The survey relied on doing interviews in "clusters" of homes. From the report:
By confining the survey to a cluster of houses close to one another it was felt the benign purpose of the survey would spread quickly by word of mouthThey also tried to verify each death; amazingly, in 90% of the cases they families of the "dead" were able to provide a death certificate. The problem with this is that the surveyors didn't record the names of the dead for comparison later. In the West, this might not be a problem, but it's different in the Middle East. Arab culture is built around the extended family; therefore, many people are likely to say that someone who had died (especially a "heroic" death) would have lived in their home. This is especially true when Iraqis know that Americans often pay "blood money" to the families of those who died accidentally in coalition attacks. So here's an Iraq family, and these Americans come to their door and ask if anyone has been killed lately by Americans. How do you think they'll respond? Since they did the study in a "cluster", I could see people running a "death certificate" from house to house ahead of the surveyors -- "Hey, show this to these Americans, and we might get blood money. Those Americans sure are stupid."
among households, thus lessening risk to interviewers.
This is the problem with "moonbat" science -- failure to adjust their assumptions for clearly outlandish results. Any normal scientists, seeing results that were completely outside the expected range, would check their assumptions and methodology to see where they went wrong. (Actually, for this group, they were clearly hoping for a high number, so they had no incentive to question their methodology.) It's the same thing with the supposed "discrepancies" in exit poll numbers in the last two Presidential elections; the people running the exit polls, to their credit, took steps to fix their methodology -- realizing that, for some reason, Republican voters were more likely to refuse to answer the exit poller's questions. Moonbats, on the other hand, take the results and spin them into a huge conspiracy theory where hundreds of computer programmers (a notoriously conservative lot, I know) conspire to make electronic polling devices all change enough votes to the Republicans so that they win by just a little bit.
Those of us who appreciate real science should continue to speak out against the fake science of both the left and the right -- even if people accuse us of a "lack of regard for human life".
Update 2207 14 Oct: Huckleberries Online linked to this post (Thanks, Dave!), which generated some interesting discussion in the comments.