Students Who Frequent Office Hours

The phenomenon where a minority of people use up the majority of available resources is known by many different names: power laws, 80/20 rule, Pareto principle. No matter what it’s called, it definitely applies to how students use office hours at Berkeley. The majority of students never attend office hours or attend once or twice in the entire semester, while a small number of students use office hours weekly, if not daily, through the entire semester. Why is that, and what can we do about it (if anything)?


I have two broad theories behind why a small subset of students use office hours so frequently. They’re not mutually exclusive, and the true reason probably lies somewhere in between.

Reason 1: Students try to extract answers from staff in office hours by asking vague questions and asking questions repeatedly to different staff members until they get answered. I think it’s important to note that not all students are doing this maliciously–it often stems from not understanding how to ask the right questions to get unstuck, a misconception that learning how to debug isn’t part of the assignment, or stress from other classes or extenuating circumstances leading students to try and finish assignments as quickly as possible at all costs. Nevertheless, dealing with students attempting to extract solutions is something that we discuss frequently with TAs, including in the pedagogy class for first-time TAs.

Reason 2: Students have a fundamental misunderstanding of the material that can’t be solved in a short office hours ticket. This is related to the idea of thrashing in office hours–in a model where each student gets a small bit of advice or help each time, a student who hasn’t fully grasped the fundamentals yet will be stuck repeatedly requesting help from different TAs, instead of being able to sit down for a longer tutoring session that would help them more in the long run. There might be a correlation between students who are thrashing in office hours and students who are struggling in school in general (e.g. extenuating circumstances, being from a marginalized background, etc.). Since our classes follow a somewhat MOOC-like model that requires some self-learning, struggling students might not have enough experience with self-learning or enough free time to efficiently self-learn the material.

Allocating Limited Resources

A small proportion of students using most of the resources becomes a problem when the resources are limited. This is the case with office hours at Berkeley–it’s not uncommon to wait over 2 hours in a queue before getting any help. Even in classes that have allocated as many staff hours to office hours as possible, the average wait time is probably still around 20-30 minutes, and it’s still not uncommon to wait around an hour before receiving help.

Most classes use a first-in, first-out queue to allocate limited OH resources. This queue does not take into account how often a student has used office hours in the past–whether it’s your first time or 100th time this semester, your queue is processed with equal priority.

However, knowing that a small number of students are making the majority of help tickets, it might be worth revisiting the first-in, first-out queue and really consider what it means to fairly allocate these limited resources to all students. For example:

  • Is it fair for someone who’s gotten helped 100 times this semester to get helped for the 101st time, ahead of someone who’s asking for help for the first time this semester? If we consider this unfair, then maybe some priority for infrequent users is needed.
  • If there is a correlation between struggling students and frequent users, do we want to focus our resources on struggling students who need help the most? If so, then maybe some priority for frequent users is needed.

Also, what we as staff consider fair may not line up with what students consider fair. This might explain why the first-in, first-out queue has been around for so long, and why no attempt at changing this paradigm has succeeded–from a student’s perspective, this is the fairest way to distribute resources.

One situation where the first-in, first-out queue breaks down is at the start of office hours, when there is often a flood of students all requesting help. If help tickets are processed in order, and a dozen students make a help ticket at pretty much the same time, the first student might end up getting help an hour earlier than the last student. Some classes have tried to mitigate this by randomly shuffling the tickets when there’s a flood of tickets created around the same time, but this presents another scenario where we have to consider what “fair” allocation of resources means.

Limiting Help Tickets

If we believe that there’s a point where repeated help tickets are either unproductive thrashing or attempts at extracting answers, then we could consider an upper bound of resources that any one student can use. This could be considered fair in the sense that every student is working with the same limit, and the limit ensures more equal spread of resources to all students. However, if the students who end up hitting the limit are also the students who are struggling the most and need help, then this system might encounter other fairness and equity issues.

How would this upper bound look in practice? We could try to limit the number of tickets a student is allowed to make. If we set the limit to be high enough, only a small handful of the most frequent users would ever get close to the limit. For example, with a limit of 100 tickets over a semester, a student would have to attend office hours nearly every day to hit the limit.

There are a few problems with limiting tickets. Even if we think the limit is high enough that most students won’t reach it, the existence of the limit might discourage students from asking for help when they actually need help, for fear of “wasting” a ticket. This might disproportionately affect marginalized students who are more reluctant to ask for help in the first place. Also, the limit might incentivize students to try and maximize the help from each ticket (e.g. making one ticket to ask questions about several different assignments), which would cause each ticket to take longer. Students might also be unhappy when a help ticket is short or ends in the TA asking the student to come back later, which is often necessary when the student needs to do some individual work before getting more help.

Instead of limiting the number of tickets, we could also try to limit the time between tickets, so that a student who has already gotten help has to wait a certain number of minutes before getting help again. However, this approach may also incentivize students trying to maximize the help from each ticket.

Limiting tickets may also not solve the root problems of students thrashing between different TAs and students who try to extract answers and don’t try to solve problems on their own.

Contacting Frequent Users

Since this resource allocation issue is mainly caused by a small subset of students, there might be a solution that only involves those students, instead of a blanket policy that affects all students (e.g. limiting tickets). One possible approach is to reach out to the most frequent users, though we would have to be careful about what we say. If the wording is too strong, struggling students might be discouraged from asking for help in the future. If the wording isn’t strong enough, the message might be ignored and ineffective.

Ultimately, there isn’t one clean answer to this 80/20 power law phenomenon, and it’s not clear if this phenomenon is even a problem that needs to be solved in the first place. But it’s still worth noting that it’s a result of our current office hours model, and maybe a different model will deal with this phenomenon differently.