University of Maryland, College Park
Smith School of Business
BMGT 882
Applied Multivariate Analysis I
Fall 2004
Professor: P. Zantek
Office: 4349 Van Munching Hall
Office Hours: Wed 2:30 − 3:30
Course Meeting Times: Mon/Wed 3:45 − 5:00
Telephone: 301-405-8644
pzantek@rhsmith.umd.edu
www.rhsmith.umd.edu/dit/
COURSE DESCRIPTION
Multivariate statistical methods and their use in empirical research. Planned topics include
summarization and visualization of multivariate data, multivariate paired comparisons and
repeated-measures designs, multivariate analysis of variance, discriminant analysis, and
canonical correlation. The maximum likelihood and likelihood ratio principles are also
discussed. An important component of the course is analysis of business data using
contemporary software.
ASSUMPTIONS
BMGT 882 is designed to be taken as the third of the four “research tools” courses required of
Smith School doctoral students.
PREREQUISITES
AREC 623 or ECON 621 or STAT 450 or STAT 740 or permission of instructor
BMGT 882 assumes a working knowledge of matrices and elementary linear algebra and a
sound understanding of univariate statistics, including random variables, statistical inference,
ANOVA, and regression. For example, it will be assumed that you have studied regression using
the matrix formulation of the linear model. If you lack this background, you are likely to
encounter significant difficulty in this course and are, therefore, urged to drop. Dropping will
enable you to either acquire the prerequisite coursework or to take an alternative multivariate
course.
MATERIALS
•
•
•
Course notes (to be distributed by instructor)
Rencher, Methods of Multivariate Analysis (2nd edition), Wiley, 2002 ISBN 0-471-41889-7
Lattin, Carroll, & Green, Analyzing Multivariate Data, Duxbury, 2003 (LCG) (OPTIONAL)
2
CLASS MEETINGS
Class meetings consist primarily of lectures and discussions of examples/applications.
EXPECTATIONS
You are expected to attend each class, to contribute substantively in class, and to always exhibit
mutual respect.
SOFTWARE
An important part of the course is the use of contemporary software for statistical analysis.
Several software packages for performing multivariate analysis are available (e.g., MINITAB,
SAS, SPSS); each has its strengths and weaknesses. In this course, we use SAS for Windows
(Release 8.01 or higher). Some advantages of SAS include its numerical accuracy and the fact
that it implements a wide variety of methods. You are required to become familiar with the
aspects of the software covered in class, including the interpretation of output.
For exercises and projects, you are permitted to use software other than SAS if you wish, but
doing this has disadvantages; for example, the format of the software output will differ from that
given in class. Please note that the instructor cannot provide assistance with software other than
SAS.
EXERCISES
You will be given several exercises during the semester. Some of these will be collected and
graded
PROJECT
Each student is required to define and complete a project that entails applying at least two
methods covered in this course to a data set. The data set and the topic are of your choosing (you
are encouraged to select a topic that is related to your research interests). Two hard copies of a
report on your project are due by December 3. The report is to include the following:
introduction, research problem/question(s), model and/or hypotheses to be tested, description of
the sample and data, methodology, empirical results, summary and conclusions, and references.
The report should be approximately 6-10 pages in length (not including tables and figures),
double-spaced, and in 12-point, Times New Roman font; please use one-inch margins.
3
ACADEMIC INTEGRITY
The University’s Code of Academic Integrity is designed to ensure that the principles of
academic honesty and integrity are upheld. All students are expected to adhere to this Code.
The Smith School does not tolerate academic dishonesty. All acts of academic dishonesty will be
dealt with in accordance with the provisions of this code. Please visit the University’s website
for more information on the Code of Academic Integrity.
The University of Maryland Honor Pledge reads:
“I pledge on my honor that I have not given or received any unauthorized assistance on
this assignment/examination.”
Unless you are specifically advised to the contrary, this pledge statement should be handwritten
and signed on the front cover of all documents submitted for evaluation in this course.
GRADING POLICY
Your final grade for the course is based on your performance in the individual exercises, the
project, and two examinations. The weights given to each of these components are as follows.
Individual Exercises
Individual Project
Midterm Examination
Final Examination
15%
15%
35%
35%
No make-up examinations are provided.
COURSE WEBSITE
The Internet will be used to provide course-related information such as data sets and SAS
programs. Courses offered by the Smith School use the ‘Blackboard’ web course environment.
You may login to your course(s) by going to http://bb.rhsmith.umd.edu. Please note that your
University Directory (LDAP) username and password are required to access Blackboard courses.
To find your username, visit https://ldap.umd.edu/cgi-bin/chpwd?searchbyumid. To set or
change your password, visit https://ldap.umd.edu/cgi-bin/chpwd.
Students are required to maintain their current e-mail address in Testudo, as Blackboard uses
this address to send course-related e-mail. (Since e-mail addresses are imported from Testudo
into Blackboard, it is not possible to update e-mail addresses from within Blackboard.)
Additional information about R. H. Smith Blackboard is available from
http://www.rhsmith.umd.edu/blackboard.
4
INCLEMENT WEATHER
In the event of inclement weather, please check with the University to see if classes are
cancelled. Cancellations are announced on the University’s website http://www.umd.edu/ .
SCHEDULE
The following schedule is tentative. Please note that all chapter and section numbers refer to
Rencher. The schedule is subject to change by announcement or by distribution of a revised
schedule.
Tentative Schedule
Aug.
Sep.
30 Mon Introduction
Chapter 1
1 Wed Review of Matrices and Linear Algebra
Chapter 2
6 Mon No Class—Labor Day
8 Wed Review of Matrices and Linear Algebra (cont’d)
Chapter 2
13 Mon Random Variables, Multivariate Samples, and Sample Statistics Chapter 3
15 Wed Random Variables, Multivariate Samples, and Sample Statistics Chapter 3
Oct.
20 Mon Multivariate Normal Distribution
Chapter 4
22 Wed Maximum Likelihood Estimation
Chapter 4
27 Mon Mean Inferences: One-Sample Hotelling T Test
Chapter 5
29 Wed Multivariate Paired Comparisons/Repeated Measures
Chapter 5, § 6.9
4 Mon Multiple Comparisons and the Bonferroni Method
Chapter 5
6 Wed Two-Sample Hotelling T Test
Chapter 5
11 Mon Likelihood Ratio Tests
Chapter 5
13 Wed Testing Homogeneity of Variance-Covariance Matrices
§ 7.3
18 Mon Multivariate Analysis of Variance (MANOVA)
Chapter 6
20 Wed One-Way MANOVA
Chapter 6
25 Mon Midterm Examination
Nov.
Dec.
27 Wed MANOVA (cont’d)
Chapter 6
1 Mon MANOVA (cont’d)
Chapter 6
3 Wed Two-Way MANOVA
Chapter 6
8 Mon MANOVA (cont’d)
Chapter 6
10 Wed MANOVA (cont’d)
Chapter 6
15 Mon Two-Group Linear Discriminant Analysis (LDA)
Chapter 8
17 Wed LDA: Test for Additional Discrimination
Chapter 8
22 Mon LDA: Fisher’s Linear Discriminant Function and Profiling
Chapter 8
24 Wed Multigroup Discriminant Analysis:
Chapter 8
29 Mon
Chapter 8
1 Wed
Discriminant Functions and their Statistical Significance
Test for Additional Discrimination
Project Due Dec. 3 ! Chapter 8
6 Mon Canonical Correlation (time permitting)
Chapter 11
8 Wed Canonical Correlation (cont’d)
Chapter 11
Final Examination (comprehensive)