Abstract
Graphics play a crucial role in statistical analysis and data mining. The lineup protocol for experimentally testing graphics has traditionally used p-values to identify plots which are significantly visually distinct from randomly generated ``distractor” plots, but this approach does not easily facilitate the examination of randomly generated plots to determine the strength of the distractor effect. This study presents a Bayesian approach to visual inference, using Bayes factors to examine the difference in signal strength in two-target statistical lineups.
Location
Section on Statistical Graphics, JSM
Event Type: Conference
Location: Denver, CO