Paper short abstract:

National statistical institutes experiment with new data sources to cut costs and produce more timely statistics. I contend that, based on imaginaries of the creative and tech sectors, they reshape what counts as evidence in statistical practice and produce new professional identities.

Paper long abstract:

National statistical institutes (NSIs) are tasked with the production of rigorous and reliable data about the state of a population. Yet they are also pressured to cut costs and produce more timely statistics. Some NSIs have started to experiment with new data sources to answer these challenges. In this paper I examine experimental forms used by NSIs to test data sources such as Twitter and traffic sensors. I explore and theorise a fieldwork finding: in an environment of rigorousness, new experimental forms seem to encourage acceptance of 'imperfections' and results that are 'just good enough'. This may very well be typical of innovation and work-in-progress, but it is controversial in official statistics.

In STS the quality of evidence is generally considered to be a social achievement. Yet, 'just good enough' has received little attention as an explicit valuation. So how is 'just good enough' accomplished? And what exactly is it? I draw on ethnographic observations of experimental practices at Statistics Netherlands and other European NSIs, among which a bootcamp and an innovation lab where actors such as internet companies are invited to cooperate. Based on imaginaries of the creative and tech sectors, these formats promote 'creativity', 'agility' and 'quick results'. I contend that they reshape what counts as evidence in statistical practice and produce new professional identities (cf. Shapin and Shaffer 1989; Haraway 1997). They do not introduce radical change, however, as they incorporate existing standards, and are intertwined with ongoing cost cutting measures and disciplining managerial techniques.