Jamia moved to Lusaka, Zambia, at the age of 24 in search of a better life. Jamia grew up in a small village far from Lusaka and moved to the city to pursue exciting job prospects and a vibrant social life. Established in 1905 as a colonial hub for exporting copper, Lusaka is now a rapidly growing city of almost 2 million people.
When she arrived, finding an affordable place to live was challenging. Jamia ended up renting a flat in one of the many peri-urban slum areas around the city—areas characterized by a lack of infrastructure, low-quality housing, and weak local government and regulatory enforcement. Her room was small but sufficient. Her walk to work wasn’t too bad, either. But every day, she dreaded the thought of having to use the toilet she shared with the four other tenant households living on the same 900-square-foot plot.
The toilet’s walls were made from a combination of cement bricks and plastic tarp, and the floor was concrete. But there was no roof, and the wooden “door” was not actually attached to the structure. The toilet was poorly ventilated, rarely cleaned, and full of flies. It had barely enough space in which to turn around. The pit, which contained the waste, was simply a hole in the ground, and Jamia had heard stories of toilets in the area collapsing while in use. The pit overflowed any time there was the slightest rain.
Jamia’s landlord lived on the other side of town, and her complaints to his rent-collecting agent fell on deaf ears. The tenants blamed each other for making a mess and lacked the trust to work together to improve their situation. So each night after returning home from work, Jamia snuck onto a nearby plot to use their toilet, wary of the angry screams of her neighbors if caught.
Jamia’s situation is all too common—some 700 million people worldwide use shared sanitation in peri-urban areas, sanitation that can be dirty, unpleasant to use, and even leave its users vulnerable to assault. Poorly constructed toilets in peri-urban areas are difficult to empty when they fill up, and though many governments are planning extensive sewer networks, most existing toilets aren’t well-built enough to connect to them.
Jamia’s story illustrates the behavioral challenges associated with sanitation in peri-urban areas in lower- and middle-income countries. Recently, my colleagues Robert Aunger, at the London School of Hygiene and Tropical Medicine, and Roma Chilengi and Jenala Chipungu, from the Center for Infectious Disease Research in Zambia, and I set out to help solve this behavioral challenge.
Every day, she dreaded the thought of having to use the toilet she shared with the four other tenant households living on the same 900-square-foot plot.
Our goal was to design a low-cost, scalable intervention to improve the quality of shared peri-urban sanitation that could be adapted for use throughout sub-Saharan Africa. We worked with over 1,000 landlords in Bauleni, a typical peri-urban neighborhood in Lusaka with a population of about 64,000.
We approached the problem using behavior centered design (BCD). BCD, originally developed by Aunger and Val Curtis, is a behavior-change framework operationalized in an “ABCDE” design process: Assess existing knowledge, Build knowledge where it’s missing, Create an intervention using professional creatives, Deliver it, and, finally, Evaluate it.
Assess: In the assess phase, we discovered that much of the existing research linked better sanitation with factors like wealth and level of education. (Not exactly factors that are easily changed with a low-cost intervention.) Previous research also revealed health messaging to be not terribly effective. However, some work showed that an effective strategy in some contexts is to make toilets a status symbol and use community pressure to keep them clean.
Build: Next, we met with dozens of landlords and tenants and observed scores of toilets to build knowledge where it was missing. The BCD framework provides guidance on the kind of information to gather during this phase. The framework argues that behavior occurs at the intersection of the environment (social, physical, and biological) and the body (whose behavior is driven by a brain at the reactive, motivated, and executive levels).
The main behavioral challenges we identified were poor cleanliness and a lack of structurally sound toilets. The social cohesion among those who shared toilets and in neighborhoods was also weak. Few landlords viewed toilets as a status symbol, and caring for the health or dignity of tenants was far from their top priority. Finally, it was clear that landlords called the shots on toilet improvements, so we knew we’d have to develop a novel approach to motivate them to improve the sanitation conditions.
We found that landlords became landlords to generate income, often as a form of retirement plan, and their biggest pain point was dealing with conflicts that arose on their plots. We realized that a key to success would be figuring out how to merge quality and cleanliness of the toilets with what was important to the landlords.
