|
March 1 - 5, 1999
Adam's Mark Hotel at the Florida Mall
1500 Sand Lake Road
Orlando, Florida 32809
Categorical Data Analysis
Instructor : Alan Agresti
Start Monday (3/1/99) 8:30 a.m., End Wednesday (3/3/99) Noon
Survival Analysis
Instructor : P.V. Rao
Start Monday (3/1/99) 8:30 a.m., End Wednesday (3/3/99) Noon
Research Design: The Art of Doing Science
Instructor : Ronald Marks
Start Monday (3/1/99) 8:30 a.m., End Wednesday (3/3/99) Noon
Experiments with Mixtures
Instructors : John Cornell and Greg Piepel
Start Monday (3/1/99) 8:30 a.m., End Wednesday (3/3/99) Noon
Analysis of Repeated Measures and Longitudinal Data
Instructors : Jane Pendergast and Ramon Littell
Start Wednesday (3/3/99) 1:00 p.m., End Friday (3/5/99) 4:00 p.m.
Generalized Linear Models
Instructor : James Booth
Start Wednesday (3/3/99) 1:00 p.m., End Friday (3/5/99) 4:00 p.m.
Epidemiology Methods with Applications to the Pharmaceutical Industry
Instructor : Brad Pollock
Start Wednesday (3/3/99) 1:00 p.m., End Friday (3/5/99) 4:00 p.m.
Response Surface Methodology
Instructor : Geoff Vining
Start Wednesday (3/3/99) 1:00 p.m., End Friday (3/5/99) 4:00 p.m.
Registration Form
Lodging Information
The University of Florida
College of Liberal Arts and Sciences
Department of Statistics
The Department of Statistics at the University of Florida is among the largest in the country with over thirty faculty members, many of whom are
internationally recognized, and more than fifty graduate students. The department prides itself on the balance of theory and applications taught in
its curriculum. The strengths of the department include Categorical Data Analysis, Nonparametric Statistics, Biostatistical Methods, Linear
Models, and Design of Experiments. With a single exception, the instructors teaching these courses are members of the department and have
extensive experience both as consultants to government and industry and as instructors.
Instructors
Alan Agresti (Professor) is the author of the texts Analysis of Ordinal Categorical Data, Categorical Data Analysis, An Introduction to Categorical
Data Analysis, and Statistical Methods for the Social Sciences. He is a Fellow of the American Statistical Association (ASA). He has presented
numerous short courses worldwide on this topic to industry (particularly pharmaceuticals), professional organizations and universities.
James Booth (Associate Professor) has recently had great success developing and teaching a graduate level course in generalized linear models
for the department. Professor Booth is an active researcher in statistical methodology and has published numerous articles on a variety of
statistical topics.
John Cornell (Professor) is a consulting statistician with the Florida Agricultural Experiment Station. He is the author of the text Experiments
with Mixtures and coauthor the the text Response Surfaces. He is past recipient of the Youden and Brumbaugh Awards as well as the Shewell
prize from the American Society for Quality Control (ASQC). He is a fellow of the ASQC and ASA and is an elected member of the International
Statistical Institute (ISI). He is a past Editor and current member of the editorial review board of the Journal of Quality Technology.
Ramon C. Littell (Professor) is a consulting statistician with the Florida Agricultural Experiment Station. He is the co-author of the texts SAS
System for Linear Models and SAS System for Regression. Professor Littell is a Fellow of the American Statistical Association (ASA), and is an
associate editor for the Journal of Agricultural, Biological, and Environmental Statistics.
Ronald Marks (Professor) is Director of the Division of Biostatistics in the Health Science Center at the University of Florida. He has extensive
experience in the design and conduct of large clinical trials and epidemiological research. He has taught a popular two semester sequence in
research design and analysis methods to graduate students from the six Health Science Center colleges and other colleges across campus. He has
published extensively in the dental, opthalmologic, and geriatric areas.
