"Evaluating Paired Categorial Data When the Pairing is Lost"
Jonathan D. Mahnken, Ph.D.
Professor in Department of Biostatistics,
Director of the Data Management and Statistics Core,
University of Kansas
Thursday, August 24, 2017
12:00pm – 1:00 pm
1.104 Ewing Hall.
This seminar contributes to achieving one or more of the CTSA core competencies in the category of Statistical Approaches.
Abstract: We encountered a problem in which the experimental design for evaluation data called for the use of a paired data analysis, but the pairing between subjects was lost during the data collection procedure. Thus, we were presented with a data set consisting of pre- and post-intervention response measures on 2-dimensional ordinal affect grids, but with no way of determining the dependencies between our observed pre- and post-intervention values. We utilized a bootstrap approach to create a null hypothesized distribution for our test statistic, which we defined as the Euclidean distance between the center of mass from two affect grids (pre- and post-intervention). Using this test statistic we addressed whether a self-revelatory performance impacted perceptions of Alzheimer's disease. We rejected the null hypothesis and concluded that the intervention did influence perceptions about the disease; specifically the intervention resulted in more positive perceptions of Alzheimer’s disease.
*** Lunch Provided ***
Participating TRCC Institutions include: UTMB, UTHSC-Houston, UTHSC-San Antonio, and UT Southwestern