Really? Using the nullabor package to learn if what we see is really there?

Di Cook, EBS, Monash University

Plots of data often provoke the response "is what we see really there". In this talk we will discuss the use of the nullabor package to assess the significance of structure discovered by exploring data visually. Classically, quantifying significance with p-values required a rigorous protocol involving several steps: hypothesis formation, data collection, test statistic calculation, and comparison with a reference distribution requiring strict assumptions. The nullabor package implements the lineup protocol, which compares a plot of data with plots of null data. The lineup protocol is named after the "lineup", popular from criminal legal procedures. The nullabor package has several methods for generating null data, randomises and encodes the position of the data plot, and all the power of ggplot2 for making data plots. This package enables the data analyst to quantify their findings as different, or not, from spurious patterns.