Crash test dummies have a gender — and safety — gap
Today’s female model called ‘unacceptable,’ is slightly smaller than average 12-year-old girl
WASHINGTON — Women get in fewer car accidents than men. But when they do, they’re up to 73 percent more likely to be injured and 28 percent more likely to die, according to new data from the Insurance Institute for Highway Safety.
One reason for that might be there is no crash test dummy that represents the average female body used in car safety testing, despite women making up more than half of all licensed U.S. drivers.
“It’s completely unacceptable,” said Jason Levine, executive director of the Center for Auto Safety. “What we’re seeing in the studies that keep coming out is that the crash protection is not equal, and it should be.”
Advocates say that needs to change. They are renewing calls for federal regulators to advance studies of crash impacts on women and other groups that are poorly represented in safety testing after several reports have shown women are significantly more likely to be hurt in crashes.
More data — and updates to federal regulations — are necessary to help build better crash test dummies and models to increase vehicle safety, experts say.
Carla Bailo, CEO of the Ann Arbor-based Center for Automotive Research, said the reason there is no average-size female dummy can be traced back to “bad history.”
“It’s got to do with the fact that women didn’t drive for a long time, and once they started, it was assumed incorrectly that it was just short trips to store” and similar errands, she said.
Safety experts have known for decades that female and male bodies are impacted differently by car crashes. But the National Highway Traffic Safety Administration didn’t start using a female style dummy until 2003. The dummies are put through simulated crashes to find out whether a vehicle meets federal vehicle safety standards and in determining the vehicle’s federal safety rating.
The dummy represents the 5th percentile of women in the 1970s, meaning 95 percent of women
were larger than it. It is 4-foot-11 and clocks in at 108 pounds — slightly smaller, by today’s standards, than the average 12-year-old girl. The average American woman, meanwhile, now is just under 5-foot-4 and weighs around 171 pounds, according to the Centers for Disease Control and Prevention.
The male dummy, representing the 50th percentile of men in the 1970s, is the same height as the average man of today but is around 15 percent lighter. The female dummy is 8 percent shorter and 45 percent lighter.
The dummy also is not built like female bodies. Rather, it’s a scaleddown version of its male counterpart, despite women and men having different spinal alignment, muscle strength, responses to trauma and more, according to Stanford University researchers.
And the agency’s safety rating tests don’t include the female dummy in the driver’s seat for frontal crashes — the type that results in the most fatalities. Female dummies are in the driver’s seat for some side-impact crash tests and for frontal crashes in compliance testing, according to NHTSA.
It’s not just women: Elderly people, heavier people and pregnant people are also more likely to suffer severe injuries, according to the research compiled by Stanford.
“There’s no excuse any longer for anyone saying they were not aware that this gap existed,” Levine said. “It is a very bizarre argument that the 5th percentile female is good enough to represent the other 95 percent of the gender.”
Federal response
In response to a request for an interview with NHTSA officials about the crash test dummy program, agency spokeswoman Lucia Sanchez said via email that they use dummies “to ensure that vehicle manufacturers design and produce vehicles for crash protection not only for a wide range of occupant sizes, but also for targeted occupants at risk when involved in a crash.”
“Real-world data suggested that the smaller females (not 50th percentile) were at greater risk of incurring injuries and more likely to be seated in the right front passenger position in frontal crashes,” Sanchez wrote.
The agency didn’t respond to multiple requests for clarification about why there is not an averagesize female dummy, whether the agency believes there isn’t a need for one or why there are no plans to create one, but did note it helps fund research for computer models that may better represent all types of bodies.
A 2019 study from the University of Virginia showing significant injury disparities between men and women prompted U.S. Reps. Kathy Cantor, D-fla., and Jan Schakowsky, D-ill., to reach out to NHTSA and ask what they’re doing to address it.
In a response letter provided to the Detroit News by Cantor’s office, Acting Administrator James Owens said previous NHTSA studies had shown that “differences in crash injury and fatality rates between men and women were not statistically significant.”
Owens wrote the agency will continue evaluating crash data and developing “state-of-the-art tools” to evaluate injury causes in different body types.
His answer didn’t cut it, Cantor told the Detroit News: “We really need some answers. They just kind of glossed over that.”
A path forward
However, physical differences might not account for all of the disparities between men and women in car crashes, according to a study released last month by the Insurance Institute for Highway Safety, a research group supported by insurance companies.
Women usually drive smaller cars than men, the researchers found, making them more vulnerable to death or injury in crashes. While both genders crashed in minivans and SUVS equally, around 70 percent of women crashed in cars compared with 60 percent of men. More than 20 percent of men crashed in pickup trucks, while less than 5 percent of women did.
Women are also more likely to be in the vehicle that’s struck (rather than striking) in a front-to-rear crash or a side-impact crash.
But the types of injuries women suffered — disproportionately more leg injuries — indicate that regulators and researchers should invest in learning more about how women’s bodies react to crashes, said Jessica Jermakian, vice president for vehicle research at the IIHS and one of the authors of the study.
“We need to understand why there is a difference in injury risk before we can identify what the solution is,” Jermakian said.
“An important aspect to understand is that crash test dummies are both sophisticated and crude at the same time,” she added. They are tools to measure forces on passengers, “but they’re not so sophisticated that we can differentiate between differences in bones and ligaments and muscles of men and women.”
That’s something Matt Reed, head of the biosciences group at the University of Michigan Transportation Research Institute, and his colleagues are working to remedy.
Lack of data
There is a dearth of data on female bodies — and all kinds of bodies except the typical man’s, Reed said. Creating a female crash test dummy would be “enormously expensive” and take 15 to 20 years.
“The question is not whether it would be appropriate to improve protection for women. Absolutely it should be a priority to do that,” Reed said. “The question is what’s the best way to go about that?”
Reed and other researchers have developed computational human body models that can simulate a wide variety of bodies and their reactions to crashes without creating a physical model.
Most major automakers use a version of these models in developing their vehicles, Reed and Bailo of the Center for Automotive Research said. Several major companies, including the Detroit automakers, are a part of a consortium that funds research on human body models.
However, even these models aren’t informed by a fully representative body of data, Reed said: “The fundamental problem is they only represent a small range of people.”
There’s a “vicious cycle” in which automakers, researchers and regulators get the most value out of continuing to improve upon mid-sized male models rather than starting from scratch with other kinds of models, he said, adding more research on women and other types of people is necessary to make these more representative.
Reed’s group at the University of Michigan is using a different method that’s more time intensive but creates models that can more easily be morphed to represent a wide variety of people, he said.