In this paper we treat instructions as an experimental variable. Using a public good game, we study how the instructions' format affects the participants' understanding of the experiment, their speed of play and their experimental behavior. We show that longer instructions do not significantly improve the subjects' understanding of the experiment; on-screen instructions shorten average decision times with respect to on-paper instructions, and requiring forced inputs reduces waiting times, in particular for the slowest subjects. Consistent with cognitive load theory, we find that short, on-screen instructions which require forced inputs improve on subjects' comprehension and familiarity with the experimental task, and they contribute to reduce both decision and waiting times without affecting the overall pattern of contributions.
