tl;dr ... a minor objection tough ... it is not at all impossible for 70 percent of drivers to be better than average, you just need the lower 30 percent to (on average) be really bad
90% (actually, about 98%) of American drivers are above average, owing to a statistical quirk. There are 6 million car accidents in the U.S. each year, out of roughly 240 million vehicles. At a minimum, that means that 97.5% of drivers get in no accidents that year. They join the big bulge of people who are "average", having perfect driving records or only an accident a decade or so.
Accident frequency is a power law distribution - a large percentage of crashes are caused by a small number of drivers who habitually violate traffic laws, drive drunk, or otherwise engage in risky behavior. One of the distinctive features of power law distributions is that there's this long tail of people with very small values, and then a few people who make up most of the curve. So (made up numbers) you might have 60% of the population who has never gotten in an accident, then 35% who has gotten in one, then a tiny fraction of 1% who's been in a dozen. Over an 80 year lifetime, that 6 million accidents/year results in 480 million accidents, or 2/person, so with the hypothetical percentages above, 95% of people are above the mean and 60% are above the median.
Just goes to show that you can't assume everything is a Gaussian. ;-)
A similar phenomenom occurs in many, many other fields. The average (median) wage-earner actually makes below average (mean) wages, because the existence of Bill Gates and Carl Icahn skews the distribution upwards. The median sale price for a startup is $0, because over half of them fail. The average test scores in Palo Alto or Weston, MA or Hunter College High School really are above average, because those places already preselect for bright kids. It's quite possible for Lake Wobegone to exist: you just need to compare your kids with someone else's average.
And none of this invalidates SSB, but you picked a bad example to illustrate it. When 90% of drivers think they're in the top 50%, they're right.
Do you have evidence that most people aren't above-average drivers? I would be pretty shocked if driving skill was normally distributed. I would guess that there are few extremely good drivers (you can only do so much to improve), but significantly more extremely bad drivers. This would drag down the average.
Even if you word your survey question in terms of medians, people are bad at numbers. I don't think you can really decide if that's illusory superiority without more digging.
This assumes a lot about the correctness of people's perceptions of their own driving skills. My experience is 99% of drivers think they're above average, and that obviously can't be true.
Just wanted to add that most people might actually be better than the mean driver depending on the shape of the distribution (e.g. if a small group accounts for the majority of accidents). Only 50% can be better than the median though.
Sure, but that only works because you've chosen a discrete measure of how "good" a driver is. If you defined a "good" driver by some other measure (say, a function of risky behavior and fuel consumption), then 50% of drivers would be above the median and 50% would be below, by definition.
reply