Safety and the Perception of Safety

There's a certain brand of writing about cities that plays a particular game of sleight of hand: conflating the statistical realities of cities with what people think about cities, and acting as if any anecdotal evidence supporting the latter is transubstantiated into the former. The latest example of this comes from The Atlantic about public safety on public transit.

I have thoughts, having just been in a city that people claim often is unsafe and having never felt unsafe at all.

1. Perception vs reality

Now it's a valid point that people who don't feel safe on public transit won't go on it (unless forced). But the sleight of hand here is that treating the perception of safety as the same as actual safety is that it elides a critical question: how do you convince people that something that is safe is safe? When you conflate perception and reality you can claim that what needs to be done is to make public transit safe.

But if you admit that public transit is safe, and that the problem is perception, it's actually much harder to produce good answers to what to do. This is why Lehman's article is so dismissive of Matt Yglesias, who makes the point that transit is safe. Actually engaging with real safety data is quickly pushed aside for survey data on perception --- which is not the same!

A key tell here is moving from discussion of rates to absolute numbers. Lehman says that "large transit agencies reported just shy of 2,200 assaults last year" and while he's probably right that that's undercounted, he doesn't give any sense of what that tracks out to in terms of the actual likelihood of experiencing or even witnessing such a crime.

There were 4.7 million daily transit trips (bus + train) in New York City on the MTA alone last year, 1.3 million on trains and 3.4 million on buses.

Let's do some math. A full MTA regular bus is expected to carry 60 people, and articulated bus 85. To maximize the chance you'd see or experience a crime, let's assume every bus is articulated and full, so we estimate a lower number of bus trips for that capacity of people. That means we would need a minimum of 40,000 bus trips (3.4 million divided by 85) for those bus riders. The same math for trains gives us (1.3 million divided by about 1400) another 900 trips (aren't trains efficient?). If every single reported assault on transit happened in just 1 day just in New York, then, the 2200 assaults would be spread over 40900 trips at an absolute minimum. If we further assumed that all trips had equal chance of violence, we can just divide them equally. That's a bit over 5% chance that you'd see or experience an assault. Wow! That probably feels high! Transit is dangerous! 1 in 19 to 20 trips see an assault!

But wait. That's is every assault on "large transit agencies" was on one agency. And all of the assaults from a year were on a day. And all of the trips had maximum capacity of riders so our estimate minimizes trips (and thus maximizes the chance of seeing an assault on a given trip). 

Each of these ramps up the danger (note that the year/day one does it by a factor of 366, since 2024 was a leap year). Actually, it seems that you're REALLY REALLY UNLIKELY to see or experience assault on transit.

But that doesn't comport with people's perception. Well, that means the problem is with perception and communication, not safety. That's a different issue! But it gets framed as if transit is actually unsafe, not as if there's a fundamental disconnect between its safety and public perception.

2. Same tired answers

And as we might expect, Lehman's answer to this is a return to broken windows policing, aka overpolicing transit to make it feel safe. This is always the answer that a certain group wants to impose both on the public generally and any public amenity specifically whenever fear mongering has created a sense of public unease. It comes, as always, with a callback to the heyday of this kind of policing in the 80s to early-mid 1990s and an insistence that this period shows that this policing is more effective than anything we do nowadays.

One problem? Again, this misses statistical reality. Crime is massively down since then. Cities and the country are genuinely much safer than they used to be. Why would we police the same way now that we did when there was two to three times as much violent crime? And how can we say those techniques were more effective than what we do now?

What is needed isn't different policing; it's different communication. It's a government and a media that doesn't try to scare people. It's actual numeracy that speaks to how increasing population produces (sometimes) high absolute numbers but low rates: 4.7 million DAILY trips is legitimately hard to wrap your head around. And it's people like Lehman admitting that people like Yglesias have a point: one high-profile attack on a Charlotte, NC light rail train doesn't undo decades of falling crime and increasing transit trips.

3. More transit for more people 

Lehman is right that it will take more than just building more transit to get people to change their riding habits. But in addition to public education about the safety of transit, the best way to get people to take transit remains to make transit actually useful. It's much easier to declare that transit is dangerous, ugly, and useless when you don't actually get any benefit from it than it is when you shave half an hour off your commute with it. It's easier to dismiss it as a symptom of those dangerous cities (itself a misnomer) when it doesn't serve you. 

This also means getting transit out of a defensive crouch. Too many transit projects get watered down or even cancelled because the politicians or even the designers themselves pre-compensate for these objections. They build worse transit because they don't have or want to spend the political capital to build good transit. And then no one rides the bad transit, and then people go on not experiencing decent transit and keep their bad assumptions about transit in general.

This isn't enough alone. Public education is necessary (but sadly unlikely from a federal level). But no matter what Lehman thinks, this is both chicken and egg: we don't get people to trust transit just by building it, but since it's a matter of perception and not actual safety, that is a critical element of the process. To show people their perceptions are wrong, sometimes you have to actually show them.