#### Comment Preferences

• ##### Yes(0+ / 0-)

This is interesting

http://www.dailykos.com/...

looks like the 12 largest precincts dropped the "ones" column in the presidential vote totals.

when I add a zero in the ones column to the counts president vote counts for these precincts it looks like this:

BALLOTS   TOTAL
CAST         COUNTED
TOTAL

9640    9509
8772    8550
8383    8270
7004    6850
5748    5485
5498    5275
5246    5069
5140    4886
4015    3923
2925    2789
1684    1666
1607    1587
1540    1534

--------------

After doing this and summing the 25 largest precincts with a total number of votes for president cast of 77,712
1.3% of the state total

total the breakdown is

McCain = 42,231 = 54.3% of the vote
Obama = 35,481 = 45.7% of the vote

final state total tallies (with the correction above)

McCain = 2,708,311 = 47.00% of the vote
Obama = 2,964,298 = 51.44% of the vote

Total delta (+McCain + -Obama)

=13.07% vote swing

"They Own You." --George Carlin

[ Parent ]

• ##### huh?(0+ / 0-)

I'm not seeing a lot of zeroes in your ones column. Also, did you look at the registration figures and candidate counts in those precincts? Are you sure that there were over 77,000 votes cast for president in these precincts?

Setting that aside, do you know anything about these precincts that would lead you to suspect "vote swing"? I realize that you're just getting started -- except that, at the same time, you've already asserted your conclusions.

Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
Better Know Your Voting System with the Verifier!

[ Parent ]

• ##### Hah!(0+ / 0-)

You are right, the problem was with the reported votes, the actual votes counted was ok

Here is the new data

in the 15 largest precincts McCain had 51.82%

total votes cast was 11,075 for McCain
and 10,058 for Obama

in the 1,230 largest precincts McCain had a total of 53.3% and Obama had 45.35%.

total votes cast was 559,925 for McCain
and 476,438 for Obama

the top 1,230 largest precincts accounted for 1,050,593 total votes or 18.4% of the state total.

"They Own You." --George Carlin

[ Parent ]

• ##### OK, that looks about right(0+ / 0-)

Now, when I split the precincts into quintiles by votes counted for president, I see a monotonic increase in McCain's vote share. Yes?

Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
Better Know Your Voting System with the Verifier!

[ Parent ]

• ##### heck, we don't even need quintiles(0+ / 0-)

You stand by your graph, yes? (It looks OK to me, although the choice of scale might be construed as quirky.)

So, the relationship under discussion does not appear "only in large precincts," yes?

That might have some bearing on your attempts to explain it, if indeed you do attempt to explain it.

Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
Better Know Your Voting System with the Verifier!

[ Parent ]

• ##### quintiles(0+ / 0-)

average mccain by precinct size quintile

http://i46.tinypic.com/...

also on the blog

"They Own You." --George Carlin

[ Parent ]

• ##### It looks like the predictor(0+ / 0-)

for higher GOP returns for larger precinct sizes is scalable to the size of the precinct, on average.

so, yes not on a precinct by precinct basis but on ALL precincts on Average due to their size.

This would be the most effective way to hide a vote flipping scheme when smaller precincts are more susceptible to a recount verification.

(theory)

"They Own You." --George Carlin

[ Parent ]

• ##### really?(0+ / 0-)
This would be the most effective way to hide a vote flipping scheme when smaller precincts are more susceptible to a recount verification.

(theory)

That seems pretty close to saying, "If this were a great way to hide fraud, then it would be a great way to hide fraud!"

I think it would make some sense if the supposed "swing" were heavily concentrated in a small fraction of large precincts -- at least in states where there is some sort of simple-random-sample audit of the primary results, however many states that may have been. But that isn't at all what we see here. What we see here is a "pattern" that is trivially easy to pick up in correlational analysis, and that confers no obvious advantage to a hacker.

Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
Better Know Your Voting System with the Verifier!

[ Parent ]

• ##### Interesting(0+ / 0-)

Regardless of how easy it is to determine this bias of larger precincts toward the GOP, I would wonder what other methods could be used to skew the vote, if one were so inclined.

It seems to me that, if there was a way to flip votes, and the goal was to produce a 10% swing in the direction of the favored candidate (flipping 5% of the total votes) then, one would HAVE to focus on the larger precincts.

What is the error margin of the random sample?  I would bet that it is 90/5.

Does Ohio even perform a random sample check?

"They Own You." --George Carlin

[ Parent ]

• ##### it isn't obvious why one would have to focus(0+ / 0-)

Every time one decides not to attack one precinct, one has to flip that many more votes in some other precinct. That may or may not be expeditious depending on what one is up against.

Does Ohio even perform a random sample check?
As a result of a legal settlement, it presently does in presidential primaries and in even-year general elections including the presidential contest. This year, it's a simple random sample including at least 5% of votes in each county. So, packing the fraud into a small number of precincts could be a good play -- although Ohio also lets candidates ask for partial recounts -- but smearing it up and down the size distribution doesn't seem to confer much advantage. The specifics would depend on the state.

Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
Better Know Your Voting System with the Verifier!

[ Parent ]

Subscribe or Donate to support Daily Kos.