ParcourSup study: French higher education

Here is the URL link to the original dataset. This is my personal project about ParcourSup data, the French process to get to higher education and which type of students go to which type of higher education which will be published in multiple parts. I wanted to look at this data because higher education (also called post-bac education in France) is a huge deal and can also shows a lot about our society. It's also interesting to compare reality with what we usually view higher education. Last but not least, that's also a good exercice of data visualization and story telling which I hope you will enjoy.

How to use this file?

Most figures are interactive: you can select an item in the legend (usually on the right side) to remove this item from the figure. You can also double-click on it to show this item only and remove the others. Another double-click will display back all items. If you have any comment or problem, reach me out on my LinkedIn!

Introduction

ParcourSup Logo

Back to school! Or rather, its analysis. Last time, we got a good look at map of all kinds to see where students go to study. Most notably we learned that many formations are centered around Paris (and Lyon to a lesser extend), disproportionately so. We will now define and study some indicators that will let us good a more statistical grasp of the differences between each kind of formation.

Formations indicators

We will define 6 indicators here to go on this analysis:

To get more data points, the data is grouped not only along the formation type, but also the region (which means that each formation type should be represented 14 times as not metropolitan regions are aggregated together again). The size of each data point is the square root of the total of candidates.

Let's start with the most basic indicator: the admission ratio. It will give us an hint about how selective formations can truly be.

There are clear differences between selective and non-selective formations. Licence and PACES have an admission ratio above 10% (and Licence's ratio is dragged down by STAPS, a selective licence) whereas the others are well below 10%. All in all, selective formations are indeed selective, what would you know! However, we can see the breadth of what Autre formation represent: the ratio come from less than 0.01 (less than 1%) to 0.45 (nearly 50%!).

But this ratio alone is meaningless: an admission ratio should be high because as the formation is non-selective, there are many available places but it could also be because few students actually apply to the formations. We need to look at the pressured ratio as that is the ratio that will show us how many students apply compared to the formation capacity.

There is some correlation between these ratio, which is not too surprising. Formations handle high demand (= high pressure ratio) by simply not ranking too many students, allowing these applicants to move on without being put on a waiting list. As a very rough example, a formation with 30 available places will always rank 90 applicants, no matter how many there are. A priori if a formation have the luxury of

Primarily, the IFSI stick out has been highly pressured yet having a high rank ratio. It means that a massive amount of people are put in a waiting list. Low-selective formations (PACES and Licence) are at the higher-left of this graph because of their non-selective nature, allowing them to rank many applicants. Based on this graph alone, they might be at risk of overcrowding! CPGE, DUT and BTS all follow the middle of the curve and are more or less indistinguishable. The last point I want to address is the cluster of Autre formation at the right of the main curve which are for the most part Engineering formations (private or public) and CUPGE. These formations are one way to go into engineering / business schools and can be seen as an alternative to the intimidating CPGE. That would explain why their pressure ration is so high: many applicants put these formations instead / in addition to a CPGE.

As I find the ranking ratio very compelling about the difference between selective and non-selective formations, here is a boxplot about this indicator

The first figure tell more or less the same story: the two rates are correlated, which once again is not surprising. If there are many candidates ranked, it is usually because the formation is able to propose an admission afterwards which two big exceptions: IFSI once again (many ranked but not many proposed because there are not many available places and few accepted students) and STAPS Licence (many pink dot on the right-side of the graph) for the same reasons.

The second one is more intriguing: all formations or more less around a ~25% acceptance rate (CPGE are slightly lower, PACES higher) with some exceptions:

I have one explanation to the lower ratio in CPGE: these formations often have different but similar classes (physics focus vs mathematics focus for example). One can imagine that a student refuses a specialization to choose another but I don't have data to support this hypothesis.

The curious case of IFSI continues as they are by far the most overcrowded formation. PACES and DUT (and EFTS but the cohorts are much smaller) toe the line but IFSI cross it happily. My explanation - but I have yet to more about this - is that IFSI can be the second choice for PACES. These formationx prepare for medical careers and people may apply to both of them. But as PACES lead to more prestigious jobs (physicians especially), many applicants decline an IFSI proposition because they want to get in PACES. Yet, as PACES is extremely competitive and there is a huge amount of people applying for IFSI, it still leads to overcrowding in the latter. I am not entirely satisfied with this explanation but I can't explain this extremely low acceptance ratio combined with overcrowding otherwise.

Regional differences

To close this post, we will now mix what we learned in the last post with these new indicators. We know that higher education is heavily biased toward Paris and Ile-de-France (and Auvergne Rhone-Alpes to a much lower extend), how does it translate here? We will look at the two indicators we already singled out: the admission ratio (for one available place, how many students apply for?) and the overcrowding ratio.

There is of course one main outlier: Ile-de-France with a much lower ratio than every other regions. As expected, much more students apply compared to the number of available places. Not Metropole is another outlier in the other direction: a big admission rate implying that more students applying get accepted than in other regions. We already saw that fewer students apply there. All in all, it confirms what we know.

The results are closer but once again, Ile-de-France is overall the region with the highest overcrowding rate, and not metropolitan regions the lowest one. Formations here have to be overcrowded to handle the large influx of applicants.

Overall, we learned what kind of indicators are relevant to analyze formations. Moreover, we know how selection translates when applied to these indicators: low ranking rate to quickly remove as large of students from the process. We also saw how IFSI is an outlier thanks to a huge number of applicants compared to its low capacity. Lastly, we applied this knowledge to the geographical analysis so see that these indicators also show the Parisian centralization of higher education in France.

This brings the overall data exploration to an end. From now on, we will focus on specific subjets, the first one being the difference between public and private formations. We will focus on the Autre formation type because it is a catch-them-all label that needs to be drilled down. It should come in around 10 days so stay tuned!

Back to main menu