Ivy vs. Ivy

Our latest infographic stars the eight U.S. Ivy League colleges. We used Hunch’s API to get predictions about each of them for 14 different questions, ranging from political ideology to personality traits. (See information about the data collection methodology under the infographic.) We then pitted schools against each other and displayed results on a relative/forced rank basis. Hats off to our Column Five Media partners for the cool design.

Keep in mind that Hunch’s Taste Graph is making predictions about people who say they like each respective college, not people who are necessarily students, staff, or alums. Still, we found some pretty interesting differences that will probably raise an eyebrow or two.

Some notes on data collection methodology

To get the rankings above, we started by using Hunch’s “get-predictions” API call, which returned a predicted correlation strength associated with how the fans of each school would answer each question. Then we computed the difference between the highest and lowest correlations and used that spread to re-scale the results to a 1-10 scale. This has the effect (by definition) of sometimes exaggerating small absolute differences. But hey, welcome to the Ivy League, where forced rankings and steep competition are a fact of life. Example: the difference between a high school GPA of 3.96 and 4.00 could easily be a 10-point student ranking.

The Taste Graph’s propagation algorithms are continually trying to get an accurate ‘resolution’ to its predictions based on all the data available to them. Ultimately, that means that correlation strengths tend to resolve and cluster around poles. For example, you’d likely see a cluster of numbers predicting which colleges will be strongly favored by extroverts and which by introverts, but typically you’d see fewer data points in the mid-ranges.

If you’re still reading by this point, you might just be interested in the absolute numbers and data tables used for all of this. Here you go.