This is part three in a series of blogs tracing each step of an iterative approach to digital programme design. As part of a UNFPA-funded project run by OneWorld to explore strategies to tackle Gender Based Violence (G.B.V.) using ICTs, we decided a sensible entry-point was to address the lack of information about laws and services in rural Benin, working with stakeholders to co-create messages for dissemination via WhatsApp and SMS. This blog details the process and results of rapid-testing this model.

Some weeks after running co-creation workshops to write educational messages on the laws surrounding G.B.V., I sat crammed into the back of my mother’s car on the way up to a family Christmas event, chatting via WhatsApp with my modern-day pen pal, Idelphonse, a young peer educator who had attended our workshops.

“What does your mum look like? 🤔 He asked. “And what are we doing next about the messages?”

If like me your WhatsApp contact list is 10% family and friends, 90% “Michael, SRHR project, Lagos”, you’ll be familiar with the highs and lows of being reachable 24/7 by project stakeholders. Early on in my career in ICT4D, I decided that whilst I didn’t love being woken up by 25 floral 💐Good Morning!! 💓 image-messages, I did love being able to build valuable remote relationships with community members. By building these relationships with WhatsApp, I've been able to maintain conversations, test ideas, even run virtual focus groups. And crucially,  as Idelphonse's message shows, remain accountable to stakeholders. 

Thanks to his enthusiasm and hard work, then, we forged ahead and set up a WhatsApp group of 11 young people with whom to test the messages, alongside the group of adults (including servicer providers and community members) created after the workshops.

Process

Our process for the prototype test was to:

1. Review and rework the messages created during our workshops, getting them validated by local experts to ensure the information in relation to the laws was accurate and the final wording appropriate
2. Select 10 messages - five aimed at youth, five aimed at parents (mostly male)
3. Give group members a final chance to input into the message content
4. Send one message per day, for five days, to each group, encouraging them to share with their friends
5. Chat with group members about who they shared with, why, how many, and what reactions they got.

We had a hunch that this method would be most effective at reaching young men, but given the enthusiasm of older workshop participants for testing the idea, we wanted to give both groups the chance to try it out.

By doing this we hoped to answer questions like:

• Can participants easily receive and send these messages? Do they want to?
• Do messages get changed as they get shared? How much of a risk is this?
• What messages work best? Least?
• How many people can be reached from only a small initial group?
• How can we assess reactions to the messages? How do people react?
• How much does gender and age affect receiving and sharing of messages?

Ultimately, we aimed to develop a more nuanced understanding of the effectiveness of this outreach method in order to move forward - either by scrapping the idea, or adapting it based on our results.

The messages

We experimented with different message types and themes to see if any did better than others. There were all drawn from the recommended stories, information and wording suggested during the co-creation workshops:

• quiz type messages: make it 'fun' and encourage sharing with friends by tapping into people’s competitiveness

• story messages: a very short narrative illustrating the topic and providing key information
• "Three facts you didn’t know about..." succint, easy to understand points
• teachable moments: - tying the information into real-time events, in this case, a public holiday celebrating the Vodoun religion
• themes including sexual harassment, rape, girls right to education, and forced marriage.

All the messages included a sentence explaining the messages were validated by UNFPA Bénin and OneWorld.

Results and Learnings

Force Jeune!

Within 24 hours of starting the pilot, it became clear that the most proactive group were the youth group. By the end of the pilot, the five most active participants had shared the messages with up to 260 people each time, with evidence that the message had been passed on by these people in turn.  

However, this wouldn't have been possibly without quick thinking on Idelphonse's part. Very quickly it became clear that the pace of sending messages wasn’t suited to the realities of local WhatsApp usage.  On a couple of occasions Idelphonse had to SMS or ring his friends when the latest message was sent through, prompting them to come online and read/send on the message.

Even when they do use WhatsApp regularly, many come online only once a week, partly because of airtime costs, partly because of restrictions set by schools and parents on their phone usage. In any case, participants felt that if we were to run a more permanent campaign, we should switch to one message a week so as not to overwhelm people.

Gendered participation behaviours

The majority of active participants were young men, which is in line with what we know about digital usage worldwide. However, although there was only one active girl in the group  (Diane, a university student who had attended our initial workshops) she was the person who shared with the most people and was the most responsive: unlike her more tentative male peers who seemed to select 15-20 people to send the messages to each time, she regularly sent the messages to everyone in her contact list, including groups she was part of, therefore reaching people who weren't in her contacts.

The people she sent to were much more varied as well, encompassing  students, NGO staff, and teachers, between the ages of 17-59. These patterns are not of course linked to her sex alone, and probably have as much to do with her education level and socio-economic status. But we'd like to repeat the exercise with her to see if we can recruit more girls in the group to start with, and assess how their behaviours vary from those of young men.

Reactions and impact

In order to understand exactly how many people were receiving the messages, and how they were responding, we asked participants to tell us how many they had sent each message to, and to forward us their reactions. This was a painstaking process made much easier by using WhatsApp for desktop in order to easily follow up with people every day, and copy/paste responses into a big document.

The pattern of responses that emerged was revealing. It included:

• Various emojis showing:

  • agreement 👍🏾   ✅

  • thanks 🙏  🤝

  • emotion 😢  😂 (NB: this last one, for our cohort at least, means 'crying' and not, as I initially thought, 'crying with laughter'. An important distinction!)

