Abstract
This chapter delves into the nature and significance of evidence within the context of hypothesis testing, advocating for the use of multiple data types for triangulation. The selection of specific data sources and indicators is emphasised, tailored to the nuanced requirements of investigating realist hypotheses. Initially, parallels are drawn between evidentiary practices in courts and scientific endeavours, setting the stage for a discussion of the realist framework. Subsequently, through examples drawn from crime-related evaluations, the drawbacks and advantages of various data sources - such as recorded crime data, victimisation surveys, observational data and interviews - are explored. Notably, the potential benefits of employing unobtrusive measures, including refuse data, are highlighted. Using a project evaluating police patrols as a case study, the integration of refuse data alongside other sources is illustrated, demonstrating its utility in addressing hypotheses. While acknowledging limitations, such as those inherent in physical evidence like refuse data, it is underscored as a valuable supplementary source for confirming, refuting or refining realist hypotheses.