Like many institutions, colleges and universities are starting to use predictive analytics tools to make a variety of decisions, including to whom they should offer admission. Given that directly measuring whether or not a prospective student is a “good fit” is impossible, an admissions algorithm would have to rely on measurable proxy variables as its inputs. As DJ Pangburn, proponents of these tools argue that they offer powerful means of avoiding human bias in admissions processes. On the other hand, critics make the opposite case: they argue that data-driven tools often perpetuate bias, while giving the decisions reached a veneer of mathematical neutrality and objectivity.