Fitting Regression Lines to Scatterplots: The Role of Perceptual Heuristics

Abstract

Relations in data are described and analyzed by fitting a regression line to the data. The graphical equivalent is a line fit to a scatterplot, typically via the least squares fit. How do people visually fit a line or recognize the best fitting line for a scatterplot? The present research proposes that people use a set of perceptual heuristics – point capture, early point, balancing points and balancing distances. Experiments 1 and 2A found that participants with no training in regression drew a line and chose the best fitting line, respectively, using those heuristics, especially balancing points. The application of the heuristics was sensitive to the structure of the graphs and influenced the error in drawing the regression line. Participants in Experiment 2B who had received formal training in regression were less likely to apply the heuristics and more likely to choose a line based on the least squares fit. The discussion focuses on the value of cognitive/perceptual analyses of graph reading and proposes methods to help people overcome biases produced by the application of these perceptual heuristics.