How university endowments predict, and don’t predict, teaching loads

It’s typically exciting to find out that your hypothesis is wrong – and I was wrong! Here’s my back-of-the-metaphorical-envelope analysis of the poll that y’all completed a few weeks ago.

I predicted that teaching loads were negatively associated with endowment size. I expected that the more money an institution had, the less that the faculty taught. I also thought that this effect would be most robust at small liberal arts colleges (SLACs). But, nope, that’s not the case.

So, I’ll just show you the data in a figure. I got responses from 110 different institutions. Where I received >1 report per institution, they were identical or similar, and I used the one with more sigdigs reported. In this figure below, Rs are research institutions (R2s and R1s), Cs are Regional Comprehensive universities, and Ss are SLACs. These can be fuzzy categories, I know. Anyhow, here you go:

Here’s where I was wrong: In small liberal arts colleges, the institutions with the bigger endowments have intermediate teaching loads! Though on the low side. But, still, even if there’s a huge endowment, SLACs are expecting a couple courses per semester from their faculty. So, if we are to believe these reported data, then if you’re looking for a low teaching load in a teaching-focused institution, that doesn’t necessarily mean a high-endowment SLAC is the best way to go. There were a number of less-prestigious SLACs that people reported to have relatively low base teaching loads. I am curious to learn more about this.

Another thing that surprised me is that enrollment didn’t matter. In other words, if you consider endowment using endowment-per-student, that doesn’t explain any more of the variance, though I’ll spare you the AIC.

The effect of endowment on teaching load most closely approaches a negative linear relationship in regional comprehensives, such as the university where I work. (And the regression does have a p-value that would please a firm-threshold frequentist, by the way.) This contrasts with the situation at Research institutions, which are all over the map for the lower-endowment institutions, but once an endowment hits a billion dollars, it looks like the base teaching load hovers at 2 courses per year. Which is, I suppose, not unlike the teaching load hovering at 4 courses per year for the high-endowment SLACs.

I created a simple model to account for Teaching Load, in using endowment, enrollment, institution type (S, R, or C as above), and whether classified as a PUI by NSF (The PUI-ness can cut across categories, because for example, the private religious university where I used to work is classified as an R2, but it’s a PUI here too because concerned because the doctoral programs are not in the sciences). The whole model without a doubt account for teaching load to some extent (R2=0.32; F=9.33; df =5,101; p<0.0001)

VariableEstimateSEMTpintercept4.740.3413.9<.00001Endowment-0.00060.0003-2.280.025Enrollmentteensyteensy-0.400.69Institution Type [R]-0.330.44-7.760.45Institution Type [C]0.0390.270.140.89PUI or not-0.620.29-2.100.038

It is broadly true that the reported mean base teaching load of research institutions reported (2.9, N=22), is smaller than that of regional comprehensives (5.0, N=42), which is smaller than SLACs (5.4, N=43), but these categories don’t matter if you’re looking at endowment. However, teach load at 4-year PUIs (5.3, N=80) is greater than at non-PUIs (mean 3.0, N=27). So if your career plan is to minimize your teaching load but work in a teaching-focused institution (which, I realize, is how a lot of people roll), then looking a high-endowment regional state university is a very solid bet, or you might find a low-endowment SLAC that pulls this off through state funding or high tuition. But as folks who work at high-endowment (and hence, prestigious, because the rankings follow the money) SLACs will tell you, the expectations for teaching are high and the culture often expects you to own your courses and you can’t simply choose to not teach them because you got a grant funded, which is pretty standard at regional comprehensives.

I also did some exploratory analyses with interaction effects and didn’t find anything I thought was worth sharing with you.

Here’s an interesting tidbit. I got data from three community colleges. I excluded them from the analysis (which if you were nitpicking my DFs, this explains that) because it would mess with the power, but here’s a key find. Two of them had an endowment (both very modest in size), and one did not. The two with an endowment had teaching loads that were way lower than I would have expected! 5 and 4.5 courses per year! Is that possible? (I’m mostly familiar with community colleges here in California, which typically have a base teaching load of 10 courses per year, according to the California Master Plan for Higher Education.) Or was this an error in reporting? Hmm.

As a caveat, it might be possible that some people were reporting teaching loads per semester rather than per year. In a few cases, I did a bit of QC on enrollment and endowment figures reported and fixed some errors (mostly by orders of magnitude), so it wouldn’t shock me if the base teaching load column also had errors, but those are ones that I can’t readily check. However, nearly everybody volunteered the names of their institutions, and based on what I know about those places, I do find it credible that these low-endowment low-teaching load institutions in the SLAC category are for real. But I think doing this properly would involve interviewing people, and developing some more sophisticated and/or flexible definitions.

I hope you had a nice break over the holiday, and that you’re easing in the the new semester well!