Student driver
Credit: Smith Collection/Gado / Contributor

Abolish the Driving Test

For almost a century, U.S. drivers have performed a ritual that supposedly serves as the linchpin for road safety efforts. But we have no evidence it works, and a lot of evidence it doesn't.
Screen Shot 2021-02-24 at 3
Moveable explores the future of transportation, infrastructure, energy, and cities.

In the early 2000s—either 2002 or 2003, he can't precisely remember—Steve Bodzin was taking his driving test in San Francisco. He aced everything despite the city's notoriously tricky roads. He pulled the car back into the DMV parking lot, perpendicularly parked, and, perhaps feeling the confidence boost from a job well done, decided to straighten the car a bit so it fit perfectly in the spot. In the process, he almost ran over a whole family. 


"I hadn't even looked to see if there were people behind me," Bodzin recalled. "I felt pretty sick about the whole thing, and frankly still do."

Despite this, Bodzin somehow passed his driving test.

After living abroad for many years, Bodzin's license lapsed and he had to retake the test in Colorado. Complying with Colorado law, Bodzin came to a stop at a light with sufficient distance from the car in front of him such that he could see the rear wheels. But his examiner, who was a good deal shorter than him, couldn't see the wheels. Bozdin failed his driving test because the examiner said he didn't stop with enough space between them and the vehicle in front.

Last year, in the early days of the pandemic, Georgia governor Brian Kemp made national news when he decided to suspend driving tests, thus granting new licenses to people without having them take one. Many people thought this was deadly irresponsible. Surely, many parents and safety experts thought, this will result in irresponsible drivers causing mass carnage on U.S. roads. 

Rebecca Weast saw the news, too. She studies driver safety for the Insurance Institute for Highway Safety, with a particular focus on young drivers. But she had to admit to herself that she didn't know the impact this decision to suspend driving tests would have.


Weast decided to look at the existing research on the subject, only to find that there wasn't much. To her surprise, there were no good studies on whether driver's tests improved road safety. It seemed to her that, likely because of the expense and difficulties involved in doing a good study, no one had bothered. Still, this struck Weast as a glaring omission. 

The driver's license is the linchpin of U.S. road safety efforts. The underlying concept of the driver's test is that it weeds out dangerous drivers by granting licenses only to those who have demonstrated they can knowledgeably and safely operate a vehicle. Short of that, it at least requires everyone to learn the necessary skills in order to do so. 

Most people understand, at least on a subconscious level, the driving test doesn't really accomplish these things. Nearly all of us have taken one and have also driven like idiots at one time or another. And we see people on the roads almost every day driving as if they had forgotten everything they were ever taught; speeding in crowded pedestrian areas, controlling the wheel with their knees while eating, looking down at their phones, and so on. 

Nevertheless, as a society, we have internalized these assumptions about the merits of the driving test. For example, a 1995 National Highway Transportation Safety Administration study found that 86 percent of those surveyed considered driver education and the license test "very important" to ensure drivers behave safely. Only two percent thought they were unimportant.


Yet, there is no evidence driver's tests accomplish anything at all.

Like Weast, I became interested in the question of whether road tests accomplish anything when I wrote about Georgia's suspension of license tests more than a year ago. Since then, despite countless hours of research and interviews with automotive historians and safety experts, I've come across precious little evidence that driver's tests are good for anything except propping up the driver education industry. It is time to declare the century-long experiment with driving tests a failure.

Fixing the way we think about driving tests, and abolishing them altogether, is important for more than just having fewer people die on U.S. roads. It is emblematic of the larger American struggle to make our institutions fairer. The implication of earning a driver's license is that the license can be suspended or revoked for driving like a maniac. And, indeed, they can be, including for dangerous behavior like drunk driving. But such cases are the exception, not the rule. One study looking at New Jersey licenses found that in 2018, 5.5 percent of all licenses were suspended, but a whopping 91 percent of those suspended licenses were for non-driving-related reasons such as failure to pay fines. By and large, licenses are suspended as a punishment not for driving poorly, but for being poor. It is an extension of our national policy of criminalizing poverty and using traffic enforcement as an excuse to extract fines to pay for a bloated criminal justice system financed through those very fines. And by having a suspended license, it is harder for that person to get and hold a job, a necessary prerequisite to paying the very fines that resulted in the license suspension in the first place. 


