Ray may be right in supposing thart the 'Claimant Count' cannot be
modelled in the same way as observer bias etc.; but that is precisely
the approach used when modelling criminal statistics which are also a
creation - this time of Jack Straw's minnions rather than DSS agencies.
Ray Thomas wrote:
>
> > All data are selections (from reality if nothing else) and
> > all statistics produced
> > from data are, I would say by definition, also observations.
> > What Roy describes as
> > the social scientist's viewpoint is actually, I would
> > suggest, the basis of statistics.
>
> It is very disturbing that challenges to this position - that I suppose many
> statisticians share - is leading to a minor exodus from the list.
>
> > What is sampling theory but a means of controlling or taking
> > into account some
> > aspects of the selection process? Statisticians are well
> > aware of this problem.
>
> Sampling theory is of course fundamental to statistics as a social science
> as well as to what has become mainstream statistics. But the prior problem
> is identification of the target population. As spelt out earlier,
> statisticians in general don't seem to be aware of the importance of
> specifying the target population.
>
> > There must be hundreds of statistical texts warning about the
> > dangers of biased
> > collection invalid survey techniques, effects of extrinsic
> > variables etc. etc.
>
> But such texts have nothing to say about, for example, the Claimant
> Unemployment Count. The Count of Claimant can certainly be regarded as the
> product of biased collection, invalid survey techniques, etc. And because
> of such factors the Count is not very popular among government
> statisticians.
>
> The important characteristic of the Count of Claimants is that it is an
> administrative statistic. Statistics texts don't have much of significance
> to say about administrative statistics. Sampling theory is not usually
> relevant because admin statistics are typically available on a 100% basis.
> So it is very regretful that statisticians don't want to learn from social
> scientists in this area. What is the point of giving statisticians
> responsibilities for handling administrative statistics if their expertise
> is of limited relevance?
>
> The Count can be really understood only by taking into account its
> nineteenth century origins as a trade union insurance scheme. Of course
> the regulation are a palimpsest to which has been added innumerable small
> print amendments. But its main biases - against the young, against the
> old, and against women can only be understood in terms of its origins in an
> insurance scheme that dates from a different era.
>
> The Count is nowadays also the product of a process of negotiation between
> the applicant for JSA and the employment office. It cannot be assumed that
> the regulations within which such negotiations are conducted are interpreted
> in the same way in all offices. This helps to explain, for example, the
> 'discouraged workers' phenomenon revealed by labour force surveys. When
> people say that they would like to work, but are not looking for work
> because they do not believe that jobs are available they are classified by
> the ILO criteria as 'discouraged workers'.
>
> The extent to which discouraged workers are included in the Count is
> uncertain. But we can be reasonably confident that the relevant JSA
> regulations will be interpreted differently in different regions.
>
> > Just because a statistic is produced by the application of a
> > model or technique
> > does not make it any the less an observation. It just means
> > that the model used is
> > part of the selection process which needs to be taken into
> > account when
> > interpreting the statistic.
>
> It is difficult to see that model building has any relevance for explaining
> the selectivity of the Count of Claimants. Social science theory can help
> identify the nature of the selection factors. But consideration of the
> nature of the selection factors involved points to a different direction.
> We need to know the differences between the Count and measures of
> unemployment produced by standard statistical techniques of survey and
> random sampling, i.e. by the LFS. But, uninformed by standard statistical
> texts, this is something government statisticians have failed to do!
>
> Ray Thomas, Social Sciences, Open University
> Tel: 01908 679081 Fax 01908 550401
> Email: [log in to unmask]
> 35 Passmore, Milton Keynes MK6 3DY
>
> ******************************************************
> Please think before you press the 'Reply' button! Note that if you press
> the 'Reply' button your message will go the individual who posted this message
> not to the list. With many mailers you will have a 'Reply-to-All' button that
> will send automatically to the list address of <[log in to unmask]>. The
> Radstats list is set up for public discussion so please be generous with your
> thoughts and share them us all.
> *******************************************************
|