STAT 574: Survey Sampling I
Dr.
Richard Bolstein When: Fall, 2001
Office:157
Sci-Tech II Time: Tuesday 7:20-10:00 pm, first class Aug. 28.
rbolstei@gmu.edu Where: Robinson
Hall, A249
Phone: (703) 993-1689 Fax: (703) 993-1700
Office
Hours: M, T
4:30-6 pm & by appointment. Call or email anytime.
Prerequisite: STAT 344 or equivalent.
Corequisite: STAT 501 or experience using SAS statistical software
Note
to Undergraduates:
Students registered for STAT 474: Introduction to Survey Sampling, meet together with graduate students in STAT
574 but are graded independently. Although most assignments are the same in
both sections, there are some differences. A special program called the B.S./
Accelerated M.S. in Statistical Science
allows an undergraduate student to apply STAT 574 towards both degrees, thereby
reducing the number of graduate courses needed for the M.S. in Statistical
Science. There are certain other conditions that must be met so please check
the University
Catalog. If this program is of interest and you are currently registered
under STAT 474, you must drop/add to register for STAT 574.
Required
Texts:
1. Lohr, Sharon L. (1999): Sampling: Design
and Analysis, Duxbury Press.
2. SAS Institute (1999): Selected SAS Documentation for Survey
Sampling, prepared for A. Richard Bolstein.
3. Class notes and solution handouts will be
distributed from the instructor's notes.
Optional
Books and Software for using the SAS statistical package:
4. SAS Institute (1999): Selected SAS Documentation: Basic
Language and Procedures, prepared for A. Richard Bolstein.
5. Delwiche and Slaughter (1998): The Little SAS Book: a primer,
SAS Institute. (I highly recommend this unless you are a pro at SAS.)
6. Gilmore (1999): Painless Windows: A Handbook for SAS Users,
SAS Institute. (I recommend this if you are a relative beginner with
Windows.)
7. PC-SAS, Version 8.0 is available from Patriot Computers in Johnson
Center. A one-year licence costs $96.00. Do not purchase earlier versions as
they do not contain the new SURVEY procedures needed in this course.
Course
Description:
This course is the first of a two-semester sequence on the design and analysis
of sample surveys. (The second course, STAT 674: Survey
Sampling II, is offered in the Spring semester of even numbered years.) In the
first course, students will gain an overview of all aspects of designing and
implementing a survey, learn how to analyze survey data and practice proper
presentation of survey results. Emphasis is on the design of probability
samples and estimation of population parameters such as means, proportions and
totals. All students are required to participate in a class project involving
an actual survey. If anticipated funding is received, we will conduct a
statewide pre-election telephone survey for the Virginia gubernatorial election
in conjunction with some other classes.
Audience: Sample surveys occur
everywhere in the social sciences (especially economics, public affairs and
sociology), business (marketing and decision making), computer science,
environmental science, operations research and engineering. The course is
recommended to all students of mathematics, statistics or operations research,
and to quantitatively oriented students in the social sciences and business. A
two-semester sequence in survey sampling should definitely improve job
prospects for those seeking a career in survey research or statistics since the
demand for sampling statisticians in the Washington, D.C. area exceeds that of
any other branch of statistics.
Grades will be based on problem
sets and the class project (65%) and an in-class, open-book Final (35%).
Students must work alone on the problem sets.
Prerequisite: A solid course in
probability or statistics at the third-year undergraduate level that covers the
following material: discrete random variables including the binomial
distribution and the concepts of expectation, variance and covariance; the normal
distribution and its application to the basics of statistical estimation
including construction of confidence intervals and tests of hypotheses for
means and proportions; simple linear regression.
Computer
Skills:
Students will analyze real data sets using the computer and should have
experience with some statistical package or spreadsheet. The instructor and
textbook will supply some Excel, SAS, and Splus programs, and you may also use
your own software. The MS Windows version 8.0 of SAS is available in the
computing labs in ST1, Room 124, and ST2, Room 148. Splus is available in the
CSI lab on the second floor of ST1. A one year license for PC-SAS to use on
your peronal computer can be purchased for $96 at Patriot Computers in the
Johnson Center.
