You’ve seen the annual “Top Safety Pick” car ads and the crash-test videos, with cars slammed into objects as their own debris pours down around them like a snowstorm. Perhaps the latter even convinced you not to buy a car you so desperately wanted.
But that’s just a small part of determining which vehicles on the market are safe and which fall short—a long, messy process full of bulk vehicle purchases, bulk trips to the salvage yard, and automakers’ attempts to go back and fix what they didn’t get right the first time.
As you might remember from that crash-test video between a 2009 Chevrolet Malibu and a 1959 Bel Air that has 12.4 million views between the Insurance Institute for Highway Safety and Consumer Reports YouTube uploads alone, cars are vastly safer this century than last.
In fact, car safety is multitudes better now than it was in 2009, as a result of more evaluations and higher testing expectations.
In 2018, the U.S. National Highway Traffic Safety Administration released a study about the severities of car crashes in relation to the model years of vehicles, and found that the proportion of occupants in a vehicle who were fatally injured increased with the age of the vehicle. The full report is here.
Thanks in part to those advancements, traffic fatalities have gone way down in recent decades. In 2017 the IIHS recorded 37,133 traffic deaths, down from more than 50,000 in the late 1970s.
Now, traffic fatalities have been rising in recent years despite safer new vehicles. As noted in a 2018 NPR story, much of that has to do with more miles traveled, impaired driving, distractions and the failure to wear seat belts.
But crash-safety advancements in actual vehicles are especially easy to see in the improvements made to the strength of the passenger compartment in recent years. Whereas passenger vehicles used to crumple inward more easily, they’re now expected to be more reinforced up front—thus allowing less intrusion on the people inside—in order to get good ratings.
When the IIHS introduced a driver-side small front overlap test in 2012, a news release by the organization that August said only three of the 11 “midsize luxury and near-luxury cars” it evaluated got one of the top two of IIHS’ four ratings, “Good” or “Acceptable.”
That model year, the Audi A4 got the IIHS’ lowest rating of “Poor.” By 2017, it had the top rating of “Good,” and the differences in passenger-compartment intrusion between the two, which can be seen in the photos above, were huge.
But in the case of small-front-overlap protection, many automakers were reactive rather than proactive. When IIHS introduced the same test for the passenger side a few years ago, poor results were rampant once again even among the vehicles with good driver protection, and are steadily improving.
Obviously, crash test ratings are a big deal. Let’s look at how they happen.
There are two big independent organizations that crash test cars in America: the federal government, with its NHTSA arm, and the IIHS, a private research group set up for and funded by the insurance industry.
The IIHS was founded in 1959 by several insurance associations that represented 80 percent of the U.S. car-insurance market at the time, rebranding a decade later as an independent research organization. It opened its current vehicle research center in Virginia in 1992, where it still crashes cars today.
An IIHS representative said the data it gathers gives insurers insight into things like which cars crash more, which are more expensive to repair, and which features help prevent crashes, giving a guideline for costs and coverage. The IIHS thus built a reputation over the years with its research and crash evaluations, the representative said, and became a major source for crash data using that reputation and its “transparent relationship” with automakers.
Transparent relationship or not, the IIHS doesn’t get an endless parade of free cars to wreck from automakers. Once the organization finds a car that needs to be tested, the IIHS does what a lot of car buyers do: It goes to local dealers with some general criteria for the cars it wants.
“We like to avoid all black vehicles with black interiors because they don’t film well, or all white vehicles because they don’t stand out from the laboratory, which has a white floor,” David Zuby, chief research officer for IIHS, told Jalopnik.
There’s also always a set of specs that the IIHS needs to test in terms of options and technology, and it tries to find vehicles that fit the mold at its regular group of dealers. If not, it starts calling outside of the area.
The IIHS buys some cars out of its own budget, but not all. Zuby told Jalopnik that it occasionally changes its Top Safety Pick award criteria, but that it can’t test every car just for new award standards. If carmakers that aren’t already in line to be tested want in, the IIHS asks for reimbursement for the price of the vehicle.
Vehicles are weighed and measured when they get to the research center for the more destructive tests, and the IIHS uses those numbers to measure crash effects and keep variables constant. Sometimes equipment needs to go in vehicles for testing, Zuby said, and extra weight, like spare tires, has to get shaved off to bring it back down to what it was.
But before they get smashed, there’s some prep work.
“We drain the gasoline out and replace it with a mineral spirits that are less volatile, so we don’t have explosions or fires,” Zuby said. “We also remove the coolant and the engine oil, because often times the hoses and stuff that contain that will be broken or ripped during the crash test, and they just leak out onto the floor.”
