It strikes me that there is something muddled about comparing something like "a person's odds of being killed in a car accident" with something along the lines of "that person's odds of being killed in a plane crash," noting that the first odds are lower than the second, and then concluding that anyone who is concerned about flying but routinely drives is "irrational." No doubt it is true that people frequently overestimate risk connected with some rare but attention-grabbing calamity that has just made the news, while remaining blase about the risks involved in more mundane activities.
But my complaint concerning the above situation is that the odds of being in a car accident are not really odds at all. That's because those "odds" are arrived at by treating the occurence of accidents as purely random events that simply "happen" to drivers, like being struck by a meteor. To the contrary, any individual driver's likelihood of having a collision is affected quite significantly by how she drives -- her skill, her knowledge of her car, her degree of focus, and how much caution she exercises. For instance, one can only collide with another car after first closely approaching it. (Automobiles have not been observed making quantum jumps that instantaneously bridge the distance separating them.) Therefore, any driver who stays as far away from other vehicles as possible is far less likely to collide with one of them than is someone who habitually tailgates or weaves through traffic. And that means that it is methodological nonsense to treat "the odds" for having an accident calculated by treating all drivers as indistinguishable and without influence over the degree of risk they face as if they applied equally to each individual driver.
On the other hand, a passenger on an airplane really does have little he can do to alter the possibility of the plane he is on crashing. (Little but not nothing, e.g., he can refrain from violently taking over control of the aircraft and then deliberately flying it into a building.) In general, this is risk beyond his control, and I see nothing irrational about one being more adverse to such risks than to those one deliberately can control.