  • a promise to share  📲 🗣🤳🏾  

• Short, sympathetic answers:  “Let's dedicate ourselves to banishing these practices" 

• Beginnings of dialogue: "Have you got proof of this?"

• Requests for authentification: "Who is sending you the messages?" 

The only 'negative' response reported was in response to the message on rape, the recipient deciding to answer with a joke about sexually active girls 'asking for it'.

There was no consensus in the group about which message worked best/least - but the story message about forced marriage, as well as the message on sexual harassment in school, seemed to resonate strongly.

Excitingly, participants reported being sent some of the messages by people they hadn't sent them to originally, particularly the message sent on Vodoun Day, and the story message. This implies that, as intended, the messages entered the WhatsApp content ecosystem and continued to circulate. This also highlights the risk of releasing damaging content onto WhatsApp and similar platforms, as it will rapidly take on a life of its own.

Given this overwhelmingly positive pattern, we did ask ourselves whether we were perhaps operating in a WhatsApp 'filter bubble' - messages only being sent to those who already agree with the values implied within. Similarly, there was little evidence of real dialogue being triggered (and indeed - would our sharer group have the capacity to handle this in a safe and constructive way?)

Digging deeper

Given the positivity of answers,  and the variation in the number of people being shared with each day (15 one day, 20 the next, for example), we talked further with our initial group about what informed their decision to share, and the type of people they were sharing with. They agreed that they sent the message to people they thought would be open to the content, but they disagreed that this was a 'waste' - even for themselves, the message contained facts and arguments that allowed them to articulate things they knew or felt, but hadn't previously been able to articulate with backup from a credible source. Therefore, they sent to people who they knew this information would be useful to as well.

We also asked them to send us the first names of anyone they had sent the messages to, before selecting a random sample of 5 people to get in touch with about the messages (with their permission). We had no idea if this would work but it felt important to go beyond the potential positive bias of our first group. 

The first learning from this was that there was no significant discrepancy between the number of people shared with reported during the pilot, and the names shared a few weeks later - suggesting the initial group was being honest with us despite how easy it would have been to inflate the figures in order to make us happy.

The second was that there was no duplication in people who received the message from the first group (although as we've seen, once it entered the stream of content going around, there was a strong likelihood of receiving the same message more than once - which is not necessarily a bad thing ).

We managed to chat with four people, all male between the ages of 15-27, and asked them a) if they remembered the messages b) how they felt after receiving them c) whether they shared them with anyone else. We found that:

• all remembered the messages
• most received between 3-5 of the messages
• had never received messages of this type before
• all seemed very positive about the messages, feeling that they had learned something new, or like the initial group, felt they now had useful facts and arguments, particularly in the case of sexual harassment in school.

"This is an asset for me, because this type of thing can happen one day, and then I'll be able to correctly defend my opinions" 

However, the sample spoken to said that they didn't always forward the messages on, particularly the one about rape, because they either felt concerned that they wouldn't be able to properly continue a discussion with someone if challenged on the information, or that many of their friends might actually be 'like that' (read: the type of person who might not listen to 'No'.)

Conclusions and next steps

The elephant in the room is of course that the adults were almost entirely inactive during this pilot. It was so time consuming to manage the youth group that it was impractical to chase them up while the pilot was happening, but afterwards we  found out that they hadn't been 100% clear on what they were supposed to do, and furthermore were much more taken up with the demands of day to day life than their younger counterparts to participate.

Despite this, we feel that this model would be a great way of distributing key information to local populations, through working with young-people-as-mediators (or 'techno-educators', as they liked to call themselves) in the knowledge that they don't only share with people their own age. Given that we were able to reach about 250 people from a group of five, with evidence of more reached, there is a strong chance we can reach thousands with a standard size WhatsApp group of 256.

The changes we would make to the model would be:

• to improve the format by creating appropriate images (including experimenting with entirely image based messages with a small amount of text)
• to experiment with audio messages in local language - we received some during the week and it was interesting to be on the other end of the experiment: receiving an audio messages is more accessible on some level, but you can't always be bothered to listen to the whole thing (especially as it costs data and takes time to download the audio file)
• to experiment with parent-child dialogue creation: working out how the messages can be used as a launch-pad for conversation with non-readers
• to create a more gender balanced group to start with, and ensure that they had been selected based on a set of desirable criteria - ideally people who are key influencers in their peer group and community
• to explore the use of WhatsApp for business to provide a further level of trust.

Epilogue

A few weeks after the pilot, when we were back in Mono-Kouffo, Diane, the young woman who shared so many of the messages, stood up in a workshop to say how happy she was to see someone from the police force present, as she had been threatened by one of the people she had sent a message to: to a guilty conscience, receiving an informational message about rape can seem like a direct accusation, and a contact of hers had aggressively asked to know why she had sent him the message and warning her not to send him anything again.

She hadn't mentioned this during the pilot, and we were of course left humbled by the risks she, and we, had taken, despite our efforts to be responsible in our processes. A major and late learning, therefore, is to ensure that we run any such initiative as part of a holistic approach where participants (and recipients) are clearly and repeatedly informed about the supportive and protective structures available to them - and that these structures are themselves empowered through our work, rather than seen as tangential.

This is the focus of the next prototype we have tested, a mobile website which seeks to empower key actors with the information and resources they need to do their jobs well. Our hope is that in combination with a messaging initiative, and more in depth training for service providers on how to start addressing the issues that cause G.B.V. , we will have designed a model which has a real chance of shifting norms around violence, improving service provision, and increased access to services by victims.