Biometry 896: Variance Components

Objective

Mixed models arise in many experimental situations. While a lot of attention is focused on fixed effects, variance components are also of interest. When examining fixed effects variance components play an important role in experimental design, testing, and estimation. In addition, variance components (and functions of them) are of interest in themselves. The amount of variability present in a population is important both in determining the opportunity to improve the population through selection and in quality control.

The purpose of this course is to develop an understanding of different methods of estimating variance components in linear mixed models. To accomplish this we will need to brush up on our matrix algebra and review mixed models. The necessary matrix algebra will be introduced as needed. Appendix M summarizes most of the results we will need for this course. An introduction to mixed models is contained in Chapter 1. Chapter 2 contains a brief history and is recommended reading.

We will be using a number of computer packages. PROC IML will be used for examining the underlying mechanics of the estimation procedures. PROC MIXED will be also used to estimate variance components. While standard statistical packages are often inadequate for the analysis of large data sets we will not be using any of the more specialized programs. It is also important to note that in many cases there may not be a program that is capable of conducting the analysis.

Course Info

Instructor:	Steve Kachman
Office:	104 D Miller Hall
Phone:	472-2903
Office Hours:	MWF 11:00-12:00 and by appointment
Text Book:	Variance Components, Searle, S.R., Casella G. and McCulloch, C.E. (1992)
Prerequisites:	Biometry 970 Linear Models Matrix Algebra Experimental Design
Homework:	Approximately once a week 20% will be deducted for each day late.
Exams:	Two exams, will be announced at least one week in advance
Final Exam:	Thursday, May 6 at 10:00 Revise travel plans accordingly.

Course Material

Course Syllabus
To view the Course Syllabus click on . If Acrobat Reader is not installed on your computer click on .

Overheads

	Screen Format	8.5x11 Format
"Complete"
Introduction to `IML`
One-way Classification
ANOVA
ML and REML
Prediction

SAS Modules
- IML basics, modules, and debuging: Small.sas
  To download Small.sas under Netscape right-click on Small.sas and select Save Link As.
- Sire data set and ML/REML program: sire.sas
- Base IML modules for 970: 970module.sas
Data Sets
Abbreviations
Homework
- Assignment 1 Due 1/19/99
- Assignment 2 Due 1/27/99
- Assignment 3 Due 2/19/99
- Assignment 4 Due 3/12/99
- Assignment 5 Due 4/7/99
Practice Problems
FTP Server
Support Material
- Biometry 970: Linear Models
- An Introduction to Linear Algebra
- Linear Algebra Web Notes
- Distribution Theory and Statistical Inference
  Web Notes based on Hogg, R.V. and Craig, A.T. Introduction to Mathematical Statistics, 5th edition, 1989, Macmillan.

Steve Kachman <skachman@unl.edu>

Last modified: Wed Dec 6 08:45:00 CST 2000