Learning a new statistical technique can be like acquiring
the proverbial hammer -- all problems then resemble nails.
Lurking next is the most common statistical error: asking
the wrong question (because you want to apply your
newly-learned technique). Two antidotes:
1. Our course in Generalized Linear Models (GLM), which
gives you a perspective on a number of different techniques
in a single theoretical framework.
April 1-29 with Joe Hilbe and James Hardin.
2. Our course in Categorical Data Analysis, which approaches
analysis from the other end - recognize the type of data you
have, and then assess what different techniques are appropriate.
April 1-29 with Brian Marx.
GLM: GLM extends ordinary least squares (OLS) regression to
incorporate responses other than normal. This course will
explain the theory of generalized linear models (GLM), outline
the algorithms used for GLM estimation, and explain how to
determine which algorithm to use for a given data analysis.
Continuous response variables, the log normal, gamma,
log-gamma (survival analysis), and inverse Gaussian cases
are covered. Binomial (logit, probit, and others) as well
as count models (poisson, negative binomial, geometric) are
Joe Hilbe and James Hardin are the co-authors of “Generalized
Linear Models and Extensions” (Stata Press) as well as
“Generalized Estimating Equations” (CRC Press). They have
lectured widely in these areas, and have been instrumental in
developing computer routines for these methods – routines that
have been incorporated into popular statistical software programs.
Details are here
Categorical Data Analysis: This course will cover the analysis
of contingency tables (where cells represent counts of subjects
or items falling into certain categories). Topics include tests
for independence (comparing proportions as well as chi-square),
exact methods, and treatment of ordered data. Both 2-way and
3-way tables are covered. A modeling approach to categorical
data analysis will also be presented, as special cases of the
GLM, specifically Poisson regression for count responses and
logistic/probit regression for binomial responses. The focus
will be on interpretation of models rather than the theory
Brian Marx is Professor of Statistics at Louisiana State
University, and has taught Categorical Data Analysis for
over ten years. He is currently serving as Chair of the
Statistical Modeling Society and is the Coordinating Editor
of “Statistical Modeling: An International Journal.”
Details are here
Participants can ask questions and exchange comments with the
instructors via a private discussion board throughout the
period. Each course requires about 15 hours a week and there
are no set times when you are required to be online.
Hope to see you online at statistics.com!
You may leave the list at any time by sending the command
to [log in to unmask], leaving the subject line blank.