Jane Pendergast (Research Associate Professor) is a member of the Division of Biostatistics and has extensive consulting and collaborative
research experience within the health sciences. She has co-authored two tutorials on using the SAS system to analyze repeated measures and
longitudinal data. She has published on a wide variety of topics including a recent review of methods applicable to correlated/longitudinal binary
outcome data. Dr. Pendergast currently serves on the Board of Directors of the ASA and the Regional Advisory Committee of the Eastern North
American Region of the International Biometric Society.
Greg Piepel is a Staff Scientist in the Statistics Group at Battelle, Pacific Northwest National Laboratories in Richland, WA. He has developed,
applied, and published mixture experiment techniques for a wide variety of applied problems over the last 20 years. He is the author of the
MIXSOFT mixture experiment software and, with John Cornell, received the 1994 Brumbaugh Award for a paper reviewing and recommending
various mixture experiment approaches. Dr. Piepel is a member of the editorial review board of the Journal of Quality Technology, and is a past
chair of the ASA Section on Physical and Engineering Sciences.
Brad Pollock (Associate Professor) is the Director of the Division of Epidemiology at the University of Florida College of Medicine. For the past
nine years he has served as a statistician and director of the epidemiology and cancer control research programs of the Pediatric Oncology Group, a
consortium comprised of approximately half of the North American centers that treat childhood cancer. He has extensive experience in clinical
trials and the application of epidemiology to clinical medicine.
P.V. Rao (Professor) consults extensively within the University of Florida Health Science Center. As a statistician in the Pediatric Oncology
Group he has considerable experience in the application of survival analysis methods to studies on the treatment of childhood cancers. An elected
member of the ISI, Professor Rao has published extensively in the theory, methodology, and application of statistics.
Geoff Vining (Associate Professor) is the author of the text Statistical Methods for Engineers. He is a past recipient of the Brumbaugh Award
from the ASQC and a past recipient of a College of Liberal Arts and Sciences teaching award. A Senior Member of the ASQC, Dr. Vining is the
current editor of the Journal of Quality Engineering. Dr. Vining is an active industrial consultant and a frequent presenter of industrial short
courses.
Dennis Wackerly (Professor, Coordinator) is a coauthor of the best selling text Mathematical Statistics with Applications. Professor Wackerly
has extensive consulting experience with several organizations including Harris Semiconductor. He is a past recipient of a College of Liberal Arts
and Sciences teaching award. Professor Wackerly oversees a Week of Short Courses and may assist in teaching some of these courses as
needed.
COURSE DESCRIPTIONS
Categorical Data Analysis
March 1-3: Alan Agresti
The past two decades have seen an explosion in the development of new methods for analyzing categorical responses. This short course is
designed to introduce attendees to these methods, focusing on the most useful of the new model-building procedures. Attendees will learn how to
fit and interpret logistic regression and loglinear models. Specialized models are described for ordinal responses, repeated measurement data and
small samples. The course also covers older but still important methods such as chi-squared, Mantel-Haenszel, McNemar, and exact tests.
Several examples will be given of the analysis of real data sets and the use of computing software (primarily SAS) for conducting analyses.
No previous knowledge of categorical data analysis is necessary. Some models will be taught by analogy with regression and analysis of variance
(ANOVA) models, so prior background in these methods is helpful. The text Categorical Data Analysis, by Agresti, (Wiley, 1990) is required. If
you do not have one, please order it on the registration form.
Main topics covered are:
1.Two-Way Contingency Tables
2.Logistic Regression Models
3.Loglinear Models
4.Model-Building Strategies
5.Models for Ordinal Responses
6.Repeated Measurement Models
7.Exact Small-Sample Methods
8.Computer Software
Survival Analysis
March 1-3: P.V. Rao
This short course will cover statistical methods for analysis and interpretation of survival data. Even though survival analysis can be used in a
wide variety of applications (e.g. engineering, sociology and insurance), the emphasis of this course will be on methods suitable for analyzing
clinical trials data. Starting with the notions of survival and hazard functions, the attendees will be introduced to the methods of estimating
parameters and testing hypotheses that are currently being used in the analyses of survival data. Familiarity with the use of multiple linear
regression analysis and such basic nonparametric tests as the Wilcoxon Rank Sum test will be assumed. The text, Statistical Methods for
Survival Data Analysis , 2nd Ed., by E.T. Lee (Wiley, 1992) is required. If you do not have a copy, please order it on the registration form.
