I don't think we can establish anyone as having statistically relevant data on the subject. I suspect no organization outside WotC has such data, and if WotC has it, they aren't publishing it.
Nor do I think personal anecdotal evidence is useful other than for dispelling absolutes. If someone asserts, "This never happens!" then an anecdote can get us past that. But, no matter how long you've been gaming, personal anecdote won't tell us if a thing happens often enough to be relevant for design of a game that wants broad market reach.
That being said, here's a thought on the matter:
Yes, many people don't consider average damage per round when choosing a character. Many folks just go for concept, and don't worry about effectiveness.
That's fine, so long as the designers have worried about effectiveness for them. If the designers have not, and the GM does not, then you can have an issue at the table. The fact that you didn't worry about effectiveness when you created the character, and don't think you really care about it at the table, does not mean you can't have a bad or boring game experience resulting from your choices.
As an example that has nothing to do with combat - I have seen three different occasions in which a player chose a character style they thought would be cool, but then had a horrible play experience based on the choice. In each case it was a choice of the "Laconic, lone wolf character". Gee, if you play a character who doesn't talk or cooperate much with others, you won't have fun in a game based upon social interactions between players! Go figure!
In a well-designed game and/or a well run game (adjustment can be made in design or runtime, or both) it is possible to make sure everyone has fun, in theory.
That it is possible to do without it in theory does not mean we should not worry about it in practice.