George Mason University
Volgenau School of Information Technology and Engineering
Department of Statistics


STAT 657: Nonparametric Statistics

Fall Semester, 2018
Thursdays from 7:20 to 10:00 PM (starting Aug. 30, with other dates given below)

Location: room B203 of Robinson Hall


Instructor: Clifton D. Sutton

Contact Information (phone, fax, e-mail, etc.)
Office Hours: 6:15-7:00 & 10:00-10:30 PM on class nights (more information)


Text: Nonparametric Statistical Inference, 5th Ed., by J. D. Gibbons and S. Chakraborti (CRC Press, 2011)

(This book should be considered as required, as some of the homework problems will come straight from this book, and my lectures will be based heavily on the book.)

Software: StatXact 11

(Read the bottom portion of this web page in order to learn how you can obtain a copy of the required software at a very low price. This software should be considered as required, as I think it will be very hard to do a lot of the homework and pass the course without having access to StatXact.)


Prerequisite: STAT 544 and STAT 554


Description:

This course is a graduate-level course on nonparametric statistics for students who already have a decent knowledge of basic statistics and probability. The focus is on standard nonparametric procedures useful for the analysis of experimental data. One-sample, two-sample, matched pairs, one-way layout, and two-way layout procedures are covered. Tests for lack of independence, tests of randomness, and goodness-of-fit tests are also covered. Applications are emphasized, but theory is not completely neglected. State-of-the-art software for exact nonparametric inferences is to be used throughout the semester. (Note: Nonparametric density estimation, nonparametric methods for classification, nonparametric regression, nonparametric methods for survival analysis, and computer-intensive nonparametric methods such as bootstrapping are not covered. These are topics that should be covered in other courses, and there isn't enough time to cover them adequately in this course. Also, additional prerequisites would make sense for some of these topics.)

Basically, the course will cover, in order, most of the first 13 chapters of the text book (except parts of Ch.1 will be skipped, and material from Ch. 13 will be moved up and inserted sooner). Additional material related to StatXact will supplement the material from the text book. (Note: Ch. 14 of the text book will not be covered since it is material that should be covered in GMU's graduate-level course on categorical data anlysis.)

(Approximate) Class-by-Class Content:

(I think that during the first portion of the semester we'll be behind the pace indicated below, but then we'll catch up and for most of the semester we'll be ahead of the pace indicated below.)
[1] Aug. 30:
introduction (class policies, strengths and limitations of nonparametric procedures); review of some basic probability concepts with an emphasis on order statistics (including distributions of order statistics)
[the first portion of Ch. 1 of the text and the first portion of Ch. 2 of the text]
[2] Sep. 6:
more on order statistics (with an emphasis on asymptotic results); tolerance limits
[the second portion of Ch. 2 of the text]
[3] Sep. 13:
tests about randomness (tests based on runs for lack of independence, tests for trend)
[the first portion of Ch. 3 of the text]
[4] Sep. 20:
more varieties of tests about randomness
[the second portion of Ch. 3 of the text]
[5] Sep. 27:
goodness-of-fit tests
[the first portion of Ch. 4 of the text]
[6] Oct. 4:
more varieties of goodness-of-fit tests
[the second portion of Ch. 4 of the text and some supplementary material]
[7] Oct. 11:
one-sample and paired-samples procedures (inferences about quantiles, the sign test, rank-order statistics, the signed-rank test)
[Ch. 5 of the text]
[8] Oct. 18:
more on one-sample and paired-samples procedures (the one-sample normal-scores test, Fisher's permutation test, and StatXact's handling of ties), locally most powerful tests and ARE
[part of Ch. 13 of the text and some supplementary material]
[9] Oct. 25:
the general two-sample problem (with an emphasis on omnibus tests)
[Ch. 6 of the text]
[10] Nov. 1:
linear rank statistics based on two samples; linear rank tests for the two-sample location problem
[Ch. 7 and Ch. 8 of the text]
[11] Nov. 8:
tests for the two-sample scale problem
[Ch. 9 of the text]
[12] Nov. 15:
tests for the general k-sample problem (tests about the equality of 3 or more distributions), more procedures for 3 or more samples (tests based on pairwise comparisons, comparisons with a control, tests against monotone alternatives)
[Ch. 10 of the text and some supplementary material]
[**] Nov. 22:
(No class due to Thanksgiving Break)
[13] Nov. 29:
measures of association for bivariate samples and tests about the independence of two variables (Kendall's tau coefficient, Spearman's coefficient of rank correlation, associated tests for lack of independence)
[Ch. 11 of the text]
[14] Dec. 6:
nonparametric procedures for two-way layouts (Friedman's test, Page's test, coefficients of concordance)
[Ch. 12 of the text]
[**] Dec. 13:
Final Exam (note: exam period is from 7:30 to 10:15 PM)

Learning Outcomes:


Grading:


Additional Comments: