Elite Colleges & Universities with Highest Percentage of Small Classes (under 20 per class)

@theloniusmonk: If CS (computer science) classes ranging from 200 students to 650 students are “about right for most schools”, then what is the value in attending a university in person ? Couldn’t distance learning serve the same purpose ?

It would depend on how the class was structured but sure it could definitely serve the same purpose. And if it’s more affordable, that’s better all around. Stanford offers CS classes through Coursera, an online leaning platform, if it’s good enough for them, I’ll take my chances it’s good enough for others.

I think distance learning is great for lectures, but there is more to an education than lectures. Required discussion groups, projects, and poster sessions add considerable value. With CS the need to have access to computers including CPU and GPU clusters is an increasingly important component of colleges. Anyone interested should check out www.edx.org just to see some of whats available online from leading universities.

In a 650 student class on campus, you may have one faculty member, ten TAs, 649 other students, and some number of other students who have already taken the class whom you can ask questions with.

The primary reason why CS often has larger classes is because it’s such a popular major, and most highly selective private colleges have near open major enrollment. As an example, the majors with the largest 2018-19 at Stanford are listed below. CS has a similar enrollment to the rest of the top 5 most popular majors combined, and a larger enrollment than any other major during the 50+ years for which major stats are available (being a history major during Vietnam is 2nd).

  1. CS -- 739 Students
  2. Human Biology -- 219 Students
  3. Engineering -- 205 Students
  4. MS & Eng. -- 181 Students
  5. Symbolic Systems -- 170 Students

Stanford caps enrollment in few classes so they have little control of how large lectures get. However, Stanford has a greater degree of control in how large to make their post-lecture sections, which often does vary significantly from major to major, as well as from class to class within the major. How much Stanford and professors teaching the class feel that students will benefit from more personal interaction in smaller groups probably contributes to this section size.

A summary of the average section size for CS undergrad classes (numbered below 200) with 200+ enrollment during fall 2018-19, which have section information listed. CS 106A seems to have an especially small section size, which may relate to typically being the first intro to CS type class that students take, including students from a wide variety of backgrounds and planned majors (many taking the class are not planning to major in CS). Stanford seems to feel there is benefit to smaller section size and more personal interaction in CS, including things like making it easy to ask questions and review coding work in a comfortable and more personal environment, even though they still have the traditional large lecture format.

CS 106A: Programming Methodology – Average section size = 6 students
CS 106B: Programming Abstractions – Average section size = 12 students
CS 109: Intro to Probability for CS – Average section size = 10 students
CS 161: Algorithm Design – Average section size = 14 students

Average section size in CS is often similar or larger in less STEM focused classes. For example:

English 91: Creative Writing – Average section size = 13 students (no core lecture class)
Philosophy 150: Mathematical Logic – Average section size = 21 students
Psychology 30: Perception – Average section size = 30 students

Stanford seems to think there is less benefit to being able to easily interrupt the large lecture and ask questions, but some other colleges approach this differently. For example, I took some of my intro math classes at SUNY and some at Stanford. SUNY tended to use the approach of smaller classes and being less picky about who teaches. For example, my MV calc at SUNY was taught by a PhD student. Other PhD students taught similar MV calcs that semester. My SUNY MV calc class had ~30 kids in the class without sections. The corresponding class at Stanford was taught by a well known professor, rather than a PhD student. It was much larger with over 400 students enrolled and was divided in to many sections, led by graduate students. I don’t think one approach is necessarily better or worse than the other. Instead I think it depends on the specific class, as well as what type of learning a specific student prefers.

The greatest benefit of small classes typically isn’t in CS or other STEM fields, but in the humanities and social sciences, where meaningful discussion involves more than simple Q&A.

A future “thought leader” can expect much of one’s most important work to occur either alone, with a partner(s), or in small teams. You can expect board, department, or project team meetings to address problems that don’t have clear-cut answers and that must be addressed from multiple perspectives. Large lecture classes have their place, but aren’t the best preparation for that kind of discussion and decision-making.

Nor is the main benefit of small classes just so a professor can keep track of whether students have done their homework. At most “top” colleges, most students most of the time will do the reading before they come to class. The question is whether they’ve thought about it. In a small discussion class with a good mentor and rich primary source material that does require some thought, it will be painfully apparent if you haven’t.

All else being equal, smaller classes are better, even in STEM. I recall taking an abstract algebra course with a great math professor in a relatively small class (no more than 25 students, I recall). The lectures were interactive. Students were encouraged to ponder along with the professor, offering proofs or alternative proofs, for problems or theorems during the classes.

If LACs believed STEM classes should be as small as some here are advocating because of some perceived benefit, they would make them so. Yet most of them don’t. They might not be as large as identical courses at a flagship, but they aren’t 20 students. Not even close. Same for some intro social science classes.

@itsgettingreal17 I think I’d amend your statement a little: If LAC’s believed that the benefit was worth the cost, then they would make them so. Yet most of them don’t.

I can see a situation where the cost of doing so outweighs the benefit, but there still is a benefit. Nobody can convince me that Mark Z. teaching a class of 500 would be a better learning experience then him teaching a class of 20. If Mark Z. gets paid $100k for each class, then obviously the cost might far outweigh the benefit.

