[_delirium]

> commuter rail kills .09 people per 10 million passenger miles.

The non-suicide number is quite a bit lower, though, by (probably) an order of magnitude. Pedestrians on the tracks in an area they aren't supposed to be, who are struck and killed by trains, are classified in the bucket category of "trespasser" fatalities. Unless they left a note or something it's usually impossible to conclusively say that they were suicides, but the assumption is that the vast majority of trespasser fatalities are suicides, because the number of people struck while trying to steal things or take illegal shortcuts is relatively small. And, trespasser fatalities account for 88% of total commuter-rail fatalities (p. 13 in that report). All other commuter-rail fatalities (passengers, collisions with automobiles, etc.) add up to 0.011 people per 10 million passenger miles.

[trothamel]

The way I'm reading table 18, of the 526 commuter-rail fatalities, at most 285 of them are suicide-by-train. As you point out, trespassing is a wide category - it includes things like highway-rail collisions. That 285 number might be high, as it also includes things like a child straying onto the tracks that was hit by a train.

I'm not sure it matters. The automobile statistics seem to include various methods of suicide involving cars. If rail inspires suicide, that's a safety problem.

[_delirium]

> The automobile statistics seem to include various methods of suicide involving cars.

I'm not sure they include them equally, though. The most obvious way of killing yourself on train tracks is standing in front of a train, but the most obvious way of killing yourself on a road usually does not involve a car, but involves jumping from a bridge (which doesn't get counted as an automobile fatality).

> If rail inspires suicide, that's a safety problem.

I'm not too sure about that one. If there are a significant number of people who, absent trains, wouldn't kill themselves, that would be significant. But I would guess a larger proportion are just choosing a suicide modality, and if it weren't trains, it'd be jumping from the Golden Gate Bridge, taking pills, etc. It's hard to come up with any real data for that counterfactual, though. I suppose a blunt correlational approach would be to see if there's any correlation between availability of commuter rail and suicide rates, after controlling for other confounding variables. I would guess that there isn't one, but could be wrong.

[trothamel]

You're just as dead if you jump from a railroad bridge, though.

http://www.fra.dot.gov/downloads/safety/tdreport_final.pdf

is a Federal Railroad Administration report that breaks down the cause of trespasser fatalities. This isn't 100% comparable to the previous report, but it does claim that 23% of trespasser fatalities are suicides.

Appendix H breaks down the reported fatalities - it looks like walking and sleeping (drunk?) on tracks each claim more lives than suicide.

One really interesting story is the massive increase in car safety. Deaths per 100M vehicle miles have dropped from 1.73 in 1994 to about 1.09 in 2010.

Fundamentally, I'm not sure any of this matters. We're talking a very small difference in safety - to the point where convenience is probably more important to most people.