Recently an article has been doing the rounds that “too much running will kill you!!!!” and has been picked up by various newspapers that rely on making their readers feel good by justifying their insecurities
It defines ‘too much’ as more than 20 miles a week, saying the optimum is 5-20.
Here’s a good rebuttal: http://www.runnersworld.com/health/will-running-too-much-kill-you
The tl;dr is that the study’s data is normalised to stop certain unrelated things from affecting the result. It’s good to try to remove the effects of unrelated factors like drinking and smoking (well… you could argue these should be left in since someone who cares about their ability to run is probably less inclined to intentionally poison his or her body), but the study also removes the effects of BMI, and a few other things that are clearly influenced by running (or exercise in general), which is pretty daft. If running decreases your risk of premature death by lowering your BMI, you don’t want to remove the effect of BMI from your study on running and premature death!
Also, the study fails to show an increased early mortality rate, it just looks at some things here and now and extrapolates that they will result in one later. But that’s not really so important because the process to filter people into this section is questionable anyway, as per the point above.
Remember: statistics is hard and just because the study is flawed doesn’t mean the conclusion is wrong.