[EM] Children and other animals

[Bart Ingles]

I have a friend who gets daughter dressed for preschool every morning by 
presenting her with exactly two complete outfits, and then telling her 
to "choose".  Apparently after making a choice, the girl is more likely 
to accept getting dressed without a fight, and so far hasn't demanded a 
third option.  Sound familiar?

  * * *

I've been thinking about what happens when voter motivations are not 
strictly utilitarian.  Suppose, for example, voters of one party react 
to a "scary" opponent by gravitating toward the opposite extreme, even 
though a more moderate choice would be expected to have a better chance 
of winning.  This may well be the innate human reaction, much as a 
frightened animal in a windowed enclosure will repeatedly dash itself 
against the glass rather than stop and search for a less obvious means 
of escape.

This could actually be self-reinforcing, since voters of the opposite 
party might then react to the first group's extremism by rallying even 
more strongly toward "scary" candidate who triggered the situation in 
the first place.  This process would continue back and forth until the 
electorate was completely polarized. [Sound familiar?]

Not even Condorcet-compliant methods would be immune to distortion under 
this scenario.  Antagonized voters might tend to rank an extremist at 
the top of the ballot, and bury the feared opponent.  By implication, 
the "rational" Condorcet candidate would suffer as voters migrate toward 
more extreme "champions", and unknown candidates would benefit by not 
being ranked last.  Approval voting would probably be the least affected 
by polarization, since the ballots don't contain room for exaggeration 
(either sincere or calculated).

On the other hand, this kind of voter response is not necessarily 
irrational--if a voter thinks that the country has been moving too far 
to the right of the political spectrum, he may try to compensate by 
supporting someone from the extreme left in an attempt to balance things 
out.  Merrill & Grofman actually consider strategies of this type in 
their book, "A Unified Theory of Voting".

I've been kicking this one around for a while, but haven't had time to 
research or develop it further.  Does anyone know of any other work 
along these lines?

Bart