GOT something big to say? Grab a statistic, it lends authority. Most people know perfectly well that statistics can support different angles on the same story. But who has the time to put together the whole picture for themselves?
In the UK, concern for the health of statistics 鈥� not least when it comes to statistics relating to health issues, such as patient waiting times, public-private health financing, and more 鈥� has led to attempts to rescue the field from the spin of political and industrial lobbies. The UK Statistics Authority now oversees the production of official statistics independently of government, and is answerable only to Parliament.
Straight Statistics, a pressure group of legislators, statisticians and journalists, presenting statistics that have been 鈥渟traightened out鈥�, having previously been twisted for political, business or personal advantage 鈥� a practice the organisation says is 鈥渨idespread 鈥� and often undetected鈥�.
Advertisement
Partiality is deeply embedded in statistics. Data collection generally costs so much that little will be counted unless it is a government priority. This holds as much for questions in fundamental surveys such as the 10-yearly national census as it does for one-off inquiries. So it is not only the twisting of statistics that needs addressing but the way they are produced.
鈥淒ata collection generally costs so much that little will be counted unless it is a government priority鈥�
This is a starting point for the UK group Radical Statistics, which held its 35th annual conference in London last month. Among the highlights was a lecture by journalist and freedom of information activist Heather Brooke, best known for her work exposing the House of Commons鈥� resistance to disclosing figures about the expenses claimed by its members. Another was provided by David Miller from Spinwatch, who outlined how the organisation鈥檚 new website monitors public relations activities and spin in political and corporate life, and how supposedly independent institutes and disinterested experts are used as a front by vested interests to gain media coverage and policy clout.
The audience and contributors were bound together not just by an understanding of statistics: these were people who produce and live by statistics, a diverse bunch of researchers and commentators who like what they do but don鈥檛 always like what is done with their skills. Or what is not done with their skills: that is the extra ingredient. If current statistics are shaped by their funders鈥� priorities, who will produce statistics to fulfil other priorities 鈥� and what would those statistics look like?
Take the 鈥渓eague tables鈥� that rank the performance of UK schools. Decades of work has shown beyond doubt that these divisive tables are statistically meaningless. Once standard errors of uncertainty are attached, the vast majority of schools are very similar. Only the extreme outliers remain interesting. The best use of the tables would be to screen for a few examples of best or worst practice, or ones which suffer from poor measurement.
So what might statistics aimed at improving schooling achieve? In their recent book The Spirit Level, Richard Wilkinson and Kate Pickett argued that societies with the most unequal incomes have the worst health and most crime. If they are right, we might look to reduce inequalities between and within schools, and later at work.
It鈥檚 the same when it comes to employment. Figures purporting to measure unemployment tend to hide the full cost, not least by omitting people not claiming state benefits. Statistical skills could be applied to estimating how many of these missing unemployed there are, and how the figures are affected when governments change the rules.
Worldwide, the tussle between statisticians鈥� paymasters and democratic need plays out in different ways, and arrangements for producing official statistics vary considerably. In Denmark and Finland, for example, the national statistical bodies are almost autonomous, while in the US they are under close government control.
Can the priorities underlying the collection of statistics be changed? This is where it gets interesting. What might a new statistics of social justice look like? And is it possible to convey the meaning of statistics to ordinary people without so many intermediaries, or is a different kind of intermediary needed?
A commitment to using statistical skills in areas away from state or business agendas has taken Radical Statistics to all sorts of places. Its 2000 book Official Health Statistics: An unofficial guide was the product of a decade working with public-sector health campaigns. And challenging official figures on immigration and nuclear weapons has been a big priority. We may not have transformed government practice, but we have given confidence to voluntary campaigns and established statistical practice among new, usually receptive, audiences.
Forays by Radical Statistics into local campaigns to define community needs or to challenge a school closure, say, have led us to pioneering grass-roots statistical education. This can involve going back to basics: doing away with computer power, and replacing it with people totting up questionnaire results. Some members now think 鈥渂arefoot statisticians鈥� practising in the community could prove a high point for our profession.
Whether on employment statistics, official statistics independent of government, or meaningful performance statistics, Radical Statistics often lines up with majority thinking in the more formal Royal Statistical Society. Perhaps the lining up is in both directions: over 35 years, some angry young statisticians have risen to senior positions in the profession. However you look at it, the burning moral issues 鈥� war, corruption, climate change and inequality 鈥� keep the search for radical statistics bubbling.