All the examples of “my son’s favorite class was the big lecture on x,” or “my best class was a huge CS class on y” would honestly ask themselves “Would I or my son have enjoyed or learned more if the subject was delivered in a small format would it have been better?” I really can’t see a scenario where that answer is usually no. Maybe, people who don’t like discussion or intimacy, but then I struggle to see them successful in the work environment. Very few people work daily in 500 person meetings.

So, if small classes are better, then schools that offer the majority of their classes in a small format I think should get the benefit of that in the ranking algorithm, which I think is the point the OP was making.

As a counterexample, Hamilton caps its introductory computer science classes (CS110, 111 and 112) at 20 students.

https://www.hamilton.edu/academics/departments/Courses-and-Requirements?dept=Computer%20Science

Adding: Bowdoin caps most upper level CS classes at 16-22 students. Intro to CS, of which there are 4 section offerings this semester, is capped at 30.

Amherst limits even Intro CS to 20 per section. Currently it looks like they offer 3 sections per semester which is one more than last year. If it’s over-enrolled first and second years get preference.

That doesn’t seem to be unusual based on other posts here.

“All the examples of “my son’s favorite class was the big lecture on x,” or “my best class was a huge CS class on y” would honestly ask themselves “Would I or my son have enjoyed or learned more if the subject was delivered in a small format would it have been better?””

People really don’t ask these kinds of questions after a great college class. They fill out an evaluation that says the class was great, the professor was great, I learned a lot. Then the college or dept offers more large classes (or large lecture with smaller sections for problem sets). Anyway, why would you revisit the choice of class size after a good experience like this? What they do instead is say, hey class size doesn’t matter at least for STEM.

“So, if small classes are better, then schools that offer the majority of their classes in a small format I think should get the benefit of that in the ranking algorithm, which I think is the point the OP was making.”

Any ranking that says choose another college besides Stanford or MIT for computer science or engineering is seriously flawed if they use class size percentage as any kind of factor. It doesn’t happen, if you get into either, you pretty much go and well, deal with those large lectures.

“All else being equal, smaller classes are better, even in STEM.”

Ok, just because people keep posting this doesn’t make it true. And what do you mean by all being equal, affordability, or quality of department? Again, Stanford has 68% and MIT 70%, comparing them to any college ahead of them on the OPs, is not all things being equal, because the prestige is not the same.

@BrianBoiler I don’t think that smaller classes are inherently better for STEM. Maybe a weighted score that included size, prof expertise, and support.

What does it do if more students want to take those courses than they planned for?

  • Increase class size?
  • Cancel an upper level course to have a faculty member teach another section of the lower level course?
  • Hire an adjunct to teach another section of the lower level course?
  • Ration seats in those courses somehow, so that not all interested students can take them (and therefore may not be able to choose the major)?

@Rivet2000 and @theloniusmonk I agree one should not just use class size. I don’t think the OP thinks that either. I have 2 STEM degrees, my son is close to a CS degree. I disagree that size doesn’t matter. I believe it does (insert popular Michael Scott Quote here). I also believe that it might not matter enough to overcome the cost to a university and they are smarter than I am and have decided to go big class size for STEM.

There is also agreement between me and you that Stanford and MIT overcome their decision to have larger class sizes. I also agree that given the chance to study CS at Stanford vs. a small class size school only because of class size would be foolish. But, there could be other reasons to overcome the “prestige” factor.

And finally, “Ok, just because people keep posting this doesn’t make it true.” The same for your side of the discussion.

If you are looking for some bona fide research on it. 2 minutes of Googling popped up the following article:

“Small Classes Reduce Performance Gaps in Science.” The University of Minnesota, June 27, 2018

So maybe there is more “than people keep posting it?”

Why is the discussion stuck on CS. I said STEM generally. I spent 30 seconds and found Amherst intro bio class to be 96 students. It’s the same at most LACs.

And how about this:

De Giorgi, G., Pellizzari, M., & Woolston, W. G. (2009). Class size and class heterogeneity. IZA Discussion Papers, No. 4443. “Our baseline results suggest that increasing class size by 20 students reduces a student’s wage by approximately 6%. If we trust such estimate, it would be hard to dismiss class size reduction as an ineffective and inefficient policy….Such an intervention [reducing average class sizes to 20 students] would generate a gain of 80 euros x 1,500 students, or 120,000 euros in total each month, which are likely to be more than enough to pay the costs of acquiring the additional resources necessary to activate the two extra classes.”

This is not STEM specific as the above study is.

@BrianBoiler here’s info from the article itself. It doesn’t constitute what is small.

So is 500 better than 100? Perhaps But doesn’t necesssrily mean 30 better than 100.

And actual the article addresses performance gaps between genders and URMs as sub groups.

And was their own conclusions clears up little imho. It confirms what many have said here. it depends. And also the data isn’t clear. And does it really matter? Or is everyone backing into corners justifying predispositions or personal choices.

But. Yes of course smaller is better,most likely, but not tiny versus reasonable like the arguments are evolving here.

And schools like Amherst as proxies with tiny enrollments and homogeneous academic abilities isn’t useful for a broad discussion in the real world. Amherst is better than large state u? Depends on who’s paying, what is the mission and who is going. Look at what the researchers below state.

From quoted article…

“small class sizes closed the gap in academic performance between men and women;

smaller classes did not impact underrepresented minorities, who underperformed compared to students from well-represented backgrounds in STEM regardless of class size;

the impacts of class size do not generalize to all students, suggesting other characteristics of the education environment and incoming preparation affect learning.”