Impact Evaluation of Girl Effect’s Multimedia Campaigns to Improve Sexual and Reproductive Health Outcomes in Kenya

The Impact Evaluation will use an individual-level RCT design to randomly assign users recruited to participate in the study to either receive the intervention (i.e. a variation of the GE campaign) or to receive an information-only control. The evaluation focuses on the following content produced by the GE campaign:
(1) Drama clips: GE has produced a TV and web series called Tuki to empower young people in Kenya on reproductive health through education and entertainment. Collaboratively, we have produced short videos from the Tuki drama series that thematically address contraception use, SRH services, and interpersonal relationships. We present thematic videos for testing the impact of the content’s condensed narrative on individual norms and SRH behaviours.
(2) Wazzii chatbot: The Wazzii chatbot is an innovative tool designed to engage users in conversations about SRH. The pre-programmed chatbot aims to create an engaging user experience, offers a private forum for asking sensitive questions, and provides valuable insights into the effectiveness of interactive approaches in promoting SRH awareness among individuals.
(3) Information-only Campaign: Information based campaign to spread awareness related to SRH topics through infographics/flyers.

The impact evaluation (IE) will utilize Meta’s social media platforms.

2026-03-23

2026-09-23

1. Knowledge of SRH outcomes:
Knowledge of contraceptive methods (e.g., types, effectiveness, proper use)

2. SRH Behaviours/Intent
Contraceptive use at last sexual encounter
Intention to use contraception
Encouraging peers to use contraception
STI testing history
Likelihood of clinic visits for SRH services

3. Descriptive and perceived social norms related to SRH outcomes:
Social Norms
Peer norms supporting contraceptive use
Gender responsibility for contraception

Knowledge Index: Knowledge and familiarity with types of contraceptives
Risky Sexual Behaviour Index: Based on actual contraceptive usage, sexual concurrency and STI/HIV testing.
SRH Behaviour Intentions Index: Based on intent to use contraceptives in the future, visit health clinics, encourage friends to use contraceptives.

1. Self efficacy
Confidence discussing contraceptive use and consent
Comfort discussing SRH needs with reference groups (partner, peers, parents, providers, community leaders)
2. Social norms:
Attitudes related to menstruation and mental health
Community attitudes towards menstruation (e.g., stigma reduction, comfort discussing)
Perceived norms around mental health.

3. GNORM sub-scale items:
Injunctive norms on gender roles regarding male authority and women’s autonomy across domains including fertility decisions, family
planning, employment, education, and access to household resources.

Self-Efficacy Index: Based on accessing contraceptives effectively (knowledge/purchasing) and suggesting its use to a partner.
Openness in discussing SRH Index: Combining 4 reference groups-partner/healthcare provider/parents and community leader.
Gender Norms Index: Progressiveness levels based on GNORM subscale items

The evaluation will study the impact of a two-tiered intervention using a factorial design (2X2):

The intervention format differs across arms, involving different modes of media engagement. The intervention arms are as follows:
Group 0 / Information-only control (Flyers): This group will serve as the control arm of the study. This group will receive informational online flyers featuring information on the key themes of the interventions (SRH, menstruation, healthy relationships etc).
Group 1 / Videos only: This group will receive video clips featuring abbreviated scenes from the web series Tuki.
Group 2 / Chatbot only: This group will receive an embedded link to access and sign up to interact with the Wazzii chatbot.
Group 3 / Videos and chatbot: This group will receive both video clips and links + reminders for the chatbot. This group will provide estimates on the additive effect of a combined multimedia intervention.

All participants complete the same baseline and follow up surveys (1 month and 3 months post baseline) to capture the incremental impact of multimedia engagement on SRH outcomes. Across experimental arms, the methods used for recruitment, data collection, and intervention delivery are identical.

These trials will be designed not only to measure the effectiveness of individual online campaigns, but also to generate actionable evidence for refining the design, targeting, and scalability of future interventions. Which treatment variations are most effective—for example, web- based television series versus chatbots compared with control groups—and which offer the greatest impact when comparing across treatment arms?

Not available

Upon entering the survey, participants are assigned a random number, which is used within the survey logic to allocate them to one of four treatment groups with equal probability.

The study will use a simple individual-level randomisation. Recruitment will target individuals aged 18-30 and will be stratified such that the sample achieves a roughly 70-30 split of women and men respectively.

No

Randomization is at individual level. Sample size will be roughly 4,000 complete individual responses (factoring in attrition)

4,000 complete individual responses (factoring in attrition)

1000 complete individual responses in each of the 4 treatment arms (total sample: 4000 observations factoring attrition)

The indicative sample target for this study is 1,000 per study group (or a total sample size of 4,000), assuming 80% statistical power and a 20% type II error rate to capture an indicator of contraceptive prevalence. A pooled study sample of 4,000 participants post attrition will allow the evaluation to capture a minimum detectable effect (MDE) of 6 percentage points and allow for meaningful sub-group analysis by participant gender (assuming a 70-30 split by female and male participants) with a minimum detectable effect of approximately 8-11 percentage points per sub-group.