Graham, D.J. et al. 2005. Risk of acute myocardial infarction and sudden cardiac death in patients treated with cyclo-oxygenase 2 selective and non-selective non-steroidal anti-inflammatory drugs: nested case-control study. The Lancet 365: 475-481. Read Graham et al. (2005) with an eye toward finding examples of the key concepts in the chapters on on models and error from Science for Business, Law and Journalism. We read the models chapters last week, and I am asking that you read the error chapter this week.
Pease and Bull. 2000. Science for Business, Law and Journalism. Chapter 9. Error is unavoidable.
You may also find it helpful to review the two chapters on models assigned for the previous class.
2. In the first part of the Discussion of their paper, Graham and his colleagues lay out the most serious problems with their paper. This is quite different than the approach taken in the law---the authors of a scientific paper are expected to fully disclose the most serious problems with their own work. Come to class prepared to offer a simple intuitive explanation for each of the problems that Graham and his colleagues identifies. Don't try to use scientific terminology with which you are unfamiliar. Rather, explain the problems with the Vioxx paper using common, everyday, non-scientific language.
3. Find examples in Graham et al. (2005) of each of the following types of error: (i) measurement error, (ii) sampling error, (iii) human error, and (iv) bias.
4. What steps did Graham and his colleagues take to minimize each of the preceeding four kinds of error? How successful were they? That is, of the four types of error listed above, which casts the most doubt over the conclusions of Graham et al. (2005). (This is not an easy question to answer. But it goes right to the heart of what scientists do when they critique a scientific paper.)