Ray wrote:
> Bravo for Deming! It is an important step forward for a statisticians to
> recognise that statistics are the product of definitions. Usually
> statistican regard statistics as the (objective) properties of
> populations.
>
> Social scientists, and most of the refugee statisticians on this list,
> would
> characterise the definitions used to produce statistics as artefacts
> rather
> than as operational. 'Operational' implies that statisticians can
> control
> the definitions. But social scientists would recognise that these
> artefacts are social products, usually governmental or organisational
> products.
>
I'm sure Ray's right in thinking that most members of this
list would subscribe to the above -- with the proviso that where he writes
"statistics", it is actually *data* that is meant.
I'd say that statistics proper are, clearly, artefacts, but
nonetheless objective. Thus the mean, mode and median are all attempts to
give precise content to the socially-produced notion of the typical or
normal, but the arithmetic mean of a particular set of numbers is an
objective property of that set.
Incidentally, it has often struck me that, given the
contents of the journal, Radical Statistics is something of a misnomer:
"Radical Data" would capture its nature more accurately.
Julian Wells
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|