From Ranks to Intensities
Ennis, D. M. (2004). IFPress, 7(4) 2-3.
In a classic voting problem, 24 people are asked to choose among three restaurants – one specializing in steak, one in Thai food, and one vegetarian. Nine vote for steak, eight for Thai and seven for vegetarian. The vegetarians would much prefer Thai to steak and the Thai choosers would prefer vegetarian to steak. Since the steak lovers won, they all go out for steak and most of them are dissatisfied. A ranking approach in which some account is taken of second and third choices would lead to a fairer outcome. Voting for candidates in the U.S. suffers from the same shortcomings and some other countries attempt to address these problems using various methods that incorporate ranking-type features.
Figure 1. The correlation or dependency among samples is greatest for the two freshest samples and least for the freshest and least fresh ranked samples.