Sometimes it pays to be nice, and sometimes it pays to be a jerk. It shouldn't be a surprise that neither kind of behavior dominates every situation. The trick is knowing when to act how. Apparently the important aspect of jerkiness is the confidence, not the cruelty.
The problem with competence is that we can't judge it by looking at someone. Yes, in some occupations it's fairly transparent--a professional baseball player, for instance, cannot very well pretend to have hit 60 home runs last season when he actually hit six--but in business it's generally opaque. Did the product you helped launch succeed because of you, or because of your brilliant No. 2, or your lucky market timing, or your competitor's errors, or the foundation your predecessor laid, or because you were (as the management writer Jim Collins puts it) a socket wrench that happened to fit that one job? Difficult to know, really. So we rely on proxies--superficial cues for competence that we take and mistake for the real thing.
What's shocking is how powerful these cues can be. When Anderson paired up college students and asked them to place 15 U.S. cities on a blank map of North America, the level of a person's confidence in her geographic knowledge was as good a predictor of how highly her partner rated her, after the fact, as was her actual geographic knowledge. Let me repeat that: seeming like you knew about geography was as good as knowing about geography. In another scenario--four-person teams collaboratively solving math problems--the person with the most inflated sense of her own abilities tended to emerge as the group's de facto leader. Being the first to blurt out an answer, right or wrong, was taken as a sign of superior quantitative skill.