It is time to drop the pretense that driving tests and the associated bureaucracies that administer them have societal value, and well past time to imagine a better way to grant—and revoke—permission to drive vehicles on government-funded roads that accord with the 21st Century. 

Or, as Bodzin put it after recounting the test he passed despite almost running people over and the test he failed because his examiner was short, "drivers' tests are a disaster."

I failed my first two. I understand why I failed the first one. Legit have no clue what I did wrong the second time, to this day. -Ryan O'Hanlon, Brentwood, Long Island DMV. Licensed driver since 2005

It is very hard to imagine a world without driving tests. In the year or so I've been working on this story, the most common reaction I get when I tell people about it is, "So, what, anyone can drive?" 

As if that's not the case already.

We like to believe the licensing process stops dangerous drivers from taking the road. If they drive badly, they don't get a license. If they don't have a license, they can't drive. But that simply isn't true.

In fact, the driving test is one of the lowest-stakes tests you will ever take in your life for the very simple reason that you can retake it as often as you like until you pass. California has some of the strictest rules here, limiting people to three retakes. But even if a person fails all three, it just means they have to start the process and pay the fees again. In any state, a person can also re-take it anywhere they like. They can venue-shop if they didn't like the first DMV they took it at. They can go to the suburbs or the exurbs or a rural area. 


Odds are a friend or family member will tip them off to a lenient DMV that is easy to pass, because some DMVs are tougher than others. And, like any test that is judged by arbitrary metrics, some examiners are easier than others, evaluating differently depending on the kind of day they're having or how they feel about the driver personally. The outcome is, in a word, arbitrary.

The outcome also varies wildly between states, in part because the tests vary wildly between states, even though passing in one state means a person can drive anywhere in the country and, if renewed regularly, applies for the rest of their lives. Each state sets their own licensing requirements and testing standards. A 2011 paper by the National Highway Traffic Safety Administration (NHTSA) provides a comprehensive overview of the rules by state, which doesn't even mention the fact some states "on-road" testing component takes place in the DMV parking lot with some road cones. These may have changed slightly in the decade since but generally states do not change licensing requirements often, itself a testament to how disconnected from reality the driving test is given how much cars have changed over the decades.

It is obvious states want you to have a driver's license by making it quick and easy to retake the test if you fail. Nearly half of states essentially have no waiting period, allowing failed testers to retake the next day if they can get an appointment. A third of states require a week's break, and only three (Oregon, Arkansas, and in some cases Tennessee) make people wait a month. Nearly all driving tests are scheduled to take a half hour or less. Some, like in Texas, Ohio, Pennsylvania, and North Carolina take just 15 minutes.  


That same study also examined the failure rates by state, although not all state DMVs provided data. If you want to pass your driving test, take it in Idaho, which has a failure rate of just four percent. Utah (33 percent), Minnesota (33 percent), Connecticut (35 percent), New York (38.8 percent), and Maine (40 percent) had the highest failure rates. But that doesn't mean one out of every three people fail the test in these states since one person can fail the test multiple times. Nothing is stopping them from taking the test over and over again until they pass. 

In and of itself, this isn't necessarily a bad thing. In theory, drivers might learn safer behavior between tests. But the history of driving tests and how they came to be suggests that is rarely if ever the case. 

I learned to drive as an adult woman (age 29), not a teen, and took a driving test in the Bronx…I failed on parallel parking (apparently if your wheel hits the curb it's an instant fail…) The guy told me I 'wasn't safe' as a driver and was unbelievably condescending…I went to schedule another test and this time got one in New Rochelle about three weeks later. The suburbs were a whole new world…passed easily." -Torie A., licensed driver since 2014

To be sure, the concept of a driving test makes intuitive sense. So it was perfectly reasonable that, as the dangers of widespread vehicle use became painfully clear in the 1920s, many cities and states moved to implement some form of driver testing.


Then as now, the tests themselves varied across the country. But two key aspects of those early tests still serve as the basis for the logic of driving tests today.

