Some people like to characterize science as a religion. In this formulation, scientists are the modern equivalent of medieval theologians. They hand down precepts and dogmas that we take on faith, because there's no way for laypeople to keep track of all the exciting study results flooding out of laboratories every day.
But science doesn't have, say, a Pope in charge of the show. And there's no Vatican Council calling the shots either, telling us what to believe. What science does have is a method for investigating the world around us. That method has brought us modern medicine and all the high-tech accoutrements of the 21st century. But the method isn't foolproof and, with alarming frequency, some of those exciting study results prove unreliable.
The thing is, when done right, science isn't a matter of faith at all. It's a matter of doubt. Peer researchers should be able to replicate the results of a study. If they can't, the study results are in doubt. If there's enough doubt, the study is retracted. But by then, it might be too late. The original study results are out there, being cited and discussed in the scientific community and in the public sphere. Not everybody can — or will — take the time to go back and double-check that the study they're citing hasn't been withdrawn.
All of that could change, though, because the largest database of scientific retractions just went live and makes the process a whole lot easier. Retraction Watch Database is designed expressly for finding out whether any given study is still legit. The next time you read an article or hear someone say, "studies show that talking is bad for you," you can head over to the site and see what's what.
The database is an offshoot of a blog started in 2010 by two medical reporters, Ivan Oransky and Adam Marcus. One of the highlights of the blog is a list of the 10 studies most often cited even though they've been retracted. The notorious and long-since debunked study linking autism to vaccinations is there, as well as a 2013 paper called "Primary Prevention of Cardiovascular Disease with a Mediterranean Diet" (sorry foodies!).
If you question the need for such a database, consider this: Some studies, such as one claiming to have discovered a protein that mimics insulin, have actually been cited more often after they were retracted than before.