8/28 Ch.1 Design
and Implementation of Sample Surveys.
9/4 2.1-2.2 Fundamentals
of Probability Sampling and Estimation.
App. B
9/11 2.3-2.7,2.9 Simple
Random Sampling: The simple expansion estimator.
9/ 18 3 Simple
Random Sampling: Ratio Estimator.
9/25 3 Simple
Random Sampling: Regression Estimator. Domain Estimation.
10/2 4 Stratified
Sampling.
10/9 No Class (Columbus Day
break).
10/16 5.1-5.2 One-Stage
Cluster Sampling with Equal Probabilities.
10/23 5.6 Systematic
Sampling. Bernoulli Sampling.
10/30 Pre-election Poll Results. Review
and catch up.
11/6 5.3-5.5 Two-Stage
Cluster Sampling with Equal Probabilities.
11/13 6.1-6.3 Sampling
with unequal probabilities with replacement.
11/20 6.4-6.5 Sampling
with unequal probabilities without replacement.
11/27 Notes Estimation
of Proportions in Small Samples. Rare event estimation.
12/4 Review and preview of Survey
Sampling II.
12/11 Final Exam. 7:30-10:15 P.M.
* Additional reading
material from instructor's notes will supplement text material.
Additional Reference Books.
Biemer, P.P., et al (eds.)
(1991) Measurement Errors in Surveys, Wiley.
Cochran, W.G. (1977), Sampling
Techniques, 3rd ed., Wiley.
Cox, B.G., et al (eds.)
(1995), Business Survey Methods, Wiley.
Groves, R.M. (1988), Telephone
Survey Methodology, Wiley.
Groves, R.M. (1989), Survey
Errors and Survey Costs, Wiley.
Hansen, Hurwitz, and Madow
(1953), Sample Survey Methods and Theory, Wiley.
Jessen, R.J. (1978), Statistical
Survey Techniques, Wiley.
Kish, L. (1965), Survey
Sampling, Wiley.
Levy and Lemenshow (1991), Sampling
of Populations, Wiley.
Raj, D. (1968), Survey
Sampling, McGraw Hill.
Sarndal, Swenson, and
Wretman (1992), Model Assisted Survey Sampling, Springer-Verlag.
Schaefer, Mendenhall, &
Ott (1990), Elementary Survey Sampling, Duxbury.
Thompson, M.E. (1997), Theory
of Sample Surveys, Chapman & Hall.
Thompson, S.K. (1992), Sampling,
Wiley.
Tryfos, P. (1996), Sampling
Methods for Applied Research, Wiley.
Williams, B. (1978), A
Sampler on Sampling, Wiley.
Yates, F. (1960), Sampling
Methods for Censuses and Surveys, Griffen Publishing Co.
Cochran's book is considered the bible of the field and is
still widely used even though it is somewhat dated. The Raj book is outstanding
and is a major reference to some survey researchers, but it is out of print.
The book by Hansen, Hurwitz and Madow was the first devoted entirely to
sampling and is a classic. The book by Sarndal, Swenson and Wretman is modern
and comprehensive and a must for statisticians who work in sampling. (The text
by Lohr chosen for this course uses the same notation as Sarndal.) Kish (1965)
and Yates (1960) are also classics and contain material not in the other books.
Levy & Lemenshow (1991), Tryfos (1996), and Schaefer (1990) are written in
a cook-book style for undergraduates, but are useful sources. Williams (1978)
is written so that students with little mathematical background can get a feel
for the subject. Nevertheless, it is
good introductory reading for all and is highly recommended.
Groves (1988 and 1989) deal with the practical side of
selecting samples in the real world. They discuss current research in telephone
survey methodology and nonsampling errors in surveys. The book edited by Biemer
and others contain the proceedings of a conference on a type of nonsampling
error known as measurement error. The book edited by Cox and others contain
proceedings from a conference on surveys of businesses and is highly
recommended.