I’ve started working for a new university and there are lots of differences between here and ‘the old place’. As I started to meet our students one of the very striking things I noticed was how far away many of them lived.
There’s not a lot of official student accomodation (810 rooms for an UG student body of 11,000 (includes quite a lot of part time), and most of the students I chatted to lived across London, or further afield (e.g. Nottingham!).
Students were spending hours getting to and from classes and I started to wonder if this might be affecting their levels of engagement and persistence. A recent dutch study found some evidence that increased commute time led to students spending fewer days on campus, with an increase in the hours spent each day and a decrease in average grades. The article does discuss some confounding factors (e.g. more committed students might move closer to the uni) but the results were interesting, and suggested that this was a relationship to explore for my students.
The dutch study was based on reported commute times, attendance and marks, and had a response rate of 12%. Being vaguely technically minded I wanted to take a different approach and use the data we have to see if there was a relationship worth exploring.
I was able to get the term time postcodes for the 530 students on a foundation year (thanks Charles!). I could use the free version of BatchGeo to plot up to 250 points on a map, so did this for the foundation year students of 2 of our 8 schools.
View Location of some of our foundation year students in a full screen map
You can see students are spread across London. You can also see some addresses further afield (e.g.Birmingham). This highlights a potential issue with the data, we continue recruiting until very late in the year and I know of students who have started their studies before moving to London so I assume these very long distance students aren’t commuting but in the process of moving.
There is certainly evidence that a lot of our students are commuting a great deal, so the next question is how long is it taking them? For commute times I knew it *should* be possible to take the term time postcodes and run them through the Google Maps API to get an estimate for the public transport (there is no student parking) commute time to the building that the student’s school was based in. After a fruitless few hours on my laptop I am very grateful for the help of a colleague from my days as an SSI fellow Barry Rowlingson for doing the actual leg work and giving me a list of journey times (assuming a 9am Monday morning arrival time for maximum effect).
The numbers confirmed that a lot of students have a big commute. In the chart below you can see the frequency distribution of commute times of our students, grouped into 20 minute bands (the output from google maps was in seconds).
Visually you can see that the majority of our students commute between 20-120 mins. In fact the median commute time for foundation year students was 57 minutes, which is a really interesting finding in itself.
At UWL we have an attendance monitoring system (SAM) which whilst not perfect does give some hard data. It’s also the basis of a scholarship I’m administering, which is being used to support student engagement and retention, and so this should be fairly reliable (it’s not, but that’s a separate blog post).
I wanted to do a scatter plot of the SAM data with commute time. I excluded those with a ‘commute’ time of over 10,000s (roughly 2hr 45) as this quickly included some very large and unlikely commutes (Whitley bay?). I also excluded an anomalous programme of ours, with a very different student body and where almost all of the students were living in student accommodation. Finally I excluded students with 0% attendance at this point (7 teaching weeks) as they probably never finished enrolling and so never actually commuted. I’ve left in students with very very low attendance, I’m not really sure whether to exclude these (later in the year I will be able to do this again with these students withdrawn from the data).
It’s been a long time since my economics degree but on the face of it that looks like there isn’t much of a relationship (at this stage in the year!) between commute times and attendance.
I should be able to do this later in the year when the effect would be more pronounced (if there is an effect it is reasonable to assume it increases as the year progresses). I also think it might be worth exploring differences between schools. I started doing this by trying to label the points by school but this was hard to visually represent through excel (it could do it, but it was hard to see anything).
Of course I might have missed something, which is why I’m writing this up, I’d welcome any thoughts or comments!