NCPA - National Center for Policy Analysis


November 16, 2005

"Why Most Published Research Findings Are False" by John P. A. Ioannidis -- an epidemiologist at the University of Ioannina (Greece) -- presents convincing evidence that an alarmingly high number of scientific "findings" are eventually proven false, says Wendy McElroy, a research fellow at the Independent Institute.

Ioannidis first published his controversial claim in the Journal of the American Medical Association (JAMA, July 2005). Ioannidis focused on original research published by influential journals and widely accepted as accurate by other scientists. (These studies are the cream-of-the-cream and can be expected to return the highest accuracy rate within medical research in general.)

Their data impact the health and, sometimes, the life of patients:

  • Of 49 studies, 45 "claimed that the intervention (examined) was effective," which means they changed the day-to-day decisions of medical care.
  • Fourteen of those 45 studies (about 32 percent) were subsequently refuted.
  • Twenty (44 percent) were replicated or validated.
  • Eleven (24 percent) remain unchallenged and, so are neither validated nor refuted.

Some refuted studies haven't received much publicity. Nor has Ioannidis' claim that fully 50 percent of medical research is wrong, with approximately the same chance of accuracy as flipping a coin.

Ioannidis' conclusion is speculative, but given that so many prestigious studies have been contradicted, it is not wildly improbable, says McElroy.

The 50 percent figure came from asking, Why are so many prestigious research proven wrong?

In response, Ioannidis designed a mathematical model; he expressed the general character of medical research in mathematical terms. The model allowed him to manipulate variables in order to determine how changing circumstances impacted research, especially with regard to well-known sources of error, says McElroy.

Source: Wendy McElroy, "Miracle Cure, or Murky Research?" Independent Institute, October 26, 2005; based upon: John P.A. Ioannidis, "Contradicted and Initially Stronger Effects in Highly Cited Clinical Research," Journal of the American Medical Association, Vol. 294, No. 2, July 13, 2005; and "Why Most Published Research Findings Are False," Public Library of Science, Vol. 2, Issue 8, August 2005.

For McElroy text:

For JAMA abstract:


Browse more articles on Health Issues