Mission or Ambition? Rethinking Who We Recruit to Mentor Young Workers

Intervention 1: Motivational Framing via LinkedIn Recruitment Ads
The first intervention targets the mentor recruitment stage. As part of Mentor Together’s recruitment process messages were varied – one highlighting the social impact of mentoring and the other emphasizing professional growth, resulting in variation in motivational framing. One set of ads highlights the social impact of mentoring to attract intrinsically motivated or other-oriented individuals. To track which message led to a given mentor’s enrollment, each ad is linked to a unique landing page, which allows respondents to be assigned to groups based on their click-through source. While the framing of the ads varies, the program content, structure, and requirements remain identical. This allows us to cleanly compare how mentors with different initial motivations engage with the program—across metrics like participation rates, session quality, content focus, and dropout, without altering the organisation’s modus operandi.


Intervention 2: Six-Month Mentorship Program
In the second stage, mentees are matched to mentors from either the intrinsic framed or extrinsic framed recruitment pool via Mentor Together’s standard matching process. Within each pool, Mentor Together’s matching algorithm assigns mentees to mentors based on availability and shared observable characteristics, ensuring comparability across arms. Critically, because the entire program is delivered via Mentor Together’s digital platform, we can leverage rich administrative metadata—that automatically gets collected through the app—on every aspect of onboarding, training, session participation, quiz completion, module selection, and mentor–mentee interactions. The program aims to improve labor market readiness, confidence, and job search outcomes among young women navigating the transition from college to work. By randomly assigning mentees to mentors from different motivational arms, we assess how mentor motivation influences engagement, mentoring style, and mentee outcomes.

2025-06-20

2026-03-30

1. Mentee satisfaction
2. Mentee learning and work readiness
3. Mentee labor market outcomes

1. Mentee satisfaction - measured at midline and endline 1.
- Self-reported satisfaction with the mentor and overall support quality (1. Psychosocial support scale 2. Emotional support scale 3. Career support scale 4. Autonomy support scale -5. Networking support scale)

2. Mentee learning and work readiness – measured at endline 1.
- Module comprehension (overall quiz score)
- Expectations (time to employment; perceived competitiveness; wage expectations—own and peers’)
- Labor market readiness (updated CV; planned use of network connections; forward-looking behaviors related to career planning)

3. Mentee labor market outcomes – measured at endline 1 and endline 2.
- Job search behavior (intensity, breadth, and efficacy, following Alfonsi et al. 2025)
- Employment outcomes (employment status, days worked, earnings, sector match)

3. Mentor Take-up – proportion of leads who complete the full recruitment process (onboarding, training, verification) after engaging with an advertisement, complete the recruitment process. We will be able to track (source: MT Administrative Data):
- Sign-up rate
- Screening survey completion rate
- Verification completion rate
- Number of self-learning sessions completed
4. Matching Process – pickiness in matching and preference over mentees
- Number of mentees rejected
- Characteristics of mentees rejected
- Time taken to match
5. Program Engagement – this is the level of sustained participation exhibited by mentors after they have been matched with a mentee and the program has started. Unlike take-up, which captures pre-program recruitment behavior, engagement reflects in-program commitment and effort. We will be able to track (Sources: MT Administrative Data; Endline 1 Survey):
- Number of completed sessions
- Average session duration
- Mentor/mentee speaking ratio
- Sentiment analysis
- Number of interaction outside of structured sessions
6. Mentorship Content – We will be able to track through session recordings, administrative data on module choices and curriculum knowledge:
- Sessions content (calls transcript)
- Sessions content (type of support)
- Mentor module-comprehension overall Quiz score
- Choice of modules, among the available ones

1. Recruitment Intervention: Randomizes exposure to two types of LinkedIn advertisements. One set emphasizes the social impact of mentorship and the opportunity to support underrepresented students (“other-oriented” benefits); the other highlights personal professional returns, such as signaling, skill development, and career advancement (“self-oriented” benefits).

2. Program Intervention: To evaluate whether mentors recruited under different framings vary in their program performance we randomly assign mentees to one of the two mentor pools and subsequently match them to mentors in the pool using the platform’s standard compatibility algorithm. Once matched, mentors and mentees complete a 6-months mentorship program aimed at supporting mentees in their college-to-labor-market transition.

Not available

Done in office by a computer.
Mentees are randomly assigned to one of two pools. Randomization is stratified at the college level.

Individual (mentee)

No

0

1200 mentees 600 mentors

0