Radical Statistics, or just radical data?
If there is one thing that nearly everyone in the street knows about statistics, it is that it is one of the three kinds of mendacity.
And if you were to ask them to give an example of a statistic, the chances are that they will mention either the unemployment count or the rate - though you are likely to hear statements like 'the unemployment rate is two million'.
This popular scepticism about statistics is, of course, a symptom of the mystificatory role its practice plays in society, and which Radical Statistics makes it its business to combat.
So it is surprising that RadStats contributors devote almost all their efforts to critiquing the counting processes of official statistics - appropriate definitions, mechanisms for collection - rather than the evaluative: hypothesis testing, or attempts to derive indices of inequality (such as the Gini co-efficient).
It is almost as if the name of the enterprise was Radical Data, rather than Radical Statistics. Only in the De-Mystifying Social Statistics volume is any consistent attempt made to throw a critical light onto the kind of things that occupy professional statisticians' training and constitute their stock in trade.
Responses, when this complaint was aired on the RadStats electronic discussion list, were various.
One was the claim that the (or rather, a particular) dictionary definition of statistics refers only to "collection and arrangement of facts", and not to any further calculations. Another made the point that the English word is ultimately derived from German Statistik, which has roughly the connotation of "numbers used as part of statecraft".
While the first response might best be described as naively pre-intellectual, the second one at least has the merit of meshing with ordinary people's perceptions.
What characterised nearly all of the responses, however, was a defensive tone - as if the respondents knew in their hearts that something has been missing from RadStats practice.
My contention is that this is justified, and that the pre-occupation with what data is collected and how ultimately hinders the aims of the RadStats project.
Although the aim - to point out that data are not facts, but artifacts - is laudable in itself and in principle tends to de-mystificatory ends, by itself it has the ironical result of endorsing the view that there are indeed 'facts', which only need objective and impartial 'collection and arrangement' for correct policies to follow.
This kind of naive empiricism - what one might call the stamp-collecting approach to statistics - is both mother and father of all mystification (as Marx remarked, if things were as they seem on the surface, all science would be superfluous).
From this point of view even such fundamental statistics as the mean, mode and median will indeed appear to be mystifying and arcane issues of technique - and best left to experts - rather than choices about concepts which express different social viewpoints.