?Disentangling confounding andnonsenseassociations due to dependence?
Betsy Ogburn, PhD
Department of Biostatistics
Johns Hopkins Bloomberg School of PublicHealth
Abstract: Nonsense associations can arise when anexposure and an outcome of interest exhibit similar patterns of dependence.Confounding is present when potential outcomes are not independent oftreatment. This talk will describe how confusion about these two phenomenaresults in shortcomings in popular methods in two areas: causal inference withmultiple treatments and unmeasured confounding and causal and statisticalinference with social network data. For each of these areas I will demonstratethe flaws in existing methods and describe new methods that were inspired bycareful consideration of dependence and confounding.
Biography: Betsy Ogburn is Associate Professor ofBiostatistics at Johns Hopkins Bloomberg School of Public Health. Currentlymuch of her research focuses on causal inference in the presenceof unmeasuredconfounding, causal and statistical inference using data with complexdependence, and the efficient use of randomized trial data to findeffective COVID treatments. Betsy completed her Ph.D. in Biostatistics atHarvard University and is a 2016 National Academy of Science Kavli Fellow.