“Big data” is particularly useful for demonstrating variation across large groups. Using administrative tax data, for example, Stanford economist Raj Chetty and his colleagues have shown big differences in upward mobility rates by geography, by the economic background of students at different colleges, by the earnings of students taught by different teachers, and so on.

The basic message here is that when it comes to understanding how to promote more opportunity, it’s dangerous to make generalizations. For any given national trend or picture, there will be places, people, or institutions that do much worse than the average—and plenty that do a lot better.

A few weeks ago, Raj Chetty stopped by Brookings to present on his most recent research. After his presentation, he joined me and Adrianna Pita on the Intersections podcast to discuss how big data helps us understand diversity within populations—or as academics would say: helps us demonstrate heterogeneity.

You can listen to the podcast here. If you missed Chetty’s presentation, you can watch video from the event.

Read the full article about heterogeneity in data by Richard V. Reeves at Brookings.