Digital Mapema Model

01 · The model

Proven pieces, joined into one system

Today the evidence sits in separate programmes. A tested curriculum, a proven behaviour method and a character-led production style each work alone, and none reaches Kenyan primary-age children together. The model joins them, delivers through three channels, and loops evaluation back into the content every term.

Long-term impact

Fewer Kenyan children aged 7 to 12 harmed online, and more of those who are harmed tell someone and get help.

02 · How change happens

Three audiences, one pathway

Each audience moves from what the programme does, to what it produces, to a change in what people know and then what they do. Harm comes mostly from people children know, so caregivers and teachers carry as much of the change as children do.

Activity Output Short-term outcome Intermediate outcome

Childrenages 7 to 12

ActivityTerm-long classroom lessons and character episodes

OutputPupils complete the full set of lessons

Short-term outcomeKnow the tricks, what is private, and feel able to handle a problem

Intermediate outcomeTell a trusted adult; less unkindness and bullying online

Caregiversparents and guardians

ActivityCreator-led clips, school parent sessions, radio

OutputCaregivers reached with the full parent series

Short-term outcomeKnow the known-person risk and where to report

Intermediate outcomeTalk with their child monthly; settings applied at home

Teachersprimary schools

ActivityTraining, ready-made lesson kits, disclosure protocol

OutputTeachers trained; kits in every pilot classroom

Short-term outcomeConfident delivering lessons and handling a disclosure

Intermediate outcomeLessons delivered as designed; disclosures handled correctly

03 · The experience

A week of Digital Mapema

Most Kenyan children aged 7 to 12 do not have a phone of their own. Among internet-using 12 to 17 year olds, four in five smartphone users share the device, only 5% can go online whenever they need to, and nearly half of their caregivers have never used the internet. Younger children have less access still. So the child's experience runs through the classroom, TV and radio, and short moments with a caregiver. It is not an app.

Step 1

Watch or listen

A short episode in which a character faces one online problem. The safety lesson is the main plot. Radio versions pause so listeners answer aloud.

TVRadioClass screen

Step 2

Play in class

Teams work through scenario cards and role play: what should the character do next? Children collaborate inside teams and compete lightly between them.

ClassroomNo device

Step 3

Take it home

A family challenge card and a message to the caregiver set one short task, such as agreeing house rules for the shared phone.

CardSMSWhatsAppRadio

Step 4

Write it down

The child adds a page to their own Safety Book, including the trusted adults they would tell. The book is theirs to keep.

PrintHome or class

Step 5

Celebrate the skill

The class unlocks the next chapter of the story. The teacher names the skill each child showed. Nobody is ranked.

Classroom

Repeats every week. One school term is one season of the story.

One season, five skills

Adapted from the five Be Internet Awesome themes, then renamed and voiced with Kenyan children in co-design sessions. Telling a trusted adult runs through every episode because disclosure is the primary outcome.

Share with careWhat stays private

Spot the fakeTricks, scams and strangers

Guard your secretsPasswords and accounts

Be kindStand up for others

Talk it outWhen unsure, tell

Tell a trusted adult: closes every episode, every class game and every family challenge.

04 · Design principles

Eight rules for the form

Each rule comes from a programme or study that measured what children did, not only what they watched.

Design for the shared phone and the classroom

The child never needs a device of their own. Phones appear only in moments shared with an adult.

EvidenceDisrupting Harm Kenya: four in five smartphone users aged 12 to 17 share the device, and only 15% had ever gone online at school.

Put the lesson in the main plot

The story's central problem is the safety skill. Side messages are cut.

EvidenceShuga's subplot messages did not change behaviour. In a gamification meta-analysis, game fiction raised effects on behaviour.

Practise choices, not just watch

Every episode ends in a decision the child makes and sees the result of.

EvidenceActive, skill-based lessons are a key marker of effective prevention (Jones, Mitchell and Walsh). Minecraft CyberSafe builds each scenario around ignore, correct, mute or report.

Play together

Teams cooperate and compete lightly with other teams. No child plays alone against a score.

EvidenceCollaboration combined with competition outperformed competition alone on motivation and behaviour (Sailer and Homner). Relatedness drives the pull of games (Ryan, Rigby and Przybylski).

