Walk With Makmende: Part 3

In our previous post, we described our prototype for a system of community-based walking groups that would escort women safely to and from their homes. But would an idea developed by a group of students really make sense in the context of an informal settlement in Kenya? How could we be sure that we had captured all the variables—variables that could either derail or bolster our project?

We had tried to validate our idea with partner organizations in Kenya, especially to make sure that our solution resonated with local residents in Mathare (our “users,” in d.school speak). Nonetheless, 10,000 miles and 11 time zones separated them from us, and we could glean only so much during midnight Skype calls to Nairobi. As a result, Makmende v0.1 was based on a set of assumptions and anecdotal evidence (such as the idea that women do indeed feel safer walking in groups, that people know how to send text messages, and that smartphones are ubiquitous). But before we felt comfortable implementing a pilot, we needed to know more.

We also recognized that the success of Makmende would require a strong human element: Would we be able to find people who were well-known and trusted by the community and who would be willing to participate as escorts for the walking groups? We would be asking people to wake up at 4:30 am and stay out until 11:00 pm, 6 days a week, and to walk through the most dangerous parts of their communities—all in an effort to provide a public benefit for their neighbors. Our main local partner, Mathare Youth Sports Association (MYSA), already told us that it felt uncomfortable with the idea of having their members participate as escorts, partly because they feared the escorts would be attacked. Instead, they suggested that we sound out the police force and local community-policing groups.

But we knew essentially nothing about these institutions—would our project be something they wanted and had the ability to take on? Police corruption is often a serious issue in informal settlements—would local residents trust us if we worked with the police? Would our new partners be able to find appropriate individuals to serve as escorts?

Over the summer, we traveled to Kenya to try and fill these gaps in our knowledge. First, we conducted a survey of 440 households in Mathare to collect some quantitative data about our users (and since Mathare is an informal settlement, such data are virtually unique). The survey in itself was quite a challenge, as we had to rely on local enumerators who had no previous data collection training. In addition, language and cultural barriers complicated our work in unexpected ways. For example, the survey asked respondents whether they personally owned a mobile phone, but in Mathare, sometimes a whole household shares access to a single phone. Phrasing the question as we did was akin to asking an American: “Do you personally own a landline?”

These challenges aside, we did learn several important things. First, about half of our respondents told us that they did indeed need to leave their homes during dawn and dusk hours, and women overwhelmingly felt that walking in a group improved safety. These findings gave us confidence that both the safety problem as we had framed it and our proposed solution to that problem resonated with the community.

On the technical side, we confirmed that mobile phone penetration is high, but we were surprised to find that, even in Mathare—one of the poorest settlements in Kenya—people overwhelmingly prefer (more expensive) voice calls to text messaging. Given this finding, we thought it would be a good idea to give our users the option of calling the escorts directly to find out the location of walking groups, as opposed to sending a text message to the Makmende hotline as we originally envisioned. This was one area where a simple local solution made more sense than what we envisioned in our class.

During our summer trip, we also met extensively with the police and community-policing groups. We were gratified to learn that both groups were enthusiastic about deploying Makmende as a tool to improve neighborhood safety and that they could help us identify trusted local community members who would be willing to play the escort role. Furthermore, our new partners agreed that the community would be introduced to the escorts at a community forum (baraza), where community members could veto any candidates they disliked.

Our partner meetings also helped generate design improvements. For example, the police district officer asked us: if we could use technology to let users know about the real-time location of escort groups, could we share that information with the police as well? And what if we gave the escorts a quick and easy way to alert the police if they ran into trouble—say, a panic button that, when pushed, could draw police to a particular GPS location? Such a tool would be useful in Mathare, where police response times are often excruciatingly slow, in part because officers have trouble locating where the trouble is (there are no street names or addresses in Mathare).

As we began to tweak our project design, incorporating the new information we learned during our trip, we started to think differently about our approach. Slowly, we realized that the technology we were bringing to Kenya was less useful for distributing location information to users (who could, after all, simply call the escorts directly) and more useful for alerting the police to crime hotspots in real time. Group walking still made sense, but we were less sure about how necessary our fancy gadgets were to the endeavor. Certainly the involvement of technology has galvanized the community around the idea of an escort service, but in the end, Makmende will stand or fall because of community buy-in, not because of any value a smartphone can add.

In our next post, we will cover our efforts to launch a pilot and evaluation study to precisely gauge the impact of Makmende and its potential to improve safety in one village in Mathare.

Read more stories by Nan Zhang.