CR pairwise

Roy royone at
Tue Sep 4 09:01:25 PDT 2001

Forest Simmons wrote:
> Dyadic Approval requires the voter to understand a process of
> assigning smaller weights to weaker preferences.

Only if you expect the voter to approach the eleciton with Dyadic 
Approval in mind. I don't think that's necessary. Sophisticated 
voters, of course, would want to know the details of vote-counting, 
so that they could maximize their votes' impact. Ordinary voters 
could simply vote intuitively (CR style), and their votes would 
reflect, to some degree, the strength of their expressed preferences.

The unsophisticated voter would be horrified to learn that a rating 
of 51 beats a 49 by the same amount as a 99 beats a 1, so the 
adjustment to DA that I would make would be to use the difference 
between the candidates instead of the absolute rating: 51 vs 49 is a 
1/32 win, just like 99 vs. 97 or 3 vs. 1. In general, a ratings 
difference of 50 (half the total range) points or more is a 1, 25-49 
is 1/2, 13-24 is 1/4, etc.

The unsophisticated voter might expect 90 vs 20 to be more 
substantial than 80 vs. 30, but when rating candidates that far 
apart, they're probably more concerned with how they compare to those 
with closer ratings. In any case, the actual scored difference is 
within a factor of 2 of what one would expect, and those that are 
rated farther apart are always scored at least as far apart as those 
who are rated closer together.

The sophisticated, strategic voter would approach the election as 
Dyadic Approval, but after ensuring 50 points between approved and 
disapproved, would have some compromises to make.

More information about the Election-Methods mailing list