Beyond eyeballing: Fitting models to experimental data

Abstract

The oldest and most commonly used tool for examining the relationship between experimental variables is the graphical display. People are very good at recognizing patterns, and can intuitively detect various modes of behavior far more easily from a graph than from a table of numbers. The process of “eyeballing the data” thus represents the experimenter’s first attempt at understanding their results and, in the past, has even formed the basis of formal quantitative conclusions. Eyeballing can sometimes be assisted by judicious application of a ruler, and often the utility of the ruler has been enhanced by linearizing data transformations. Nowadays it is more common to use a computer-based cur..