Decision Mechanics

Insight. Applied.

  • Services
    • Decision analysis
    • Big data analysis
    • Software development
  • Articles
  • Blog
  • Privacy
  • Hire us

Spurious correlations

November 27, 2016 By editor

chart plotting divorce rate in Maine with per capita consumption of margarine

Most people know that correlation doesn’t mean causation. Some people are fed up of hearing it.

When there are studies showing that people who have more sex earn more money you can see why people really want to make the inference.

I find that most of the much-maligned link bait articles reporting fascinating correlations don’t actually claim any causality. They leave that to the febrile mind of the reader.

In the majority of cases, such as the sex and money one, making the link is a bit of harmless entertainment. However, it’s potentially less harmless when people start to draw political or economic conclusions from such findings.

Tyler Vigen’s spurious corrlations website is a fun demonstration of the dangers of conflating correlation with causation. The r = 0.9926 correlation between per capita consumption of margarine and the divorce rate in Maine (chart above—from his site) stands out.

Of course, there’s always the danger that we are too quick to dismiss causality when it’s not immediately obvious. If my mother switched my father’s butter for margarine divorce would certainly be on the cards.

Print Friendly, PDF & Email

Share this:

  • Email
  • Twitter
  • LinkedIn
  • Facebook

Filed Under: Data science Tagged With: causation, correlation, spurious correlation

Search

Subscribe to blog via e-mail

Subscribe via RSS

Recent posts

  • Data Wrangler
  • The Trolley Problem
  • Counting votes using Excel
  • Accuracy vs precision
  • It’s not because we have insufficient data…

Copyright © 2025 · Decision Mechanics Limited · info@decisionmechanics.com