Driving along Texas highways, drivers will likely see electric signs that provide real-time traffic alerts, weather information or unique public safety announcements. While these signs are designed to increase public safety, new evidence suggests that one type does more harm than good.
A recent study from Joshua Madsen and Jonathan D. Hall outlines how dynamic message signs (DMSs) displaying the year-to-date number of fatalities actually distract drivers and cause more accidents.
“The intention of these messages is to hopefully reduce crashes and encourage safe driving, and our findings are showing that it’s backfiring,” said Madsen, a professor at the University of Minnesota. “We’re finding an immediate increase in crashes, very small, but its clearly going in the wrong direction.”
After analyzing data on 880 fatality signs and all crashes occurring in Texas from 2010 through 2017, Madsen and Hall found that the number of crashes increased as drivers got closer to the signs.
The number of crashes increased by 1.35% within 10 kilometers of the signs, and raised to 1.52% within five kilometers of the signs. Those numbers only increased in areas with complicated road segments.
“What’s the cost of a two-second distraction?” Madsen asked. “If I’m on a straight highway between Austin and Houston, there’s not many consequences to a two-second distraction. If I’m dealing with an interchange, there’s five lanes of traffic. I need to be switching lanes and getting out. That would be a much more complicated road segment and having a two-second distraction could certainly be more costly.”
The research suggests that the signs cause an additional 2,600 crashes and 16 fatalities every year in Texas, with an annual cost of $377 million. The study says the effects of displaying fatality messages are comparable to raising the speed limit by 3-5 miles per hour or reducing the number of highway troopers by 6-14 percent.
In a statement to the Chronicle, a TxDOT spokesperson said the “real issues around traffic fatalities in Texas are speed, distracted driving, impaired driving and people not wearing seat belts.”
“We appreciate any focus on safety and the critical need to inspire drivers to make the best decisions behind the wheel,” the statement reads.” In relation to this particular study, there are too many unknowns to draw any firm conclusions, to include assumptions made by the study authors regarding driver psychology and behavior.”
“We continually evaluate the effectiveness of our safety messages, and for quite some time now, we have not shared fatality numbers on the dynamic messaging signs (DMS). We look for every way to make our roads as safe as possible, and to use effective measures to remind drivers that most of the time they have the power in their hands to help prevent fatalities on our roadways.”
The study is here. I read through the abstract, and if I’m reading this correctly they are comparing collisions during the weeks that TxDOT is displaying these dynamic messages (called “campaign weeks”, which as the authors note has been one week per month since August 2012) to the weeks when it is not. The comparison is for the areas near the signs in the campaign weeks to the off weeks. The method seems reasonable to me, and the time span is long enough that there ought to be enough data to draw conclusions, but I don’t fully buy it. The large area in which the crash data was measured, which I presume is to allow for a sufficient number of crashes to measure, is broad enough that I don’t think you can assume enough of these drivers even saw the signs in question on the journey that included the crash.
I don’t want to speculate about what else might be in play here. The way they defined the data sets does a pretty good job of eliminating a lot of randomness, and the hypothesis that the signs can be distractive has merit. I just feel like this is a broad conclusion to make from inferential data. I can certainly believe that the signs don’t have any positive effect, and I can believe they could have a negative effect. I’d just like to see some more data before I’m convinced.