[EM] winning votes vs margins...

[Elisabeth Varin/Stephane Rouillon]

Say that there are N voters.  N(a>b) + N(b>a) = N

In this case you are right, winning votes and margins produce the same
result.
The problem is when N(a>b) + N(b>a) + N(a=b) = N
Then results are different.

Further meanings can be implied when people do not indicate
preferences.  It could mean a=b (my opinion is that candidates a and b
are interchangeable clones) or a?b (I do not have any opinion and I rely

on the other voters judgement):
N(a>b) + N(b>a) + N(a=b) + N(a?b) = N

I explained why this leads to the
use of relative margins intead of margins or winning-votes in order to
match optimally the ordering result with the smaller probability of a
reverse pairwise comparison using ranked pairs.

Refer to paragraph 2) on May 22, message:
http://groups.yahoo.com/group/election-methods-list/message/9774

Steph.

----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em