Matching Markets at Stanford
This article discusses the roommate selection process at Stanford University, which is conducted by two undergraduate housing coordinators. Incoming students answer a number of questions about their level of desired social interaction, cleanliness, extra-curricular activities, music preferences, study habits and sleeping habits. Then, two undergraduates—Kelsey Davidson and Dan Miller—look first at the bedtimes, cleanliness levels and desired social levels of incoming students and pair roommates with similar criteria. The next criteria used to pair roommates is placing students together with similar interests, but also keeping enough differences and diversity so that the students can learn from each other and grow as individuals.
The process of selecting roommates relates to the concepts of matching markets and matching with preferences as discussed in lecture. The idea behind matching markets is to place each person with their favorite item and create a perfect matching. In a perfect matching, there is a set of edges with all nodes in exactly one edge. In this example, a perfect matching occurs when each incoming student is paired with another incoming student as roommates. However, with such diversity and large number of incoming students, finding a perfect matching can be challenging.
The housing coordinators could assign values to potential roommates with certain characteristics, which would allow the housing coordinators to determine the quality of each roommate assignment. Values could be determined by creating preferences for a roommate based on the criteria listed above (such as study and sleep pattern preferences). Roommates who best meet the desired criteria of the incoming undergraduate would be assigned the highest values. Then, students would be matched with a roommate to whom they have the highest value (optimal assignment) until every incoming student is assigned to a roommate. Therefore, the roommate selection process at Stanford is one real-life example of matching with preferences. In this example, individuals are not acting in their own self-interest to maximize their valuations for a roommate to produce optimal assignment. The housing coordinators make the executive decisions about creating roommate assignments by collecting data from incoming students. If individuals chose their own roommates, it would be interesting to see how values and “prices” would be assigned in order to clear the market.
http://news.stanford.edu/news/2010/september/freshman-roommate-match-091410.html