Some examples of unrecognized confounding by researchers
*William Lawrence Bryson, Boehringer Ingelheim Vetmedica, Inc. 

Keywords: confounding

Frequently statisticians not involved in the design aspect of a study are requested to conduct an analysis of the data collected. Upon inquiring about how the study was conducted, the statistician may find that confounding factors existed in the study that were unrecognizable to the researcher, but clear to the statistician. This impacts the analysis as well as any conclusions. Several examples of these situations will be presented along with researchers’ justification as to why the confounding was unimportant or unavoidable. Suggestions as to the actions of the statistician in these situations will be discussed.