The first aspect is the technical and engineering knowledge required to operate a vehicle. In the early days of the automobile, starting and driving a car was a true skill that required instruction and practice. For example, here is a nine minute video demonstrating the basics of how to operate a 1916 Model T. Twice during the demonstration of how to start the car, the host Todd Schelling warns not to put your arm or thumb in a certain position or else you might break a bone. And stopping the car was not as simple as pressing the brake pedal. There wasn't one. It instead involved pressing a combination of the three pedals and, perhaps, engaging the handbrake lever.

Modern cars are still very complicated machines; way more complicated, in fact, than the Model T in that video. But most of that complexity is in service of making driving as easy as possible. Driving a car is, mechanically speaking, one of the simplest things most people do on a daily basis. It is hardly more complex than running the dishwasher or washing machine, though it does require more constant attention. 

A driving test designed to prove someone is physically and technically capable of driving a car made sense in 1920 and even 1940 and 1960, but it makes little sense today. Or, as University of Chicago public health sciences professor Kavi Bhalla told Motherboard, "I’m certain I can teach my 4-year old to drive my Prius!" If that sounds unlikely, consider the hundreds of battery-powered miniature car children's toys available for purchase.


Mine was 15 years ago in Old Saybrook, CT, but I still remember thinking, "boy, that was easy," after I took it. I remember the route, and just plotted it on Google Maps. It was 5 miles, and took about 10 minutes, with short sections on Route 1, a busy commercial route, and I-95. Mostly right turns, and just a few stoplights. -Christopher Kennedy

Beyond proving people knew how to drive, the test was also designed to weed out dangerous drivers, a purpose we still ascribe to the driving test today. 

According to automotive historian Lee Vinsel and author of one of the only books about the history of automotive regulations, this was based on a burgeoning field of social science that believed people could be categorized by psychological tests. The central concept for driving tests was that certain people were accident prone, and by preventing them from getting a driver's license, roads would be safer.

During the 1920s, the National Research Council pumped money into programs designed to identify such accident-prone people. But after psychologists engaged in this research started engaging in blatantly profiteering behavior like patenting psychological testing instruments and taking lucrative consulting contracts with insurance companies, the NRC began to suspect this was all an elaborate scam. 


In the mid-1930s, the NRC hired a psychologist named Percy Cobb to conduct a study that would fact-check all this accident proneness stuff by examining Connecticut's driving records. As Vinsel wrote, "Using those records as a statistical database, Cobb found that the standardized driving tests being pushed by some psychologists correlated with accident records only 35 percent of the time. He showed these driving tests to be worthless for identifying accident-prone drivers, if such characters even existed."

By the 1940s, Vinsel found, the concept of accident proneness was thoroughly discredited. The standard driver's license test lost one of its main justifications for existence.

But, as humans so often do when a reason they believe in something is disproven, people simply invented new reasons to believe it. Instead of weeding out accident-prone drivers, driving tests became a method of evaluating whether a driver had been sufficiently trained, something that could be improved and refined with instruction. While this may sound like a subtle shift, it meant the driving test was no longer a means to keep people from ever being licensed due to their personality type, but merely a way to delay licensure until they had their instincts corrected through training.

An illustrative example of this transition comes via a February 23, 1941 Washington Evening Star article titled "How to Pass A Driving Test." The article begins with a father of a young driver lamenting that his son, who is an expert on automobiles, somehow failed his driving test. How can this be, the father wonders? 


So the writer sought out the examiner, who reported the kid drove like a total asshole. The examiner goes on to explain this was all too common. People thought that because they knew how to drive that they could drive however they wanted:

“In other words,” I said, "they’re all right mechanically but below par mentally?”

“Not mentally,” corrected the examiner, “psychologically — or perhaps it is sociologically; I don't know. They’ve just got to master their own behavior as well as they’ve mastered their car's, and it seems to me that this matter of attitude is the key to the whole traffic problem. It goes back to instruction, naturally. And I don’t mind saying that if young Mason had learned to drive in one of the high-school courses sponsored by the A. A. A. and other organizations, he’d have his license now."

