[extropy-chat] IRAQ: Weapons pipeline to Syria

[Alfio Puglisi]

On Mon, 1 Nov 2004, Mike Lorrey wrote:

>
>--- Alfio Puglisi <puglisi at arcetri.astro.it> wrote:
>
>> Except that they are cherry picking the data to lower the total
>> amount, the way someone trying to make the war look better would
>> do. So if this is to fit some political agenda, this agenda would
>> be fairly aligned with Bush.
>
>On the contrary, their rationale is that they can reliably avoid the
>most violent areas where they would be at risk of kidnapping and avoid
>all the areas where nothing is going wrong and just pick the low
>intensity conflict 'middle' areas, but they are screwing up because
>they are making that selection based on no prior data about what the
>'middle' really is, or whether areas of medium violence actually
>reflect the mean for the whole population.

As I see it, it's very simple. There was a group of houses in Falluja
that was by far the worst and gave very high death counts like 200,000+.
Removing this group gets the average down to 100,000 and this is what they
published. There are no other assumption on what the relative violence
rate is here and there, just that they had an unusual data point and they
removed it. Keeping it would have mean an even higher count, that could
render unbelievable the study.

So they made a mistake (ignoring data) to avoid negative publicity. Their
total body count is lower than it would be if all of the households were
included. There's no need for elaborate thinking about hidden motives or
whatever.

Of course the standard deviation is so high that the actual number could
be anywhere.

Alfio