You may be right for a select group, but you can't have 90% of everyone be better-than-average; studies show this closer to 50%.
My wife Jessica teaches quantitative reasoning at Wellesley, and actually used the driving thing as an example last year. I'll get details wrong, but there's some study that says roughly: most people judge themselves to be above average because, in accidents they are involved in, they are "at fault" less than half the time. In the simple model where every accident is 2-car and one driver caused it, this seems reasonable at first glance -- and you get the 90% result if each bad driver causes nine accidents.
Sure you can... Studies or not, in theory, you could have everyone *but*one* be above average.
(Jessica also ends her introduction to averages with the classic "Almost everyone has an above-average number of legs.") --Michael Kleber kleber@brandeis.edu