PJ Media

“80% of 650k had death certificates = 520k.”

Not really. They did not interview every household in Iraq.

There were 1,849 interviews recorded for the study. Supposedly 80% of those had death certificates to back them up (47 clusters, about 40 interviews each cluster). The 650k number is extrapolated from those to represent all of Iraq.

Extrapolate: infer unknown values from trends in the known data.

Does anyone know how many deaths by natural causes, accidents or non-terrorist murder were recorded by the study in addition to those attributed to war?