5 ms·

The Economist Fails at Statistics.

4 points by pitdesi 15y ago

This is why the Economist is not the Statistician.

pitdesi 15y ago
The Economist fails at Stats. Quite embarassing, really. You can draw a link between any 2 increasing variables if you'd like, or between an increasing one and a decreasing one.
- scottkduncan 15y ago
  It's not much of an article, but it's also not really a stats fail. They show an example of strong correlation, warn the reader that correlation doesn't equal causation, explain that they didn't control for several other variables that might be correlated with their result, and also mention that they didn't explore reverse causality. As the title says, they're highlighting a "striking relationship" and don't claim to be proving a fact.
  - sukuriant 15y ago
    So they said wrote a linkbait article to improve their site's view rate, and then, in the article said: "this and this are going up, but we can't relate any of it. Everything we've just told you is potentially crap and it's purely coincedence. We didn't check anything, but ... Eee!! numbers that indict driving and obesity!" That's terrible, sensationalist journalism. News at 11: icecream sales strongly linked to forest fires in summer.
- mattdeboard 15y ago
  So why did you submit it here? If you knew it's a terrible article with linkbait headline, then why the hell would you submit it?
  - pitdesi 15y ago
    To show exactly that, decline in journalism over there... Perhaps I should've made the headline "The Economist Fails at Stats?" But then it would likely have been corrected. Not sure what the right way to handle this is. Regardless, I think it's interesting to people.
- jbm 15y ago
  http://www.economist.com/blogs/freeexchange/2011/06/nonsensical-correlations http://www.economist.com/blogs/freeexchange/2011/06/nonsensi... There is an interesting follow-up to this piece that you may find interesting.
zwieback 15y ago
Agree, it's not a very good article but if you're a regular Economist reader you know that they frequently put out charts like this exactly to highlight interesting but potentially misleading correlations. Click on the "Daily Chart" link in the header to see more examples.

4 points by pitdesi 15y ago

This is why the Economist is not the Statistician.

The Economist fails at Stats. Quite embarassing, really. You can draw a link between any 2 increasing variables if you'd like, or between an increasing one and a decreasing one.

scottkduncan 15y ago
It's not much of an article, but it's also not really a stats fail. They show an example of strong correlation, warn the reader that correlation doesn't equal causation, explain that they didn't control for several other variables that might be correlated with their result, and also mention that they didn't explore reverse causality. As the title says, they're highlighting a "striking relationship" and don't claim to be proving a fact.
- sukuriant 15y ago
  So they said wrote a linkbait article to improve their site's view rate, and then, in the article said: "this and this are going up, but we can't relate any of it. Everything we've just told you is potentially crap and it's purely coincedence. We didn't check anything, but ... Eee!! numbers that indict driving and obesity!" That's terrible, sensationalist journalism. News at 11: icecream sales strongly linked to forest fires in summer.
mattdeboard 15y ago
So why did you submit it here? If you knew it's a terrible article with linkbait headline, then why the hell would you submit it?
- pitdesi 15y ago
  To show exactly that, decline in journalism over there... Perhaps I should've made the headline "The Economist Fails at Stats?" But then it would likely have been corrected. Not sure what the right way to handle this is. Regardless, I think it's interesting to people.
jbm 15y ago
http://www.economist.com/blogs/freeexchange/2011/06/nonsensical-correlations http://www.economist.com/blogs/freeexchange/2011/06/nonsensi... There is an interesting follow-up to this piece that you may find interesting.

zwieback 15y ago

Agree, it's not a very good article but if you're a regular Economist reader you know that they frequently put out charts like this exactly to highlight interesting but potentially misleading correlations. Click on the "Daily Chart" link in the header to see more examples.