Time until graduation as a proxy for picking between (German) universities

post by Morpheus · 2022-01-24T18:27:32.984Z · LW · GW · 6 comments

Contents

  Data
  Other Considerations
  Guess
  Things I'd do with more time/motivation:
None
6 comments

If you are in high school and struggle with decision-making, please don't take this too seriously. If you're into EA maybe apply for 1-1 advice here?

Choosing a university without knowing what metrics to base this decision on is hard. When I did this one year ago, I was mostly comparing German universities as it was my country of residence (I debated going abroad, but covid and other things made this option unattractive for me). Since most universities here are free and rankings not really popular, there isn't as much of a prestige difference between universities to cut down the possibilities. So in the end, I based my decision mostly on cost of living and location/city.

I didn't expect the differences in universities to be large enough to be worth looking further into it. I've changed my mind about that, mostly based on the fact that I could share what I've learned with other aspiring undergraduates [LW · GW].

Data

I looked into which metrics were used for ranking universities by the CHE-Ranking which is the most popular (which ranks universities for every metric and every degree separately and refuses to aggregate them to give universities an overall rank. The idea is "make your personal ranking according to your priorities". It's like they actively refuse to play the prestige game). Most of their metrics are either subjective ratings by a few students (how useful are these ratings if these students have never been to a different university to compare it to?), or things like money per researcher, which seem rather useless for finding the best place to study as an undergraduate. The only thing that stood out to me was "% of graduations in appropriate time". Unfortunately, universities in Germany have only recently (2017)^[1] started to collect this data in a central place, so I rely on the summary statistics from the CHE-Ranking-Site and hope that they did a good job.

As you can see in the plot below, there are lots of universities where you would be above average (median) if you finished your degree in your eighth semester. It turns out, the difference how many students (in Computer Science [most popular on lesswrong]) graduate after their eighth^[2] semester ^[3] is pretty large. Some of this may be due to random variation in student ability or curriculum, but even excluding universities with less than 1000-students (so at least (1000/4) * (around 50% passing rate) = 125 students passing every semester) the differences are quite large:

Next I looked whether the same was happening for business administration, because it is the most popular in Germany and I would have to worry less about statistical significance (I'll leave the statistical tests to people who've actually taken a statistics class). It does look quite similar (although overall it does not look quite as bad as for CS):

One thing that is more pronounced here is that the upper outliers tend to be private universities, which makes sense to me (Left as an exercise for the reader Lol). I am still surprised by the variation, before I looked into this I thought that most universities are roughly equivalent.

Other Considerations

• Different universities have different specialized undergrad subjects (like "data science", "bioinformatics" etc.) under the lable computer science or business administration. Some of these tend to be more competitive (My impression from going through the CHE data) and thus consist of more able students
• If it's just about more able students, then your particular subject might be less important to you compared to the whole university

I was kind of skeptical that I might be misunderstanding something basic here, so I tried to check whether the numbers from the CHE-Ranking add up to the same as when I look at numbers from my university. (Caution, this just looks at who registered for the semester, so it includes students who never actually graduated!)

Assuming everyone after the fifth semester eventually graduates^[4] I got ~50% (excluding last semester due to covid) which is pretty close to the 58% according to the CHE data, which means at least in this case the data is probably accurate.

Guess

Interpret this as my wild guess. If you have a better interpretation, share it in the comments! I've done stupid/obvious mistakes before. If you find any please point them out!

Average time until graduation is probably proportional to (personal ability) and/or (hurdles of the university in order to pass). If you are in university mostly for the signalling and the interesting people, then both, smaller hurdles and smarter people implied by more "graduations in appropriate time" would be good news. So maybe include it in your decision matrix.

Things I'd do with more time/motivation:

• Figure out how to get more data on the whole distribution for time until graduation. To identify universities where it is easier (or students smarter) to pass not only in "appropriate time" but maybe even in the advertised 6 Semesters, you'd actually have to know how many pass at that time.
• Get the CHE data from past years and investigate how persistent "time until graduation" is over time, which would be rather important as we are necessarily trying to extrapolate here. I had a hard time finding past data with google.
• Look at data from other countries

---

1. See here for more info: Statistisches Bundesamt. “Studienverlaufsstatistik 2020.” ↩︎

2. CHE's description: “Proportion of undergraduate degree course graduates who completed their studies in an appropriate amount of time, namely the standard period of study plus one or two semesters (dependent on the subject area).” ↩︎

3. (for a “6-semester”-degree) ↩︎

4. They don't but it is good enough. ↩︎

6 comments

Comments sorted by top scores.