Big Data Meets Survey Science

10Big Data in Official Statistics: A Perspective from Statistics Netherlands

Barteld Braaksma, Kees Zeelenberg, and Sofie De Broe

Statistics Netherlands, The Hague/Heerlen, The Netherlands

10.1 Introduction

In this chapter, which builds on Braaksma and Zeelenberg (2015, 2017), we describe and discuss opportunities for Big Data in official statistics. Big Data may be characterized by their high volume, high velocity, and high variety. These aspects have advantages and disadvantages when used in official statistics. On one hand, high volume of Big Data may lead to more detailed and more accurate statistics; on the other hand, Big Data may be very selective and the estimates based on them severely biased. Similarly, their high velocity may lead to more frequent, more timely statistical estimates, but discontinuities may occur as well as sudden jumps in time series that may be hard to explain. Also, with the advance of Big Data and open data, the risk for disclosure of individual data is much higher, which poses new challenges for statistical institutes.

Using such sources in official statistics requires innovative approaches not based on surveys and censuses. This chapter focuses on methodological challenges, in particular, on how official statistics may be produced from Big Data. Specifically, we discuss the best strategies for using Big Data in official statistics and develop a strategy, which may be used for assessing quality aspects and gives guidance how to develop official ...