This attitude is a perfect example of how safe driving transitioned in the minds of Americans from being a psychological trait to a matter of training and education. The examiner concludes by relaying a message to young Mason who flunked the test. "You tell him that I know he knows how to drive. But when he comes back and shows me that he knows how not to drive, I'll pass him."

"I made some goofs on my test but still managed to pass…I hit the curb while parallel parking and made a left turn onto Sunrise Highway after the other left-turning car instead of before." -Jason Gers. Took the test in Freeport, LI. Licensed driver since the mid-1990s


Showing an instructor you know how to drive—or how not to drive—is very different from driving that way all the time. This is why, in the second half of the 20th Century, so much emphasis got placed on the driving test's close cousin, driver education, just as the anonymous examiner in the Star article wanted.

Even though driver education is different from the license test itself, it's still worth exploring what we've learned about driver education over the decades, and not just because it has been studied more than the test itself. Because driver education takes longer, involves more instruction, and is all about passing the driving test, we can reason that if driver's education doesn't make drivers safer, then it is probable the driving test doesn't either.

So it is especially noteworthy that decades of studies have failed to find much evidence that driver education works at all. 

One review of several studies found "there is little or no compelling evidence" that driver training among teen and novice drivers has any positive effects on road safety. Another literature review funded by the pharmaceutical company AstraZeneca—which wanted to know if it should keep paying for employees to receive driver training—also concluded "the effectiveness of current driver education programmes is as yet unproven."


But the most famous study about driver education was the Dekalb County, Georgia experiment. In this landmark study from the early 1980s, researchers randomly assigned approximately 16,000 teens to two types of driver education and a control group that received no education over a period of five years. The initial study found training worked, in that it significantly reduced the rate at which the teen drivers were ticketed. 

But a subsequent study that analyzed the same data in a different way found the opposite, that in fact the trained group had eight percent more crashes and 11 percent more road violations. These differences in findings come down to complex statistical analysis questions, but researchers have generally found the second study more convincing because it adjusted for the fact that trained drivers were able to get their licenses sooner. That being said, the study had other flaws—for example, about a third of students assigned to training didn't complete it and therefore weren't counted, leaving the students who completed the training more likely to be responsible individuals more generally—but it is still one of the best we have on the subject. 

There are many other studies that have looked at driver education, but the upshot is researchers are unconvinced at best and skeptical at worst about whether it actually does anything. The fact that it is so hard to tell probably indicates driver education doesn't matter much one way or the other. If nothing else, that finding would be consistent with other teen education efforts in areas like drunk driving and drug use, which have also found widespread education efforts to be ineffective. After decades of careful study, it turns out teens don't listen to a bunch of adults telling them what to do.


Still, these are all studies about driver education which is related to but different from the driving test itself. During my research, I feared the scientific literature simply had nothing relevant to say about whether driving tests matter. 

But then, I stumbled across a recent study from California.

In 2015, California changed its laws so undocumented immigrants became eligible for drivers licenses under a law called AB 60. Prior to the bill becoming law, many arguments were made both in favor and against the bill on the grounds that it would either make roads safer or more dangerous. Nobody really knew for sure. 

The law was a big success. More than 600,000 licenses were issued under AB 60 in the first year alone. And in a very clever study published in 2017, researchers looked at crash rates in California both before and after AB 60, including at the county level, to see what impact this had on road safety. 

This study is so important because it removed one of the key variables when studying teen drivers, where it is impossible to decouple their lack of driving experience with the impact of whatever they learn in driving schools. We know experience matters a lot when learning how to drive. Weast said experience is important, but so too is a diversity of experience, meaning driving on all types of roads in various conditions, which driving schools often don't do, and that experience occurs at the same time as kids are learning how to drive. So those studies cannot separate the benefits of practice driving with parents from the effect of driving school classes and formal instruction. 

But this study could. It used vehicle registration rates to empirically prove nearly all of the AB 60 licenses were given to drivers who were already driving before they got their licenses. After all, they lived in California, how else were they supposed to get around?

In this sense, the study had something extremely valuable to say about the license test. If crash rates increased, that would suggest the AB 60 population drove more recklessly because they were no longer as afraid of being pulled over. If crash rates decreased, it would suggest the licensing process somehow makes drivers safer. If crash rates stayed the same, it would suggest the licensing process doesn't do much of anything, and that people drive because they have to whether it's legal or not and, in the process, drive however they want.

