Science reporting in the news sure is broken. Do enough studies and eventually you'll happen upon one that gives the wrong (and newsworthy) answer, just by chan

Science reporting in the news sure is broken. Do enough studies and eventually you'll happen upon one that gives the wrong (and newsworthy) answer, just by chance. Here are some studies asking the same question that weren't mentioned, and their reported effect on academic performance (in order of my encountering them on scholar.google.com): 1. http://psycnet.apa.org/journals/pha/9/2/163/ http://psycnet.apa.org/journals/pha/9/2/163/ (positive) 2. http://psycnet.apa.org/journals/ccp/70/2/320/ http://psycnet.apa.org/journals/ccp/70/2/320/ (positive) 3. http://onlinelibrary.wiley.com/doi/10.1111/j.1469-7610.1993.tb01071.x/abstract?deniedAccessCustomisedMessage=&userIsAuthenticated=false http://onlinelibrary.wiley.com/doi/10.1111/j.1469-7610.1993.... (positive) 4. http://psycnet.apa.org/psycinfo/1993-46419-001 http://psycnet.apa.org/psycinfo/1993-46419-001 (positive in short term, neutral in long term) 5. http://link.springer.com/article/10.1007/BF00925819 http://link.springer.com/article/10.1007/BF00925819 (positive) I stopped there, but I think the trend is obvious. One thing I think is worth noting is that the short-term studies very directly measure academic performance by actually observing the patients. The long-term study cited in TFA - and I suspect most other long-term studies - relied on questionnaires instead. It seems probable to me that the problem here is that questionnaires are simply too noisy an information source to obtain a statistically significant result, given the additional noise created by overdiagnosis.