One For All

2021-07-02
19 min read
I'm recruiting a Ph.D. student. Find out more here.

Teaching large courses offers the potential for huge impact. If you can teach 1,000 students as effectively as 100, you’ve increased your impact by an order-of-magnitude. Multipliers like that don’t come around too often in life. And in a day where college costs and inequality are both skyrocketing, teaching more students more efficiently can also be very meaningful. In Capital in the Twenty-First Century, Piketty identifies technology and education as the two primary factors pushing back on inequality. Using technology to educate more students more efficiently is my (small) way of joining this important effort.

But of course there’s a catch. As class sizes increase, it’s common for instructor impact to actually decrease. Avoiding this is the key to effective teaching at scale.

It’s one for all versus all for one.

An effective way to think about this dynamic is to divide the work that you do as an instructor into two parts: things that benefit all of the students in your class, and things that benefit a single student. Of course there are things that fall somewhere in the middle. But this is a useful dichotomy for considering what happens when courses scale. It’s one for all versus all for one.

When a course size grows, the things that you do that benefit all of the students in your class scale to benefit even more students. If you spend an hour preparing lesson materials, then it benefits 100 students if the class size is 100 but 1000 if the class size is 1000. One for all scales.

But when your course gets big, you may also find yourself needing to spend more time on the things that benefit only a single student. If you spend one hour answering student email for a class size of 100, you’re going to be in big trouble when that class grows to 1000. All for one doesn’t scale.

Minimize all-for-one so you can maximize one-for-all.

So here’s the one quick trick for effective teaching at scale. Control the amount of time you spend helping individual students, so that you can preserve time to spend on things that help all of your students. Minimize all-for-one so you can maximize one-for-all. Do that, and you can teach larger and larger courses with ease, and have more and more impact as you do.

Last year I was the single instructor assigned to a 1200-student course in the fall and a 600-student course in the spring. I had a blast! My goal here is to share some tips with you that you can use to help improve your own ability to teach large courses and increase the impact of your own teaching.

Figuring out things to do that benefit all of your students isn’t usually that hard: creating new instructional content, updating assignments and assessments, monitoring student performance, studying your own course to glean new insights about how to improve. This category also rightly includes work outside your assigned courses, including longer-term projects that improve the educational experience for students in ways that touch multiple courses or address the entire curriculum. These are extremely valuable activities and equally vulnerable to all-for-one demands.

In any case, I’m going to guess that you already know what one-for-all activities to pursue. Let’s focus on how to keep the all-for-one activities from taking over your schedule.

I should note that one of the tools I employ here is my technological prowess. There are definitely places where I’ve created software to address all-for-one scaling bottlenecks. That approach may or may not be available to you.(1)

However, in my experience the most valuable approaches for addressing all-for-one scaling bottlenecks are not technological.

However, in my experience the most valuable approaches for addressing all-for-one scaling bottlenecks are not technological. They comprise policy and behavioral solutions that you can and probably should deploy if you are or plan to teach a large course. These are the ones that I’ll list here. And while I’ve used them to allow me to happily teach 1000 students per semester, they should be equally valuable if your goal is to teach 100.

Let me also point out that no one quick trick replaces sufficient staffing. There is important instructional work that does scale with the number of students in a course—particularly some of the most valuable student-facing components, like providing one-on-one support. That said, reducing the administrative burden of the all-for-one tasks will leave your entire staff with more time to assist students and work on more meaningful instructional tasks. Since this is what your staff signed up to do, freeing them to do more of it improves staff motivation and morale. And, inevitably, even if you hire a fantastic course staff, there can be certain tasks that you just can’t delegate, and will find their way up the food chain.(2)

When I tell colleagues that I teach 1000 students, frequently a look of horror passes over their face. “Wow. I’m so sorry…” they stammer as they start backing slowly out of my office. I have a feeling that I know what immediately comes to mind, what they think I must need to get back to right away, and what they worry would rule their life when teaching huge courses. Email.

