Speaker: Dr. Emily Robinson (Cal Poly) and Dr. Miguel Fudolig (UNLV)

Title: Can you see the change? Testing change point problems with statistical lineups

Abstract: Change point problems can be found in many disciplines such as economics, public health, and behavioral analysis. Visual inspection is often used as a tool for exploratory data analysis to check for change points which could affect the subsequent statistical analysis. This talk will outline the motivation for and development of a statistical lineup protocol to detect a single shift in the mean of a data set.  We designed a graphical experiment to determine which visual characteristics and model parameters affect the perception of translational change in the plots of independent and normally distributed data sets. These characteristics include the magnitude and direction of change, coefficient of determination, and location of the change point. We will present example statistical lineups and discuss areas of possible improvement before we conduct the graphical testing for change point detection at a larger scale.