Encouraging hands-on job experimentation among teenagers

To test whether experimentation with the real-life work environment in non-considered occupations affects students’ occupational search, we designed the following intervention in collaboration with schools and firms in multiple towns in Switzerland. In each town, the school organizes a one-day school event (henceforth event) devoted to occupational choice. During this event, students visit 4-5 different firms where they get to experience one occupation in each firm. We refer to each of these visits as a trial apprenticeship (TA). In addition, students also attend 1-2 workshops during the event. Each TA and workshop have a time slot of around 50 minutes and are attended by a group of students.

Schools collaborate with local firms to create the TAs and the workshops. Before the event, students get a list of all available TAs in their town and indicate 5-6 occupations they would like to experience during the event. To coordinate all the visits, schools use a computer program with a specific algorithm to allocate students to firms based on students’ preferences and firms’ availability. The two workshops are typically unrestricted in size and work as a buffer to mitigate logistical difficulties in creating the program.

Students who are randomized into the control arm get to experience 4 TAs and 2 workshops; these 4 TAs are selected by the algorithm from the list of 5-6 occupations they indicated at signup. Thus, control students only get to experience occupations they are interested in. Students who are randomized into the treatment arm will similarly get to experience 4 TAs they signed up for and 1 workshop. On top of this, treated students will experience a 5th TA which they did not sign up for. In other words, the treatment is to experience an occupation they had not considered before. We categorize occupations based on their gender composition (female, neutral, male) and type (working with hands/machines, working with people, working with computer). Treatment occupations are randomly drawn from a category which is different to the categories of the students’ preferred occupations.

2023-08-26

2025-10-31

Our primary short-term outcomes (measured 1-2 months after the event through surveys) are the following:
1) Search breadth index: it measures the breadth of a student’s apprenticeship search in terms of occupations
2) Occupational type index: it measures how different the apprenticeships they are searching for are from their preferred occupations at baseline

In the mid-term (by the end of 9th grade), we will measure similar survey outcomes and also collect data on online apprenticeship search behavior.
In the longer term (up to four years after the event), we will collect admin data on apprenticeship contracts and turnover.

Please see our pre-analysis plan (PAP).

1) Beliefs about own skills fit
2) Beliefs about work tasks
3) Beliefs about work environment
4) Beliefs about employer demand

Please see our pre-analysis plan (PAP).

Randomization happens in two steps.

In the first step, we assign to all the students a “potential” treatment occupation in the following way. We consider the 5 to 6 occupations that the student ranks as their favourites and assign them to the corresponding occupational categories we defined. These are the categories that the student is interested in. To pick a treatment occupation, we randomly choose 2 of the occupational categories that the student is NOT interested in and select all the TAs in these two categories that are in the student’s own or an adjacent skill category. The student is then shown the list of TAs belonging to these two non-preferred categories and can veto up to two of these TAs. We then randomly select one TA among the non-vetoed TAs in these non-preferred categories to be the 5th TA which treated students experience during the event. We also pick a backup treatment-TA which we could use in case there is an organizational problem to use a specific TA. For specific deviations from these assignment rules, see the PAP.

In the second step, we then assign students to the treatment or control stratifying by school, classroom and gender.

Not available

Randomization done in office by a computer.

Randomization is at the individual level. We stratify randomization by school, classroom, and gender

No

We expect 2,300 individuals based on current agreements with schools, but we may get between 1,500 and 3,000 students.

2,300 individuals (same as number of clusters)

We expect 1,150 individuals in each treatment arm.

Please see our pre-analysis plan (PAP).

Pre-analysis plan