- Science: Beyond prediction: using big data for policy problems
- Harvard Business Review: the best ways your organization can support working parents (and work-life balance, in general). “Before launching any support programs for working parents, gather the relevant data: Where do parents sit within the organization? What are their attrition patterns? What information can you gather from annual performance reviews?”
- Thoughts on statistical consulting: the why, the how, and the difficulties
- To live your best life, do mathematics: how math promotes human flourishing
- A consideration of design as a form of participation in complex adaptive systems
- Why football teams should go for a 2-point conversion when trailing by 4 at the end of a game
- The meaning of spurious correlations
- Spurious correlations, visualized. My favorite is below 🙂
February 3, 2017
February 5th, 2017 at 12:54 pm
First, you’re in Wisconsin now; that should be cheddar cheese, not mozzarella.
Second, my go-to example of spurious correlation when teaching MBAs was an example from George Udny Yule from 1925. He presented two annual time series, both covering 1866-1911 inclusive, both with pronounced negative trend (and hence a significant positive correlation). One was the proportion of Church of England marriages to all marriages. The other was mortality rate per 1,000 population.
Happily, there was one semester in which one of my students had attended the wedding of her sister, performed in the Church of England. She was able to provide first-hand verification that there is no human sacrifice during such weddings (other than the groom, of course), confirming that the correlation is spurious.