Answering individual student email is a great example of an all-for-one task that simply doesn’t scale as the number of students in a class grows. And things can get ugly quickly. If you are spending 15 minutes per day on student email with a 100-student class, look forward to spending 2.5 hours per day on student email once that class grows to 1000 students. Yes, you can find more people to help answer email with you—and perhaps this explains why some institutions feel the need to throw more instructors at large courses, just to help answer email. But wouldn’t it be better for everyone if we could just reduce the amount of student email, rather than recruiting an expensive army of course staff to reply?

As it happens, I do get a handful of student emails. But not per minute, per hour, or per day. Per week. I do get more earlier in the semester. But it turns out that training students to not use email as the primary form of course communication is surprisingly easy.

When students email me early in the semester, 99% of my replies are one of two things: a link to the syllabus, or a link to our course forum with a message politely asking them “Please ask questions like this on the forum.” These replies identify the two core strategies that you can use to drive email to near zero.

First, you need good online materials that address most if not all student questions.

First, you need good online materials that address most if not all student questions. That includes a detailed syllabus, course calendar, and anything else needed to address contents of the emails you receive. There’s a fairly simple approach to take to the questions that relate to course policies: if a student asks, redirect them to the appropriate part of the syllabus or other online materials. If that part doesn’t exist, write it, and then redirect them to it.

Your goal should be to make it easier for students to find what they need rather than email you directly. So you’ll probably also need to address other usability issues with your online materials. Are they buried behind a confusing LMS system that is difficult to navigate, or that requires some complex login process? Do they require downloading and opening a PDF? Are they searchable—nothing fancy necessary, just using standard Control-F? If not, then I would suggest that you improve the accessibility of critical information about your course. This is not rocket science, and there is no reason getting huffy about it. Yes, a student could have found that answer themselves. But the question that matters is: Will they.

Making it easier for them also allows you to respond more directly. Sometimes all I send in reply is a link to the syllabus. That’s both faster for me, and also carries the implicit message: Maybe you should have looked this up yourself?(3)

The second thing that you need to help decrease email is a good forum. (Sadly, Piazza is neither a forum nor good.)

Redirecting student email to a forum does two things. First, it teaches them the right way to ask that kind of question. That’s valuable in and of itself.

But it also accomplishes something even better: It transforms an all-for-one activity (responding to an individual email) into an all-for-one activity (responding to a post on the forum). I will even use this trick when students ask me an interesting question in person. I’ll respond: “Post this on the forum and I’ll answer!” That frees me to write to devote more time to a thoughtful response, since I know it benefits all students, rather than just the one standing in front of me or involved in a specific email exchange.

And by redirecting questions to the forum, I also pull in my excellent course staff, which frequently chime in with their responses. Frequently by the time I arrive the student’s original question that I received over email has been answered multiple times in way more depth and detail than I would have, and all I need to do is do some clapping and maybe further endorse the existing responses. Teamwork really does make the dream work.

Do these strategies work? Yes. Do students tolerate them? Yes.

It also helps to have very clear policies regarding course communication. We have very specific guidelines about when students are allowed to directly email the course staff in our syllabus. More importantly, we also have language explaining these policies, and how they end up benefiting the entire class. It’s not really that hard to explain to students that me spending 40 hours a week responding to email is not a good use of my time, or the best way for me to teach them all as effectively as I can.

I almost never make exceptions.

But there is another reason that I don’t get much email, and it leads us to the second strategy for limiting the impact of all-for-one interactions. I almost never make exceptions. I provide absolutely zero exceptions on homework deadlines (30%), few-to-no makeup quizzes (40%), and very few extensions on our long programming assignment (30%). As you would expect, this reduces the number of these requests, and since handling these requests is another all-for-one activity, it leaves me more time for one-for-all work.

I can already hear you thinking. But! But! But! But what if a student gets sick? Or has a death in the family? Or is out of town for a school-sponsored event? Or oversleeps? Refusing to consider requests for exceptions seems like a heartless choice that ignores the vicissitudes of life that students experience each semester.

But instead of providing individual exceptions, I’ve developed an exception policy. When grading we drop multiple low homework and quiz scores. Typically we’ll drop 12 out of the 75 daily homework problems—essentially meaning students don’t need to complete them, since these are all-or-nothing grading—and the lowest 3 out of 15 quiz scores. I can’t claim credit for this particular policy. But I love pretty much everything about how it—particularly when teaching large courses.