Reward skill with feedback, never prizes

Story chapters unlock for the class. Feedback names the specific skill. No points for time spent.

EvidenceExpected tangible rewards cut children's later interest, more than adults' (Deci, Koestner and Ryan, 128 experiments). General praise had no significant effect for children.

Bring the adult into the play

Episodes are written to watch together, and every home task needs a caregiver.

Evidence55% of Akili Kids viewing was a child with a parent or caregiver. Interactive radio, built on adult-facilitated participation, raised achievement in at least 25 countries.

Little and often

Short weekly doses across a full term, not one long session.

EvidenceOne 16-minute play of Friendly ATTAC raised confidence but did not change behaviour. Khan Academy Kids gains came from about 13 minutes a day over ten weeks.

One code, sung

Five short rules set to a song children can repeat, in Kiswahili and English.

EvidenceNSPCC's PANTS rule and Pantosaurus song passed 1.5 million online views and 2.4 million cinema viewers within three weeks of launch.

05 · Play mechanics

What to use, what to refuse

Gamification works, but modestly. A meta-analysis found small positive effects on learning (g = 0.49), motivation (0.36) and behaviour (0.25), and only the learning effect held in the most rigorous studies. Commercial children's apps use the same tools to keep children hooked and sell to them. The difference is in which mechanics are chosen.

Use

Story chapters the class unlocks together.Progress belongs to the group, so no child falls behind in public.
Choice scenarios with visible consequences.Children rehearse the decision before they meet it for real.
Team challenges between class groups.Collaboration inside teams, light competition between them.
Skill levels shown in the Safety Book.Competence is shown by what the child can do, not by points.
A weekly family ritual with no penalty for missing it.Duolingo found a low barrier builds habits: cutting its streak to one lesson raised new learners on a streak by 19%. Digital Mapema keeps the habit and drops the loss.
Characters who model asking for help.The hero's strongest move every season is telling an adult.

Refuse

Individual leaderboards.Public ranking shames the children who most need the programme.
Daily streaks that punish a missed day.Children on shared phones do not control when they get access.
Points or prizes for time spent.Expected rewards undercut children's own interest.
Ads, purchases and paid tiers.A 2021 complaint to the US FTC by 22 groups said children playing Prodigy could see up to four times more ads than maths exercises, with rewards locked behind membership.
Return prompts, countdown timers and characters who guilt children.Four in five apps used by young children in one US study contained manipulative design like this (JAMA Network Open, 160 children).
Open chat, profiles or tracking.The UK Children's Code bars nudging children to share more data and sets location off by default.

06 · What works elsewhere

Platforms that moved children

Fourteen platforms, what made each work, and how strong the evidence is. Several were tested with younger children or adults, so the lesson travels better than the number.

