MB 876 Spring 2007
T 1:30-4pm; Fulton 310; http://www.analytictech.com/mb876
This course provides an introduction to multivariate statistics, with an emphasis on descriptive and exploratory methods. We begin the semester with basic mathematical concepts such as vectors and matrices. We then move on the measurement theory and measures of similarity between variables and cases. At this point we begin working with analytical methods including multidimensional scaling (MDS), correspondence analysis, factor analysis and clustering. The course is a mix of pure theory and hand's on application to empirical data. One thing the course is *not* is a survey of how these methods are used in the management literature -- that you should be getting in MB 871 Quantitative Methods.
A key deliverable in the course is a way of seeing data -- a matrix way. It is incredibly useful.
Note that in one semester we cannot cover all of multivariate statistics. In addition, the department offers another course on the general linear model. So this course essentially ignores methods in which the user specifies a dependent variable, such as HLM or discriminant analysis. This still leaves a number of topics that we don't have time to cover, such as loglinear modeling, which is a shame. The course has no formal pre-requisites but I do assume an understanding of regression and correlation.
The course is intended for Ph.D. students, though others are welcome to take it. Drop-in auditors are also welcome any time (auditors don't have to come to every class).
This is a hands-on class in which you will learn to work with data. Bring laptops.
The course is largely lecture-driven and playing-with-data-driven. However, questions are encouraged and appreciated, both on and off the immediate topic. Class participation is a part of your grade. Points are given for relevant participation that seems to further learning.
IMPORTANT NOTE: You must bring a laptop to class. We will almost always "play" with data in class.
The principal assignment is an empirical research paper that features prominently one or more of the methods discussed in class. The paper should be at least 25 typed pages (12pt font, doubles-spaced) and make liberal use of figures and tables. The paper should be delivered electronically in a zip file that also contains all relevant data files. This is an individual rather an a group assignment. The paper is due May 5th, 2005. In addition, you will briefly present your findings in class on May 3rd. A description of the project may be found at:
There will be a set of short homework assignments due every week or so. Some of these are labeled "quizzes" which means that you should do them by yourself without help from others (but you can consult books & notes). The other homeworks are typically data collection and analysis oriented.
The relative weight of assignments is as follows:
Assignment Weight Research Project
- Written report
Important Note: Please submit all written work via e-mail -- no hard copies.
The required textbooks for this course are:
Lattin, Carroll and Green. 2003. Analyzing Multivariate Data. Duxbury Press.
Weller and Romney. Metric Scaling. Sage.
Kruskal and Wish. Multidimensional Scaling. Sage.
Important note: The Lattin et al book is expensive. Can you get through this course without it? Yes.
I also recommend:
Green. Mathematical tools for applied multivariate analysis. (buy used online)
Harris. A Primer of Multivariate Statistics.
In addition, there will be other readings on the web, notably articles that illustrate the application of the methods.
Note: A few days after each class, I will often post a handout containing key points from the lecture (if appropriate). Find them on the schedule (www.analytictech.com/mb876/schedule.htm)
We will be using SPSS and UCINET. I will have both installed in the computers in Fulton 214, but I recommend buying the student version of SPSS to have on your personal machines. I will provide UCINET for free. Please bring your laptops with this software to class. Note: There is no MAC or LINUX version of UCINET.
The schedule of assignments and readings is online at www.analytictech.com/mb876/schedule.htm.
Please note that the schedule changes throughout the semester to reflect the needs and interests of the class. That's one of the key benefits of having it online. It is therefore your responsibility to check the online schedule constantly -- several times a week, especially the night before class. Do not print it out once and assume it is valid for the rest of the semester. And do not assume that I will announce in class when assignments are due. It is your responsibility to check the web. You have been warned!
IMPORTANT: Please bring your laptops to every class! This is a very hands-on kind of class. Also, be prepared to bring your own data to play with. These data will be shared among the class (with the understanding that no-one is to use it outside of class).
You should always feel free to email me (email@example.com) and to call me at home (781-400-5505, after 9am and before 10pm). I often work at home, so it is not an imposition to call me there. My school office is in Fulton 430.
|Copyright ©2005 Stephen P. Borgatti||Revised: January 22, 2007||Home Page|