Statistical Fibs Will Not Make Pedestrians Safer
Is it true that pedestrians die in San Francisco at a 70% higher rate than the national average?
I saw this claim come across my Twitter feed today from Walk San Francisco, the local pedestrian advocacy organization, and it immediately got my attention. It didn’t smell quite right to me. More after the jump…
I’m a proud pedestrian and have been car-free for nearly 14 years since first moving here. I support efforts to increase the safety of pedestrians, even dumb ones who don’t pay attention to what they’re doing while they walk. But it also gets my nostrils a-flaring when an advocacy group resorts to shady use of statistics to increase the odds of receiving funding for their cause.
When dishonest claims are made for short-term benefit, nobody benefits in the long-term (Wall Street bankers probably have a mathematical formula that disproves this).
If you go around Walk San Francisco’s own report and press release, and to the source, a report called Dangerous by Design by Transportation For America, you’ll see right away that out of the 52 metro areas with greater than 1 million residents, the “SF Bay Area” (SF, Oakland, and Fremont) is ranked 40th most dangerous. Meaning: it’s the 12th safest.
The Walk San Francisco press release (PDF) mentions this, and proceeds to generate some San Francisco-only stats. Which is where things get dicey, since you would think that, if anything, Oakland and Fremont must be dragging us down with their hyper-automobilization. Walk SF came to the opposite conclusion, though:
Nationally there are 1.53 pedestrian fatalities per 100,000 people. In San Francisco, that rate is 2.60 pedestrian fatalities per 100,000 people, 70% higher than the national average. Nationally, 11.8% of all traffic deaths are pedestrians while in San Francisco, that number is 47.7%. At the same time, the report found that nationally only 1.5% of federal transportation dollars are spent on pedestrian projects, and in San Francisco it’s even lower at 0.5%.
More fatalities per 100,000 people is easily explained by the fact that people walk at a much higher rate in San Francisco than the national average. In fact, the Pedestrian Danger Index (PDI), which the Dangerous by Design report uses to calculate the rankings, explicitly and clearly accounts for this:
The PDI corrects for the fact that the cities where more people walk on a daily basis are likely to have a greater number of pedestrian fatalities, by computing the rate of pedestrian deaths relative to the amount of walking residents do on average.
When I emailed Manish Champsee at Walk San Francisco to get his response to this, he said the following:
There is no doubt that a large part of the reason why our pedestrian fatality rate is so high is because so many people walk here… However, this shouldn’t be a reason to rest on our laurels as all too often happens… With regards to spending, I think that nationally we are spending far too little on pedestrian safety.
To be fair, Manish’s email was thorough, and he gave a well-meaning example that seemed to endorse a more utilitarian allotment of federal and other funds:
[Say we] have funds to fix 1 intersection and we have our choice. An intersection where 100 people walk through that has had 2 fatalities or an intersection where 1000 people walk through that has had 5 fatalities. On an exposure basis, the first intersection is the more dangerous, but fixing the 2nd intersection will improve the situation for many more people.
He does not address the fact that his statistical claim about SF’s rate of walking fatalities relative to the national average is not based in reality. While I might (and do, actually) agree that funding more-walkable cities all over the country is better than lots of other funding projects, this rhetorical approach ain’t kosher. Aside from the simple factual challenges, there will be a backlash, such as the one drug warriors have to face after trying to indoctrinate kids with “Just Say No” falsehoods.
To claim, as Walk San Francisco does, that “the pedestrian fatality rate in San Francisco is 70% higher than the national average,” and that this means SF is more dangerous to pedestrians, is to stretch the truth a great deal. It will be self-defeating to the cause it is meant to support.
UPDATE: Manish responds with a follow-up email, which I post here unedited:
I guess I’m a little taken aback by your assertion that we are being dishonest. We are small local non-profit that has been handed a report by a larger national organization. The report gave an analysis based on us being in a much large Metropolitan area that includes many cities that bear no resemblance to us like Fremont. Based on that, we tried to compile some numbers based on what we had to compare ourselves to what the report gave us. We simply didn’t think that using the San Francisco-Oakland-Fremont Metropolitan area where San Francisco only comprises 20% of the population had much relevance to our constituency. If I had my way, we would have just compared ourselves with other walkable cities like New York, Boston, etc., but the report created Metropolitan areas that included these cities as well as their suburbs so there was no good way to compare ourselves to those cities without digging up numbers for all of them as well.
Reasonable people can disagree on what are relevant statistics and what are not, but we’re certainly not being dishonest about anything. There is actually vigorous debate within the pedestrian community on exposure-level numbers vs. population based numbers. In the example I gave you, the academic might deem the intersection with fatalities and 100 people crossing as more dangerous, but if you are one of the 5 people killed in the intersection with 1000 people crossing you don’t really care that 995 others made it safely. It can be argued both ways, but I wouldn’t say that the person arguing for exposure based analysis is being dishonest.
The question is which number is more relevant to you. I would argue that perhaps if you lived in Tampa that the exposure-based numbers might be more relevant as you chances of getting hit by a car in Tampa are either 0 if you like most Tampa citizens don’t walk but high if you are a pedestrian in Tampa. However, in San Francisco, everyone walks. You would have to have a garage, drive to work where you had underground parking in your building and never leave to have lunch. In the evenings and weekends you could only go to places with a parking lot or valet parking. In San Francisco, the question is how likely is that I won’t come home tonight because I got hit by a car and given that everyone is at risk to some level the population-based numbers make more sense.
Beyond that is the question of how do you measure how much walking is happening in San Francisco vs. other places. Most people (including t4a) use Census walk to work data, but that doesn’t capture non-work trips (which are around 80% of all trips). It also doesn’t cover things like walking to and from your car, transit stops, etc. Beyond Census data, there is no other exposure-type of data that is collected nationally. Some cities do their own studies on how people get around, but their methodologies aren’t consistent so you can’t compare them apples-to-apples.
UPDATE II: Michael Rhodes over at SF Streetsblog has a well-rounded post on this issue. However, he does problematically highlight the “low levels” of spending on pedestrian improvements in San Francisco. He says:
While the report found an average of 1.5 percent of federal transportation spending is focused on pedestrian and bicycle safety, Champsee said San Francisco is spending only 0.5 percent of its federal transportation dollars on that purpose (MTA spokesperson Judson True said he couldn’t immediately confirm that number.)
Even if these numbers are accurate, so what? They are unrelated stats. Should SF’s budget items directly shadow that of the federal government, and if so, why? Is it not possible that, even if SF’s spending on ped improvements is somehow “low,” it’s because we already have a robust ped safety infrastructure, and therefore relatively less need for improvements?