[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

#4813: more on election count method (fwd)




From: steven white <stevelong@spyral.net>

I had a short discussion off list with Joanne Mariner of Human Rights
Watch, and  I thought I might submit it (with Ms. Mariner's OK).  This
issue has been run over so often on the list but it might still be of
interest for those who are still trying to figure out what the heck
happened.  here is my letter to her, followed by her response.


Dear Ms. Mariner

Your essay omits two facts.  First, that the method used by the CEP was
not new, but was the same used in previous elections.  Indeed, you imply
the
opposite in the phrases: "novel counting method" and "Haitian election
officials tallying this year's results ..." (my italics).  Second, each
voter was allowed to
choose two or three candidates. This makes the possibility of a
candidate getting a majority of eligible voters much easier than in a
multiple candidate race in
which each voter casts only one vote.  Thus your comparison to the US
presidential election is comparing apples to oranges, because a vote for
one
candidate does not preclude voting for another- you could vote for Nader
and Gore.  In the Haitian election system, even if every single eligible
voter in the
district cast a vote for a candidate, he could still not possibly get
50% +1 of votes cast, because every one of those voters also got a
second vote to cast.
Although there may be a more mathematically correct way to resolve this
dilemma than the one used by the CEP, any method proposed after the
votes have
been counted is more suspect of partisan favoritism than one designated
before the results were known.


Dear Mr. White,

Thanks for your interest in the Haitian election situation, and my
article commenting on the calculation method used for senate seats.
Just to respond to your comments:

First, the CEP asserts that its method was used in previous elections,
but there is no evidence to support this claim.  Indeed, just the
opposite.  As I explained in the article, under the method used in this
election, the leading candidate in each department is mathematically
bound to win.  (Because the math used is the following, as you can see
if you check the numbers released by the CEP:  The total votes of the
top four candidates in each department are counted as equaling 200%,
since there are two posts in each department.  Any candidate with 50%
plus one vote wins in the first round.  One-quarter of 200% is 50%.   In

other words, the average of all four candidates is necessarily 50% under

this method.  The leading candidate necessarily has more votes than the
average.  You can try any numbers and the leading candidate will always
win in the first round.  The math is slightly different in races with
six posts, where the six leading candidates are counted, but the result
is the same -- the leading candidate is bound to win.)

Thus, under the CEP's current method, one slot per department will
always be won in the first round (and the other slots are extremely
likely to be won in the first round).  Since Haiti has 9 departments,
that means 9 candidates will always win in the first round.  Yet in 1990

elections, only 5 of 27 senate seats were won in the first round.
Therefore, it's is clear that a different method was used.

Similarly, in the 1995 elections, even though 11 of 18 seats were won in

the first round, there were a few departments in which no seat was won
in the first round.  This is mathematically impossible under the method
used by this year's CEP.  Again, a different method must have been used.

The above should also answer your other comment about the impossibility
of getting 50% plus one of the votes cast.  It's not impossible, because

under any system used, two posts are considered 200%.  The question is
whether all the votes are counted to get this 200%, or only the votes
accruing to the top four candidates.  Indeed, under the normal voting
system, several senatorial seats in this election would have been won in

the first round -- but not all of them.

I hope this makes clearer what I was trying to explain in the article.