Economists Do It With Models

Warning: “graphic” content…

Bookmark and Share
Battling Bad Science, One Comedian At A Time…

May 11th, 2016 · 3 Comments
Fun With Data

You didn’t think I was going to miss an opportunity to bring you John Oliver talking about p-hacking, did you?

Unfortunately, i think many well-intentioned researchers do this without even thinking about it- they get a dataset and start poking around with it and then decide to focus on the relationships that appear significant (as opposed to having a- singular- theory to test before exploring the data). This doesn’t sound so bad until you consider that 1 out of 20 spurious correlations is going to look statistically significant at the 5% level, which makes the “looking at a bunch of stuff and writing about what looks interesting” approach inherently questionable.

In related news, a picture really is worth a thousand words.

Tags: Fun With Data

3 responses so far ↓

  • 1 spaghetti // May 16, 2016 at 11:59 pm

    If you don’t already, you should read Andrew Gelman’s blog. Or, specifically, read his paper on “The Garden of Forking Paths”.

  • 2 Harshini // May 25, 2016 at 1:52 am

    Nice to see your blog post.. I really enjoyed by reading your blog post. Thanks a lot for sharing this with us.. Keep on sharing like this informative post.

  • 3 Aaron // May 31, 2016 at 11:39 pm

    I just want to tell you that was a really good video, I think there is more to talk about the problem of the scientific studies and their impact on the society.

Leave a Comment