Platform	Form	Result	Evidence	Take for Digital Mapema
Be Internet Awesome and Interland (Google)	Classroom curriculum on five themes, paired with a free browser game	Trial with 1,072 pupils in grades 4 to 6 raised knowledge and confidence. Interland was played for over 163,000 hours.	RCT	Pair the game with teacher lessons; localise the five themes.
KiVa (Finland)	Recurring lessons, school routines and bystander roles for ages 7 to 11	Reduced online victimisation in Finland. Bullying fell by 13% to 51% across UK, Finnish, Dutch and Italian trials.	Multiple RCTs	Recurring sessions and roles for bystanders, not one-off awareness.
Akili TV (Kenya)	Free-to-air children's channel on every Kenyan operator's basic package	Reached 36% of Kenyan children within six months of launch; most-watched channel in homes with children by end of 2022. 55% of viewing was with a caregiver.	Reach	First broadcast partner to approach; write episodes for watching together.
Ubongo (Tanzania, airs in Kenya)	Animated series in Kiswahili and English across TV, radio and mobile	Akili and Me trial with 568 children raised early learning skills in four weeks.	RCT, ages 4 to 5	Regional co-producer with Kiswahili characters and a research habit.
Eneza Education (Kenya)	SMS lessons and quizzes on basic phones, KES 3 a day, with Safaricom	10.2 million learners reached, 70% rural. Pupils reported 23% better performance after nine months.	Reach, self-reported	Feature-phone channel for caregivers who are not online.
NSPCC PANTS and Pantosaurus (UK)	A five-letter rule, an animated dinosaur and a song, for ages 4 to 8	1.5 million online views and 2.4 million cinema viewers in three weeks. NSPCC says over 400,000 parents used PANTS to talk with their children.	Reach, self-reported	One memorable code, sung, that parents use too.
MTV Shuga (first season filmed in Kenya)	Drama series with the health message in the main plot	Raised HIV testing and lowered chlamydia among women viewers in Nigeria. Subplot messages had no effect.	RCT, ages 18 to 25	Keep every lesson in the main plot.
Khan Academy Kids	Learning app with characters and short daily activities	Trial with 49 four and five year olds: literacy rose from the 34th to the 47th percentile in ten weeks at about 13 minutes a day.	RCT, small	Short, frequent doses work.
PBS KIDS transmedia (US)	Episodes paired with games built on the same characters	Over 450 children given both outperformed controls in science and engineering after eight weeks.	RCT, ages 4 to 5	Every episode gets a matching game or activity.
Minecraft Education CyberSafe	Story worlds where players choose to ignore, correct, mute or report	Each challenge ends in reflection that builds the player's own Good Game Guide. No outcome data published.	Design only	The Safety Book: children write their own rules.
Friendly ATTAC (Belgium)	Single-player game: respond to cyberbullying posts as a bystander	One 16.6-minute play raised confidence and intention to help (g 0.26 to 0.30) but not behaviour, in 216 pupils aged 13 to 14.	Cluster RCT	One session is not enough; build dosage.
ParentText (South Africa)	WhatsApp chatbot for caregivers, built on Parenting for Lifelong Health	Randomised trial with more than 20,000 caregivers; results not yet published.	Pending	Model for the caregiver channel; track the results.
Duolingo	Streaks and daily goals, an adult product	Learners on a 7-day streak were 2.4 times more likely to return the next day.	Company data	Habit design works; strip the loss mechanics for children.
Prodigy	Free maths game in school, paid membership at home	22 advocacy groups asked the US FTC to investigate its ads and its rewards locked behind membership.	Cautionary	No ads, no tiers, the same game for every child.

What this research changes in the model

The child layer is device-free. Phones appear only in moments shared with an adult.
Every week follows one loop: watch, play, take it home, write it down, celebrate.
The caregiver channel splits: WhatsApp for families online, SMS and radio for the many caregivers who have never been online.
Akili TV and Ubongo join the partner list for broadcast and Kiswahili co-production.
Design guardrails join the safeguarding guardrails: no ads, rankings, loss-based streaks or manipulative prompts.

07 · Outcomes framework

What we measure, and how

The primary outcome is disclosure: children telling a trusted adult. Reach is reported but never counted as success. Targets are set after the baseline with the evaluator, so no figure here is a promise.

