Class Sizes – Better Summary Measures than USNWR?

College Search & Selection
1

<p>Class size is one (among many) comparisons that people make between various colleges/universities, in particular between small liberal arts colleges (LACs) and national universities, and also among various universities. But I’ve been frustrated by the slipperiness of the data provided in the headline statistics. A summary measure is provided in US News’ tables, where they give (1) the % of classes of 50 or more, (2) the % of classes of 20 or less, and (3) the median class size. For instance, Stanford has 10% of its classes over 50 students. But this doesn’t mean that the average Stanford student will be in a class of over 50 students only 10% of the time – actually, they will have a class over 50 students almost 50% of the time.</p>

<p>Why the difference? Intuitively, it’s because there are a lot more students in a large class than a small one, and therefore the large classes carry a lot more weight than smaller classes in the overall experience of the student body. The US News indicators do not perform such a weighting. So what I’ve done below is to show summary results below obtained by appropriately weighting class size data (from the Common Data Set for each institution; Harvard’s is from their viewbook data based on Fall 2006 courses). I’ve presented the results for some colleges and univs below, and followed with some more details on the rationale and methodology (and in case I’ve made some mistakes, any comments or corrections are welcome).</p>

<p>The numbers below present answers to the following questions:
What % of an average student’s classes would be over 50 students? Below 20 students? And what would be the average class size encountered? (although I think the averages are less useful than the >50 or <20 questions)</p>

<p>Some top LACs:
Swarthmore: 9% >50, 48% <20, avg. 27
Pomona: 5%, 50%, 24
Carleton: 5%, 42% 25
Grinnell: 0%, 43%, 22
Amherst: 16%, 49% 33
Oberlin: 13%, 44%, 33
Wesleyan: 20%, 36%, 38</p>

<p>HYPS:
Yale: 40%, 38%, 74
Princeton: 47%, 33%, 85
Stanford: 49%, 31%, 83
Harvard: 55%, 26%, 120</p>

<p>An explanation of the rationale/methodology: For a simplified example, assume that a college has 50 classes with 2 students in each class (i.e. total of 50<em>2=100 “student experiences”) and 10 classes with 90 students each (i.e. total of 90</em>10=900 “student experiences”). So there are a total of 1000 “student experiences”, of which 90% are in classes of 90 and 10% in classes of 2. In other words, the average student will have a 90% chance of being in a class above 50, a 10% chance of being in a class below 20. This is in contrast to the US News manner of presenting the numbers: they would say that only 17% of the classes were above 90 (i.e. 10 out of 60 classes) and 83% below 20. While their statement is true, I don’t think it is as useful as the other way of looking at the data.</p>

<p>Another way to think about this is that if each student has for example 10 classes (and hence there are 100 students), you could think of each student having, on average, 9 classes of 90 students each and 1 class of 2 students. And so the weighted average class size encountered by each student would be approx 81.</p>

<p>You can also calculate weighted average medians if you prefer those to the mean (in doing so, you should also interpolate the results so that you don’t have large discontinuities).</p>

<p>btw one assumption I had to make was for the average class size above 100 (since the CDS just says greater than 100). I assumed 150 for the LACs and 200 for the national univs (200 was Harvard's average class size above 100).</p>

2

<p>Very interesting. Do you have a longer list of universities?</p>

<p>I have sometimes wondered how much schools fudge their class size numbers. For example, how widespread is the practice of assigning more than one course number to the same class? I know it is done. If the course number is then used in calculating class size, the class size would be underestimated.</p>

3

<p>Problem is large classes are fine for most people taking an intro level survey course which is typically where such classes exist. You don't get 300 people in an advanced biochem class. There is no study that shows any educational disadvantage for such classes. More advanced classes are where some individual attention, research projects, and discussion becomes important.</p>

4

<p>collegehelp:
I'd done a couple more
Brown 45%>50, 30%<20, avg 79, similar to non-Harvard Ivies
I was surprised btw that Princeton's class sizes were not smaller by these measures than its peers (other than Harvard, which is no surprise) - and in fact Yale's was somewhat smaller than Princeton's. Would have guessed the opposite based on general chatter on Princeton.</p>

<p>And in between the LACs and the Ivies:
Rice 35%>50, 28%<20, avg 51
Northwestern 39%, 35%, 57</p>

<p>Among the ones that I'm interested in, I wasn't able to find CDS data for Vassar, Columbia, UChicago. If anyone finds CDS-style class size distribution data on these, pl let me know.</p>

<p>And on your question on course numbers, collegehelp: the Common Data Set specifies several criteria in computing these numbers, including: "....Each class section should be counted only once and should not be duplicated because of course catalog cross-listings." Not sure they've managed to plug all the loopholes, but they do seem to be trying.</p>

5

<p>I think this is an IPEDS statistic, and am sure that it's a Common Data Set statistic. College institutional research officers often discuss how to do these counts on an email list for Common Data Set users. The underlying data are only so good, so these comparisons among colleges are only so meaningful.</p>

