|
Return
to Syllabi List
Summer,
2003
PM603.
FACTOR ANALYSIS
Instructor: Chih-Ping Chou
(626)457-6649, cchou@hsc.usc.edu
TA: Bin Xie
(626)457-6604, bxie@hsc.usc.edu
Class hour: Wednesday 9-12
Office hour: Friday 10-11 (Alhambra Office 5114)
This course will
deal with the statistical techniques for factor analysis and structural
equation modeling. Both factor analysis (FA) and structural equation
model (SEM) are multivariate statistical techniques to analyze association among
multiple measures through latent variables (or constructs, factors).
This course will
start with relationships among measurable variables (or the Pearson type correlation),
regression models, and path models. Basic concepts of Exploratory Factor Analysis
(EFA) and Confirmatory Factor Analysis (CFA) will then be discussed. All of these
models can be considered as special cases of SEM. SEM can be conceptualized as
a path model built upon latent factors.
The purpose of
Factor Analysis is to use a set of factors with smaller dimension to represent
a set of measurable variables. The EFA tries to explore the underlying factors
to explain the relationships among observed variables. The CFA evaluates the relationships
through the covariation among factors. The SEM, furthermore, hypothesizes causal
relationships among factors to account for the relationships among measured variables.
Both FA and SEM offer statistical control of the measurement error in the analyses.
The SAS program will be used for the EFA. The EQS program (Bentler, 1995) will
be used for CFA and SEM.
This is a 12-week
course with 4 units. The first 3-4 weeks will be devoted to basic concepts and
geometrical representation of FA techniques. The following 2-3 weeks will concentrate
on the application of EQS program. The rest of the semester will focus on the
statistical and empirical issues of SEM. Students are encouraged to bring in their
own data sets and apply FA or SEM to develop a final term paper. A data set will
be available for those who don't have their own data set. There will also be homework
assignments and quizzes. Grade will be determined by the homework, final paper,
and possible one or two quizzes.
Three textbooks
have been ordered through the bookstore:
A. Bentler, P.
M. (1995). EQS program manual. Los Angeles: Multivariate Software Inc. (required).
B. Byrne, B. M. (1994) Structural Equation modeling with EQS and EQS/Windows.
Thousand Oaks, CA: Sage (required).
C. Bollen, K. A. (1989). Structural Equations with latent variables. Wiley (optional).
Tentative
weekly schedule:
| Week |
Topic |
Chapter |
| 1-2 (May 21, 28) |
Introduction of
FA, correlation, basic concepts of vector, matrix algebra, and geometrical representation
of FA. |
C. Appendix |
| 3 (June 4) |
Factor extraction
and rotation |
|
| 4 (June 11) |
EQS Program |
A1-A4 |
| 5-6 (June 18,
25) |
Confirmatory Factor
Analysis, Structural Equation Modeling, and EQS |
B1-B7 |
| 7 (July 2) |
LISREL Program |
C2, C3 |
| 8 (July 9) |
Model Identification
|
C4 |
| 9 (July 16) |
Model estimation
and Distribution assumption |
C5 |
| 10 (July 23) |
Model modification |
A6 |
| 11 (July 30) |
Moment matrix |
A8 |
| 12 (Aug 6) |
Multiple group
approach |
A7, B8-B10 |
Grade:
Homework (4-5)
20%
Quiz 5%
Midterm 15%
Term paper 60%
Return
to Top
|
