[EM] Voting machines

[Blake Cretney]

Janet R. Anderson wrote:
>> Since in the US, the
>> districts buy their own machines, richer districts can afford top-of
>> the line, first class democracy.  So, in the US, voting power is
>> partly tied to wealth.  This helps out the Republicans, and is quite
>> possible the reason that George W. Bush is now President Elect.
>
>Blake,
>You may know your math, but you sure have mixed up facts to arrive at
the
>partisan conclusion you wanted!  I believe most, if not all, elections
are
>conducted by Counties, not districts.  

Sorry, that is correct.

> Most counties were formed at
>statehood and were related  more closely to geography than population.
>County income from taxes is influenced by the amount of commercial
activity
>in that county.  The "richer" counties usually host the largest cities
which
>tend to vote heavily Democrat.  The rural and suburban areas of a
state,
>usually host far less commercial property and by your definition rank
among
>the "poorer counties" and tend to lean Republican.  

I never defined poor counties in that way, or in any way.  I would say
that a poorer county would be a county that is unable to provide as high
a level of services for the tax rates it imposes on residents.  For
example, if a County had more of its money taken up by policing and
social services, and had less money for schools than another county, I
would consider it to be poorer, even if its tax base was actually
higher.  Similarly, if a County chooses to have low tax rates and few
services, I don't think that makes it poor either.  

In a poor county, using my definition, buying new voting machines is
likely to be a hard sell, since voters will see the county as having
more pressing problems.

> Let's try to hold down
>the misleading partisan rhetoric.

I'll give it a try.  By the way, I'm not a registered Democrat.  I'm not
even an American, so I don't see how I can be a Democratic partisan in
any traditional sense.  Of course I may be a naive foreigner, duped by
Democratic partisans.

Anyway, to get back to the subject at hand, I found the following
statistics at:
www.sun-sentinel.com/news/daily/detail/0,1136,36000000000129817,00.htm
These statistics are for all of Florida.

pen-marked (I think this is the same as optical scan)
presidential vote: over 99% (percent of votes with a vote read for
president)
Bush 52.9%
Gore 44.6%

punched
presidential vote: 96.1%
Bush 46%
Gore 51.8%

So, it's clear that for whatever reason, the Republican voters tend to
have better machines.

According to:
http://washingtonpost.com/wp-dyn/articles/A36760-2000Nov16.html

Not even all punch card systems are the same.  

In the "Votamatic" system, voters punch holes with a stylus.  In the
"Datavote" system, they use a hole-punching machine, which eliminates
the chad problem.  According to the Washington Post, in Datavote
Counties, Bush lead Gore by 90,000 to 60,000, while in Votamatic
Counties, Gore leads Bush by 1,825,000 to 1,600,000.

There is another explanation besides wealth for these disparities. 
Urban areas (which tend to be more Democratic), were more quick to adopt
voting machines, since the efficiency is more significant for more
ballots.  Unfortunately, this older technology doesn't work very well. 
Rural and suburban areas (which tend to be more Republican) have adopted
machines more recently, so the machines are better.

So it isn't simply a case of wealthier Republicans buying better
machines.  Although, since the poor are disproportionately Urban, they
are affected negatively by the difference in machines.  It certainly
doesn't change the point that a County with money to spend can increase
the voting power of its electorate.  

For example, Brevard County (which voted Republican) managed to increase
the percentage of votes with ballots read for president to 99.7% from
97.2% by upgrading its machines from punch card to optical scanner. 
It's possible that voters using punch cards are just not as interested
in the presidential race as are voters using optical scanners, but it
seems more plausible that that the 2.5% difference is caused by the
tabulation method.

Of course, whether these differences in machines made the difference in
the election is harder to say, and I currently don't have enough
information to state with confidence one way or the other.

However, here is a plausible guess.  Let's call P the percentage of
ballots with a vote for President.  Now, the best methods seem to get a
P of over 99%.  This suggests that over 99% really do want to vote for
President, or that there votes will be spoiled in even the best methods.
 However, lower numbers are likely the result of the machine or the
ballot.

So, to determine the likely scores that would have been obtained under a
different method, we can multiply them by 99/P.  So, based on the
Votamatic Counties, the expected increase in votes for Gore minus the
expected increase of votes for Bush is:

(1,825,000 X 99/P - 1,825,000) - (1,600,000 X 99/P - 1,600,000)
-> 1,825,000 X (99/P - P/P) - 1,600,000 X (99/P-P/P)
-> (1,825,000 - 1,600,000) ((99-P)/P)
-> 225,000 X (99-P)/P

Unfortunately, I don't know what P is for Votamatic counties.  However,
P for all punch systems is 96%, so we can expect that P for Votamatic is
even lower.  Setting P to 96% gives 7031 votes for Gore above the number
it gives to Bush.

Of course, depending of the accuracy of Datavote machines, this number
might come down.  On the other hand, if my 96% number is an
over-estimate, then the number would go up.  Note as well that my figure
includes votes that were spoiled as a result of the ballot, so a hand
recount would not necessarily have given this many votes.

---
Blake Cretney