To the Editor:
I read with interest the recent paper on causal inference by Holman et al. 1 In their study, the authors mailed sets of 12 simulated scenarios to members of the Australasian Epidemiological Association (AEA) and asked them to indicate whether they were likely to attribute causality to the association described.
As current President of the AEA and as a participant in the study, I would like to point out an error in the authors’ description of the membership policy of the organization, and therefore their description of the study population, as this may affect some readers’ interpretations of the results.
The authors indicated that full membership in the AEA was granted after review by an election committee and that membership at the “associate” level was available for those who did not qualify for full membership. Although this was true in the early years of the AEA (which was formed in 1987), it has been AEA policy since the early 1990s that full membership is open to anyone with an interest in epidemiology. This includes epidemiologists and non-epidemiologists from a wide range of backgrounds. Our records indicate that the change to open membership was approved in 1992, well before the study that was conducted in 1997–1998 by Holman, et al. 1
The authors also indicated that the study participants included epidemiologists in both Australia and New Zealand. Because of the history and structure of the organization, we maintain separate databases for Australian and New Zealand members, based on country of residence. Given the very small number of respondents (four) who reported New Zealand as their country of training (see Table 1 in Holman et al. 1), I think the authors may have been using a list of Australian-based members, rather than a list of the total AEA membership.
While I don’t think either of these errors detracts substantially from the value of this groundbreaking study, I thought it would be worthwhile to provide clarification. I found the study to be quite thought provoking and stimulating, and I look forward to more work in this area.
1. A psychometric experiment in causal inference to estimate evidential weights used by epidemiologists. Epidemiology 2001; 12: 246–255.