I fully accept that John Logsdon does not "consider that rape victims bear
responsibility for the crime perpetrated against them and that (he resents)
totally anyone who tries to represent that (he does)". I fully accept his
further protestations: "Nor do I consider that casualties of war are guilty
of being in the wrong place whether they be from air strikes or suicide
bombs, Afghani, Iraqi, Palestinian, Israeli or whoever, whether they are in
Afghanistan, Iraq, New York, Madrid, London, Bali or anywhere. To suggest
in any way that I do is wrong. I take extreme objection to such innuendo
and smear."


We should all appreciate that expression of attitudes in John Longsdon's
mind. But it is unfortunate that these thoughts were not expressed clearly
in the messages he posted to the list. The lack of clarity is encapsulated
by John's assertion that my repeating of a sentence he wrote amounts to
"gross misquoting".

The problem is not John's attitudes but several carelessly worded,
ill-thought-out messages from John early in the month that were tendentious
towards position different from that proclaimed in his message of 19 Nov.
Of course "Marie Curie cannot take responsibility for the deaths at
Hiroshima and Nagasaki" but the practical implication of this bit of
anachronism is to draw attention away from the fact that many atomic
scientists have had doubts and misgivings about the morality of the work
they do.


Science is not morally neutral. I believe that these doubts and misgivings
that scientists express are appropriate and should become part of public
debate.

Neither is statistics morally neutral. Statisticians who make inferences
and build models are thereby often making moral judgements. It is important
that they are aware of this. I'm not sure that John Logsdon appreciates
this. He seems to believe that because his heart is in the right place he
can use statistics in any way without compunction. Perhaps he can reassure
us?


Ray Thomas
******************************



-----Original Message-----
From: email list for Radical Statistics [mailto:[log in to unmask]] On
Behalf Of John Logsdon
Sent: Sunday, November 19, 2006 11:24 PM
To: [log in to unmask]
Subject: Re: STATISTICAL AND OTHER KINDS OF EVIDENCE


Ray's message is not only grossly misleading but wrong.

Ray quotes me (16th Nov):

'He actually says things like "The responsibility for the casualties in this
case is in some way shared, even if the family may not realise it."'

This is an example of gross misquoting to which even the gutter press in
this country rarely stoop - surely no standard to emulate.

From the preceeding sentence in my original text, it is clear that I am
suggesting that the responsibility could be shared by the protagonists and I
am suggesting that, when attributing the cause of death to the interviewer,
the family was not aware that they had been caught in cross-fire and
therefore attributed the cause to one of the parties when it should have
been both.

The full paragraph reads (my mail of Nov 6th):

'But if the air strike is as a result of previous action or intelligence or
the car bomb is aimed at the coalition forces - or even the Iraqi security
forces - then it is in some macabre way understandable as a military action.
The responsibility for the casualties in this case is in some way shared,
even if the family may not realise it.'

It is not that Ray has innocently extracted the quote as the following
sentence in his message of 16th Nov says:

'The implication of the situation he hypothecates is that a family by being
Iraqi, by being in the wrong place, by being attached to the wrong man, are
in part responsible for their own destruction. The objectionable and
immoral word here is 'responsible'.'

Maybe I should have been more specific but I did not dream that anyone on
this list would either misunderstand the simple English which I put in two
sentences rather than one, or suggest that I was blaming the family for
being hit. I do not consider that rape victims bear responsibility for the
crime perpetrated against them and I resent totally anyone who tries to
represent that I do. Nor do I consider that casualties of war are guilty of
being in the wrong place whether they be from air strikes or suicide bombs,
Afghani, Iraqi, Palestinian, Israeli or whoever, whether they are in
Afghanistan, Iraq, New York, Madrid, London, Bali or anywhere.
To suggest in any way that I do is wrong. I take extreme objection to such
innuendo and smear.

I could go on - particularly about his ability to ignore history before 2003
when convenient which would surely put the cat among the pigeons for GCSE
examiners. But there is one clear pointer to Ray's thinking that needs
examining. In his 16th Nov post, he says:

'John's basic confusion here is to fail to distinguish between descriptive
and inferential statistics. The survey published by the Lancet belongs to
descriptive statistics.'

Ray is 180 degrees wrong. The Lancet paper belongs to inferential
statistics. Since when have descriptive statistics been published with
(quite wide) confidence intervals? Perhaps Ray could point to where the
650,000 or so numbers are tallied either as bodies or some other
documentation. The numbers are estimates and even more, they are estimates
of excess deaths. So in Ray's descriptive statistics argument, the paper
presumably has taken full account of the people who would have been killed
under Saddam Hussein's regime had it continued. Preferably the researchers
would have known who they were - who was saved and who would have been
killed under the two regimes. Perhaps God was a co-author.

No, Ray. This is a model. These are estimates and as such it is quite
proper to ask whether the model can be improved not in order to allocate
blame to one quarter or another but as an excercise in finding the truth.

If Ray believes that the Lancet paper was descriptive statistics, this
explains much of his post. The sad thing is that while his posts have been
overlaid with politics and much antipathy towards one side of the conflict,
there has been no criticism of the other sides.

