Lurking Variables: Some Examples

Abstract Lurking variables are important explanatory variables that might well escape attention in a routine statistical analysis. In this report several examples of lurking variables are given. Important points illustrated include the following. Careful checking, plotting, and thinking are very important. Whenever possible, data and residuals should be examined with respect to time order and spatial arrangement. A variety of plots of the data and the residuals is virtually indispensable. In designing experiments, time order should be considered and, when practical, randomized. Such randomization is not a panacea, however, since lurking variables can still be present.