Research review
Does driver training actually make you safer?
It feels obvious that more training means fewer crashes. The evidence is more uncomfortable — and far more interesting — than that. Here is what fifty years of research really shows.
Section 1
The uncomfortable question
An entire industry was built on a single, intuitive assumption. For decades, almost nobody checked whether it was true.
For most of the last century the logic seemed airtight: teach people to drive, they become more skilful, and skilful drivers crash less. That belief built the professional driving-school industry, and by 1960 many US states required teenagers to complete a course before licensing. Insurance companies even offered discounts on the strength of it.1
But when researchers finally looked closely, the early "proof" fell apart. Enrolment was almost always voluntary, and the students who chose to train were already different — better grades, more stable backgrounds, lower risk. They would have crashed less anyway. Once those differences were accounted for, the apparent safety benefit largely vanished.1
This article does not argue that learning to drive is pointless — you obviously cannot drive safely without learning to drive. It asks a sharper question: does formal training reduce your risk of crashing, beyond simply teaching you to operate the car? The honest answer has three parts, and the third is the most useful.
Section 2
The gold-standard trial
Only one study settled the selection-bias problem properly — by assigning training at random. Its result surprised everyone.
The DeKalb County study (Georgia, USA) remains the most rigorous driver-training evaluation ever conducted. Around 16,000 teenagers were randomly assigned to one of three groups: a "state-of-the-art" safe-performance curriculum (including hazard perception), a stripped-back course covering only the licence-test basics, or no formal training at all. Random assignment removes the self-selection problem that had wrecked every earlier study.1
Across all the randomly assigned teenagers, neither trained group crashed significantly less than the untrained control group. Among those who actually became licensed, the trained group did show 13.1% fewer crashes in their first six months — but the effect was only borderline (p≈.07) and faded after that, falling to a non-significant 8.8% over two years.1
Then came the twist. Training got teenagers onto the road sooner: 69% of trained teens were licensed within six months versus 59% of the control group, accumulating on average about 23 extra days of driving over two years. More young drivers, driving earlier, means more exposure — and once that extra exposure is counted, the small per-driver benefit is offset by a net increase in crashes across the population.1
"Conventional driver training programs… probably reduce per licensed driver crash rates by as little as 5% over the first 6–12 months of driving. The possibility of an effect closer to 0 cannot be dismissed."— Peck (2011), reviewing the DeKalb trial, IATSS Research
One more sobering detail explains why this debate stayed unresolved for so long: crashes are statistically rare events, so you would need roughly 35,000 drivers in a two-group study just to reliably detect a 10% reduction. Most studies were far too small to find an effect even if one existed. And there is no evidence that simply adding more on-road hours improves the outcome.1
Section 3
When training backfires
Some of the most popular "advanced" courses didn't just fail to help. The evidence says they made young drivers more likely to crash.
Skid-control and car-control courses feel like the ultimate safety investment — learn to handle a slide, brake at the limit, control the vehicle when it goes wrong. Yet decades of evaluation reached an "overwhelming" consensus that these skills-mastery programmes were a detriment to safety, especially for young male drivers.2
Norway — skid training
A before-and-after study found participants had more crashes on slippery roads afterwards. The researcher concluded the training taught drivers to cope with a skid rather than to avoid one — so they stopped trying to avoid them (Glad).2
Scandinavia & Switzerland
Slippery-road training across Denmark, Finland, Norway and Sweden "generally failed"; safe-driving courses in Switzerland showed a zero effect on crashes. The recurring finding: manoeuvring exercises raise self-confidence and lead to higher speeds (Katila et al.; Siegrist & Ramseier).2
Why it backfires: car-control training reliably raises a driver's self-perceived skill — but not necessarily their real skill. The gap between the two is dangerous. A driver who believes they can handle anything leaves smaller safety margins, drives faster, and takes more risks. In one study, drivers trained in skills rated their own ability far higher than insight-trained drivers, even though their actual measured skill was no different.2
Section 4
Skill is not safety
If raw vehicle-handling skill kept people safe, the best-trained drivers on earth would have the cleanest records. They don't.
Consider the cleanest natural experiment imaginable. Researchers compared the road records of 3,000 members of a sports-car racing club — people with genuinely elite vehicle-control skills — against ordinary drivers matched for age and background. The racers had significantly more on-road crashes and violations, not fewer.2
"Teaching drivers to improve skills such as high-speed braking and cornering serves only to increase driving confidence to the point of decreasing road safety."— Washington, Cole & Herbel (2011), IATSS Research
This is the single most important idea in the whole field. Crashes among young drivers are not mainly caused by a lack of car-handling ability. They are caused by attitude, risk-taking, distraction, fatigue, peer pressure, alcohol and overconfidence — the very things a track day does nothing to fix, and can quietly make worse. A US national review put it plainly: the biggest barrier to effective driver education is the difficulty of changing "the attitudes, motivations, peer influences, and… decision-making skills" that actually shape how people drive.2
Section 5
What actually works
Two changes turned the picture around: a smarter licensing system, and a completely different philosophy of training.
