May I be a little mischievious?
Usually when we calculate some effect in a statistical model, we take
account of conditionals - ie interactions of some sort (well not quite the
same thing of course).
So if we wish to calculate the effect of the coalition forces in Iraq,
just taking the respondents' views as to who caused them is too
simplistic. It may be that an innocent person died as a result of
cross-fire between the two parties but which they attributed to one party
or the other - maybe not knowing that there were two parties firing even.
Or one party may have attacked an area because they had been attacked from
that area and intelligence indicated this was the target. Or a car bomb
may have been aimed at coalition forces but blown up innocent people.
Put another way, if the insurgents/freedom fighters did not shoot at the
coalition forces in the first place, then the latter wouldn't shoot back.
Or maybe if the coalition forces were not in a particular place, the
insurgents wouldn't have shot at them. Chickens and eggs anyone?
You see what I mean. Who is really to blame?
I am not trying to excuse the horrific numbers but to say that the cause
is more deeply rooted than simple marginal totals indicate. As
statisticians perhaps we should be trying to disentangle the data. Just
looking at table margins where 'blame' is attributed to one party or
another is too simplistic a solution. Of course it makes for good copy
but since when has that been an excuse for scientific rigour?
Best wishes
John
John Logsdon "Try to make things as simple
Quantex Research Ltd, Manchester UK as possible but not simpler"
[log in to unmask] [log in to unmask]
+44(0)161 445 4951/G:+44(0)7717758675 www.quantex-research.com
On Tue, 31 Oct 2006, Kevin McConway wrote:
> John Logsdon wrote (in part):
> > A small issue with Kevins comprehensive discourse on mortality is that
> > Pakistan has had quite serious wars over the past 60 years.
> >
> [big snip]
> > So while Pakistan may be a useful analogue for Iraq, comments based on
> > changes in life expectancy or death rate between the two countries must
> > take into account the histories in some way. Not an easy thing to do I
> > agree...
> >
> John's right to point out the history, of course, but I think he's
> really making my point from the other end, so to speak --- the point of
> a (period) life expectancy is that it depends /only/ on mortality
> experience in the period in question --- so if it's the life expectancy
> at birth for 2001 it depends only on the age-specific mortality rates in
> 2001 and not on the previous history. Average age at death, for deaths
> in a particular year, depends on the population age structure in that
> year, which in turn depends in a possibly horrendously complicated way
> on the history of the people who happen to be living in the country in
> question in that year (so involves previous migration, mortality and
> fertility even in an area where the borders haven't changed).
>
> Regards,
>
> Kevin
>
>
******************************************************
Please note that if you press the 'Reply' button your
message will go only to the sender of this message.
If you want to reply to the whole list, use your mailer's
'Reply-to-All' button to send your message automatically
to [log in to unmask]
Disclaimer: The messages sent to this list are the views of the sender and cannot be assumed to be representative of the range of views held by subscribers to the Radical Statistics Group. To find out more about Radical Statistics and its aims and activities and read current and past issues of our newsletter you are invited to visit our web site www.radstats.org.uk.
*******************************************************
|