Main topics covered are:
1.Survival and hazard functions
2.Types of censoring
3.Estimation of survival and hazard functions: the Kaplan-Meier and life table estimators
4.Comparison of survival functions: The logrank and Mantel-Haenszel tests
5.The proportional hazards model: time independent and time dependent covariates
6.The logistic regression model
7.Methods for determining sample sizes.
Research Design: The Art of Doing Science
March 1-3: Ronald Marks
This course will cover the fundamental issues necessary to consider in designing a biomedical research project. The course is geared toward
medical professionals (physicians, nurses, dentists, etc.) who are involved in (or responsible for) research protocol development or who wish to
critically evaluate published research. There is no statistical prerequisite but experience or interest in research is essential. The course will
present a number of important design issues essential in planned research and emphasize some of the important choices or decisions that the
researcher must make to ensure quality and avoid "fatal flaws" in the research. The course will also illustrate how appropriate data analyses are
tied to an understanding of research design.
No text is required. Comprehensive course notes will be provided. Main topics covered are:
1.Study types
2.Data types
3.Data form generation
4.Sampling methods and bias
5.Repeated measures and block designs
6.Power analysis (sample size determination)
7.Randomization
8.Linking study design to analysis
Experiments with Mixtures
March 1-3: John Cornell and Greg Piepel
People perform mixture experiments daily. Imagine, if you will, adding sugar or sweetener to coffee or tea to improve the flavor; mixing corn oil and
vinegar to make salad dressing; or, in the interest of getting better mileage or performance, adding 93 octane fuel to the fuel tank of the family car
that presently contains some 87 or 89 octane fuel. These are all mixture experiments.
This course will benefit anyone involved in the development of mixture formulations or blends. We will focus on the design and analysis of mixture
experiments by presenting the most frequently used statistical techniques for designing, modeling, and analyzing mixture data. These techniques
will cover mixture experiments in which the ingredients or components can vary from 0 to 100% and also cases where the components are
constrained by lower and upper bounds and/or where there are multicomponent constraints. The inclusion of process variables and/or a total
amount variable in mixture experiments will also be covered. Designs and model forms for use in these combined experiments will be presented.
Recommendations will be made and warnings given about relying too heavily on the available software packages for generating optimal designs.
Real examples of each type of mixture experiment will be presented and discussed. Participants should have some previous exposure to
non-mixture experimental designs (factorials, fractional factorials, response surface types) and least squares regression. The textbook required
for the course is Experiments with Mixtures, 2nd Ed., by Cornell (Wiley, 1990). If you do not have a copy, please order one on the registration
form.
Main topics covered are:
1.Introduction to Mixture Experiments
2.Designs and Models for Exploring the Whole Simplex
3.Model Building and Data Analysis Strategies
4.Additional Constraints on the Components
5.Component Effects and the Response Trace
6.Other Mixture Models
7.Designs and Models for Including Process Variables
8.Mixture-Amount Experiments
9.Recent Topics (Biplot Displays, etc.)
10.Discussion of Available Software
Analysis of Repeated Measures and Longitudinal Data
March 3-5: Jane Pendergast and Ramon Littell
Repeated measures and longitudinal data require special attention because they involve correlated data that commonly arise when the primary
sampling units are measured repeatedly over time or under different conditions. Normal theory models for split-plot experiments and repeated
measures ANOVA will be used to introduce the concept of correlated data. PROC GLM and PROC MIXED in the SAS system will be illustrated
using practical examples. Mixed linear models will be discussed. These models provide a general framework for modeling covariance structures, a
critical first step that influences parameter estimation and tests of hypotheses. The primary objectives are to investigate trends over time and how
they relate to treatment groups or other covariates.