First and foremost, it’s efficient. Students don’t need to request a drop, they are simply automatically applied when we compute grades. I’ve heard of courses requiring a part-time staff member at least in part to handle requests for extensions and exceptions. To me, this is insane, and completely unnecessary.

It’s also fair. Having to consider exception requests requires staff make strange choices. A student wants to miss a quiz so that they can participate in an e-gaming tournament. A second student wants to miss a quiz so that they can participate in a golf tournament with the university team. The second student has a letter on official university stationery. The first student does not. Why am I making this choice again? Is the official letterhead significant? Is golf (which I find boring) more exception-worthy than e-gaming (which I also find boring)? Would I give a drop if the student was traveling to see Flock of Dimes, which I don’t find boring?

The drop policy takes me and my course staff out of the business of making these decisions entirely. Not only does this free up time, but it is also psychologically healthy. You hear a lot of cynicism among faculty surrounding student excuse-making. Grandparents who die but then, miraculously, show up for graduation. Doctors notes with suspicious misspellings. The fact that the university employs people to write excuse letters for athletes.

I’ll believe all of your excuses. But then redirect you to the drop policy.

I’ve had students who contact me about a death in the family. Should I be keeping a spreadsheet to track how many of their relatives are still living? But maybe they are adopted but then found their real family later in life, and so actually have some extra relations? And shouldn’t my primary response in this kind of situation be, simply: I’m so sorry for your loss. I love teaching and my students, and don’t want to be dragged down by this kind of paranoia. I’ll believe all of your excuses. But then redirect you to the drop policy.

This policy is also easy to justify. There is a limit to the amount of work a student in my course can miss before I am no longer confident that they have met the learning objectives. The drop policy represents that limit. Dropping low scores also helps reduce testing anxiety by eliminating outliers that might result from exogenous factors, and by allowing students a bit of leeway in learning how to prepare for our assessments and do their best in the quiz format.

When you look at the data you find that most students don’t utilize all of their drops. This is both good, and makes sense, since students will naturally save their drops in case they need them later in the semester. I also remind students that their drops are a hard limit on the number of homework and quizzes that they can miss. It doesn’t mean that you can just skip out on three quizzes, since if you do and then get sick later, you’re out of luck.

It’s important to acknowledge that the effectiveness of our drop policy at nearly eliminating requests for exceptions is connected to our use of frequent low-stakes assessment: in this case, 15 quizzes together worth 40% of the student’s grade, and no high-stakes assessments like a summary midterm or final exam.

When you give high-stakes assessments, you don’t give many, so it’s hard to have a policy allowing students their three lowest scores. You end up needing to make exceptions for students to take those high-stakes exams. And it’s also worth pointing out that you may end up encouraging students to find ways to delay those high-stakes assessments to try and achieve a competitive advantage, either through a bit more study time or information about the exam from students who took it on schedule.

Things can spiral downhill here rapidly in ways that really highlight the all-for-one versus one-for-all divide. Writing good assessments takes time. If a handful of students misses your high-stakes midterm, what do you do? You can give a makeup, and just accept the unfairness inherent to that approach. Meanwhile, as the students who missed the exam stall for time, the entire class is prevented from reviewing the exam questions and answers which would really benefit their learning. Or you can write an additional exam, at which point you are repeating all of the work that you put into the first exam for the entire class for just a handful of students: one-for-all has become all-for-several. And usually the make-up exam is similar to the original, so you may still not be able to release information about the exam to students who have already completed it. It’s just a mess. No wonder faculty get cynical and frustrated by excuses, even when they are perfectly legitimate.

There really is a problematic tendency for faculty to abandon the collective in favor of the individual

There really is a problematic tendency for faculty to abandon the collective in favor of the individual in situations like this. I think that the reasons are clear: Helping one student has specific direct interpersonal benefits for the faculty member, whereas helping the collective typically does not provide the same emotional payoff. I’ve witnessed colleagues expend indefensible amounts of time and energy implementing exceptions for individual students in the midst of teaching a large course, time and energy that could easily have been put to better use. We tend to valorize this kind of behavior, rather than challenge it. But that doesn’t make it right.