1. Graduated licensing (GDL). Rather than trying to train risk out of teenagers, GDL manages their exposure — phasing in driving privileges, restricting the highest-risk situations (night-time, young passengers) and delaying full licensure. This directly fixed the "accelerated licensing" problem that had cancelled out training's benefits in DeKalb.1
2. "Insight-based" training. Around 1990, several European countries threw out skills-mastery and rebuilt their advanced training around a different goal: helping drivers understand their own limitations and anticipate risk — not master the car at its limit. The aim is a driver who leaves the course with a more realistic view of their ability, not an inflated one. On the track, instructors deliberately stage a loss of control sooner, slower and for longer than the student expects — puncturing overconfidence instead of feeding it.2
The modern model treats driving as a hierarchy: vehicle control sits at the bottom, but the factors that decide safety — self-assessment, anticipation, motives and lifestyle — sit at the top. Good training works on the top of that pyramid, not just the bottom.2
And the results from this "renewed" approach are genuinely encouraging:
| Country | Programme | Reported outcome |
|---|---|---|
| Austria | Multi-phase licence; advanced training made mandatory in 2003 (safe-driving course, group discussion, feedback drives) | Personal-injury crashes among 18–19-year-olds in the programme fell 11.2%, versus a 2.1% fall in the reference group of all other drivers (large samples, ~22,000 crashes in the comparison group).2 |
| Finland | Compulsory second-phase training since 1990 (taken 6–24 months after licensing, or the licence is suspended) | Built on higher-order skills and hazard anticipation rather than vehicle mastery; part of the cluster of programmes showing marginal-to-significant benefit.2 |
Honest caveat: these European evaluations are promising but imperfect — several lack proper control for exposure (licence-holders or distance driven) and the authors themselves call for further validation. They show a consistent positive signal, not final proof.2
Section 6
A more hopeful, recent signal
The most recent large study complicates the old pessimism — in driver education's favour.
A 2015 study tracked every teenage driver in the US state of Nebraska — 151,880 of them — over eight years. Crucially, the licensing system let researchers compare two clean groups: teens who took driver education, and teens who instead completed a supervised-driving log with an adult. They did one or the other, never both.3
The driver-education group had significantly fewer crashes, fewer injury or fatal crashes, fewer violations and fewer alcohol-related violations in both their first and second years of solo driving. That held after controlling for gender, ethnicity, household income, urban or rural home, and the age the licence was obtained.3
Why does this look better than the older studies? It compared driver education against a genuine alternative (supervised practice) rather than against nothing, in a modern graduated-licensing system. The honest limits still matter: it is an observational study, not a randomised trial, so unmeasured differences between the groups can't be fully ruled out, and it only covers the first two years. But as a large, recent, population-wide signal, it suggests well-placed driver education can make a real difference exactly when risk is highest.3
Section 7
What this means for you
The evidence doesn't say "don't train." It says train for the right things — and be honest about your own limits.
Pulling the threads together, the research points to a clear and practical conclusion:
Be wary of "skill" as a goal
Courses that just make you better at handling the car at the limit can leave you more confident than capable — the riskiest combination on the road. Confidence should follow real, calibrated ability, never run ahead of it.
Train the things that cause crashes
Hazard perception, anticipation, honest self-assessment, speed and following-distance discipline, and attitude do far more for your safety than cornering technique. This is the top of the pyramid — and it is learnable.
This is exactly why our lessons are built around coaching and awareness, not just passing the test or impressing you with car control. We work on reading the road, managing risk, and building a realistic picture of your own driving — the factors the evidence says actually keep you alive. It is also why we are firm believers in plenty of varied, supervised practice and a patient, graduated build-up of independence.
If you want to go deeper into the mechanics behind safe driving, pair this with our guides on safe following distance, the stopping-distance simulator, and the physics of braking and cornering.
Training that targets what really matters
Our coaching approach focuses on awareness, hazard perception and judgement — the skills the research links to fewer crashes, not just a quicker pass.
See how we teachSources
- Peck, R.C. (2011). Do driver training programs reduce crashes and traffic violations? — A critical examination of the literature. IATSS Research, 34(2), 63–71. (Open access.)
- Washington, S., Cole, R.J. & Herbel, S.B. (2011). European advanced driver training programs: Reasons for optimism. IATSS Research, 34(2), 72–79. (Open access.)
- Shell, D.F., Newman, I.M., Córdova-Cazar, A.L. & Heese, J.M. (2015). Driver education and teen crashes and traffic violations in the first two years of driving in a graduated licensing system. Accident Analysis & Prevention, 82, 45–52. (Open access.)
This article summarises peer-reviewed international research; the studies cited are from the USA and Europe, and findings are presented as international evidence rather than Irish-specific data. Figures and quotations were checked against the original papers. It is provided for education and does not guarantee any individual outcome.