G. David Garson, NCSU)
Multiple Discriminant Analysis (G. David Garson, North Carolina State University)
3. Estimation based on Maximum Likelihood (MLE)
a. Logistic Regression
Logistic Regression (G. David Garson, North Carolina State University)
b. Structural Equation Modeling
The Form of Structural Equation Models (Ed Rigdon, Georgia State University)
Structural Equation Modeling (G. David Garson, North Carolina State University)
LISREL (University of Texas at Austin)
If you have difficulty with any of the concepts presented in the text or lectures, please consult one or more of the links above or the syllabus for one or more of the courses (with associated links) on my home page. If you are unable to satisfactorily resolve any questions by consulting the appropriate links, please make an appointment with me at your earliest convenience.
Evaluation and assignment of grades
Questions or comments to: daniel@hawaii.edu