Ray's issue of immorality goes to the heart the ethics of science. Should
the (model) creator take responsibility for its use? No, the creator of any
scientific product must take responsibility for communicating about it. But
I cannot take responsibility for Ray if he chooses to misread a sentence I
have written. Marie Curie cannot take responsibility for the deaths at
Hiroshima and Nagasaki. Faraday cannot take responsibility for electric
shock torture. And are algorithm generators guilty for the use of their
algorithms? Maybe NaG should hand their heads in shame as many of their
programs are used in designing weapons of war.

No, the responsibility is of the user of the science. This is true of
statistical models. In particular of course statistical models generate
associative rather than causative links so using them to allocate blame is
not right. Temporal models can be used to eliminate possible causal
relationships. But does that mean that they should not be used to study
things? Should we be gagged thus?

I was puzzled why Ray should have so much difficulty in understanding my
original question so I extract here the essence.

1) Suppose a small group (A) of children are playing peacefully in a school
playground when a stone deliberately thrown at them by a member of a group
(B) of other children hits one of the group A children on the head. Who is
at fault?

2) Suppose stones were just being thrown about haphazardly by group B and
one of these hits a child in group A. Who is at fault?

3) Suppose group B are actually throwing stones at another group (C) of
children in some sort of gang war. Who is at fault?

4) Suppose group C is not visible and group A did not know they were there
- they are over a fence or something. Who is at fault?

I asked my 5 year old son without prompting. His answers were respectively
B, B, B and C, B and C. I agree with these. My question in this playground
context was can we use tools to distinguish these cases.

I agree that this was not exactly a scientific experiment - maybe the
questions could have been better formulated, asked in a different order, a
randomly selected sample could have been taken etc etc. But if, for
scenarios 3 and 4, a 5-year old can assign the fault to two groups, whether
or not one of them was visible, why cannot Ray who is much older and should
be much wiser understand that my question was an academic one?

There are deeper problems in Ray's response to my post. I am not (in my
original question) interested in the politics. I stated so many times.
I do not think that the RadStats list is the proper place to air them.
But the vitriol and vile poured on my head have been laced with politics.
I have been accused of wooly thinking, mumbo-jumbo, immorality. These are
strong words - the Send button is not always your friend.

The Radstats 'mission' is "We believe that statistics should be used to help
change society for the better" or "We believe that statistics can be used to
support radical campaigns for progressive social change. Statistics should
inform, not drive policies. Social problems should not be disguised by
technical language". So far, so good. Motherhood and apple pie.

But what is radical? Is is knee-jerk agitprop reaction to the state, big
business or is it clear thinking? Does it take its lead from what everyone
else is saying or is it innovative and questioning? It is possible that
sometimes the state and big business can be right? Is it possible that the
opposition to the state and big business can be wrong?
Is this allowed in RadStats or is considered axiomatic that state wrong,
opposition right? Have we arrived a little late at 1984? If I want your
opinion I will give it to you.

If you throw a lot of mud at the wall, some of it sticks. But does it make
falsehood into truth or vice-versa? No - it just means that emotion has won
over facts and emotion is often only a short step from fiction. Confusing
fact from fiction is fundamentally non-scientific. So unsupported
accusations of lying (aren't we innocent until proved guilty?), ranting on
about Iraq, Blair this, Bush that does nothing to help clear thinking nor to
help in statistical interpretation of the facts. In my view, it has no
place on this list and members who wish to vent their spleen on other people
should do it elsewhere - alt.politics.iraq or some such place if it exists.

No-one has pointed out the fundamental errors in Ray's 16th Nov message -
neither the mischievous selective quote nor the simple error on what type of
statistics is involved. Why, I ask myself? These are the possible reasons
that I can see:

1) People are not interested in Iraq or body counts etc.

2) People have too many other things to do.

3) People agree with Ray anyway.

4) People do not want to get into arguments.

Maybe there are others but of these, the first two are not supported by the
evidence of posting frequency - even over the weekend. Joining Voltaire, I
would defend Ray's right to his opinions and by extension anyone else on
this list. But I don't think political opinions that are not directly
concerned with statistics should be on this list - today's post on the
al-Jazeera Blair interview is an example.

I find the last postulated reason the most worrying as it implies that
members are being bullied into silence. In other words we are accepting a
censorship of which Goebbels would have been proud. Think like I do or get
rubbished. Don't raise issues that may be contentious and may lead to
unpopular conclusions. Don't use rational argument because that is not
allowed. In the last century maybe 100 million people died fighting over
such issues - that is an estimate of course.

So read 'extreme' for 'radical' and only converse with those that agree with
you. Close your mind. If this list descends to such levels there will be
little point in it continuing.

I apologise to list members for occupying your bandwidth. I have responded
cooly in the past but have come under personal attack and I defend my right
to reply. I am happy to respond to private mails but unless I am unfairly
attacked I will not post on this topic again. Oh dear - is that bullying or
pre-emptive self defence?

Best wishes

John

******************************************************
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.
*******************************************************

******************************************************
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.
*******************************************************