September 15, 2007
estimate of total Iraqi deaths from violence since the invasion,
made by the British polling company ORB using standard sampling and
interview methods (sample 1720, standard margin of error 2.4% - I
assume at 95% confidence).
and Burnham, authors of the famous Lancet studies, must be relieved.
The pinko mantle of leading pacifist fraudsters has now passed to
the British suits, just as the Iraq Body Count estimates suddenly
became respectable when the Johns Hopkins team published a much
updated the synthetic chart I blogged
in June, adding the ORB as a fourth survey result.
For explanations, see my previous post.
ORB estimate of 1.22 million is very close to Lancet 2 updated
according to the IBC body count timeline - 1.16 million. So they
reinforce each other.
now have four
survey estimates from three
independent teams of professionals using two
different good-practice methods. They all say that the excess deaths
in Iraq are hugely greater than the IBC body count, let alone the
numbers from the MNF or the Iraqi government. The mean estimate,
combining the ORB result with my extrapolations from the three older
ones, is 782,000.
can of course quibble with the ORB survey and many will. There were
no checks with death certificates, as in the Lancet
query about the definition of households
is answered by ORB - household was defined to respondents as "under
one roof". The point that worries me is Kevin
Drum's observation that 20% of the deaths were attributed to car
bombs, which would make 32,000 or so of them since 2003, mostly
unrecorded by the press.
conceivable that respondents were exaggerating deaths to make a
point. But it seems more likely to me that respondents would have
good reason not to get into a conversation with strangers about
shootings - and who might be responsible for them. Saying "car
bomb" could be a polite way of saying "I'm not going to
tell you how". The Lancet interviews were carried out by
doctors in the privacy of the respondents' homes, so they would be
more likely to be frank.
ORB carried out another
survey in March, asking Iraqis whether a "family member or
relative", or "colleague or friend", had been
murdered. Since ORB didn't define these terms, the results can't
generate an overall estimate, just an impression. But the responses
- 26% had lost a family member to violence since the invasion - are
consistent with Lancet 2 and the new ORB survey if extended family
size is a plausible 10; but to be reconciled with IBC, it would have
to be 170, which is absurd.
sheds a curious light on the weakness of Iraqi civil society that
the response for "colleague or friend" was much lower than
for relatives - 17% for men, 7% for women. Imagine trying to
estimate the casualties of 9/11 by asking a similar question of a
random sample of New Yorkers. Their non-family circle would be much
bigger than their family one, and the ratios reversed.
AM | TrackBack
War in Iraq