Models and estimation methods applicable to non-normal data will also be presented. An overview of the generalized linear model (GLM) will
serve as background for extensions which can accommodate correlated observations. (See the description of the Generalized Linear Models
course below.) In particular, the Generalized Estimating Equation (GEE) methodology and the the Generalized Linear Mixed Model (GLMM)
will be discussed. In the situation where the mean at each time point can be modelled using a GLM, the GEE approach utilizes the correlation over
time in the estimation scheme, providing more efficient estimators. The GLMMs are extensions of the mixed linear model described above to the
larger family of distributions covered by the generalized linear model, such as the binomial, Poisson, normal, gamma, and inverse Gaussian
distributions. GEE examples will be illustrated using the SAS PROC GENMOD, and (time-permitting) the gee() function in Splus. Examples of
GLMMs will be illustrated using the SAS GLIMMIX macro
Participants should be familiar with basic matrix algebra and should have some knowledge of regression and analysis of variance. No textbook will
be required. Detailed course notes will be provided.
Main topics covered are:
1.Balanced split-plot and repeated measures designs
2.Modeling covariance structures of repeated measures
3.Repeated measures with unequally spaced times and missing data
4.Generalized linear model (GLM)
5.Generalized estimating equation (GEE) method for marginal models
6.Generalized linear mixed model (GLMM).
7.Computer implementation using S-plus and the SAS system.
Generalized Linear Models
March 3-5: James Booth
The generalized linear model is possibly the most important development in practical statistical methodology in the last twenty years. Generalized
linear models provide a versatile modeling framework in which a function of the mean response is "linked" to appropriate covariates through a
linear predictor and in which variability is described by a distribution in the exponential dispersion family. These models include logistic regression
and log-linear models for binomial and Poisson counts together with normal, gamma and inverse Gaussian models for continuous responses.
This course will provide an introduction to generalized linear models. Practical examples involving real data will be used to introduce different
topics, and attendees will learn how to perform the analyses using the SAS computer package. No previous knowledge of generalized linear
models will be assumed, but a background in normal theory linear models is required. Generalized Linear Models by McCullagh and Nelder
(Chapman & Hall, 1989) is a useful reference but is not required. If you would like a copy, please order it on the registration form.
Main topics covered are:
1.Review of normal theory linear models
2.Definition of generalized linear models (GLMs)
3.Inference and diagnostics for GLMs
4.Binomial regression
5.Poisson regression
6.Methods for handling overdispersion
7.Generalized estimating equations (GEEs)
8.Generalized linear mixed models: subject specific versus population average inferences (time permitting)
Epidemiology Methods and Applications to the Pharmaceutical Industry
March 3-5: Brad Pollock
This course will provide a general introduction to epidemiology methods and their application to the pharmaceutical industry. Assessment of the
impact of medical exposures in treated populations and alternate study designs including clinical trials and observational designs will be discussed
along with their inherent advantages and disadvantages, and the strength of evidence provided by each. Measurement problems unique to the
pharmaceutical industry including the assessment of the validity of exposure to new drugs and disease outcome will be presented. Examples of
new drug and new diagnostic test assessment will be provided in such clinical areas as such as cancer, heart disease and AIDS. The textbook
Pharmacoepidemiology, 2nd Edition edited by Strom, B.L. will be used (Chichester, England: John Wiley & Sons, 1994).
No previous knowledge of epidemiology is necessary. The text Pharmacoepidemiology, 2nd Edition edited by Strom, B.L., (Wiley, 1994) is
recommended. If you wish, you may order the text on the registration form.