Create: We were next ready to create an intervention to change behavior. The first task was to improve cleanliness. Toilet users usually took turns cleaning the toilet, and other programs had encouraged group meetings where tenants and landlords encouraged everyone to try harder and do their part. But the existing system—a daily rotation with responsibility passed along verbally—led to confusion over whose turn it was and conflicts over whether someone had cleaned when it was their turn.
BCD’s behavior-change process is a basic reinforcement-learning process—an external stimulus (a message, an object, or a sensation) may increase the perceived value of a target behavior, and then when the behavior is performed, rewards that are different from what was previously expected cause a person to change the perceived value of that behavior, and a feedback loop drives behavior change. We needed to make sure tenants saw the value of cleaning the toilet, didn’t think that others were cheating the system, and thus learned to continue the behavior.
To do so, we introduced a weekly rotation that was marked by a symbol hung outside the door of the responsible household. Whereas the previous system had been discouraging (when faithful cleaning appeared to make little impact) or difficult (when trying to pin down noncleaners), this system was simpler and improved accountability. Those now tasked with keeping the toilet clean were rewarded with praise from landlords and other tenants, and those who failed to clean were easily identified.
The more complex challenge was motivating landlords to improve toilets’ structural quality. We measured how much tenants were willing to pay for better sanitation and calculated that in cases where there were at least three tenant households, it was a better investment to improve the toilet rather than build more rooms (as was commonly assumed) to increase rental income. Financially, then, it made sense for landlords to invest in toilets. But practically, this didn’t ensure better sanitation. Tenants rarely felt comfortable talking to landlords about an awkward subject like toilet cleanliness, so a question loomed: how would landlords interpret and react to this information?
Tenants rarely felt comfortable talking to landlords about an awkward subject like toilet cleanliness, so a question loomed: how would landlords interpret and react to this information?
Deliver: We partnered with a local marketing firm to produce surprising, hidden-camera-style videos of tenants’ complaints about toilets (the stimulus), games and demonstrations (increasing value), and accountability mechanisms (facilitating performance). We collected extensive baseline data and delivered the program as a series of four meetings occurring every two weeks. The meetings started by targeting small improvements and each week grew to encompass bigger and more expensive ones; even this was structured to facilitate the reinforcement learning feedback loop.
Landlords were quick to persuade each other of the truth of the math, chiming in with relevant experiences they’d had with making these improvements and helping each other troubleshoot. As landlords rarely met as a group to discuss this business they were all in, creating a new social environment—one that encouraged deliberation and reflection—shattered the assumption that building more rooms was the best way to increase rental revenue.
Evaluate: When we evaluated the program through a randomized controlled trial, we found that landlords had improved the hygiene (9 percent), privacy (11 percent), accessibility (10 percent), and sustainability (10 percent) of their toilets (publication under review). Not only did more landlords institute cleaning rotas, but many switched to weekly cleaning systems with visible reminders, which led to higher reported rates of cleaning and visibly cleaner toilets.
Landlords also made structural improvements, some of which went beyond the scope of the study; while we did promote adding locks to the toiled doors, some landlords also installed new, sturdier doors. Some landlords made small changes, and others totally rebuilt toilets, with strong pits, pristine structures, and flushing toilets. Most of all, by helping communicate that tenants are willing to pay for sanitation, we hope that market forces will help drive further improvements as word spreads about the financial, social, and health benefits of better sanitation.
The program didn’t solve all of Lusaka’s sanitation problems; the government is still building sewage lines and treatment plants, and some tenants can’t afford better toilets. But for Jamia and many others, they now have the ability to take charge of improving their own health, by paying a slightly higher rent to have a cleaner, more pleasant to use, and more sustainable toilet.
If you’re trying to solve your own behavioral challenge, we have three recommendations based on BCD and our findings. When looking for drivers of behavior change, we recommend using a framework of determinants that captures those in the brain, body, and environment, rather than focusing only on psychological drivers. When looking for how to sustain behavior change, we suggest looking at target behaviors as learning failures to be corrected by the power of reinforcement learning. And when looking at how to apply these findings in practice, we believe that a systematic process of design and evaluation will lead to the best results.