#### Comment Preferences

• ##### Excellent analysis!(4.00)
Any way to graph it?  (A picture is worth a thousand words.)

IMHO, the voting machine game and deliberate vote suppression is just a big a case of fraud as any 'counting' issue.

Never underestimate the power of stupid people in large groups

[ Parent ]

• ##### Ideas for graphing this(none)
I have neither the time nor the skill to do this right, but I think that it's critical that this not just end up on a bar chart.  It needs icons.

What we need are rows representing precincts that are colored red or blue based on whether they went for Kerry or Bush.  Each square is split horizontally in two; on the left half there's an icon for each voting machine in the precinct, on the right half there's a little person icon for say every 10 people who had to wait in line at that precinct.  The rows should be sorted by ratio of line length to number of machines.  If there's a giant splash of blue at the top, there's an obvious problem.

So if X is a machine and i is 10 people:

--------------------------------------
Fake Kerry precinct
XXX    | iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
Fake Bush precinct
XXXXXXX| iiiiiiiii

Just a thought.  Hope the formatting comes out ok.

• ##### For graphs, see below n/t(none)

The man who trades freedom for security does not deserve nor will he ever receive either.-Benjamin Franklin

[ Parent ]

• ##### REQUEST(none)
Will someone please summarize this data in the simplest possible terms and post it up here near the top?  Not everyone has the time to do more than skim a diary full of statistical arguments like this one.  It would help to have a simple explanation so that someone can decide if it's worth their while to look very closely at the statistical argument in the diary.  The summary could be titled something like "WHAT THIS DATA SHOWS."

"Now watch this drive."

[ Parent ]

• ##### One body of graphs and analysis(none)
By new kossite 'Febble' (Joe Knapp) here (and referenced in other diaries). He derives an estimate of ~7,000 discouraged Kerry votes (net of Bush discouraged votes).

Be advised, Knapp's initial "smoking gun" enthusiasm is tempered considerably as a result of our colloquy here. The data may not support the weight of conclusions some would lay on them.

• ##### My error and apology(none)
... for confusing independent analyst kossite 'Febble' with copperas author kossite 'jmknapp'.
• ##### she(none)
she,actually. More soon
• ##### A slighlty different view of the data(none)
I've done some graphs and will post the URL as soon as it is up.  There is an alternative explanation (to deliberate malpractice)of the data although it still looks like culpable negligence to me. See (for now) the bottom of this page: linked text
• ##### Some graphs and analyses here(4.00)
I've written up some statistical analyses I've done on the Franklin county data and it's posted here:

It has some graphs, which make the statistical points fairly clearly.

My findings are similar, if more conservative than those of Richard Phillips, and I have used statistical tests of the effects..

To summarise:

1. Inadequate machine provision DID depress turnout*
2. This dispoportioniately affected Kerry voters*
3. The reason for this was at least in part that Kerry precincts were allocated fewer machines*

I also offer an alternative cock-up rather conspiracy theory, which is only slightly less damning: essentially they did not buy enough voting machines, and had to cover up.  They therefore factored in turnout in 2000 when allocating machines.  As turnout was more depressed in Democratic precincts (another statisically significant effect), this was tantamount to discriminating against Kerry voters in 2004.  And because half the county had too few machines to process every voter, this amounted to a "cap" on turnout at 2000 levels, and undermined the GOTV efforts of Democrats.

Finally I estimate the total loss of votes overall and to each candidate using a more conservative method than Phillips: I estimate that 18,5000 votes were lost, of which 66% would have been for Kerry and 30% for Bush.  This translates into a net loss of 7000 votes for Kerry in terms of Bush's final margin.

It is still possible that the machine allocation strategy was fiendishly clever rather than terminally stupid, but if it comes to court, they have a sort of defence, so it as well to know what it might be (and what is wrong with it).  The fact is that at even at this conservative estimate, around 18,500 voters were disenfranchised, and because the error was not randomly distributed between presidential candidates, it seriously distorted the result in favour of Bush.

* these results are statistically significant, at a probability of one in 10,000 or less.

• ##### Sorry:(none)
* these results are statistically significant, at a probability of one in 10,000 or MORE (ie 1 in more than 10,000 of the effects occurring by chance)

Sorry, I've got flu.

Even after assiduously attempting to destroy the effects in various ways (as any good statistician should do, and in line with suggestions by RonK Seattle) they were barely dented. And I should here thank RonK Seattle for his/her help and for the suggestion of looking for the algorithm likely to have been used to allocate the machines. Sorry, should have done that before.

• ##### Ballot Spoilage Map by Race(4.00)
This map shows the Cuyahoga County precincts where spoiled ballots (those deemed to not have a valid vote for president) exceeded 3% of the total ballots:
• ##### A picture tells a thousand words n/t(none)

"Now watch this drive."

[ Parent ]

• ##### This is starting to add up(none)
The Repub Challenges and thus provisional ballots

Unnecessary poll closings and switches

Voter/Machine Ratio Turnout Robbery

Warren County Lockdown

Long Lines

Plus all the Florida Shenanigans (see Bev Harris' page at http://www.bbv.org)

= Lots of people in on the Republican Fix, who would be paid off.

Wayne Madsen says he has the Election-Rigging Check for \$29.6 million "the check for \$29.6 million, which is marked "Not to exceed fifty million dollars," is drawn on the Laurentian Bank of Canada's Toronto branch. Its serial number is 317675450 3 and the bank number is 23-97/1020. The bank instrument is issued by Integrated Payment Systems, Inc. of Englewood, Colorado, and Bank One, NA, Denver, Colorado."

But KO and many here at KOS doubted there would be one check.  Too convenient.

I don't know, this must have been funded somehow.  How better than offshore money laundering?

• ##### Supporting data for the above graph(none)
Election data from problem precincts:

http://copperas.com/cuyahoga/cuyprobs.zip (zipped Excel file)

EIRS case details for the reported problems:

http://copperas.com/cuyahoga/eirs_cuyahoga.zip (zipped Excel file)

Subscribe or Donate to support Daily Kos.