Defending Johns Hopkins Iraq Numbers

Article excerpt


Four population-based surveys (two by Johns Hopkins University) have shown a consistent pattern to the scale of Iraqi deaths in the current conflict. The article by Neil Munro and Carl Cannon in the National Journal contains numerous misquotes from a variety of people, as well as incorrect numbers and unfounded assumptions ("Johns Hopkins' Iraq numbers," Editorial, Tuesday).

One principle above all in any human study is the ethical protection of participants. Recording or distributing information that could identify participating households, including the photographing of death certificates, clearly would violate this principle. This is but one of the ways surveys in the midst of conflict present challenges not there in safe circumstances.

In the 2004 survey, deaths in a quarter of clusters could be verified in quality checks, and results from the 2006 survey tracked precisely with those results for the 17 months covered by both studies. The accusation of data-heaping in this study was based on incorrect interpretation of the data that was provided by the investigators. Missing verification of death certificates was spread across a wide number of clusters. At the conclusion of the interview where a death had been reported, the certificate was then requested. Some households did not have certificates readily available. In some cases, selected neighborhoods proved to be very tense, and questioners dropped this part of the interview. Epidemiologists know that data-heaping (frequent recording of certain values) is likely to come from either respondents or interviewers. Expanding the demographic data recorded for each household is not a safeguard against fraud.

A low number of unoccupied houses is to be expected. The low refusal rates we observed are similar to rates for surveys in Kosovo and Congo and are common in surveys during emergencies. …