Data Reanalysis: A Tool for Manufacturing Uncertainty
Michaels, David PhD, MPH, published in Epidemiology,
November 2006 - Volume 17 - Issue 6 - p S43
ISEE/ISEA 2006 Conference Abstracts Supplement
Abstract:
Data reanalysis is a tactic used by product defense consultants to assist efforts to manufacture uncertainty to oppose imposition of public health regulation. Corporations in the tobacco, chromium, beryllium, plastics, and mining industries have used this tactic. The reanalyzes do not test prior hypotheses; often they simply substitute different comparison populations or divide the study population into different categories; the result is usually a lowering or the disappearance of an elevated risk reported in the original study. These methods used in the reanalysis are generally selected after the results of the original analyses are reported. For the most part, the reanalyzes have been conducted using raw data from publicly financed studies. Although access to data from publicly financed studies is freely available, epidemiologists working for these same industries are under no obligation to share data from their studies, and, in general, they do not. Resource-rich parties unhappy with the results of an original study pay for most reanalyzes. The overall effect is to fill the biomedical literature with studies with contradictory conclusions, manufacturing, or magnifying uncertainty.
© 2006 Lippincott Williams & Wilkins, Inc.