Psychology 607

Multivariate Data Analysis

Focus

This course is focused on methods and techniques for analyzing multivariate data. Emphases include both conceptual and computational aspects of the most commonly used analytic tools when one has multiple measures on the same experimental units. Derivations and advanced mathematical and statistical concepts will not be featured parts of the course but students will be expected to master the rationales behind the methods that will be covered to the extent that they can generalize the applications to novel problems and contexts. This course hopes to avoid the extremes of ``cookbook analyses'' on one hand and theorems and proofs on the other to provide generalizable working knowledge of multivariate statistics.

The initial part of the course is committed to the essential operations of matrix algebra, a key language of multivariate analysis. Subsequently, a close look will be taken at the nature of linear combinations of variables. The remainder of the course will feature the application of multivariate data analysis methods and models.

Computer work associated with the course will involve the SAS and Splus statistical packages. It is expected that students will learn to be sufficiently familiar with modern computational software such as SAS and Splus that they can access available routines to perform matrix operations and various multivariate analyses (e.g., multiple regression, canonical correlation, principal components and factor analysis, etc.). Access to specialized computing routines may be made available as needed.

Prerequisites

Psyc 507-508 (Quantitative Methods I and II) or equivalent is required. Familiarity with regression, ANOVA and simple correlational techniques is assumed. Occasionally I will be posting messages to the class via email, so everyone must have an email account and read their account on a daily basis. I will be assigning homework programs on the class web page, so students must be able to surf the web.

Student Evaluation

Evaluation and course grades will be based on homework, a midterm exam and a final exam. The homework will include computer exercises using Splus and SAS. The student is expected to compile the results of their homework exercises into a notebook. The homework notebook will be due at the beginning of class on October 15th. The midterm exam and final exams will be take-home exams.

Textbook and Handouts

The textbook will be Tabachnick & Fidell (1996) Using Multivariate Statistics. A set of handouts covering the basics of Matrix Algebra will be provided to supplement the textbook. Students are strongly encouraged to apply the approaches and methods under discussion to their own data and to bring results to class for discussion and critique.

• This is an pdf file of the overheads used in class and in a longer workshop on mixed models. The SAS programs discussed in the handout are First Example, Second Example, and More Examples.

• AppendTo.cshrc is a Unix textfile that should be appended to the file called ".cshrc" in your home directory.

• A38.txt is the data set for IntroToMatrices3.S.

• twincog2.dat is the dataset for input to SAS for the principal components homework.

• twincog.dat is the dataset for input to SAS for the factor analysis homework.

• twincog1.dat is the dataset for input to SAS for the canonical correlation homework.

• quest62005.dat is the data set for Final Exam Question 6.

Steven M. Boker; Room 118-C; Department of Psychology; University of Notre Dame; Notre Dame, IN 46556 Phone: 574-631-4941, FAX: 574-631-8883 e-mail: sboker@nd.edu