MATCHING TAs TO COURSES IN THE CS DEPARTMENT ISMAIL KHAN MOHSAN ELAHI Preview 1. Market Overview
MARKET OVERVIEW UTA Market Two-Sided Market Applicants (aspiring TAs) Courses (represented by professors & HTAs) Many-to-one matching
Market size (Spring 2012): 94 TAs hired 20 courses hired TAs Many rejections: CS 4 hired 9 from 63 applicants CS 16 hired 12 from 44 applicants Administered by Meta TAs
Stages of the market 1. Application stage 2. Ranking stage 3. Hiring Meeting and Notification Application Stage Applications submitted online, available for 12 days
Applicants apply to as many courses as they like Applicants strictly rank all courses that they applied to These rankings are hidden at this stage Other information includes: Courses taken Prior teaching experience Grades Schedule for coming semester
Ranking Stage, Part 1 Courses have 15 days to submit a rank-ordering of all applicants Courses must interview all candidates Professors and HTAs hold the interview Test for enthusiasm, knowledge, teaching skills
Interview summary: Why do you want to be a TA for ____? What improvement to the course do you suggest? Mock office hours demonstrate teaching skills by pretending to talk to student Ranking Stage, Part 2 Courses rank all applicants
Courses rankings must be strict Courses may provide a threshold, under which all students are deemed unacceptable i.e., they would rather have a position unfilled than to hire the applicant At this stage, the preferences of both courses and applicants are hidden
UTA Hiring Meeting Hiring Meeting All HTAs meet after courses have submitted preferences Preferences for both applicants and courses are revealed Hiring Process: 1.
2. 3. 4. Simple case: All applicants that are acceptable to only one course, and who find only one course acceptable, are placed in a course
Complex case: For all applicants who are acceptable to multiple courses, applicants are placed in the course that is highest on their preference list. Every-man-for-himself: If a course did not meet its quota, it negotiates with other courses to get more TAs. Settle: Courses settle for the top available candidates that are within their quota
Notification Notification Applicants are informed of their match Applicants can accept or reject the offer, but rejecting an offer is looked upon unfavorably in future applications. Proposed Change Current mechanism: Professors judge candidates using
interviews and personal information of candidate Private value - Courses preferences are hidden until the Hiring Meeting. Proposed change: Introduce TA evaluations All professors have access to evaluations
Professors now have much more information during the ranking phase, and they share this information Common Value Professors now know each others preferences FITTING THE MARKET TO A MODEL Stability and Preferences
Assume strict, responsive preferences Hiring meeting is a stable matching mechanism Individually rational: No student is matched to an unacceptable course and vice versa Stable: No blocking pair Due to strict preferences, no blocking pair is formed and resulting matching is group stable (Lemma 5.5)
No blocki DIVING INTO THE LITERATURE References
Possible directions? We focused on having complete information what about having similar preferences as well? Procedure 1. Apps due by 12 Nov No signals
State preferences (invisible to professors) Courses, experience, grades, schedules 2. Interviews Test for enthusiasm, knowledge, teaching skills Dont know preferences Must interview all candidates 3. Rank candidates Strict preferences
Rank applicants: some applicants may be unacceptable Other courses preferences are unknown 4. Submit preferences by 27 November 5. Hiring Meeting All HTAs meet & fight over applicants (individually rational) Resulting matching is stable (otherwise, courses would collaborate and swap)
