Monday, 7 April 2014
Webinar on exact tests for correlated data (Tuesday, April 8, 11:00 am - 12:00 US Eastern time)
This coming Tuesday, the folks at Cytel will be giving an instructional webinar on exact tests for correlated data.
Webinar: Tuesday, April 8, 11:00 am - 12:00 US Eastern time (NO Charge)
Register from the StatXact web page: http://www.cytel.com/software-solutions/statxact
Correlated data are common in many research areas, but especially in multicentre clinical trials, genetics, epidemiology, ophthalmology, and teratology. Correlated data arises often where multiple outcomes are measured on an individual over time, or on several individuals sharing common genetic or environmental exposures.
Conventional statistical methods for analysing correlated categorical outcomes rely on large-sample distributional assumptions (e.g., approximate normality) to justify their results. These approaches work poorly for small or sparse samples. That's when you need exact methods.
A recently introduced StatXact® module provides a suite of exact tests for analysing correlated data tables. These tools provide correlated-data analogues for common exact tests, including Fisher’s test. We’ll illustrate use and results interpretation using real-world examples, including ophthalmological, developmental toxicology, family-based Alzheimer’s genetics studies, two multicentre clinical trials, and brain pathology research applications.
We’ll analyse the examples using:
• Trend tests for ordered binomials
• Wilcoxon test for ordered multinomials
• Kruskal-Wallis test (for a two-way table with one ordered and one unordered variable)
• Fisher’s exact test (for a two-way table with two unordered variables)
• Stratified 2 x 2 tables
• Exact test for clustering
• Exact trend test for multiple binomial outcomes
Interested but can't attend? Email firstname.lastname@example.org for the slides.
The lead presenter is Dr. Christopher Corcoran, Ph.D., Associate Dept. Head, Utah State University. Dr. Corcoran received his B.S. from USU in Statistics and Computer Science in 1995. He earned his Biostatistics and Genetic Epidemiology doctorate from Harvard in 1999, then joined the faculty of USU's Department of Mathematics and Statistics. At USU, Chris has collaborated on several large research projects, including studies of aging, dementia, cardiovascular disease, cancer, and autism.
Chris focuses on genetic causes of disease, and how genetic and environmental factor interactions alter disease risk. Considering thousands or even millions of genes simultaneously requires carefully designed statistical and computational methods. Chris has steadily developed, implemented, and documented such approaches in StatXact, LogXact, and in SAS-compatible functions.
Call: 020 6680 0300 / 322