(441-442 or 861-862 at MSU). Linear models are among the most frequently used, well and poorly, of statistical models. Simple and multiple regression analysis and the analysis of variance are statistical methodologies based on linear models. These methods provide a means of studying the relationship between one dependent variable and one or more explanatory variables when the dependent variable is also affected by some random error.
Chapter headings are: 1) Linear Algebra, Projections, 2) Random Vectors, 3) The Linear Model, 4) Fitting of Regrssion Models, 5) Simultaneous Confidence Intervals, 6) Two and Three Way Analysis of Variance, 7) Miscellaneous Other Methods. Another chapter, 8) Log-linear models, on the analysis of frequency data, is also included in the text. That material will be discussed in Statistics 842, to be taught Spring Semester by the same instructor.
The theory depends substantially on the very intuitive theory of orthogonal projections
in n-space, and employs linear algebra, probability, and statistical inference.
Applications wiil be introduced as the theory is developed.
PC's in the Computer Laboratory, and the software package S-Plus will be used to
illustrate the theory and perform the computations. No previous experience with the
software is necessary.
Grades will be based on the number of points accumulated, with perfect scores being
approximately as follows:
Problem Sets 300
2 Mid-Term Exams 200
Take-Home Final Exam 200
Total 700
Copies of old mid-terms exams will be distributed before the corresponding exams in this course. Problem sets may be of many different types, from just one or two problems due the next class meeting to much longer collections of problems due a week or more later.