Figure 1. Conceptualization of passive experiments within wider evaluation framework.

A conceptual model displaying different types of experiment based on allocation of an intervention (null, natural, or experimental) and assignment to an intervention (randomized or not). Although randomized experiments are generally considered to be least susceptible to bias, passive experiments (both natural experiments and experiments of nature) enable the evaluation of changes to a system that are difficult or implausible to manipulate experimentally. Adapted from de Vocht et al (86) and Remler and Van Ryzin (294).