INTRODUCTION

Experimenting with crime and social programs has a rich tradition. Some notable developments include the “lost letter” experiments of the 1970s (Farrington 1979) and the experimenting society concept advanced by Donald Campbell and others during the 1960s and 1970s (see Campbell 1969, 1979). Researchers used the lost letter experiments, which involved leaving cash in an apparently lost letter on the street, and other similar techniques to examine public dishonesty and minor deviance. The experimenting society concept is rooted in the large-scale social programs of the Kennedy and Johnson administrations and the need to identify valid and rigorous methods of evaluation. It has generated profound interest in the use of experiments to test the impact of criminological interventions.

Present-day experimental criminology very much has its roots in these developments (Sherman 2010). At its heart are the methods of experimentation introduced in science in the seventeenth century. The defining feature of an experiment is control of the independent variable. An experiment is designed to test a causal hypothesis about the effect of variations in one variable on variations in another. A hypothesis cannot be tested experimentally unless it can be expressed in these terms. The methodological adequacy of any test of a causal hypothesis can be assessed on four major criteria (Campbell and Stanley 1966; Cook and Campbell 1979; Shadish, Cook, and Campbell 2002). Statistical conclusion validity refers to whether the two variables of interest truly are related. Internal validity refers to whether a change in one variable did produce a change in another. Construct validity refers to the theoretical constructs that underlie the measured variables; and external validity refers to how far the results can be generalized to different persons, settings, and times.