STA 303H1S / 1002HS --
Methods of Data Analysis II
-- Winter 2012
Suggested Readings:
-
Logistic Regression:
-
Sheather:
Chapter 8 is good. Note that Sheather starts with binomial and then goes to binary outcomes while I'm doing the reverse.
Skip sections 8.2.3 and 8.2.4.
-
Agresti:
The chapter titled "Generalized Linear Models" is a good introduction to the
concept behind the logistic and Poisson regression models. The chapter titled
"Logisitc Regression" is good, but I like the approach in Sheather better.
(Chapter numbers vary with the edition of the book.)
-
Poisson Regression and Log-linear models:
-
Agresti:
Part of the chapter titled
"Generalized Linear Models" introduces the
the Poisson regression model.
There is relevant material in the section on chi-squared tests of independence
in the chapter titled "Contingency Tables".
In the chapter "Loglinear Models for Contingency Tables" you can skip the material on confidence intervals for odds ratios, 4-way and higher dimensional tables,
independence graphs and collapsibility, and modelling ordinal association.
- Repeated Measures / Mixed models:
-
Sheather:
Chapter 10: Read pages 331-349. You can ignore: the discussion on shrinkage on pp. 337-339, and empirical Bayes residuals on p. 346. You can also ignore Cholesky residuals and section 10.2. Note that AIC is defined in a different way in Sheather than how we've defined it to be consistent with SAS.