Then, they get smashed—five vehicles, five times, if necessary. The organization currently does three frontal crash tests, a side test and a test for roof strength, and every crash test requires a fresh, clean, new car to destroy. Some tests aren’t destructive—like the ones for braking systems and headlights—and, if scheduled, cars do those tests before getting crashed.
“There’s a lot of broken glass, broken plastic [and] chunks of metal,” Zuby said.
The crash tests done by the IIHS and NHTSA differ slightly, in terms of type and test criteria. NHTSA does four compared to the IIHS’ five, and breakdowns of each are below.
In terms of the speeds cars are crashed at, which appear low, a spokesperson for NHTSA said impact speeds represent “the real-world crash environment to which passenger vehicles are exposed” and a “severity level where a significant number of vehicle occupant fatalities and serious injuries occur.”
NHTSA doesn’t test rear impacts because it “has a limited budget” and thus focuses on “front and side-impact crashes that are responsible for the highest percentage of deaths and serious injuries,” as it explains in its own FAQ.
Particularly newsworthy at the moment are headlights, one of the trickiest parts of safety tests at places like the IIHS. Crash tests are straightforward: If a vehicle structure caves, it caves—something that’s understood regardless of trim level. But a vehicle model could have the best headlight ratings in its top trims and the worst at the its bottom trims, and, even if legal for sale, the IIHS describes bottom-rated headlights as “inadequate.”
Automakers have the ability to make incredible HID, LED and even laser headlights these days, which make, for lack of a better phrase, a night-and-day difference from the base-model halogens most of us are probably accustomed to. But many automakers put top-rated LEDs on more expensive trims and bottom-rated halogens on lower ones, even if it is feasible to put highly rated headlights on cheaper new cars.
The 2019 Mazda 3, for example, starts at $21,000 and has headlights with IIHS’ second-highest rating on every trim level. It’s possible, but not the norm yet.
Dummies in the crashed cars have sensors, and IIHS sometimes puts cameras on board to see what happens on the inside. People then pull the dummies out after the test, remove the equipment and take photos to document what did or didn’t happen, like if an airbag didn’t deploy or if a door came open.
“Once the equipment is out, we usually keep the vehicle around for several weeks in case the automaker wants to come and see it—especially if there are any questions about the results that we got in comparison to [what] they got when developing the vehicle,” Zuby said. “Ultimately, we sell it off for salvage.
“When [vehicles] get close to being salvaged, we move them out into what we call the boneyard, which is the big, fenced-in parking lot which is where they sit and wait for the salvager to come and get them.”
For vehicles it crashes in house, the IIHS bases ratings on the results it gets from dummy sensors, measurements of how vehicle structure holds up, and video footage.
“One of the things we frequently downgrade a vehicle for is if the [seat-belted] dummy’s head slides off of the front airbag and into a gap between the curtain airbag,” Zuby said, reaffirming that looking at video footage helps researchers to see potential safety issues that measurements can’t show. “That may not show in the dummy measurement, but it’s the kind of thing where if it were a real person of a slightly different size, their head might hit something in the vehicle because it’s no longer in contact with the airbags.”
Zuby said the IIHS tries to set up ratings to be as objective as possible, even when it comes to looking at footage. Still, there are times when reviewers can’t quite land on a rating.
“When we have that situation, we usually involve additional people to try to come up with some consensus about ‘This vehicle in this test deserves a downgrade for this observation,’” Zuby said. “But that doesn’t happen too often, because we put some effort on the front end to make the things that we’re looking for objectively definable.”
The measurements go into a formula the IIHS sets up for every crash, letting it give ratings for how the car performs and how the dummies fare. Those ratings are Good, Acceptable, Marginal and Poor, and the scores a car gets in the different categories go into an overall score on the same scale.
But not every category has the same weight in the overall rating.
“The most important aspects are the head and the chest, because there are vital organs in there, and then the vehicle structure and then the extremities,” Zuby said. What’s interesting is that each new test is designed to not be impossible—at least, not as far as ratings go.
“We set up the rating boundaries in a way that when a test is new, there are at least some people who can earn the best rating,” Zuby said. “That becomes the target performance for everyone else. So, there aren’t sort of absolutes we’re looking for when we create a new test, but we’re just trying to bring the market up to the state of the ones that seem to be offering the best protection.”
From there, it’s less about making the standards for existing crash tests harder, and more about inventing new ones that address safety gaps elsewhere.
