Unintended consequences

Topics/tags: Miscellaneous, Grinnell, academia, scheduling

This semester, Curriculum Committee [1] is asking the faculty to consider a new timetable for class offerings. This is the third-or-so proposed change to the timetable we’ve received over the past six or so years [2]. The more I think about the proposal, the more I realize that it represents the rule of unintended consequences.

As is often the case, some backstory is likely necessary.

About six years ago, Curriculum Committee revised the class timetable to achieve two goals. First, they hoped to better support the faculty who were not satisfied with Grinnell’s standard 3x50 [3] class schedules and who preferred 2x80 or 2x110 classes. Second, they hoped to decrease conflicts between classes. That new schedule had one major problem: It moved Scholars’ Convocation from 11:00 a.m. on Thursdays to noon. But the faculty agreed to a few-year trial with an understanding that (a) someone would study the effect on conflicts and (b) the timetable would revert back to the old form if the faculty did not vote to continue it.

As many of us expected, that schedule had negative impacts on Convocation. Attendance decreased. Costs also increased, since it now became important to feed people lunch and College Catering is quite expensive. As I understand it, no one followed up on the promise to study conflicts. That may not be surprising; such a study is complicated to design.

So, about three years ago, we were ready to roll back to our old timetable. However, many faculty members appreciated the availability of longer class slots. And there was a push from someone in administration to add another reserved time for talks. So, instead of rolling back, a newer timetable was proposed. This one kept the longer time slots. It restored Convocation to its long-standing slot of Thursdays at 11:00 a.m. It also added a new Community Hour on Tuesdays at 11:00 a.m.

Now, that one extra hour may not seem like a lot to you, but it had a significant impact on a few departments. You see, people who offer 4x50 classes that are available to first-year students in the fall used both MTWF at 10:00 and MTWF at 11:00 [4]. 8:00 and 9:00 were not available because Tutorial meets 8:00-9:50 on Tuesdays and Thursdays.

I expect that those planning the schedule said to themselves Not many people use the 4x50 slots; it shouldn’t make that much of a difference. And it is correct that not many departments use those slots; if I recall correctly, it’s mathematics and statistics, computer science, and many of the foreign languages. But to those departments, the change had a huge impact. Particularly for math, stats, and CS, it’s not reasonable to move most of your 4x50 classes to the afternoon: those classes serve science students and 4x50 afternoon classes are much too likely to conflict with labs [5]. I raised that concern but was told that it was less important than the other issues that Curriculum Committee was addressing.

What do you do when you are now stuck with only one morning slot for your 4x50 classes? One possibility is to switch some sections to a different model, say 3x80. But it takes a lot of work to switch from a 4x50 model to a 3x80 model. So if you do it for one section, you will probably want to do it for every section.

And that’s what Mathematics did. Every section of Calculus switched from 4x50 to 3x80. That gave them two morning slots (8:30-9:50 and 10:00-11:20) and two afternoon slots (1:00-2:20 and 2:30-3:50).

CS tried to maintain the 4x50 model. But we quickly discovered that the morning section of CSC 151 filled much too fast and the afternoon section had difficulty filling. From our perspective, the new schedule did not adequately support student needs for that course. But change is hard for workshop-style courses; when you’ve written a lab for each fifty-minute class period, rearranging from 56 classes to 42 classes requires rewriting about forty labs. Nonetheless, in developing a new model for CSC 151, we switched CSC 151 to the 3x80 model last year. I believe the folks running CSC 207 also took the time to rework the class last year. This year, we also switched and CSC 161 to the 3x50 model; I believe Henry Walker spent much of the summer restructuring the class.

If Tuesday at 11:00 a.m. becomes available again, will we switch back to 4x50? I’d guess that the answer is no. Let me make that clearer. I won’t switch back: 3x80 is a better format, and I don’t have the time or energy to rewrite these classes again. I assume my fellow faculty won’t switch back, either.

But CS isn’t the issue at play, Calculus is. There are nine sections of Calculus this fall, four of MAT-131, Calculus I, and five of MAT-133, Calculus II. Under the old 4x50 model, they would have occupied nine 50-minute MWF classroom slots. Under the new 3x80 model, they occupy sixteen of those same slots. That’s an increase in seven slots. And, while seven slots may not seem like much, I’m pretty sure that it has an effect, particularly at the tight times. If they are spread out evenly across the day, it’s a whole additional classroom [6]. More likely, it means an additional demand of two large classrooms during high-usage times [7]. But I’d have to look at the old and new schedules to be sure.

More generally, I expect that a gradual change from 3x50 or 4x50 to 3x80 or 2x80 or 2x110 across the College has had an impact on our use of classrooms. And, unfortunately, I don’t think a minor change to the timetable is likely to mediate that impact. It may be that policies can help. For example, it could be that we’d resolve stress points by moving more 2x80 and 2x110 classes to TuTh. Perhaps we also need other restrictions on the two-day-per-week and one-day-per-week classes. I’d need to see the class scheduling data to be sure. And that’s a subject for another musing.

In the end, I see this as an instance [8] of the rules of unintended consequences or cascading consequences. That is, what seems to have been a minor change (eliminating one hour out of thirty-four) has led to a bigger impact (additional demand for between one and two more large classrooms). I don’t know how to avoid such unintended consequences in the future. However, this example suggests that we need to think carefully about potential consequences.

Postscript: Now that I’ve done some additional analysis [9], I’m less certain that the demand for two additional classrooms is quite as significant as I thought. What is one or two classrooms out of sixty? On the other hand, Calculus requires large classrooms. And, when you’re close to reaching your classroom capacity, even a little change can make a big difference.

Postscript: At some point, we also cut fifteen minutes out of lunchtime and made the day end fifteen minutes earlier. I don’t like the shorter lunches, but I appreciate giving more freedom to our athletes and the opportunity for afternoon meetings that end at a reasonable time. Those changes weren’t relevant to the main point of this musing, but might be of interest to some readers, so I’ve included them here.

[1] While the request is coming from CC, I believe that the Registrar’s office is the primary force behind the request.

[2] There have been proposals that have been defeated, so third or so proposed change. I’m too lazy to figure out when the first change happened, so six or so years.

[3] Three days per week, fifty minutes per section.

[4] I believe Mathematics, which was one of the primary users of the 4x50 model, also used MWF 9:00 + Thursday at 10:00.

[5] Looking at this semester’s schedule, it appears that the morning sections of Calculus are slightly less popular than the afternoon sections. Given our experience in CS, that surprises me.

[6] We have seven fifty-minute slots each day, four in the morning and three in the afternoon.

[7] My quick estimate is that we have about sixty simultaneous classes at the busiest time slots.

[8] Or perhaps multiple instances.

[9] I’ll report on that analysis tomorrow.

Version 1.0 of 2018-10-16.