Code	Level	Outcome	Indicator	How measured	When	Evidence basis
C1	Primary	Children would tell a trusted adult when something goes wrong online	Share of children who name a trusted adult and say they would tell them, across set scenarios	Scenario survey, read aloud for younger grades	Baseline, end of term, 6 months	32.7% of Kenyan incidents never disclosed (UNICEF)
C2	Intermediate	Children actually disclose	Disclosures recorded by school focal points and referrals to Childline 116, per 1,000 pupils	Anonymised school records	Every term	A rise is expected first and read as a good sign
C3	Short-term	Children recognise tricks, risky requests and private information	Share of scenario items answered correctly	Assessment adapted from the Be Internet Awesome trial	Baseline, end of term	Knowledge gains in the BIA randomised trial
C4	Short-term	Children feel able to handle a problem online	Mean score on a self-efficacy scale	Validated scale, localised	Baseline, end of term	Self-efficacy gains in the BIA trial
C5	Intermediate	Less unkindness and bullying online	Self-reported frequency of being treated badly online	Victimisation items used in KiVa trials	Baseline, end of term, 6 months	KiVa reduced online victimisation in Finland
P1	Intermediate	Caregivers talk with their child about online safety	Share of caregivers doing so at least monthly	Caregiver survey	Baseline, end of term	Metric used by O2 and NSPCC
P2	Intermediate	Safety settings applied at home	Share who applied at least one device or account setting	Caregiver survey with a guided check	Baseline, end of term	Parent controls are a core O2 and NSPCC action
P3	Short-term	Caregivers know harm often comes from people the child knows	Share answering the known-person items correctly	Caregiver survey	Baseline, end of term	Most Kenyan incidents involved known people (UNICEF)
P4	Short-term	Caregivers know where to report	Share who can name Childline 116 and the school focal point	Caregiver survey	Baseline, end of term	Reporting gap in Disrupting Harm Kenya
T1	Intermediate	Lessons delivered as designed	Share of planned lessons delivered in full	Teacher logs, spot observation	Every week of the term	Dosage drives effect (Jones et al. review)
T2	Intermediate	Disclosures handled correctly	Share of teachers following the protocol in a scenario check	Structured scenario check	After training, end of term	Safe handling is required for C1 to work
S1	Guardrail	Every creator and staff member vetted before any production	Share vetted, with Certificate of Good Conduct and safeguarding training	Safeguarding register	Before each production cycle	Design standard: 100%, no exceptions
S2	Guardrail	Children's data minimised and consented	Share of assessed children with recorded parental consent and child assent	Consent register, data audit	Before each data collection	Data Protection Act 2019; design standard 100%
S3	Intermediate	Content adopted by the system	KICD content approval; CA recognition under its COP programme; funders committed	Signed instruments	After the pilot	Path to national scale
S4	Reported only	Reach at full dosage	Pupils completing all lessons; caregivers completing the parent series	Platform and school records	Every term	Context for outcomes, never a success measure
S5	Reported only	Families complete the weekly loop	Share of pupils whose household completes the weekly home challenge	Safety Book check, SMS replies	Every week	Dosage check for the home channel, not an outcome

Impact measure: online victimisation and technology-facilitated abuse among children aged 7 to 12 against the baseline, read alongside the CA and University of Nairobi national survey once published. Most Kenyan evidence covers ages 12 to 17, so the pilot baseline is itself a new contribution.

08 · Pilot evaluation

Prove it in one term

Schools are randomly assigned to start the programme now or after the pilot. Waiting-list schools receive it later, so no school is left out, the same design KiVa used in the Netherlands. An independent evaluator sets the sample size and the smallest effect worth detecting before the baseline.

Schools are drawn from urban, peri-urban and rural counties so results hold beyond Nairobi. Any disclosure during data collection follows the protocol, with referral to Childline 116.

BASELINEBefore term starts

ENDLINEEnd of term

FOLLOW-UPSix months later

09 · Guardrails

Non-negotiables

No creator or mentor has a direct contact channel with any child.
No child-targeted brand sponsorship or product placement in any content.
The safety lesson sits in the main plot. Shuga found that messages in subplots did not change behaviour.
Children are reached through schools, TV and radio, never on social platforms that set 13 as the minimum age.
Children's data is collected only for evaluation, with parental consent and child assent.
Governance stays politically neutral; no partisan figure fronts the programme.
No ads, purchases or paid tiers. Every child gets the full programme.
No individual leaderboards, loss-based streaks, countdown timers or characters that guilt children into playing.

10 · Delivery arc

From concept to a decision on scale

Each phase ends in a named instrument with an owner. Durations are proposed and set with partners.

Days 1 to 30

Set up

Constitute the advisory board, appoint the safeguarding lead, confirm the Be Internet Awesome adaptation licence.

InstrumentLetter of Intent with CA under its Child Online Protection programme

Days 31 to 100

Build and baseline

Localise one term of lessons, produce episodes and parent clips, run the baseline in pilot schools.

InstrumentContent review agreement with KICD; evaluation agreement with the evaluator

One school term

Pilot

Deliver in programme schools; waiting-list schools continue as usual. Weekly delivery logs track dosage.

InstrumentPilot agreements with participating county education offices and schools

After the term

Decide

Publish results, adjust content, spin out the programme, and agree funding for scale.

InstrumentFunding compact with telcos and platforms; incorporation as a company limited by guarantee

The ask

Four moves this month

Approach CA about the under-12 data in its national survey with the University of Nairobi.
Confirm the Be Internet Awesome adaptation licence and test Google's appetite to co-fund localisation.
Open conversations with Akili TV and Ubongo on broadcast and Kiswahili co-production.
Brief an independent evaluator on the pilot design before any content is produced.

Evidence base