Image: Center for Neighborhood Technology, I-GO
Lies, Damned Lies, and Statistics
Do you want the good news or the bad news first? The good news: Hybrid registrations were up 38% in 2007, over 350,000 hybrids replacing internal combustion motors on the US roadways. The bad news: Headlines are touting a new study by the National Highway Traffic Safety Administration (NHTSA) finding that hybrids are hitting pedestrians and bicyclists more than traditional cars, especially when operating at low speeds, presumably with the engine in electric-only mode. Are hybrids a hazard? Or should we consider the opinion of Mark Twain, that statistics cannot be trusted?
For example, if hybrids spend more of their kilometers driving in metro areas, does that change the conclusion? What if hybrid drivers are more likely to report minor accidents than ICE drivers, being naturally more civic-minded? Conversely, what if hybrid owners are so smug about driving their eco-car that they feel entitled to ram the occasional walker or sever a bicyclist's leg? Let us examine the statistics that so easily make nameless and faceless the dark tragedy each case represents. We will start with kudos to the developers of the study: it carefully evaluates many confounding factors, such as the speed of vehicles at the time of the accident, the lighting and climate conditions, and maneuvers at the time of the crash (starting, slowing to stop, backing up, turning, etc). But the study nonetheless suffers from several significant flaws. Before headlines scare potential buyers away from the hybrid market, the validity of the data and the best path forward should be considered.
Weak Points of the NHTSA Study
First, the achilles' heel of many a study: the amount of data available is limited. The study uses data from only 12 states, because these are the only states which report the Vehicle Identification Number to the database, allowing identification of the make and model of car, and differentiation between hybrids and Internal Combustion Engine (ICE) vehicles. Furthermore, only incidents involving vehicles from model year 2000 or newer were used, to ensure comparability between the construction standards. While this tends not to restrict the data set on hybrids, it does limit the data available for study. The results is that there are only limited analyses which NHTSA claims are statistically significant: these are described in greater detail at the end of this article, for the hard-core reader.
Nonetheless, it is not quantity but quality of data that makes conclusions difficult. In accident studies, the golden standard is accidents per distance covered. That information would clearly answer the question of whether a higher percentage of metro driving relates to hybrid accident rates. But that data is extremely hard to come by: can you imagine having to file a driving report annually, like taxes, telling NHTSA how many miles you drove so they could have better accident statistics?
Sadly, the NHTSA left the Honda Insight out of the study, because early models did not have an electric-only mode at low speeds. It would be interesting to compare the accident rate of Honda Insights with those of the other hybrids, as this would control for the hypothesis that the noiselessness of the vehicle, rather than the hybrid driving style, accounts for the higher accident rate of hybrids. Probably, the small dataset available does not allow a statistically significant conclusion to be reached.
Other Statistics Worth Examining
Insight can be gained by looking at other sources of data. For example, Portland has the highest ownership of hybrids per capita, but ranks at the 9th safest city for pedestrians in America. (Sadly, this claim underpins a renewed commitment to safety in a press release after the tragic accident involving pedestrians Lindsay Leonard and Jessica Finlay. The vehicle involved was not a hybrid.)
Six blind pedestrians were killed by moving vehicles in 2007. None of those vehicles was a hybrid. But 6 is too small a sample from which to draw conclusions. Another study shows that hybrids are 74% closer before a pedestrian hears enough sound to localize the source.
We can also look forward to better "golden standard" accident rate data, accidents per distance driven, as companies acquire fleets of hybrid or electric cars. Larger companies tend to keep business accident statistics as part of their Occupational Health and Safety commitments. Perhaps groups that benefit from subsidies for such experiments can be encouraged to report thoroughly such data.
Abusing the Data
The data that will stir up the most emotions are not statistically significant. The most telling finding in the NHTSA report is the rate of accidents with peds during slow maneuvers like backing up, turning or starting in traffic. The rate of 5.3% for accidents involving pedestrian while backing up a hybrid casts a long shadow over the 2.9% of ICE crashes that occurred during the same maneuver. While going in a straight line, the pedestrian incident rate of hybrids and ICEs are almost identical (0.9% versus 0.8%).
The situation is completely turned around though for bicyclists: 0% of accidents backing up involved hitting a bicyclist, while twice as many incidents with bikes occured among hybrids driving straight, as opposed to ICES (0.6% versus 0.3%). The rate of accidents at dark or dawn/dusk is lower for hybrids (0.5% vs 0.6%), which is counter-intuitive if an assumption is made that higher accident rates are related to not hearing vehicles at times when they are difficult to see.
Also interesting is the case of entering or leaving a parking space or driveway: Hybrids hit cyclists in 3.6% of 83 crashes versus ICE drivers crashing into bikes in only 0.3% of 5870 reported accidents during that maneuver. Hybrids hit a pedestrian while entering or leaving a parking space or driveway in 1.2% of accidents studied, whereas ICE rate was 0.9%.
In all of the findings noted above, the sample size clearly does not support conclusive determinations. But since so many people can't tell a good number from a cave-man's drawing, you can expect to see crazy opinions proliferate in the coming days and months.
The Statistically Significant Findings
- Out of 8387 hybrid accidents, 77 or 0.9% involved hitting pedestrians. It can be said that is 50% more than the corresponding rate of ICE accidents with walkers: 3578 out of 559,703 or 0.6%.
- Hybrids in accidents when travelling under 35 mph hit walkers in 1.8% of cases while ICEs incidents with peds were only 1.2% of all crashes. (Above 35 mph, the results were not statistically significant. During maneuvers that imply low speed, the rates compare at 1.2% vs 0.6%.)
- While making a turn, the hybrid pedestrian accident rate of 1.8% is nearly double the ICE rate of 1.0%
- Hybrids hit cyclists at a rate of 0.6% versus 0.3% for ICE vehicles, overall.
- For crashes with bikes at under 35 mph, the stats are 1.0% versus 0.6%, with the data for faster accidents also too limited to be reliable (0.8% vs 0.5% if speed is judged by low-speed maneuver).
- Unfortunately, the number of accidents with bikes while making turns is not statistically significant. While going straight, Hybrids pegged a bike in 0.6% of crashes versus 0.3% for ICEs.
- Results for accidents on roadways (peds) or intersections (bikes), on clear days, during daylight are reliable, but since the sample sizes for other locations like parking lots, inclement weather conditions or low light situations are not statistically significant, little can be learned.
It is the findings for accidents under 35 mph that most support safety advocates campaigning for noise makers on electric cars and hybrid cars at slow electric-only speeds. Noisy electric cars are most likely an inevitability. Gotta start working on how to hack the hybrid noise maker so you sound cool when creeping past that SUV in the usual Monday morning traffic jam.
More on Hybrid Safety
Florida: Fairly Fatal for Peds and Cyclists
How Many Blind People Have Been Hit By A Prius?
Hybrid Taxis in New York City Get Challenged on Safety Concerns
Forums: Should hybrid cars be noisier to warn the blind?
5 Safest Hybrids - Save the Planet, and Yourself