Big Data Meets Survey Science

7Improving Quality of Administrative Data: A Case Study with FBI's National Incident‐Based Reporting System Data

Dan Liao¹, Marcus E. Berzofsky¹, G. Lance Couzens¹, Ian Thomas¹, and Alexia Cooper²

¹RTI International, Research Triangle Park, NC, USA

²Bureau of Justice Statistics, Washington, DC, USA

7.1 Introduction

With the dramatic growth in computational power and data storage sources in the past two decades, government agencies, organizations, and corporations have made great efforts to modernize administrative procedures that collect detailed, complex qualitative and quantitative information via electronic systems. Because this information is collected typically for administrative purposes, such as registration, record keeping, or transaction monitoring, this type of information is referred to as administrative data. Aside from administrative purposes, the increased collection of these types of data also presents exciting opportunities for statisticians, researchers, and policymakers to address important topics in a timely manner without requiring additional resources to collect the data by traditional experiments or surveys.

Many statistical institutes and researchers worldwide have explored producing official statistics with administrative data (e.g. Connelly et al. 2016; Holt 2007; Statistics Finland 2004; Wallgren and Wallgren 2007). However, like other data sources, the need to understand and control for data quality issues in administrative data before putting them ...