About a week ago, I did something New Yorkers don’t do very often: drive a car. This gave me some time to think. To quote Bob Seger: 

your thoughts will soon be wandering
the way they always do
when you’re ridin’ sixteen hours
and there’s nothin’ much to do

What I started to ponder was this: what is the safest lane on the highway? There were plausible reasons to think any lane on a three-lane highway was safer than the others:

• The left lane: on one hand, drivers are moving the fastest in the left lane, which seems less safe. On the other hand, you could argue that the left lane is more likely to be empty.
• The center lane: seems like being hemmed in by cars on both sides would be much less safe. Yet, drivers in the middle lane also have more road to play with (e.g., they could safely swerve to the left if there was danger on the right.)
• The right lane: the right lane is the slowest lane of travel. However, it is also the lane most often used to merge or exit a highway, which can get hairy.

Without a clear answer, the only option was to turn to the data. Fortunately, the National Highway Traffic Safety Administration (NHTSA) maintains the Crashworthiness Data System (CDS) as part of its National Automotive Sampling System (NASS), which is “detailed data on a representative, random sample of thousands of minor, serious, and fatal crashes.” Unfortunately, from what I could tell, the NASS CDS doesn’t track lane of travel in any structured way.

Nevertheless, one piece of the NASS CDS—the “accident description”—contains a linear narrative of the events leading up to the crash. Further, a large number of these narratives mention specifics lanes. So, I wrote a script to mine the data in two simple steps. First, a narrative had to include one of “three-lane”, “three lane”, “3 lane”, or “3-lane” to be considered relevant. Second, the script scans for the first lane mentioned in a narrative using terms like “right lane,” “lane one,” “1st lane,” “first lane,” “center lane,” and “middle lane.” 

The technique appears to work fairly well. Here are two sample narratives it identified:

Vehicle 1 was heading eastbound on a three lane highway, in lane 1. Vehicle 2, a bus, was parked in lane 1 further east, stopped  after a previous collision. Vehicle 1 failed to reduce speed and struck Vehicle 2.

V1 (1980 Chevy Blazer) was heading South in lane three of a three-lane divided highway (no positive barrier—curb on both sides of grass median) when control was lost. V1 departed its travel lane to the left (East), jumped the median curb, travelled through the grass median and entered the Northbound lanes (three-lane highway, as well) after clearing the second median curb. 

Of the roughly 100,000 accident reports in the raw data, 8,069 mention a three-lane highway. Of those, 4,028 (50%) cite a specific lane: 900 left lane (22%), 1,513 middle lane (38%), and 1,615 right lane (40%):

If we modify the script to count total lane mentions instead of just the first lane mentioned, the results are even more dramatic. The right lane jumps to 10,899 mentions (46% of all mentions), while the middle lane (8,923, 37%) and left lane (4,089, 17%) drop.

So we have our answer: the right lane is the most dangerous lane on the highway, and the left lane is the safest. Granted, the methodology isn’t 100% bulletproof, but I consider it good enough to prove what we already knew—the Eagles had it wrong.

Life in the fast lane 
Surely make you lose your mind