Richard Gonzalez
Psychology 614
Winter, 2004
The class website:

1   Analyzing Multivariate Data

1.1  Course Description

A graduate-level introduction to multivariate data analysis. The general flavor of this course will be intuitive. We will not cover proofs and time spent on matrix algebra will be minimal. The course will emphasize the application of multivariate statistical techniques.
Topics reviewed include multidimensional scaling, principal components, factor analysis, multivariate analysis of variance, canonical correlation, discriminant analysis, cluster analysis, reliability theory, and structural equations modelling.
Before getting to the multivariate material however, we need to finish the spillover from 613. We will pick up where we left off on lecture notes #8. Lecture notes #10 - #13 will be available from a copy center later in January.

1.2  Texts and Software

Required texts are:
Multidimensional Scaling by Davison
and one (only one) of the following two:
Using Multivariate Statistics by Tabachnik and Fidell
Applied Multivariate Statistical Analysis by Johnson and Wichern
The Tabachnik and Fidell book is more "cookbookish" than Johnson and Wichern. J&W use a little matrix algebra. I suggest you purchase the book that fits your own tastes for detail of mathematical explanation; browse through the two books and decide which one you'd prefer reading during the semester.
I do not have preferences for the software package used in the course. The examples I will offer in class will be computed in SPSS, Splus, Maple, Matlab, KYST, SINDSCAL, SAS and ADDTREE, but students who know other packages are welcome to use those when possible. I'll bring up the issue of statistics packages throughout the course.

1.3  Prerequisites

A strong background in data analysis is essential. Successful completion of a course on ANOVA/Regression and experience analyzing data is required. Students who took my Psychology 613 course will be adequately prepared and know about my own biases for how to do statistics. A willingness to tackle new problems and computer programs is also needed.

1.4  Grading

Approximately every two weeks there will be a problem set. At least one of these will be a group project. I encourage auditors to do the problem sets too.
There will be one midterm in the course, which will be a takehome exam.
The course grade will be a weighted average of the midterm (25%), the group project (25%) and the total of all problem sets (50%).

1.5  Outline of Topics

  1. Complete remaining topics from 613

  2. Multidimensional scaling (two-way, three-way)

  3. Tree structures and cluster analysis

  4. Unfolding analysis

  5. A little matrix algebra

  6. Principal components analysis

  7. Covariance algebra

  8. Test theory (in particular, reliability)

  9. A very basic introduction to structural equations modelling (SEM)

  10. Review of repeated measures ANOVA

  11. Multivariate analysis of variance

  12. Discriminant analysis

  13. Canonical correlation

1.6  Class E-mail List & Web Site

We will make heavy use of the class e-mail group to post announcements, hints, typo corrections, etc.
The class website:

File translated from TEX by TTH, version 3.33.
On 13 Jan 2004, 03:25.