Supporting career choices for students with special educational needs

The program under evaluation is a German-wide pilot expansion of career counseling services by the BA for students in SEN schools, implemented from the school year 2026/27 onward. The BA and its local agencies are legally mandated to provide career counseling to all students; the program expands this mandate by substantially increasing both the intensity and the structure of counseling for the SEN target group. Career counseling already exists in SEN schools but is currently less intensive than in regular schools, typically comprising infrequent career orientation events in class and mandatory individual end-of-school consultations.
Concretely, treatment schools receive 1. career orientation events in class starting already in the pre-pre-graduation year, with at least two units per year and class, 2. individual counseling sessions at the school at least once per year from the pre-graduation class onward (and earlier if needed); and 3. parental information events starting in the pre-pre-graduation class, also at a frequency of at least one per year and class. All counseling in treatment schools is delivered at the school premises and in an action-oriented format specifically adapted to the needs of SEN students.
Control schools continue to receive the standard counseling provision: 1. career orientation events in class beginning in the pre-graduation class, with the target of at least two units in total (commonly unmet), 2. individual counseling taking place as needed and typically at the end of school, when consultation is mandatory within the vocational rehabilitation framework, and taking place at the employment agency rather than at the school, 3. parental events beginning later (pre-graduation class) and being less frequent (at least one unit in total).
This contrast in counseling intensity, timing, and accessibility constitutes the treatment variation we exploit for causal identification.

2026-10-01

2028-03-31

Child administrative data:
- Career counseling take-up (individual, vocational guidance event)
- Being a registered applicant at the Federal Employment Ageny
- Transition into apprenticeship training
- Transition into preparatory vocational training programme

Child questionnaire:
- Planned Transition into preparatory vocational training programme
- Continuing school education / secondary school certificate
- Informedness about potential transitions after school
- Certainty about occupation of choice after school
- Occupation of choice after school: 1st occupation, 2nd occupation, 3rd occupation

Parent questionnaire:
- what the parent would ideally want the child to do after leaving school
- what the parent realistically expects the child will actually do after the current grade
- satisfaction with the child's plans for further schooling and occupational choice
- services the child has used so far (school-based career orientation, career counseling by the Federal Employment Agency, Integrationsfachdienst, private or other counseling centers, advice from family or friends, other services, none, don't know)

To investigate different channels and mechanisms, we use secondary outcome variables. We ask for various explanations, including to whom students reach out for information, what they already know about their preferred occupation, their own agency, their beliefs about future prospects, and satisfaction with educational decisions. We also ask parents for their view on support measures, information given about the future of their children, and their expectation for their child.

To implement our treatment, we cooperate with the headquarter of the Federal Employment Agency and local employment agencies (Agenturen für Arbeit) in multiple states (Bundesländer) across Germany. The intervention is carried out by local career counselling teams, which are specialized on special needs students. In our study, we include all special needs schools focused on learning disabilities and physical impairments. For more information on the randomization, please consult the "randomization method" section. The treatment intensifies the career counseling that the employment agencies are legally mandated to provide. In treatment schools, career orientation events in class start one year earlier and take place more frequently, individual counseling sessions are offered annually and on school premises rather than at the employment agency, parental information events begin two years before graduation, and all activities follow an action-oriented format adapted to the target group. Control schools continue to receive the standard provision.

To estimate the causal effect of intensified career counseling, we employ two complementary regression strategies. Our primary specification follows the standard approach in the RCT literature and regresses the outcome of interest on a binary treatment indicator, controlling for employment agency district and school support domain fixed effects. As a second specification, we exploit the multi-year panel structure of our administrative data by adopting a difference-in-differences framework that incorporates pre-treatment outcome values as an additional control. Using the years prior to the start of the intervention with a school fixed effect, this approach differences out time-invariant unobserved heterogeneity across schools and accounts for pre-existing trends in outcomes such as counseling take-up and transitions into vocational training. By leveraging the pre-treatment variation available in the administrative data, this specification substantially increases statistical power and the precision of our treatment effect estimates – which is particularly valuable given the constraints on sample size imposed by the fixed number of participating districts.

Not available

Randomization by computer (Stata), using a fixed, documented seed. Each school is assigned a single uniform random draw prior to assignment; schools are ranked on this draw within each eligibility pool defined by employment agency district and support domain, and the domain-specific treatment quotas are then filled sequentially, with schools selected as treated for one domain removed from the remaining pools.

School

Yes

189 Schools

8,505 pupils

120 schools control, 69 schools treatment

Power calculations follow Duflo, Glennerster, and Kremer (2007) and assume 189 schools, 8,505 students, an intracluster correlation of 0.06, 80 percent power, and a significance level of 0.05. For entry into vocational training, with a control-group mean of 10 percent (estimated from administrative data), the minimum detectable effect is 3.6 percentage points, or 36 percent of the control mean. For attainment of at least a lower secondary school certificate, with a baseline of 27 percent (KMK 2024), the minimum detectable effect is 5.3 percentage points, or 19.6 percent of the control mean. Conditioning on pre-treatment outcomes in the difference-in-differences specification is expected to reduce these values further, by an amount that depends on the serial correlation of the outcomes across cohorts.