But if the promise of not having to read doctors’ notes encourages you to give frequent low-stakes assessments a try, then great! Whatever works! Frequent low-stakes assessment has a lot of benefits. This is definitely one of them.

The next all-for-one activity to discuss is recommendation letters. This topic has a bunch of sharp edges. Let me see if I can cut myself on all of them.

My policy is that I do not write letters of recommendation for students that I have only had in class.

My policy is that I do not write letters of recommendation for students that I have only had in class. I just don’t know them well enough.

I used to say yes and prepare a form letter, which didn’t really say much more than what someone could have easily found on the course website. But here the problem is that the systems that are used are incredibly disrespectful of faculty time. If all I had to do was write a letter, I’d probably do it.(4) But there are endless accounts to create and stupid similar-but-not-the-same forms to fill out. It quickly becomes a bigger time sink than it should be.

I do write straightforward letters for the hundreds of students who have served on my course staff.(5) And I will write excellent, strong, and very detailed letters for senior course staff who I know well and whose contributions I have had a chance to observe directly. I once wrote one so strong for an internal award that someone emailed me to complain about it.(6)

It’s hard to raise this topic without commenting on the insane proliferation of letter requirements which now seem to accompany applications for almost everything—or at least everything in academia. Does a rising sophomore applying for a summer research experience really need multiple or even a single faculty letter? How, exactly, are they supposed to obtain that letter? I think we’re well past the point where recommendation requests are just generating more mad-lib letters that add nothing to the students resume and transcript. This insanity seems also designed to enforce the myth that the most valuable relationships that students can form in college are with faculty, which is flattering but absolutely false.

As you teach larger courses, you also recognize the fundamental paradox at the heart of this economy of recommendation letters. As you receive more letter requests due to teaching large courses, your letters are inherently less valuable.

I can’t unilaterally stop this madness. But I can control how I choose to participate. My responsibility is to all of the students that enroll in my course. Writing pro forma recommendation letters just robs too much time from my one-for-all activities.

Those are my three primary strategies for reducing all-for-one activities that don’t scale as the course size grows:

  1. Eliminate email.
  2. Excise exceptions.
  3. Refuse recommendation requests.

Together, these put me in a position to devote more time to one-for-all activities that benefit all of my students and let my course scale further. I think I could comfortably teach 10,000 students per semester. I’d like to try that one day.

Maybe this all seems a bit or even quite impersonal to you. Maybe you enjoy interacting with students and worry that you’ll miss that if nobody is emailing you about anything. A lot of us got into teaching because we like working with students one-on-one. Scaling your course does involve forgoeing some of that kind of interaction in favor of one-for-all work. But surrendering all of it seems like a very high price to pay.

By reducing low-value all-for-one activities you create space for more meaningful all-for-one activities

But here’s the best part: By reducing low-value all-for-one activities you create space for more meaningful all-for-one activities without jeopardizing your one-for-all goals.

As an example, I still hold a few office hours a week. Very few students attend, but that’s fine, since it creates the opportunity to actually get to know the ones that do that much better. From the perspective of our all-for-one versus one-for-all framing, this is clearly more on the all-for-one side of the scale. Even if it’s open to all students, only a few are actually benefiting. And yes, I could use those hours to work on things that would benefit the entire class.

So this is hard to justify from our all-to-one versus one-for-all optimization perspective. But that’s OK! Because I enjoy office hours and the chance to meet and interact with students in my class. I’m in my class, too, and feel free to arrange at least a few things for my own benefit and enjoyment. And I feel even more comfortable engaging in this kind of indulgence knowing that I’ve created enough time for the one-for-all activities that benefit everyone: even the students I’ll never meet or hear from.

I remind myself of this a few times each semester: Optimizing the 99% means that I have more time and energy to spend on the 1%. So when I get an usual or interesting email from a student, I have time to respond fully and thoughtfully. I have time to write excellent recommendation letters for students I do know well. I have time to have great conversations with students who stop by my office during office hours. And I have time to work on higher-level issues like working with colleagues to improve the entire curriculum. And I have time to write a lot of code! Meanwhile I’m teaching 1000 students extremely well, and enjoying every minute.

Thanks for reading! I'd love to hear your take. Feel free to get in touch.