The problem with having such options in surveys is that everyone then picks it, even though it is statistically unlikely. So better to force them to make a decision instead.
I think four bins of below average, slightly below average, slightly above average, and above average would be reasonable. It would show how many people think they fit into the interquartile range versus the extremities.
But then it will just take longer. You'll still have to pick which side of the average you're on, plus how extreme (or maybe you'll just never pick the extreme).
