There’s much that is surprising about the city of Mumbai, but one of the most astonishing facts is the leading cause of unnatural death in the city: being hit by a suburban train. On average, trains in this Indian metropolis kill 10 people every day who are attempting to cross the railway tracks.
This shocking figure remained stubbornly high for years, in spite of numerous efforts by railway and city authorities to bring the number of fatal collisions down.
Final Mile’s successful attempt to reduce the number of deaths on Mumbai’s suburban rail tracks is the best known of our projects, though our organization was barely a year old when we took it on as a pro bono experiment.
By studying the data carefully and investigating the psychology of risk perception, we were able to save numerous lives with simple, science-informed, and cost-effective design interventions. It was an experience that shaped Final Mile’s journey and changed our trajectory as an organization.
Mumbai’s trains – essential, but sometimes lethal
To understand why so many people get hit by trains in Mumbai, it’s important to consider the city’s unique and complex geography. Mumbai’s suburban railway network dates back to the 19th century, and now carries a whopping 7 million people a day from the suburbs into the narrow peninsula that is the heart of the city. By helping Mumbaikers manage their daily commute to work or school, it’s very much a lifeline. But the railway’s layout also makes it a potential death trap. There are almost no stretches of underground or elevated track, so the multiple lines traverse the city mostly at ground level, running right through the dense residential and commercial neighborhoods that have sprung up over the years.
Crossing the train tracks is an unavoidable peril for thousands
Every day, people of all ages must cross from one side of the railway lines to the other to get to school, to shops, or to work – and then cross back. There are few footbridges, and they’re arduous to climb because they have to be tall enough to avoid the power lines running above the trains. Inevitably, most people who need to get to the other side take the “easier” route of crossing the tracks themselves. But the danger is immense: there are typically four sets of tracks to get over, with trains passing on average every minute, 21 hours a day, and often at speeds of 60mph or more.
The need for a new approach
When the railway authorities briefed Final Mile on this challenge, they told us about their attempts to reduce the tragic loss of life on the network. They had posted warning signs by the tracks. They had run daily print and radio advertisements warning people about the dangers of trespassing on the lines. They had built walls and fences in some places to physically prevent people from crossing. They had tried enforcement drives to catch and fine people who went across the tracks. And none of this had worked. The authorities were looking for a different solution to changing this potentially deadly behavior.
Attempts to prevent trespassing had failed because they had not considered how people make decisions
Getting to the roots of the behavior
The structural problems causing the high number of fatalities were clear to us, but the psychological factors were not. The authorities proposed a number of hypotheses for us to investigate. Did most deaths occur at railway stations, where the largest numbers of people crossed the tracks? Were the people being struck by trains thrill-seekers? Or was it elderly people with poor reaction times who were getting killed? Was it due to bad light or poor visibility? Or was it simply that some people lack common sense?
We reviewed data from railway police logs, talked to train drivers, met with people who had lost friends or relatives on the tracks, and carried out our own detailed observations. What we learned surprised us. First, despite the large number of track-crossings at stations, almost no fatalities occurred there: people were getting hit when crossing the lines between stations. Most of these deaths happened in broad daylight, where visibility was not an issue. And most of the victims were not elderly, but healthy young men who gave no evidence of doing it for thrills. They were just normal people crossing the tracks for everyday needs. But one observation that struck us was that those who died had crossed the tracks not because they didn’t see the oncoming train, but in spite of clearly spotting it. They were aware of the risk, but overconfident about their ability to cross safely. It was notable that although in our observations men made up about 60% of trespassers, the data showed that they accounted for 90% of the fatalities.
People crossed tracks even when they could see a train approaching
Our research led us to three initial conclusions. First, the current system of warning signs, using text and stick-figure illustrations, was ineffective. Second, the train drivers’ practice of sounding the horn didn’t succeed in alerting people: either it was too familiar to catch people’s attention, or they honked too late to make a difference: once people were on the track and heard the horn, they decided to run for it rather than to turn back – but they couldn’t outrun the train. A third, related observation was that people crossed even when they could clearly see the oncoming train. We discovered a reason for this in what’s known as the Leibowitz hypothesis, which states that people underestimate the speed of oncoming large objects. In the case of trains, we may perceive their speed to be as much as 40% lower than it actually is. If someone thinks a train approaching at 60mph is traveling at less than 40mph, they’re more likely to think they have enough time to cross the tracks safely – a miscalculation that can prove fatal.
Leveraging our insights for simple but effective solutions
In designing our solutions, we had to work within several constraints. First, whatever we came up with had to be cheap to implement. Second, anything that required installing could only be put in place during the three-hour stretch each night when no trains are running, so it had to be quick and simple to do. Third, we couldn’t use materials that would be liable to be stolen in a city where plenty of people make their living by recycling. Finally, warnings with written text were impractical, because some Mumbaikers are illiterate, and because so many different languages are used in the city.
We designed solutions that worked at a nonconscious level
We went through several rounds of design before settling on three interventions. First, the existing text-based warning signs needed to be replaced with something that would work instinctively and at an emotional level – essential when every millisecond counts in the decision to cross a track or stay back. We proposed putting up posters at trespassing points with a photograph of an actual person, capturing his face in close-up the moment before he is hit by a train. The goal was to provoke an instant and visceral fear in viewers. (They didn’t need to know that the unfortunate victim was really an actor!)
Second, we recommended changing the way drivers used their horns to warn people. From scientific studies, we determined the best distance from trespassing points for the approaching train to honk, so that it would be heard neither too soon nor too late. We added a further refinement based on a study showing that brain attention peaks during the gap between two musical notes: the listener’s attention is grabbed and held while they wait for the second note. Inspired by this finding, we recommended two staccato horn blasts instead of a single long hoot.
Finally, we had to tackle the problem of incorrect perception of the train’s speed. Since this is not a conscious process, the solution would have to work at an unconscious level too – and instantaneously. A sign saying “The train’s going faster than you think” would not do the job. Our solution was to paint groups of railway ties yellow at regular intervals, on either side of trespassing points. The bright stripes attract attention, and as they disappear under a moving train, the brain quickly and unconsciously recalibrates its estimate of the train’s speed.
Taking the concepts to the tracks
Since all these interventions were new, we had to show proof of concept. We piloted them at a trespassing point that saw an average of 40 deaths a year. After one year implementing our three interventions, the number of fatalities had fallen to 10 – a reduction of 75%. This so greatly surpassed everyone’s expectations that some members of the railway authorities were skeptical that our interventions could have been responsible for the decrease. So we scaled up the solutions to five other locations, and saw reductions of between 40% and 70% in the number of deaths there.
Our first pilot reduced fatalities by 75% in a year
The solutions were subsequently scaled up to multiple locations around the suburban railway network. In the most recent evaluation, in 2019, a 5km stretch of the Central Line that had seen 88 deaths in the previous 11 months (the highest number on that particular line) experienced a 44% decline in fatalities in just six months. At each of the 22 trespassing locations where the solutions were implemented, the number of accidents fell by 80-90% during the same period.
Final Mile learned a lot from this project. Despite our success, a shortcoming that became apparent in retrospect was that we had focused only on the end user – pedestrians – and did not invest time in understanding the decision-makers in the railway system. Implementation of our solutions suffered numerous delays and bottlenecks because we lacked good insight into the motivations and decision-making processes of the officials who were ultimately responsible for putting the solutions into practice across the network.
The Mumbai railway project was a case where everything came into focus for Final Mile. A problem was understood and redefined. Research identified the real causes of behavior and the solutions, and there was a great marriage between scientific principles and design concepts. The experience shaped our approach to the work we have done since.