The researchers found crash rates stayed the same. AB 60's effect on crash rates per capita were "statistically indistinguishable from zero." The authors further concluded "obtaining a driver's license did not change their routine driving behavior," the exact opposite of what most people think the driver test is supposed to do.

It was a joke. All I remember having to do is drive around the block and park in a parking lot. The DMV employee was talking on her cell phone and not paying any attention. -Rebecca Hahn, New Orleans. Licensed driver since 2002

Still, recognizing the driving test is a waste of time, money, and energy is one thing. Coming up with a sensible replacement that balances safety concerns with the economic reality that pretty much everyone needs to drive to be a functioning adult in the U.S. is another.

It is very hard to reconcile these conflicting goals. In the U.S., to withhold a driver's license from someone—or, for that matter, revoke it—makes it difficult for them to be a functioning member of society. I personally do not believe failing to pay a fine or even texting while driving should condemn someone to unemployability because they cannot reliably get to and from their jobs. But, on the other hand, what's the point of a driving test if everyone needs a driver's license? The U.S.'s auto dependency is the strongest barrier to a better licensing process, which itself is the strongest barrier to a comprehensive road safety program. 

Often when reporting this story, I'd hear an anecdote or see a study about the driving test in other countries. Some, like the UK and Japan, are much harder. Others, like in Russia or much of the global south, are some mixture of perfunctory or outright corrupt. People often like to assume these differences are terribly meaningful. Obviously, they say, driving in Japan is safer because the driving test is so hard! 

There might be some truth to that, but it's hard to know if that's because of the driving test or if the driving tests reflect cultural attitudes towards driving that manifest in other ways. In other words, countries that take road safety seriously, from road design to investing heavily in public transit and so on, might also be more likely to have tough driving tests. Or perhaps having all those viable alternatives to driving to begin with make those societies view driving more as a privilege than a right, which then informs their road safety policies. What is certain is it would be unfair and counterproductive to both have a strict licensing program rigorously enforced and no real alternative for getting around.

This leaves the U.S. somewhere between a rock and a hard place. Our farce of a driving license system reflects the choices our country has made to structure itself around the automobile. It's why getting a license is such a huge deal for teenagers, and why the legal driving age crept downwards until states introduced graduated drivers licenses in recent decades, one of the few license-related interventions that actually works. With graduated licenses, teen drivers are only allowed to drive at certain times or under certain conditions while they gain experience. Even after they pass their driving test, there are still limitations until they've been driving for a certain amount of time or turn 18. It's not a perfect system, but it is one that de-emphasizes an all-or-nothing approach to driving rights around a bullshit test in favor of scientifically-backed rules and restrictions that ease over time.

A better system is likely one that embraces the lessons of GDLs, jettisons the idea of a one-size-fits-all test to prove competency, and instead demands ongoing responsibility and care from drivers throughout their entire driving lives. Here, I envision a system that more readily suspends licenses for increasing durations for repeated dangerous traffic violations like speeding or running red lights. And fines, if they continue to exist, ought to be indexed to the driver's ability to pay like they are in much of Scandinavia.

But I am not naive about the prospects of that happening here. Namely, they are zero. Even in New York City, one of the few places in the U.S. people can live productive lives without a license, holding dangerous drivers accountable for their actions is a quixotic quest, whether they kill a pedestrian or cyclist or simply drive irresponsibly as a matter of course. For example, a program that would force people with 15 speed camera or five red light tickets in a 12-month period—a tiny fraction of the population considering the high bar for recklessness—merely to attend a driver safety training was barely passed, defunded during COVID, and is still a much-weakened program from what was initially proposed. Eight states have outright banned red light and speed cameras, which would be critical enforcement tools to have fair and equitable traffic enforcement. The U.S. is locked into an unhealthy system where driving is both one of the leading causes of injury and death and also an economic and social necessity. It is an irrational and unsustainable system, both economically and environmentally. So perhaps it is no surprise at all that the driving test is such a disaster. It is just one small part of a disastrous system, with no real prospects on the horizon for significant reform. It is, in that way, a familiar problem.