Bad Science

Posted in Uncategorized by Pierre Jacob on 27 November 2010

Just a quick link to a blog called “Bad Science“, written by Ben Goldacre for the Guardian. It’s about studies that are supposedly backed up by science, but that actually present huge flaws. These flaws are often linked with a poor understanding of basic statistics, hence I think it is a pleasant blog to read for statisticians in particular.

The last entry for instance, is about a study from Stonewall, the lesbian, gay and bisexual charity in the UK.“the average coming out age has fallen by over 20 years”. As Ben Goldacre explains it, the study is wrong because of arguments that are obvious to whoever knows the basics about survival analysis. To summarize, when you ask a N years old guy “At what age did you experience event E ?”, his answer is going to be less than N. In survival analysis, it is called right censoring. If you don’t take that censoring into account, for instance by ignoring the answers from people who didn’t experience event E yet, then obviously your estimation of the mean age at which event E usually occurs is extremely wrong (ie it’s going to underestimate the mean age massively).

And that’s too bad because there are a lot of ways to correct this bias, lots of models used by economists and epidemiologists every day… speaking of which I think the title of this blog entry, “A new and interesting form of wrong“, is not quite right because it’s a very common mistake for people who are not familiar with survival analysis. I remember having made the same kind of mistake before discovering the Kaplan-Meier estimator for the first time. Anyway, it’s a great blog!



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: