However, investigators have a natural preference for problems they can solve, and approaches to studying interactions have historically focused on specific statistical models. We begin by briefly considering the differences between how a biologist might think about interaction and how a statistician might operationally define it.
