Motorists are inundated with road signs. These may include speed limit signs, construction warnings, directions to important landmarks, and safety warnings (eg “Icy road ahead”).
In a interesting article recently published in Scienceresearchers Hall and Madsen argue that some road safety warning signs can backfire and in fact increase — rather than reducing — the risk of fatal accidents.
As part of a road safety campaign, the State of Texas (like many states) regularly posts “dynamic message boards” showing the number of recent traffic fatalities – for example, “1,669 DEATHS THIS YEAR ON THE HIGHWAYS OF TEXAS”. However, Texas displays these signs one week per month, providing a natural experiment for how well these signs have reduced accidents.
To their surprise, Hall and Madsen found that the signs did not reduce the number of accidents, quite the contrary. Over a period of several years, on the weeks the signs were posted, car crashes increased by 4.5% on stretches of road 10 km (6.2 miles) after the signs. This translated to 2,600 additional crashes and 16 additional fatalities each year in Texas, with an estimated annual socioeconomic cost of $377 million.
The researchers also performed extensive analysis to control for other confounding factors. For example, they checked death rates for similar stretches of road ahead of the signs, as well as data for comparable weeks before the start of the public safety campaign. They also controlled weather, holidays and other factors.
Hall and Madsen argue that these signs caused more crashes due to a combination of increased cognitive overload on drivers and high “saliency” (attention effect) of an extremely negative message.
I liked their article because they backed up their reasoning with some interesting data analysis. For example, a numerically higher number of fatalities on the sign (i.e. potentially more eye-catching) was correlated with a higher accident rate. Additionally, the signs apparently caused more crashes on more complex stretches of road, consistent with a cognitive overload hypothesis.
Additionally, the signs increased the number of multiple-vehicle crashes, but not single-vehicle crashes. (Single-vehicle crashes are more likely to be due to large errors such as running off the road, while multi-vehicle crashes are more likely to be due to multiple small errors such as drifting out of the lane, usually observed with distracted driving.)
Hall and Madsen also explored seven other hypotheses, such as the possibility that display weeks are inherently more dangerous, that any message increases crashes etc. and found that none of them were consistent with the data.
I find it entirely plausible that an outright negative message (e.g. “A lot of people have already died doing what you do right now! “) could aggravate driving errors, especially on an already difficult stretch of road that requires a high level of concentration. There’s a reason coaches try to give athletes positive encouragement during pressure situations, rather than negative encouragement. If I had to make a pair of crucial free throws to win a basketball championship game, I’d much rather hear my coach say, “You get it, kid!” rather than “Don’t choke and blow up the game, you loser!”
The Texas Department of Transportation clearly had good intentions with its campaign, with an explicit goal of reminding motorists “that driving deserves their full attention every time they get behind the wheel.” Unfortunately, this particular effort had the unintended consequence of diverting drivers’ attention, sometimes fatally.
The effect is also not limited to Texas. The researchers note that “most damage is done within the first few days the message is posted…and in places where death messages are posted 1 day a week, such as Colorado, the effect could be worse. “. I might be a little more careful driving in Denver on those days.
Hall and Madsen draw several conclusions from their work, and I encourage others to read the full paper.
One of the most important lessons applies to policy makers who love psychological interventions and “nudges” to shape the behavior of ordinary people: “[M]measuring the effect of an intervention is important, even for simple interventions, because good intentions do not necessarily imply good results.” [Emphasis mine.] This is an important principle that all government decision makers could best take to heart.