The letter by Dr Schuman implies that for science to be credible only "positive" and perhaps also statistically significant findings should be reported. We do not agree. We strongly believe that often a reporting bias exists, whereby statistically significant findings are more likely than "negative" findings to be reported in scientific publications.
We concur that the epidemiologic method performs best at identifying risk factors with high specificity and convincing strength of association. Our decision to report our findings of weak or no associations stemmed from our belief that the credibility of science is enhanced when the scientific community is acquainted with facts, regardless of their interpretation. As long as methods are thoroughly described—and Dr Schuman does not offer any criticism in this regard— description of findings (be they "positive" or "negative," significant or nonsignificant) should only assist in advancing science.
Two other issues raised were (1) whether "small communities" are