Administrative data is a specific type of big-data, which derives from the operation of administrative systems in the public sector. Record linkage from welfare administrative systems was recognized as a powerful tool to understand long-standing needs and for evidence-based policy design. At the same time, the utilization of big-data, and LAD in particular, is potentially a controversial political topic since it involves monitoring individuals' behavior and often transferring data outside government systems. This raises ethical dilemmas concerning privacy and data ownership.
LAD monitor front-line workers' decision making and services consumption, and therefore, equip evaluators with longitudinal data-sets, facilitate real time evaluation, and enhance the understanding on how interventions at the national and practice levels operate in achieving outcomes. Despite its recognized potential in informing policy making, the utilization of LAD in the field of social welfare is very modest, and is focused predominantly on child welfare. Few efforts have been made to harness LAD to evaluate national welfare programs for families.
Our research aimed to conceptualize a theoretical model to understand the process of LAD utilization for national welfare program evaluation, by applying Kingdon's (1995) multiple streams model, and by comparing the Californian and Israeli contexts due to their similarity in welfare regime and technological capacity, but difference in LAD utilization in the social welfare domain.
We applied grounded based theory and case study as mixed methods, and as a suitable tool for exploring emerging complex phenomena, and since it is acceptable when applying existing theory to understand issues related to information systems. We conducted in-depth expert’s interviews with key informants at the national and street bureaucracy levels including leading researchers, in the field of social welfare; and content analysis of relevant regulation, government-academia agreements, and protocols of parliament committees.
Findings revealed that the mechanism of LAD utilization in national program evaluation comprises of multiple power relations between four types of policy actors and entrepreneurs at the national, local, and street-level bureaucracy; and uncovered the barriers and facilitators that led to the gaps between the potential to the actual use of LAD. The four types are government elected and non-elected officials, researchers, mid-level managers in the public services and front-line service providers. We identified key factors for utilization that operate as "given" or pseudo barriers, such as lack of technological capacity and lack of governmental regulations. We also identify facilitators such as government-academia trust, compliance of service workers to complete documentation in the agency’s information system, and innovative attitudes toward research methodology in academia and public administration. Barriers and facilitators tend to affect the relationship between policy actors, in a specific way and to a difference level.
Conclusions and Implications:
Research findings expand the understanding of big-data utilization as an outcome of the interaction between diverse policy actors and levels. Discussion deals with the strategies, derived from the emerging explanatory model, on the ways evaluators should operate to influence facilitating processes of harnessing big-data for social good.