STAT 402A: - Statistical Design and the Analysis of Experiments:
Agriculture and Biological Sciences
Spring 2015

Philip Dixon

Syllabus and other useful information

## DRAFT: Information here may change before the start of ths semester.

Useful Information

• Lecture: MWF 8-8:50am,
• Instructor: Dr. Philip Dixon, pdixon@iastate.edu, 2121 Snedecor, 4-2142
• Office Hours:
• Questions?: Please feel free to e-mail me (pdixon@iastate.edu) or anytime with questions or comments.
I often check e-mail in the evening, especially the night before homework is due. I don't promise to be up past 9pm, so I recommend you ask questions earlier rather than later.
• Class Web site: http://www.public.iastate.edu/~pdixon/stat402
Copies of the syllabus, homework assignments, SAS code and data sets will be available here.
• Text
SAS for Linear Models, 4th edition, ( Littell, Stroup, and Freund).
The fourth edition is important. The 3rd lacks an important chapter.
This will be supplemented by readings from various sources.
• Reserve materials (all 2 hour in Parks Library):
• Ramsey, F.L. and Schafer, D.W. The Statistical Sleuth
This was the text for Stat 401, sections A,B. If you have it, it provides an alternative treatment for simple ANOVA.
• Kuehl, R.O. (2000) Design of Experiments: Statistical Principles of Research Design and Analysis
I have used this as a text before. Good source of details.
• Goals:
• Understand the principles of experimental design.
• Be familiar with some standard experimental designs and how to analyze data from those designs.
• Be able to apply statistical design principles to novel problems.
• Prerequisites:
The official prerequisite is Stat 401. I will assume that you are comfortable with t-tests, 1-way ANOVA including contrasts, and assumptions, i.e. the concepts covered in Chapters 1 - 3, 5, and 6 of Ramsey and Schaefer, The Statistical Sleuth. I do not expect you to be able to do calculations by hand or remember the formulae (we'll use the computer for all that). I realize that some of you took 401 (or its equivalent) a while ago, so this material may be rusty. There is also considerable variation between sections of 401. Please ask questions if I am assuming something you haven't seen or don't remember. Office hours are a great place to review background material.
Homework: 100 pts
Two Midterms: 100 pts each
Final: 100 pts
• Computing:
This course will emphasize the use of SAS for statistical computing. We will heavily use proc glm and proc mixed. I will not emphasize traditional formulae for Sums-of-Squares and other quantities. Instead, I will focus on the principles of experimental design and analysis, the use of SAS to compute the needed quantities, and the appropriate interpretation of output. Data files and example SAS programs will be distributed via the class web pages.

For those wishing to use R, JMP, or SPSS, I will be happy to help during office hours.

It would be best if you had access to the SAS package on a relatively powerful computer. At ISU, SAS is available on many PC's, some MAC's, and on a campus-wide Linux server (sas.iastate.edu). A current list of public computing labs with SAS can be found by going to http://www.it.iastate.edu/labsdb/, entering SAS in the search box, then clicking the Search button. In addition, many departments maintain departmental computing labs where SAS is available.

If you want your own copy of SAS, the Statistics Dept. computing group has a multi-user student license. This is free for students on their personal machine. Go to http://www.stat.iastate.edu/resources/software/sas/ and read the section on "Student Windows SAS Installations". Installing SAS on an ISU-owned PC requires a license, which costs ca \$100 per year from Statistics. E-mail Kathy Shelley to get more information. Licenses are also available from the Agronomy Dept. These are cheaper if you're associated with Agron.

Finally, if you want to use SAS at home or in your office, but do not want to install a copy, you can use a Windows remote desktop connection to connect to sas.iastate.edu. SAS runs on the server and displays on your computer (the client).

Course Outline (details may change depending on class pace)
The course is structured around a series of commonly asked questions by people doing lab or field experiments. . To answer each requires developing one or more statistical concepts.
```
Week   Dates        Topic

1   Jan 12-16      Intro / How Many Replicates? (sample size and power)
2   Jan 19        ML King Birthday - no class
Jan 21-23     How many Replicates, continued
3   Jan 26-30     Should I subsample? (variance components)
4   Feb  2- 6     Subsampling, continued
5   Feb  9-13     How can I reduce unwanted variability? (blocking)
6   Feb 16-20     more blocking (Latin Squares and incomplete blocks)
7   Feb 23        Midterm I due (takehome, thru end of week 5)
may be delayed if lecture schedule gets behind
Feb 23-27     What does "interaction" mean? (factorial trt design)
8   Mar  2- 6     Factorial designs continued.
9   Mar  9-13     What's the right error? (split plot designs)
Mar 16-20     Spring Break
10   Mar 23-27       split plots, continued
11   Mar 30-Apr 3    other designs with multiple error terms
12   Apr  6-10     Any problem with repeatedly measuring the same plots?
13   Apr 13        Midterm II due (takehome, thru end of week 11)
Apr 13-17     How do I analyze data from multiple sites and years?
14   Apr 20        Likely no class
Apr 23-25     Should I worry about spatial correlation?
15   Apr 27-May 1  Should I measure a baseline value? (ANCOVA)
May  4-May 8  Finals week (final will probably be takehome)
```

Details:

Readings: Assigned readings will be posted regularly. My lectures will cover the same concepts, but I will use different examples and a different presentation. It will help if you read the assigned material before the appropriate lectures.

Supplemental readings, from other books, will be made available when useful. These are optional. They will cover the same topics as the assigned readings and are given for your information, in case you prefer another book's style.

Homework: Statistics is best learnt by doing. The homework problems give you the chance to do. Problems are written to give you practice analyzing data or to illustrate points made in lecture. The intent is to help you understand and apply lecture concepts. Discussion with friends and classmates is encouraged. Copying papers does not contribute to learning and will not be tolerated.

Homework will be assigned weekly, except when there is a midterm, and due at the start of Friday's lecture. No late homework will be accepted. The lowest homework score will be dropped.

Exams: The two midterms will be takehome exams. They will be handed out at the Wednesday lecture and are to be returned at the following Monday's lecture. If my lectures are slower than scheduled, the midterm may be delayed. The format of the final will be discussed in class. I prefer to give a takehome exam, but as a class, we may decide on a traditional 2 hour in-class exam.

All exams are open book. You may use your textbook, your lecture notes, or other material from the library reserve desk. You may not use your friends or classmates. Please e-mail me if you have questions. This includes questions about SAS coding. Ask me, not a classmate! I'll help you.

• Midterm I is tentatively scheduled for Weds, Feb 18, - Mon, Feb 23 , covering material through end of week 5 (blocking), but probably not incomplete blocks or Latin Squares.
• Midterm II is tentatively scheduled for Weds, Apr 8, - Mon, Apr 13 , covering material through end of week 11 (probably designs with multiple errors).
• Final: Date and time depend on format and the registrar's schedule
Makeup exams or alternate due dates will be given only if you contact me and get approval prior to the scheduled exam.