Main topics covered are:
1.Definition of basic epidemiology concepts
2.Measures of effect, exposure assessment and outcomes
3.Alternate study designs for clinical research: observational versus experimental approaches
4.Sources of bias
5.Integration of laboratory assessments into clinical research studies
6.Statistical associations, weight of evidence, and causality
Response Surface Methodology
March 3-5: Geoff Vining
This course is intended for people who have had some exposure to experimental design. Attendees will learn the basics of response surface
methodology (RSM). RSM is a set of tools for planning and analyzing experiments to fit models for a response or responses of interest. Analysts
then can use these models for process optimization.
This course discusses designs and analyses for both first and second-order models. This course then illustrates how we can use these models for
process optimization. Real life examples illustrate these techniques. Attendees will get hands-on experience applying these methodologies
through two in-class projects. The textbook required for the course is Statistical Methods for Engineers, by Vining (Duxbury, 1998). If you do not
have a copy, please order one on the registration form.
Main topics covered are:
1.An Overview of Experimentation
2.The 2k Factorial Design
3.Fractional Factorial Designs
4.The Catapult Experiment
5.Sequential Experimentation
6.Path of Steepest Ascent
7.Second-Order Experiments
8.Optimizing a Single Response
9.Optimizing More than One Response
10.The Distillation Column
11.Using RSM in Robust Parameter Design
GENERAL INFORMATION
Registration Form
Fee Structure
Statistics Week of Short Courses : $1695 includes 2 courses (2.5 days each), 5 continental breakfasts, 2 breaks with refreshments each day, 4
lunches (M,T,Th,F), 5 evening receptions. Each individual 2.5 day course is available for the fee of $895. Detailed course notes provided for all
courses. Required or recommended texts available at additional fee. See the registration form for details.
Lodging
A block of rooms has been reserved at the Adam's Mark Hotel, at the Florida Mall, Orlando, Florida. The special room rate is $95 single or
double, plus taxes.
The Florida Mall, has over 200 specialty stores, department stores and 30 eateries. The Mall also includes two fine restaurants and a Lobby
Lounge. There is a pool, a whirlpool and an exercise facility available for Adam's Mark guests who stay in one of the 496 oversized custom
decorated guest rooms in this 11 story high-rise hotel.
The Adam's Mark is an ideal meeting headquarters within easy reach of all the excitement and attractions that Central Florida has to offer.
Situated in South Orlando at Sand Lake Road and U.S. 441, the Adam's Mark is six miles west of Orlando International Airport and seven miles
from downtown. The hotel offers free parking, on site car rental, attraction tickets and transportation is available to: WALT DISNEY WORLD,
Magic Kingdom, EPCOT Center, Disney-MGM Studios, Universal Studios Florida, Sea World, Church Street Station and the Kennedy Space
Center.
Make reservations by calling (407) 859-1500 [Fax: (407) 855-1585 ].
or mail to:
Adam's Mark Hotel
1500 Sand Lake Road
Orlando, Florida 32809
In order to receive the special room rate ($95, single or double), you (or your agent) must clearly indicate that you are attending the
University of Florida Statistics Week of Short Courses. The hotel requires a one night deposit to confirm a reservation. Deadline for hotel
reservations is February 1, 1999. After this date rooms will be on a space and rate availability basis.
Note
Please make your hotel and travel reservations early. The Orlando area is a major conference and tourist area, and March is a very popular month
to visit Florida. Airlines, auto rentals, and hotels fill up quickly at this time of the year.
Refunds
Full refunds, less a $25 processing fee, will be made if written request is postmarked by January 31, 1999. The University of Florida reserves the
right to cancel this event and will not be responsible for any refunds or expenses other than those expressly stated in this brochure.
In compliance with the ADA act, participants with special needs can be reasonably accommodated by contacting Carol Rozear in the Department
of Statistics at the University of Florida before February 3, 1999. She can be reached by phone at (352) 392-1941 ext. 207, by fax at (352)
392-5175, or by calling 1-800-955-8771 (TDD).
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|