6

<p>Dadx2, feel free to do Michigan's. Class size details are at the bottom.</p>

<p><a href="http://sitemaker.umich.edu/obpinfo/files/umaa_cds2008.pdf%5B/url%5D"&gt;http://sitemaker.umich.edu/obpinfo/files/umaa_cds2008.pdf&lt;/a&gt;&lt;/p>

7

<p>Dad:</p>

<p>Aren't you really saying that 50% of students at Stanford will incur a class of 50+ students during their four years? It's inconceivable that they will incur a class of 50+ 50% of the time, i.e., two years' worth of classes.</p>

8

<p>this is very interesting, first time i've seen this, if you could, please do this for all top 50 USNEWS schools. that'll be interesting to compare.</p>

9

<p>Dad:</p>

<p>Wow. That's a lot of work. I had once tried to think about how to approach the CDS data from that angle, but was never confident in how to do it.</p>

<p>USNEWS could do everyone a big service if they would just present the full CDS data on class size distribution. The arbritrary way that USNEWS simplifies the data is specifically designed to blur major distinctions.</p>

<p>BTW, I don't think Swarthmore counts any of their one-on-one Directed Study "courses" in these class sizes. I believe that the only courses over 50 there are intro Psych, which has discussion sessions led by other department faculty, and the intro science courses (two semesters of Bio, two of Chem, one of Physics, Organic Chem). These all have lab sections led by faculty of 20 students or less. All other departments (Econ, Poli Sci, English, etc.) break there intro courses up into multiple small class sections. So, in reality, there's probably no "average" student. You are either going to get a several large courses (sciences) or none.</p>

10

<p>i-dad:</p>

<p>UChicago just started counting its one-on-one tutorials last year, which caused their jump into the top 10. Swat might want to consider it, not that they need the accolades. :)</p>

11

<p>The Common Data Set instructions specifically state not to include any one-on-one tutorials or instruction in the class size statistics. </p>

<p>Likewise, the instructions say not to count thesis advising. At Swarthmore, the thesis counts as one of your classes for both semesters senior year. There is a ton of one-on-one advising from two faculty members.</p>

12

<p>^^ sry, my bad. I do remember reading an article about Chicago and how they met with USNews to understand the ranking system better. The net result was more smaller classes (%) and a jump in ranking. (too lazy to google the article)</p>

13

<p>The three Cs (Cal, Chicago and Cornell) and Johns Hopkins all seem to have experienced an awefully suspicious leap over the course of one or two years. I am talking from 45%-50% of classes with fewer than 20 students to 60%-65% of classes with fewer than 20 students.</p>

14

<p>Alexandre:</p>

<p>There are a lot of games being played. It does't take very many tutorials to change the statistics dramatically, for the reason Dadx2's math points out. Viewed as a percentage of total courses, 10 tutorials counts the same as 10 lecture halls filled with 500 students.</p>

<p>There is little likelihood that Chicago's class sizes are really shrinking in the middle of a very significant increase in undergrad enrollment.</p>

15

<p>Not that it matters. I have always maintained that at top universities, introductory classes can afford to be, and generally are, large. Once passed the intro levels, classes shrink considerably. Most of my classes at Michigan had fewer than 30 students and no TAs. Then again, I only took 5 or 6 intro classes. My friends at other universities, including Cal, Cornell, Duke, MIT, Northwestern, Princeton and Stanford, did not enjoy significantly smaller classes and they were all exposed to TAs and large lectures.</p>

16

<p>I hate small classes. I tend to feel less pressure, and thus put more genuine interest into classes that are larger. </p>

<p>I don't know why people keep saying that smaller is always better... I know many people who, like me, hate small classes, and love larger ones.</p>

17

<p>^^^ I don't get that. Why would a small class cause you to feel more pressure? I don't think most people feel that way. They want to be more directly engaged with the professor and other students as part of their learning experience -- not less.</p>

18

<p>Not all classes call for an intimate setting. Some courses require a great deal of interaction and indivudal attention and some do not. I personally like classes that have 15-25 students. More than 30 is too big and less than 10 is too small.</p>

19

<p>
[quote]
Why would a small class cause you to feel more pressure?

[/quote]
</p>

<p>Small classes are more demanding of students, requiring a higher level of engagement. It's pretty easy to anonymously sit in the back of a lecture hall. </p>

<p>In a small class, especially in a discussion class, there's nowhere to hide.</p>

20

<p>Yes, I do understand that part of it. However, I'm operating from the assumption that the idea is to get the best education possible. I admit to interpreting the latter to mean some degree of rigor and intensity to which small classes are most conducive. My view of the purpose of higher education is for students to be challenged to grow. I don't know how that takes place as well when one is enjoying anonymity in the back of a large lecture hall.</p>