Chap T. Le, of the Division of Biostatistics at the University of Minnesota, will present:
“Public Health Statistics: Weighted Statistics and Weighted Statistical Methods”
In common applications of statistics or biostatistics, the basic unit of observation is a single entity, like a person or an animal; but with more recent public health research, that has changed. In many public health applications, the unit of observation is not a single entity; each unit of observation is a cluster of individuals. It could be a state, a county, a classroom, a hospital, or a nursing home. We still obtain one measurement for a variable from each cluster (mean, median, proportion) and apply common statistical methods (correlation and regression, t-tests, Wilcoxon test). Unlike single entities, clusters have different sizes; the data points (computed statistics) carry different weights. Common statistical methods might not be optimal. A few times, we do it the “right” way; for example, using the weighted average or obtaining the weighted Least Squares slope. That’s about it; but we could do more.
I like to fill out the missing pieces to complete the system of elementary weighted statistical methods; then pulling out a few applications.
All are Welcome.
For more information please call 612-624-4655