[flickr]photo:5103472743[/flickr]
CDOT, NHTSA, and Western Michigan University are involved in the installation of these optical illusion zig-zag markings, part of an experiment to see if they improve night time visibility. You can read more about them on Steven Can Plan. Since installation, I’ve asked CDOT multiple times for the results of the experiment and each time the report hasn’t been finished.
A reader forwarded me an announcement email about the National Highway Traffic Safety Administration’s (NHTSA) Traffic Safety Facts 2010 Data (.pdf) in which one of the “facts at a glance” included this one: “Close to 70 percent [of fatal pedestrian crashes] took place at night”. He asked, “What percent of ped fatalities in Chicago happen at night? 70% seems disproportionately high.”
I looked it up. “Night” in this analysis means situations where the Illinois Department of Transportation crash data is coded as “Darkness” or “Darkness, lighted road”. I’ve divided the data in a table below to show the difference between those two designations.
Table 1, night time crashes by year and by “light condition” factor for Chicago city limits
| Year | Pedestrian fatalities | Percentage that occurred on “darkness” or “darkness, lighted road” | Percentage that occurred on “darkness, lighted road” only |
|---|---|---|---|
| 2005 | 65 | 58.46% | 52.31% |
| 2006 | 48 | 50.00% | 39.58% |
| 2007 | 45 | 42.22% | 33.33% |
| 2008 | 55 | 49.09% | 41.82% |
| 2009 | 33 | 60.61% | 60.61% |
| 2010 | 32 | 53.13% | 43.75% |
| 2011 | 36 | 47.22% | 41.67% |
Table 2, night time crashes by year and by “light condition” factor for Illinois excluding Chicago
| Year | Pedestrian fatalities | Percentage that occurred on “darkness” or “darkness, lighted road” | Percentage that occurred on “darkness, lighted road” only |
|---|---|---|---|
| 2010 | 81 | 74.07% | 29.63% |
| 2011 | 99 | 73.74% | 27.27% |
Table 3, night time crashes by year and by “light condition” factor for Illinois excluding Chicago
| Year | Pedestrian fatalities | Percentage that occurred on “darkness” or “darkness, lighted road” | Percentage that occurred on “darkness, lighted road” only |
|---|---|---|---|
| 2010 | 113 | 68.14% | 33.63% |
| 2011 | 135 | 66.67% | 31.11% |
The NHTSA email took the same wrongly-directed tone as the City of Chicago’s pedestrian safety campaign (which continues today). It’s the tone that places the responsibility of one’s safety on one’s self, instead of on those who can (and do) inflict the greatest harm. The email’s prominent headline, in large, red letters, was “Watch Where You Walk: Pedestrian Deaths Up in 2010”. The city’s latest contribution is a yellow decal pasted on the sidewalk.
The first ever Chicago pedestrian plan doesn’t address nighttime crashes involving pedestrians, but it does focus on dark underpasses (I don’t know of a link between dark underpasses and car-pedestrian crashes as the two are usually separated by concrete columns).
The NHTSA report (.pdf) has other information, like ranking all of the states in the number of pedestrian fatalities per 100,000 residents. In 2010, Illinois had 0.90 pedestrian fatalities per 100,000 residents, while Florida, leading the pack, had 2.58 pedestrian fatalities per 100,000 residents.
Do you have an idea on why pedestrian fatalities is up from 2010 to 2011, but the share of those that happened at night are down?
Updated September 21 to add Tables 2 and 3.
Is the sample size really big enough to draw any conclusions?
I’m not asking anyone to make any conclusions.
Another question I could ask is, why is there such a large difference between the number of nationwide fatalities that occur at night, and the number of Chicago fatalities that occur at night?
Is there data for other large cities? I would guess that rural areas without street lights have more nighttime accidents and skew the average.
I can think of two reasons. The first is, as Jared suggests and I’m most likely to buy, sample size. When you’re talking 36 deaths in 2011, the difference between 47.22% and 70% is 7 people. That’s a very small number. Over the six years of data we have, we see that the average is 51.53% of the 314 deaths. The difference between that number and the 70% national average is 58 people … closer to a good pool, but still small.
The second reason is simply the nature of when people are traveling. Saying we have a smaller percentage of fatalities at night implies we have a greater percentage of fatalities during the day. A large city like this will have a far greater percentage of trips made my pedestrians during the day … a far larger portion of us get to work by walking. Smaller communities where commutes are made almost entirely by car don’t do nearly as much walking in daylight, so we have greater exposure. We have a greater proportion of people available during the daytime to get hit.
I think I will run the analysis again, but exclude the fatalities in the City of Chicago (or exclude Cook County, or both). It will only be for years 2010 and 2011, as I don’t have Illinois data for the prior years (personal choice to keep my crash database easy to manage).
I added Table 2, the same chart but for all of Illinois excluding Chicago. It seems that whether or not a road is lighted makes a difference in predicting fatal crashes.
Ah ha! Well, it seems that if you remove Chicago, the Illinois average more closely resembles the national norm. And the second analysis strongly suggests the reason for the difference, which is that Chicago is much better lit. Good analysis.
It does make me curious about two things. One, does the light condition factor hold true in other states, and two, what happened in 2009, when it seems all fatalities occurred on lighted roads? That can’t be right. It seems anomalous enough to suggest some sort of data error.
Good catch on 2009 data. I double checked the query I used to get the data, then ran it again and checked it against my spreadsheet I used to calculate these figures and create the table. The results are the same. In 2009, in Chicago, ALL pedestrian fatalities at night happened on lighted roads. There were no fatalities on “darkness [, unlighted roads]”. It could be a data entry error by the responding police officer, a processing error at IDOT, or no error at all.
I’d guess the number of ped crashes that happen involving cabs in the loop skews it.