STAT368 handout

STAT 368: Introduction to Design and Analysis of Experiments

INSTRUCTOR: Ivan Mizera (contact); office hours MW 13:00-15:00 in CAB 441, or by appointment.
TIME AND VENUE: MWF 10:00-10:50, CAB 273.
CALENDAR DESCRIPTION AND PREREQUISITES: Basic principles of experimental design, completely randomized design-one way ANOVA and ANCOVA, randomized block design, Latin square design, Multiple comparisons. Nested designs. Factorial experiments. Prerequisites: STAT 266 and a course in Linear Algebra; MATH 225 recommended.
COURSE OBJECTIVE: To give students solid foundations in understanding techniques of design and analysis of experiments with response involving uncertainty, with strong emphasis on hands-on computational experience.
TEXTS: Statistics for Experimenters: Design, Innovation, and Discovery, 2nd Edition by G. E. P. Box, J. S. Hunter, W. G. Hunter. Wiley, 2005. (Required).
An Introduction to R, by W. N. Venables, D. M. Smith and the R Development Core Team. (Recommended. Available freely from the Internet, as well as other R documentation and texts.)
Design and Analysis of Experiments by D. C. Montgomery. Wiley, 2008 (7th edition). (Supplementary. A more mathematical textbook, available in many editions.)
ASSIGNMENTS AND DATES: Homework 35% (five assignments, equally weighted: due Jan 21, Feb 4, Mar 6, Mar 20, Apr 3)
Midterm exam 15% (Monday February 25, in class)
Project 20% (report due April 12)
Final exam 30% tentatively looks like Tuesday, April 23, 9:00-12:00.
There are no classes February 18-22 (reading week), March 29 (Good Friday), April 1 (Easter Monday).
IMPLEMENTATION OF THE GRADING SYSTEM: The raw cumulative score will be computed at the end of the term, with the weights as indicated above, and used to determine the letter grade using the combination of absolute and relative standing as follows: everybody with 91% or more percent will be guaranteed to receive A; everyone with 76% or more B; everybody with 51% or more C. The standards for + and - modifications will be set similarly. All the cutpoints may be somewhat lowered (but not raised), should the situation require that.
Samples of past or representative evaluative course material will be made available through the course web page.
COMMENTS ON THE WORK IN THE COURSE: Attendance: although no marks are given for class participation, students are expected to attend classes and actively participate in discussions.
Homework: the weight of the assignments signals that it may take a couple of days to complete; plan accordingly. The papers do not have to be typed, but must be readable and document all the steps necessary to obtain the solution.
Exams: closed book, closed notes. The policy regarding use of simple calculators will be announced if necessary.
Project: will consist in designing, carrying out, and analyzing an experiment, and submitting a printed report documenting all its aspects: the purpose of the experiment, the design used, the analysis of the experimental output and the conclusions. The evaluation of the project will be based on originality, relevance of the subject to the course, quality and sophistication of the statistical analysis, and clarity and professional appearance of the report; the latter is normally supposed not to extend 10 printed pages (and not to contain any direct computer output except for the relevant graphical displays). The work on the projects will be done in groups of up to three students; the composition of the groups has to be finalized before the Reading Week. Rather than a one-time assignment, the work on the projects should be seen as a continuing process; I will be happy to discuss plans and look at the partial results at any time.
SOFTWARE: All computations in the course (textbook, lectures, exams) will be done in the statistical language R, an open source software available at CRAN and its mirrors. R may be downloaded and installed for free, on a variety of platforms; it is also available in CAB 341 and 345 labs. The use of R is aimed at gaining a working understanding of the course concepts, and allows for possible bypassing of cumbersome mathematical details without compromising the quality of the learning process.
The computations may require installing certain add-on packages, either from CRAN, or from my private repository; the way how to do it is described in Lecture 2. The package BHH22, developed specially for and during the course, should be reinstalled frequently, so that students use always the latest version.
TOPICS: General paradigm of deductive-inductive iterative reasoning. The components of experiment: response, factors, replication.
Comparing two entities: randomization and blocking. Review of significance tests. Permutation tests, their approximations by random sampling. Tests using normal and t distributions. Discrete and count data.
Estimation by least squares, decomposition of the sum of squares, ANOVA tables. Randomized experiments with a simple factor. Graphical analysis of variance. Inferences: tests, multiple comparisons, confidence intervals.
Randomized block designs, Latin and Graeco-Latin squares. Incomplete blocks.
Factorial designs. Analysis of variance. Two factors, Tukey test of additivity. General factorial design. Factorial designs with two levels. Blocking and confounding.
Fractional factorial designs, nested and split-plot designs, experiments with random effects.
Also, some prerequisite concepts will be reviewed, if necessary: these include notions of random sampling, frequency distribution, probability density, expectation, covariance, and correlation, and various probability distributions.
STAT 501: Students taking this course as STAT 501 will be graded separately. Otherwise, the same rules apply for them, except for the project, which they will do individually and besides the report prepare also a presentation, which will count for 5% (out of the total 20%) of the mark for the project.
LINKS: Lecture transparencies (will be available after classes): 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22; Assignments: 1 (solutions) | 2 (solutions) | 3 (solutions) | 4 (solutions) | 5 (solutions); Midterm review sheet | Midterm warm-up |; STAT 368 at U of A eClass (for grades) | Normal probability paper
FINE PRINT (OBLIGATORY ANNOUNCEMENTS): There are neither deferred term exams nor extended deadlines for the homework. A student who cannot complete the assignment or misses a term examination because of an incapacitating illness, severe domestic affliction or other compelling reasons can apply for a deferral of the missed component weight to the final exam.
A student who cannot write the final examination because of an incapacitating illness or is suffering from severe domestic affliction or other compelling reasons can apply for a deferred final examination. Such an application must be made to the student's Faculty office within 48 hours of the missed examination and must be supported by a Statutory Declaration (in lieu of a medical statement form) or other appropriate documentation. (University of Alberta Calendar 23.5.6.) If granted, the deferred final exam will be held from 9:00 to 12:00 on Saturday, May 4, 2013 (meeting outside of CAB 357).
The University of Alberta is committed to the highest standards of academic integrity and honesty. Students are expected to be familiar with these standards regarding academic honesty and to uphold the policies of the University in this respect. Students are particularly urged to familiarize themselves with the provisions of the Code of Student Behaviour and avoid any behaviour which could potentially result in suspicions of cheating, plagiarism, misrepresentation of facts and/or participation in an offence. Academic dishonesty is a serious offence and can result in suspension or expulsion from the University. All forms of dishonesty are unacceptable at the University. Any offence will be reported to the Senior Associate Dean of Science who will determine the disciplinary action to be taken. Cheating, plagiarism and misrepresentation of facts are serious offenses. Anyone who engages in these practices will receive at minimum a grade of zero for the exam or paper in question and no opportunity will be given to replace the grade or redistribute the weights. As well, in the Faculty of Science the sanction for cheating on any examination will include a disciplinary failing grade (no exceptions) and senior students should expect a period of suspension or expulsion for the University of Alberta.
Cell phones are to be turned off during lectures, labs and seminars; they are not to be brought to exams.