On the potential of incorporating administrative register data into randomized experiments

Abstract

Administrative register data offer an opportunity to increase the range and nature of experimental evaluations. In addition to providing a range of offending and non-crime outcome data, the uniformity, longevity, and scope of registers support a broader conceptualization of policy assessments. Although found in several countries, the Scandinavian registers provide a unique opportunity to conduct innovative and impactful randomized studies. Here, and with a focus on Norway, these prospects are explored in detail. We rely on official information and policies on national-level administrative data, as well as data retrieved from Norwegian registers, to explore the possibilities and difficulties in using these unique sources of information within an experimental framework. We provide a descriptive analysis of two groups of offenders in Norway to both support these assumptions and illustrate the potential value-added. Employing Norwegian register data in a between-group comparison illustrates that register data would be particularly useful for several key areas of an RCT, including as a source of pilot data, to expand the range of variables on which between-group equivalence can be demonstrated, to increase the length of the observation period, to expand the number and type of individual-level outcomes available for consideration, in addition to providing a foundation for more theoretical, universal policy assessments. Overall, this overview suggests that there is a significant and underutilized potential for the use of administrative register data to develop experimental tests of criminal justice and social welfare programming. Value is added to RCTs by increasing the quality of data available and reducing costs, relative to primary data collection, associated with such efforts.