Hrm. I would guess that effect tends to result from people associating generally with other folks in the same social tier. Like, to side step wealth entirely, let's say you're someone who values physical fitness. You will generally hang out with other folks who value physical fitness. When asked to rate the people around you of the gender you're attracted to, you'll reply by finding the differentiating features, and going from there. But, since you've self-selected your social circle, they'll all tend to have values that you find important. (The undergraduate studies the author speaks of are limiting for the same reason.) So when you've selected people in whom you are relatively more interested, you'll magnify the importance of the unique things about them that you like, but those unique things aren't exactly what made the two of you compatible in the first place.

Put another way, I'd be surprised if this finding still held true if you took a a random hundred people from the general populace and put them in a dome. I predict you'd see "mate value" very clearly delineated in the pairings up that resulted.

I'm not arguing that uniqueness isn't a factor. Maybe I am just nitpicking here -- it's unclear whether the author intended to convey that uniqueness was one factor among many, or an overriding factor. If the former, then my posts here have probably not been useful, but if the latter, I doubt it.