Abstract

This chapter argues first that the notion of “big data” is misleading in most human rights work, and recommends other ways to think about human rights data; second, it argues that the limited, incomplete nature of data about the social world means that using raw data to calculate statistics about the social world can be deeply confusing. In human rights data, especially, much of the universe of violence is deliberately hidden. Third, it explores two examples showing how data on homicides in the context of political conflict are partial, and how in some cases data can be adjusted with statistical models to get less biased estimates. After some reflections on common misunderstandings about what statistics mean, this chapter concludes by recalling that much of the debate about the legitimacy of “big data” is rehashing a debate settled 80 years ago about whether bigger samples are necessarily better.