FSU Statistics STA 4202/5206 - Analysis of Variance-Design of Experiments Spring, 2002

Instructor: Duane Meeter 201D OSB 1-1:30 MTWTh meeter@stat.fsu.edu phone: 644-8402

Grader: Han Yu 104E OSB 1:25-3:30 W, 12:20-1:20 F yu@stat.fsu.edu phone: 644-1372

Text: Notes at Target Copy, 623 W. Tennessee St., about $6.

OUTLINE

Elements of a Designed Experiment

Completely Randomized Experiment or Study

Checking the Assumptions; Transformations

Multiple Comparisons

Contrasts

Randomized Block Design

Normal Probability Plots

Factorial Designs

Split Plot and Repeated Measures Designs

Hierarchical Designs

Latin Square Designs

Analysis of Covariance

GRADING

Data Analysis and Homework 50% (A = 93%, A- = 90%, B+ = 87%,

First Test 25% B = 84%, B- = 80%, etc.)

Last Test 25%

There will be four or five data analysis assignments. These will be written up as reports to an employer. Computations and graphs may be easily done in Minitab, and cut and pasted into a report. Minitab is available in the computer labs in Strozier, Carothers, and the Union. The student version may be ordered from www.e-academy.com/minitab/rental.cfm for $25.99/semester, or $99 forever. A trail copy may be downloaded for one month of use.

OBJECTIVES

I expect students to be able to

- State models, assumptions, advantages and disadvantages of a design

- Discuss seriousness of assumption violations and how to check for them

- Analyze data and write out results & conclusions in both technical and non-technical terms

- Given a verbal description or plan of experiment, describe design and ANOVA table

- Given a research hypothesis and experimental constraints, specify a design, including treatment levels, controls, and replication, which will answer the given question, or specify a regression study.

Original Work: Students in this course may assist each other with performing computer runs. They may even discuss with each other what graphs, plots, or analyses they have produced. All interpretations, writeups, discussion, etc. of their results must be their own work.