Choosing your apprenticeship with real employment data

Before choosing an apprenticeship program, the most useful question isn’t “does this sound interesting?” but “what actually happens to the people who complete it?” Good news: this isn’t a matter of opinion. The Ministry of Higher Education publishes these figures, program by program, through the InserSup program, as open data. Bad news: read carelessly, these figures can tell you almost anything. Here’s how to read them correctly.

What InserSup actually measures

The InserSup program measures the employment outcomes of higher-education graduates at several points after graduation: 6, 12, 18, 24, and 30 months depending on the survey wave, with data updated twice a year on the ministry’s open data platform. It covers, among others, bachelor’s degrees (licences générales), professional bachelor’s degrees (licences professionnelles), master’s degrees, BUT degrees (a three-year technological bachelor’s), engineering degrees, and business-school degrees at the five-year level, at universities, engineering schools, and business schools.

For each program, the typical indicators are the rate of salaried employment, the share of stable jobs (permanent contracts in particular), and the median salary, with the option to compare graduates who went through an apprenticeship track to all graduates of the same program. This last point is exactly what matters most when you’re torn between an apprenticeship track and the same program taken through regular full-time study: the dataset lets you compare, program by program, whether the apprenticeship route actually leads to employment faster or more solidly, rather than relying on a general impression.

What “nd” actually means

Browsing an InserSup table, you’ll inevitably run into cells marked “nd” instead of a figure. This isn’t an oversight or a hidden number: “nd” stands for “non disponible” (not available), and this label appears according to a precise threshold. A rate is marked “nd” when the survey’s response rate and the number of respondents are too low to publish a reliable figure, with thresholds set below a 30% response rate and below 20 respondents respectively. A minimum threshold of 20 responding graduates per degree program is also applied to salary data, precisely to prevent the displayed rate from swinging erratically from one year to the next because of a very small number of responses, and to preserve the statistical confidentiality that protects respondents’ anonymity.

In practice, an “nd” is neither a good nor a bad sign about the program itself: it’s missing data because too few graduates responded to the survey, or because the cohort in question is too small. Reading an “nd” as a zero, or worse, as a warning sign about the program’s quality, would be a fairly common misreading.

The small-sample trap, even when a figure is published

The “nd” threshold protects against the most extreme cases, but a figure published just above the minimum threshold is still fragile. A program that graduates fifteen people a year, even if the response rate clears 30% and allows a figure to be published, will see its employment rate swing sharply from one year to the next if only two or three individual situations change. This is actually why the program provides, when a cohort is too small to reach the minimum respondent threshold, for its data to be pooled with the previous year’s cohort.

The useful habit here is to look not just at the displayed rate, but also at the number of graduates it’s calculated on: a 95% employment rate calculated on 200 respondents doesn’t carry the same weight as a 95% rate calculated on 22 respondents just above the threshold.

Compare, don’t rely on a single isolated figure

An employment figure takes on meaning through comparison: the same program at another institution, the same program via apprenticeship versus full-time study, or the field’s national figure to place a given institution relative to the average. An employment rate that looks good in absolute terms might sit in the low range for its field, and conversely a rate that looks modest can be solid if the entire field sits at that level. Isolated, a figure doesn’t tell you much; compared against a relevant benchmark, it becomes real information for making a choice.

Key takeaways

• InserSup measures graduate employment at 6, 12, 18, 24, or 30 months depending on the survey wave, by institution and by program, with a comparison between apprenticeship graduates and all graduates.
• “nd” means data not available, applied when the response rate is under 30% or the number of respondents under 20 (and a threshold of 20 graduates for salary data).
• A figure published just above these thresholds remains statistically fragile for a small cohort: always look at the sample size behind the percentage.
• Small cohorts can be pooled with the previous year’s to reach the publication threshold.
• An isolated figure is hard to read on its own: compare it against another institution, apprenticeship vs. full-time study, or the field’s national average.

This is exactly what the insertion-pro connector allows: querying these official figures by program and by institution, comparing apprenticeship and full-time study, and situating a field against national figures, without having to dig through the raw table yourself.