Teaching Thursday: Thinking about Teaching Data

Over the past two weeks, I’ve been playing around with a decade of enrollment data in our department. This amounts to 17,111 students in 683 sections since 2015. This feels like “good work” to me.

I thought I would blog about some of this stuff while trying not to disclose too much nitty gritty data. To be clear, none of what I’m going to be writing about is individual student data. It was all given to use as aggregated data. I likewise won’t talk about individual classes or instructors. The point of my post is to articulate some of the hypotheses and broad conclusions that I’ve been able to draw. Some of these might be relevant to other folks in other situations.

Understanding the Data

Our main data set was superficially straight forward. It was the number of students in each class taught by our department, the number of majors and minors, and how fully enrolled the class was as the percentage of available seats. There are two complications to this data. One is the COVID year (and immediate aftermath) and the other is that in 2018-2019 we changed our major from 39 to 33 credits meaning our majors took fewer upper classes. This latter complication has an impact on one of our hypotheses. 

In general, our number of majors has seen a steady, if small decline over the past decade but our department size has declined faster than our number of majors suggesting that this is at least partly a supply problem (as in number of courses offered) as well as a demand issue (fewer declared majors). Like all programs, we’re interested in growing our number of majors even if we have to acknowledge that we have very little slack in the system to do this. In fact, there is a strong correlation between number of upper level courses and program growth.

Some Hypotheses

We started out with a fairly simply hypothesis. It felt like in recent years we had fewer and fewer majors in upper level courses and more and more “essential studies” students. These are students who take upper level courses to fulfill university mandated requirements that we call “essential studies.” As a result, we found that we had many students in upper level courses who were solid and even interested students, but had little background in history. 

We tested this and there does seem to be a trend where our 400 level courses have seen a decline in the number of majors to fewer than 20% per class. Our 300 level classes hold closer to 30% majors. To be clear, the difference between our 300 and 400 level classes is insignificant. The different numbers are the artifacts of a time when we had a graduate program that could use 400 level courses. Indeed, when we combine 300 and 400 level courses in our analysis, we find that the decline in number of majors in upper level courses is insignificant over the past decade. 

What is interesting is when we exclude online classes, the percentage of majors in upper level courses has increased slightly (2.5%) over the past decade. In fact, online upper level classes, which are a fairly recent addition, attract around 17% fewer majors.

What is suggestive is that some preliminary analyses have suggested that European courses have a much higher percentage of majors than American history courses. The reasons for this are probably not hard to figure out: non-majors are likely drawn to classes on familiar topics in US history. It may also be that our data is skewed by two very charismatic faculty members who were Europeanists and just recently retired.

Some Other Observations

It is disheartening, of course, to see our number of majors decline and our program to be effectively running at capacity. What is more interesting is to see that our program has enjoyed very high retention rate since COVID.

The calculations to discern retention rates without access to individual student data are kind of cool. Students have to take H240 (methods) and H440 (capstone). They generally take H440 as seniors and take H240 as freshmen (rarely), sophomores, and juniors. So we had to consider the total number of majors in the pipeline at any give time and whether H440 represented an appropriate (roughly 40% of that group of students).

What’s interesting is that pre-COVID when we had more majors, more faculty, and more classes, we had worse retention (closer to 80%). This made me wonder whether what we have lost in recent years are students who started a history major but did not complete it. These students represent a kind of churn — the “casual major” — that can be frustrating (if our goal is retention), but also — in a crassly utilitarian way — useful because these students count toward our number of majors but often do not add a significant burden to our upper level classes. Right now, since we are operating nearly at capacity (albeit with more elasticity in the potential percentage of majors in our courses that I am going to assume here), the goal might be to attract more casual majors which will impact retention but not enrollment numbers (at least not in a consistent way). This, of course, is profoundly cynical. 

When we concatenate our enrollment numbers with our course schedules, it becomes pretty easy to see when our students like to take classes. Our best performing times are more or less predictable — 10-12 am MWF and TR — what is interesting is that we attract a large number of majors to late afternoon (3-4 pm) MWF and TR classes suggesting that these times might be more popular with students than we would expect.

Finally, there are some rather interesting trends concerning who teaches most of our students. We do not employ a large number of adjunct instructors. In fact, tenured faculty teach the vast number of our courses. It is interesting to consider whether our schedules are optimized to draw students from 100 level courses into upper level courses. Whether we care to admit it or not, many students are drawn to a major because they like a particular instructor. This should encourage us to pay attention to the pathways that draw students from the 100 level to upper level.

We also have to be attentive to who teaches most of our students and whether uneven work loads (usually generated by need and other factors) limits the diversity of student experiences across our curriculum and leads us to become particularly dependent on one rather than several pools of students.