When demand for TAs exceeds available TAs on staff, who gets help?
I face TA resource allocation problem in almost every class I teach. It doesn’t matter much how many TAs I have: there are always high-stress times when there are many more students wanting help then there are TAs to help them. Roughly, this arises because
TAs help with assignments
Assignments have deadlines
A large percentage of students do not finish early
I could discuss ways to mitigating these causes, but instead I want to discuss how to handle the surplus demand when it arises.
First, let me say this: raising hands is a bad way to go. There may be people who can fairly handle a sea of hands, but they are certainly not common. Raising hands is a tool for when TA supply exceeds demand, a way of letting TAs rest between helping others.
In most of my classes, we have two related approaches to fair TA allocation. The first is generally used in labs, and is what I call “walk a path:” each TA moves through the room in a predictable repeating loop, pausing to handle questions of each student in order. The second is generally used in office hours, and utilizes a digital queue: students submit an “I need help” entry into a big master list and TAs handled them in a first-come first-served manner.
Both of these techniques suffer from several shortcomings.
Everyone except the currently-being-helped students want TA-student interactions to be short, so the list does not get too long. But the student being helped has the opposite motivation, since it will be some time before the TA will be able to revisit the student.
There is incentive to create false demand, signing up for help before any help is needed and asking questions when the TA is available whether or not the student has had time to think through the question.
Students that use TA help well (to get past problem areas and learn material) and students that use TAs poorly (to try to wheedle answers out or avoid thinking themselves) are given equal priority.
There are others, no doubt, but these are the ones that come to mind.
So, what else could we do?
One option that could be added is a TA help auction system. Each student is given a fixed number of “TA help chits” and they can bid on TA help; if you offer two chits and I offer five, I get the TA first.
A problem with this is that most people are bad at valuing things, and valuing present help vs future ability to get help easily is not something I’d expect even most trained economists to be able to do well. I’d also expect some thrashing: students bid on TAs but not enough for immediate help, make a little more progress on their own, see the TA coming and drop their bid, etc. And while this could balance TA access, it leaves in the incentive for long TA visits, for asking them every Q you can think of all at once, etc.
In short, auctions do not appear to be superior to queues.
What if the queue position was based on something other than time on the queue? Particularly with a digital queue tool, we can use many criteria:
We could track the total number of minutes each student has ever spent with TAs and whoever has the smallest number goes to the front of the queue.
We could track when a student last had TA help; the longer that has been, the closer to the front of the queue.
We could have TAs rank students’ questions, awarding them some kind of “Help Karma”, increasing it for questions that suggest previous thought and progress, reducing it for questions that suggest laziness or lack of preparation.
We could tie queue position to course grade in some way (though it is not immediately evident if high or low grades should get priority).
We could measure the average duration of a single TA visit for this student, putting fast-to-help students at higher priority.
And there is no need to have just one of the above; we could, for example, say that one minute on the queue for a good-karma student advances them as much as two minutes does for a bad-karma student and that initial queue position is the inverse of the time spent being helped so far.
It seems fairly evident to me that some form of priority queue could encourage better behavior than a simple first-in, first-out system. But I cannot readily point to which priority system is best.
The longer I teach, the more often I find myself looking at various aspects of course design, pedagogy, and assessment and asking myself, “what are the incentives inherent in this structure?” What am I training my students to be?
At the largest level, I often wish that school was a place where students came to learn. But while I think most students have a desire to learn in them somewhere, the value that employers and parents put on diplomas and GPAs creates a second incentive system that it is very difficult to align with the desire to learn.
But this challenge of creating good incentives carries down to the smallest details as well. How do you allocate TA help, or answer questions, or pick between lecture presentation and in-class exercises, or any of a myriad of other topics in a way that reinforces, rather than counters, good desires?
When I was hired, I had no idea that one of the most important jobs of an educator was to be a motivation engineer. Fortunately, it is a challenge I find stimulating, even if I can’t yet answer basic questions like “what’s the right model for an office-hour queue.”