Short Course

Fall 2017 Short Course

Friday, September 29th, 2017, 1:00 PM - 5:00 PM CT


REGISTER NOW

The Kansas-Western Missouri Chapter of ASA and the University of Kansas Medical Center, Department of Biostatistics are co-sponsoring a short Course on September 29, 2017. This short course is open to anyone with an interest in designing experiments that will use Generalized Linear Mixed Models methods in the analysis.

 
Title: GLMM-Based Tools for Planning and Design of Research Studies

Presenter      :  Walter, Stroup Ph.D.

Date              :  September 29, 2017, 1:00 pm to 5:00 pm

Location        : University of Kansas, Edwards Campus (BEST Conference Center)

                       12600 Quivira Rd., Overland Park, KS  66213

Registration Fee     : $50 for ASA members, $25 for students, $75 for non-ASA members

                                Boxed lunch included in registration, limited number of vegetarian options available. There is a discount if you register by September 22nd 2017.

 

Abstract

 Power and sample-size are important aspects of research study design. Researchers are often required to provide power analysis as part of their grant proposal submission. Statistical methods textbooks typically devote perhaps a page of two to power, usually giving a simplistic formula to compute the power for comparing two treatment means. Many software packages are available, for example SAS® PROC POWER and PROC GLMPOWER. While useful, the typical textbook presentations, and available software packages, are limited in two important ways. First, power calculation is – or should be – the last step in a planning process. Power is partly a function of sample size, but it is also a function of how sample size is deployed. Researchers – often aided and abetted by their statistical consultants – tend to focus on the former and neglect the latter. Second, many designs have an error or covariance structure that is more complex than most power and sample size software packages are able to handle. This is complicated by the fact that the response variable of primary interest is often non-Gaussian. How does one decide whether to use an incomplete block design, a complete block design, or a split-plot design when the primary response variable is a discrete count? Once the design is selected, how does one compute sample size? You can miss badly unless you use appropriate tools. This workshop presents GLMM-based tools useful for planning experiments. Precision analysis – comparing the pros and cons of competing plausible designs – as a prelude to power calculation, is emphasized. The workshop begins with a review of relevant GLMM theory and methods, and then presents several examples. Given the author’s background, examples will use tools available in SAS/STAT.

 Biographical Sketch

 Walt Stroup is a Professor in the Department of Statistics at the University of Nebraska-Lincoln. He is Fellow in the American Statistical Association, and served as department chair, 2001-2010. His current responsibilities include teaching statistical modeling, design of experiments, and research on generalized and mixed models including collaboration with researchers in agriculture, natural resources, medical and pharmaceutical sciences, education, and the behavioral sciences. He participated in a multi-state mixed model project that provided motivation for the development of MIXED procedure in SAS/STAT. More recently he participated in an industry-government-academic working group focusing on pharmaceutical shelf life, and was PI on an NSF grant concerned with statistical modeling of teaching effectiveness. He has co-authored textbooks on SAS for linear models, SAS for mixed models, and GLMM Applications for Plant and Natural Resource Sciences. Most recently, he authored the text Generalized Linear Mixed Models: Modern Concepts, Methods and Applications (2013), and in co-authoring the 3rd edition of SAS for Mixed Models, scheduled for completion by year’s end, 2017.

Follow the link below to register :

 https://www.123signup.com/register?id=hqchp

For more information, please contact John Keighley at jkeighle@kumc.edu