Social media and statistics: an unlikely marriage?

29Sep12

Statistics is not just about maths… beyond chi-squares and p-values, statistics was originally created in the seventeenth century as the science (or perhaps, the art)! of describing the state, and still does so today. Governments have statistical agencies in charge of informing it on the state of the country. Among other things, statistical agencies measure GDP, the rate of unemployment, and the level of prices.

Because most of these indicators are key to policy-making (you need an accurate inflation rate to regulate the money supply, for example), and because measuring them requires the trust and active collaboration of citizens and organisations (who need to provide data, for example retail prices), statistical laws are very strict on what these agencies can and cannot do. They imperatively need to reassure citizens about the use of their individual data, and to guarantee the impartiality of the resulting indicators. By the same token, these regulations also make statistical agencies very conservative: any novelty that could undermine public trust is unwelcome.

So, using social media would seem to be worlds apart from what statistical agencies do… apart from some basic form of one-to-many, monodirectional communication where they use, say, Twitter, to announce new indicators being made available on their website. The rate of unemployment has increased in the third quarter of 2012. Exports have gone down. And so on.

The problem is the collaborative potential of social media -the fact that they may involve receiving feedback from the public, with two-way communication. And may even involve three or more actors at a time, with community-like discussion groups. Many official statistical agencies would be uncomfortable with that -do we really want to know what people think? Perhaps not: people may ask things that we are not legally allowed to provide.

Yet things are changing. The Australian Bureau of Statistics decided to engage a consultation with the public to better understand what people considered to be meaningful measures of progress, and how statisticians should measure them. They were following international intellectual policy debates on how to measure well-being (happiness vs GDP as discussed in Sarkozy’s initiative, the Stiglitz report etc.). They did so last year, by setting up a collaborative blog and involving a few prominent Australians (sports stars, popular news presenters etc.) as key influencers.

The US Census Bureau is using a blog and (to some extent) Twitter and YouTube to improve communication with data users. “Open data” and transparency are not enough to guarantee accessibility and usability of the data: social media provide a way to better understand how to make data widely available in a user-friendly way. Old-fashioned tables are not enough: today, there is a demand for data in the form of videos and data visualisations.

In Europe, Eurostat is currently setting up a collaborative wiki to discuss issues related to measuring CPI (the key indicator for inflation figures) and how to harmonise measures across countries. This approach is more cautious in that it involves only specialists and experts in the statistical agencies of European countries, not the general public; but it is still an interesting attempt at engaging with collaborating organisations.

These are just three examples, and I am sure there are more, even though they are still a minority among governmental statistical agencies. But they are a sign of the times, in the wake of the open data movement -engaging citizens with the production and dissemination of information aimed at informing policy-making. This, I think, is something that can be useful to be aware of.