Employment benefits of an educational incubator for business graduates in Senegal

The intervention is the CESAG incubator, a twelve-month, tuition-free, non-compulsory programme delivered at the Centre Africain d'Études Supérieures en Gestion (CESAG) in Dakar, Senegal. It targets final-year students enrolled in Licence 3 and Master 2 in Business and Management.

The programme comprises four integrated components delivered sequentially:

(1) Certificate of Competence in Leadership and Entrepreneurship (Months 1-2): design thinking workshops, real-world case studies, business plan development, pitch sessions with mentor feedback, and leadership and civic engagement modules.

(2) Complementary Training (Months 2-4, based on learner needs): innovation and product development, digital marketing, financing and fundraising, professional skills and job search, personal development (emotional intelligence, resilience, stress management), and digital skills and AI.

(3) Coaching and Mentoring (Months 4-9): individual and group coaching using the 4A method (Diagnostic Analysis, Ambitions, Areas for Improvement, Appraisal of Results), SMART goal-setting, personalised action plans, 360-degree feedback, structured mentoring with CESAG alumni, and networking events.

(4) Maturation and Induction (Months 9-12): project maturation, market preparation, access to FabLab, innovation competitions, fundraising support, and job insertion support (CV, interviews, employer connections).

All components are delivered in a blended format (in-person and online) and are scheduled to avoid disruption to academic obligations. Control group participants continue with the standard CESAG curriculum and have access to regular career services but receive no incubator components during the study period. They are guaranteed priority access to the next incubator cohort.

2026-06-30

2027-05-31

Binary employment status at endline: whether the graduate has secured (1) wage employment, (2) self-employment, or (3) a paid internship leading to employment.

Measurement timing: approximately 15 months after baseline (August 2027).
Measurement method: self-report supplemented by employer verification for a random subsample.

The primary outcome is a binary indicator (1/0) constructed as follows:

Employment = 1 if the respondent reports at endline that they are in wage employment, self-employment, or a paid internship that is on a confirmed pathway to employment.
Employment = 0 otherwise (unemployed, unpaid internship, still studying without employment).

Self-report responses will be cross-validated against employer confirmation for a randomly selected subsample. The construction rule is pre-specified and will not be modified after baseline data collection.

Measured at Midline (Month 6 / Month 9):
• Soft skills score: composite score from validated self-assessment scales covering communication, leadership, teamwork, and professional conduct, supplemented by peer and trainer assessments.
• Entrepreneurial intention and readiness: Entrepreneurial Intention Questionnaire (EIQ) score, self-efficacy for business creation, risk tolerance, and binary indicator of business plan or prototype completion.
• Mentorship engagement: composite of mentor-meeting frequency, self-reported relational quality, and professional network expansion.

• Measured at Endline (Month 15):
• Time to first job: months elapsed between graduation and first paid employment or paid internship.
• Type of contract: ordinal scale (permanent / fixed-term / internship / informal or self-employment).
• Job quality: monthly gross earnings (FCFA); self-reported job satisfaction (Likert scale); job security perception.

Soft skills score: composite index of self-assessment scales across communication, leadership, teamwork, and professional conduct sub-dimensions, supplemented by peer and trainer ratings. The composite is the mean of standardised sub-scale scores.

Entrepreneurial intention and readiness: composite index aggregating EIQ score, self-efficacy scale for business creation, risk tolerance scale, and a binary indicator for business plan completion. Individual scales are standardised before aggregation.

Mentorship engagement: composite of (i) frequency of mentor-mentee meetings (count per month), (ii) self-reported relational quality (Likert scale), and (iii) professional network expansion (change in number of professional contacts). Each component is standardised and averaged.

All composites are pre-specified. The exact instruments (validated scales) are described in the pre-analysis plan submitted concurrently with this registration.

The study uses a two-arm, individual-level randomised controlled trial (RCT) with a single treatment group and a single control group. Randomisation is conducted at the individual level following a stratified procedure.

Eligible participants are final-year students in Licence 3 and Master 2 in Business and Management at CESAG who voluntarily apply for the incubator programme. Students are stratified by gender (female/male), degree level (Licence 3 / Master 2), and prior work experience (yes/no) before random assignment. Stratification is implemented using minimisation to ensure ex ante balance in this small sample.

The assignment sequence is sealed until the baseline survey is complete, at which point the statistician communicates assignments to programme administrators.

Three contamination safeguards are in place: (1) spatial separation of training sessions from common academic spaces; (2) a non-disclosure protocol signed by treatment group participants; (3) contamination screening items in all follow-up surveys.

The primary estimator is Difference-in-Differences (DiD). The Intent-to-Treat (ITT) effect is the main estimate. The Local Average Treatment Effect (LATE), estimated via instrumental variables using treatment assignment as an instrument for actual participation, is reported as a secondary estimate.

Data collection: Baseline (June–July 2026); Midline 1 (November 2026); Midline 2 (February 2027); Endline (August 2027).

This study is a mini-RCT pilot designed as a stepping stone toward a full-scale RCT. No data have been collected at the date of registration.

Not available

Stratified minimisation, implemented by computer. Students are stratified on three dimensions: gender (female/male), degree level (Licence 3 / Master 2), and prior work experience (yes/no), before random assignment. Minimisation achieves balance more reliably than simple stratified block randomisation in small samples. The assignment sequence is generated by the study statistician and sealed until baseline data collection is complete.

Individual student. The study uses individual-level randomisation (not cluster randomisation), which is appropriate given CESAG's single-campus structure and is preferred for its greater statistical efficiency.

No

Not applicable. The design is not clustered. Randomisation is at the individual level. There are 290 individual participants.

290 individual students (149 treatment, 141 control).

149 students, treatment group (CESAG incubator: training + mentorship)
141 students, control group (standard CESAG curriculum + regular career services)
Total: 290 students

Primary outcome: binary employment status at endline. Baseline parameters (from Ozocakli et al., 2025, a comparable soft skills programme): • Control group employment rate (p0): 32.2% • Treatment group employment rate (p1): 57.6% • MDE: 25.3 percentage points Power calculation adjustments applied sequentially: Step 1 : Baseline: α = 0.05, power = 0.80, two-sided test of proportions leads to 59 per arm Step 2 : DiD design effect: ρt = 0.40 leads to × 2(1 − 0.40) = × 1.20 leads to 71 per arm Step 3 : Attrition: assumed rate = 8% leads to ÷ (1 − 0.08) leads to 77 per arm Step 4 : Non-compliance: assumed compliance rate = 72% leads to ÷ (0.72)² = × 1.929 leads to 149 (treatment) Step 5: Optimal control allocation (Rosenberger & Lachin, 2002): r* = √[p1(1−p1) / p0(1−p0)] = 1.058 leads to 141 (control) Final sample: 149 treatment + 141 control = 290 total. Statistical power: 80% at α = 0.05 (two-sided), accounting for DiD design effect, 8% attrition, and 72% compliance.