Some very quick reflections on the UK Statistics Authority’s seminar on Better Statistics, Better Data, which took place this morning. It gave me a slight sense of déjà vu from conversations a decade or more ago – statistics are hard to understand, most people don’t understand them (or trust them), and there are significant priming effects in opinion polls on issues.
However, there are some signs of progress, partly driven by the Open Policy Making agenda. A few highlights:
There’s clearly now pressure for equal-footing access to data, so Government and Ministers don’t get 24 hours to hone their lines (and leak them) before the rest of the world does. One version of this discussed was release to everyone at the same time – the alternative being release to media organisations on the same timetable as Ministers. The former feels far more democratic, not least because I don’t think anyone knows who “the media” for statistics is any more.
Will Moy from the brilliant Full Fact called for better referencing of evidence in political statements. That brought a lot of nods in the room, though Chris Giles from the FT pointed out correctly that there is a big difference between “true” and “fair” statistics. It goes without saying that better referencing of all evidence used, not just stats, would make policy decisions better and more open. The positive side-effect would be that focus on political philosophy.
Citizen education on statistics is a busted flush (Ben Page said Britain is the 5th most ignorant country about policy-relevant statistics in the world) and no campaign is going to change that – however passive statistical awareness and the power of social media is increasing, which enables people to challenge as well as to spread dodgy data. Hence the importance of organisations like Full Fact, but also good independent academic institutions, but they are often shy to get involved in what seems like political discussions.
The problem with increasing use of data however, is that it often turns into two pressure groups getting each other in a data headlock. The immigration debate, for instance – the data equivalent of drunks swinging punches at each other in an alley – is full of contradictory statistics, more or less reliable but all loudly transmitted. There are few trusted voices to correct or point out discrepancies – Tim Harford talked about us being in a “low-trust equilibrium” that means political statements are distrusted by default.
“We play the ball not the man” said Will Moy, and said that should be the basis for experts to get involved in into a world where data informs decisions and discussions.
That’s obviously the world we want to see – where discussions and decisions are informed by all sorts of evidence, understood by participants. It’s a test not just for UKSA and academic institutions, but for every actor in the open policy making process.
As a side-note, the event itself was quite panel-ish with perhaps a missed opportunity to get all the wisdom out of the room. I will be interested to see how John Pullinger and the UKSA are going to take the conversation on – in an open policy sort of way, of course.