Can scientific techniques stop politicians lying about statistics?


Dr Ben Goldacre, from the blog Bad Science, appeared on BBC Newsnight recently to call out politicians for manipulating statistics for their own benefit.

He argues that we need better data – and data free from political meddling – to work out what polices work and what policies fail. He calls for an "evidence-based approach" to policy-making.

He also suggests that we should see more randomised testing of policy proposals before they are implemented – similar to randomised trials of new drugs before they are released.

A simple trial he proposes is for a policy to get people off unemployment benefits and back to work. He suggests testing any proposed scheme using 1,000 people, half administered under the proposed scheme and half under the existing one, then measure the effectiveness of each approach.

That way, Goldacre says, we could gather evidence of a policy's effectiveness before spending millions of dollars on a vague hope that it will work.

More from Cosmos on how politicians misuse and manipulate data and statistics here and a humorous look at how you can make almost any data sets "correlate" if you try hard enough here.

  1. http://www.badscience.net/2015/02/i-did-a-newsnight-thing-about-how-politics-needs-better-data/
  2. https://cosmosmagazine.com/mathematics/review-statistical-flaws-age-big-data
  3. http://blog.cosmosmagazine.com/blog/2014/7/23/lies-damned-lies-and-meaningless-correlations
Latest Stories
MoreMore Articles