Texts:

1. Required:
   • Data Analysis Using Regression and Multilevel/Hierarchical Models Gelman & Hill
   • Local Course Notes
2. Recommended:
   • Faraway, Extending the Linear Model with R: Generalized Linear, Mixed Effects and Nonparametric Regression. Chapman Hall/ CRC, 2006
   • Venables and Ripley, Modern Applied Statistics with Splus Fourth Edition (MASS). Springer, 2004
   • Pinheiro and Bates, Mixed Effects Models in S and S-PLUS Springer, 2000

Topics:

1. First work through hierarchical models in Gelman & Hill chapters 12 -- 16.
2. Basic stat computing skills in R and SAS.
3. Examine the "usual" assumptions of modeling:
   • constant variance,
   • independence, and
   • normality
   and examine ways to handle exceptions to each.
4. Inferences based on normality.
5. Choices in model building: variable selection, multicollinearity, validation, power analysis, calibration.
6. Modern (meaning computer intensive) methods such as ridge regression, bootstrap, boosting, and lars.
7. Finally we'll go back to Gelman & Hill and delve deeper into hierarchical models.

Assessment:

Unless I ask directly for computer output, the output should be considered as secondary information to support the main report of your conclusions.
Homework is due in class on the day assigned. I will accept it up to 4 pm with no penalty, after that, I'll deduct 10% per day (no credit after I post solutions or return graded homeworks). Homework average will be penalized for poor attendance and or repeated tardiness. Working with others on the homework and discussing problems with others are encouraged, but you must understand everything you write down and turn in, and you each have to go through the struggle of learning the software programs.