Social Data and College Statistics

We correlate aspects of Twitter data with college statistics. We find that the amount of “buzz”about a college on twitter predicts the number of applicants to the college, even when controlling for the number of applicants in the previous year. We also explore various methods to classify the sentiment of tweets about a school. We find that the sentiment of insiders at a college predicts the freshman retention rate, but that this result is explained by average SAT score and school size. The sentiment of Twitter messages about a college does not predict the number of applicants, the acceptance rate, or the graduation rate. The paper adds to the growing literature on the predictive power of social data, documenting its strengths and limitations, and applies these techniques to a novel set of outcomes.