“A few years after we launch a new crash-test evaluation, we go into the crash reports from police and insurers,” Zuby said. “We say, ‘Let’s look at the vehicles that are doing well in our current batch of evaluations, what problems we still see, and [if we can] create a test that will get automakers to focus on those.’”
These days, safety is a key selling point for new cars, especially with regards to the latest crash avoidance systems. Every automaker wants to be on the cutting edge there.
When the IIHS rates a car poorly, then, an automaker will sometimes rush to fix the problem—even if it means putting an update out in the middle of a model year. But poor ratings on a new test are usually the catalyst for innovation that probably should’ve been made before the test came out, which can be seen in Zuby’s explanations about IIHS’ recent passenger-side overlap tests.
“[The fixing process for the driver’s side] was one of the ways that we found the need to do the passenger-side tests,” Zuby said. “We ran the small-overlap crash test on the driver’s side and within a few short months, some automakers came back to us and said, ‘Hey, we fixed car x, we put these new structures in or we strengthened these pieces, we’d like you to test it again.’”
But the carmakers themselves had only done what was necessary in the strictest sense: only beefing up the side of the car that the IIHS was testing.
“That’s when we started looking to see ‘Are they making the same changes or similar changes on the other side of the vehicle?’” Zuby said. “The answer was, not so much.”
When automakers go back to fix vehicles that have done poorly on a crash test, though, that doesn’t mean the cars with poor structures disappear. They’re still out there, meaning it’s important for buyers to know which is which.
“If an automaker comes to us in the middle of a model year and says, ‘Starting in June, cars rolling off the line have this improvement,’ we’ll often do that test and then we’ll publish the result,” Zuby said. “What you see on our website is: ‘For the 2018 model something or other, the current rating is Good, but that only applies to vehicles built after June of 2018. Prior models rated Marginal,’ or whatever the lower rating was.”
Surprisingly, not every car you see on the road is independently crash tested.
“When it comes to a sports car, there are sports cars in the U.S. that are not tested for a multitude of reasons,” a source who’s worked at major automakers told Jalopnik, asking not to be named due to their current job. “[A sports car] sells on the way it makes you feel. It’s not a car that necessarily sells on safety.”
Although not every model has to undergo crash testing for a public rating, there are legal standards to ensure we’re not all rolling around in the equivalent of life-sized soda cans. Those are set by NHTSA, which also handles recalls. But NHTSA mostly regulates individual elements of vehicles that are on sale through what are called Federal Motor Vehicle Safety Standards.
These standards govern everything from the basics, like rearview mirrors, door locks, seat belts and windshield mounting, to crash specifics, like side-impact protection, roof-crush resistance and general occupant protection.
The Karma Revero got recalled for side-impact protection in June, even, and the company had to issue a stop-sale on it until everything could be fixed.
A NHTSA spokesperson told Jalopnik that the agency doesn’t approve vehicles for sale in the U.S.—it’s a self-certify deal. Manufacturers certify compliance with federal standards with a label on the vehicle before it gets sold. The Office of the Federal Register says self-certification means automakers must confirm that if a model were tested under the procedures in the standards, it would “would meet or exceed” what’s expected.
NHTSA does crash test cars, and rates them on a scale of one to five stars. But models that get a one-star crash rating through NHTSA are still legal, as NHTSA said its ratings system is “designed to evaluate the occupant protection performance of vehicles beyond minimum federal requirements.” It has another branch, the Office of Vehicle Safety Compliance, which tests vehicles on a pass-fail basis in terms of complying with federal requirements.
The compliance office, NHTSA said, “takes into consideration vehicles with prior compliance test failures, consumer complaints, and special emphasis areas or areas of high public interest” when choosing which ones to test.
What all of this means is that there’s a huge gulf between what’s safe enough to legally go on sale, as far as the government is concerned, and what American consumers feel comfortable buying. It’s the difference between making it onto a showroom floor and making it into a driveway—depending on the vehicle’s target market and whether it’s rated at all, that is.
“If you have a four-star in a five-star market, that automatically puts you at a disadvantage,” the source said. If a vehicle gets a bad rating in a crash test, they said, it “almost immediately” starts a chain of fixes across the world.
“The quality engineers usually go back to the folks who were responsible for that deal, whether it involves the structural engineers anywhere from Dearborn, to China, to Japan, wherever,” they said. “They will go over that data with a microscope to figure out what exactly happened and how they can fix it.
“Sometimes, that’ll mean that a car will be refreshed a little quicker.”