[EM] cycle of defeats (sort of) in Zogby poll results

[Jan Kok]

At http://zogby.com/news/ReadNews.dbm?ID=1252

1. A plurality poll of people who said they would vote in the
Democratic caucuses or primaries found:

Clinton 33%
Obama 25%
Edwards 12%
...
Not sure 20%

So at first glance it seems Clinton pairwise defeats Obama. (This is
admittedly a pretty shaky conclusion because the comparison was _not_
done pairwise, and because only Democrats were included in this poll.
Would a pairwise poll of all subjects - Dems, Repubs, and others -
have found that Obama pairwise defeats Clinton?)

2. Pairwise polls of all subjects showed:

Giuliani 47%, Clinton 40%
Giuliani 40%, Obama 46%
McCain 47%, Clinton 39%
McCain 40%, Obama 44%

Conclusion: there appears to be a cycle of defeats Clinton > Obama >
(Giuliani, McCain) > Clinton.

This poll is weak evidence that cycles of defeats can occur in real life.

If the presidential election used a Condorcet method, the
cycle-breaking method could become important!

Cheers,
- Jan