Update on New Hampshire Recount (With Links)

[nashuaadvocate]

Here: http://www.sethabramson.blogspot.com/

Based on what I've seen, I now support this recount. Not because I think the vote was hacked, but because of the numbers tracked down by the Election Defense Alliance's Bruce O'Dell (find them at the link), which are singularly creepy (as in, the strongest case of possible vote-flipping ever seen in American politics), and as such cause me worry that there was a computer glitch amongst the 81% of New Hampshire votes cast on Diebold-manufactured machines. The article at the link has links to all the major news stories written on this topic in the last ten hours. The other reason I'm for this is because I am a New Hampshirite, and we are not racists up here. I want that story put to bed forever and all time, because it offends me personally. And so, for me, this goes beyond Obama or Hillary or even protecting the franchise, but protecting the good name of the 1.2 million folks up here in the Granite State.

(UPDATE: Ron Paul campaign supporters claim they have, independently of Kucinich, raised enough to fund their own recount in New Hampshire).

With the Dems having nothing to do until Nevada, and no major primary until South Carolina much later this month, it strikes me that Democrats--and the media, who should really be behind this--have plenty of time to figure out why Hillary's own internal polls show a 14-point swing in her vote in less than 24 hours from January 8th to January 9th.

Every major media outlet is now all over this story--check the links yourself--and DU should be, too.


S.

[deminks]

1. Thank you. Reading the links now. Why Joe and Willie just this morning were saying how

on MSNBC that Dennis only had 200 bucks and a couple of Mr. Cheesie coupons in his wallet to pay for the recount. Funny how time and a few facts change the perspective.

[K Gardner]

2. I am so glad you posted this. I got up this morning feeling like Mel Gibson in

Conspiracy Theory. I've been waffling between: "I'm crazy and paranoid", "This is a trick, we shouldn't recount" or "It's over, there's no use fighting it anymore."

This at least allows me to see that, no matter what the cause, it just looks funny. Nothing more, nothing less. No one is alleging voter fraud. We just want to know why it looks "funny".. why every single pollster and pundit and citizen was caught off-guard and left breathless as results trickled in.

So.. no allegations. No boogeyman hysterics. Let's just figure out what, if anything at all, happened. If nothing did, fine. If something did, well that's fine, too. And while we're at it, let's get rid of those freakin Diebold machines. Good to see this in major media print, so we don't look like "fringe people" :-)

Way to go, guys. :yourock:

[papau]

3. the 04 NH vote was the "strongest case" but on audit it proved to be correct. Only audits

settle this kind of question - but the left rejected the nationwide audits of 08 that were in the HR811 bill (congrats to Bev/Brad)-

In any case, I certainly support these audits - and expect they will show that nothing happened. Ohio in 04 was stolen - NH was not.

[stillcool]

5. Why does an election need to be stolen...

to hand-count a percentage of votes? Why can't a bill be produced that merely legislates paper ballots, a percentage of hand counted ballots, and open-source-code? Why do we have to wait for 'Political Will' to make sure our votes count? Why can we not implement measures now...that insure these races will have accurate, verifiable tabulations?

JANUARY 2008

* January 3: Iowa (caucuses)
* January 5: Wyoming (GOP caucuses)
* January 8: New Hampshire (primary)
* January 15: Michigan
* January 19: Nevada (precinct caucuses), South Carolina (R primary)
* January 26: South Carolina (D primary)
* January 29: Florida

FEBRUARY 2008

* February 1: Maine (R)
* February 5: Alabama, Alaska, Arizona, Arkansas, California, Colorado (caucuses), Connecticut, Delaware, Georgia, Idaho (D), Illinois, Kansas (D), Massachusetts, Minnesota, Missouri, New Jersey, New Mexico (D), New York, North Dakota (caucuses), Oklahoma, Tennessee, Utah
* February 9: Louisiana, Kansas (R)
* February 10: Maine (D caucuses)
* February 12: District of Columbia, Maryland, Virginia
* February 19: Hawaii (D), Washington, Wisconsin

MARCH 2008

* March 4: Ohio, Rhode Island, Texas, Vermont
* March 8: Wyoming (D)
* March 11: Mississippi

APRIL 2008

* April 22: Pennsylvania

MAY 2008

* May 6: Indiana, North Carolina
* May 13: Nebraska (primary), West Virginia
* May 20: Kentucky, Oregon
* May 27: Idaho (R)

JUNE 2008

* June 3: Montana, New Mexico (R), South Dakota

[yodermon]

10. such legislation will not be produced because then elections couldn't be

stolen anymore.

[mopinko]

13. you know, many states do audit, they just don't talk about it.

i was surprised to find out that illinois has a 10% audit about/within 30 days of all elections. they are carried out by the county boards of election. they are done pretty much in secret, for all practical purposes. no one knows, so no one is there to witness. awareness of this is the key. i went out to witness one, but did not have all day to watch, which is the key. this one was staffed mostly by county personnel, not by citizen judges.
all of which is to say that people who are concerned about this need to dig deep into their own states election laws, and make sure they know it inside and out. and make sure they show up.

[papau]

38. true - but it is the random choice of precincts that's needed - and most audit states do it only

if requested or a very narrow announced victory

HR811 was the solution - now I'm afraid we will never have a national solution - the GOP conned the left and won the war.

[mopinko]

67. no. my state has a random 10% audit.

every election. deep dark secret, tho.

[EFerrari]

4. There you are!

:hi:

Have you read the concerns raised by DemocracyForNewHampshire? mod mom and kpete both have threads up about them.

[smoogatz]

6. It'll be an interesting—and maybe very revealing—test of the Optiscan

machines. My guess is they're about as accurate as supermarket scanners; which is to say, not very.

[OnTheOtherHand]

7. "the strongest case of possible vote-flipping ever seen in American politics"

Uh, no.

Actually, the case would be much stronger if the hand-count percentages matched the op-scan percentages, seeing as they come from very different places.

More here.

[Maven]

8. So in order to prove that New Hampshirites are not racist

you have to show that Obama actually won the primary?

Why does this have anything to do with verifying the results?

[bleever]

9. I still can't find this breakdown on the SOS website,

even though multiple sources claim as much, e.g., "...these actual New Hampshire vote totals from the New Hampshire Secretary of State's website."

My goal here is just to confirm the provenance of the basic facts, and I'm frustrated that it's been this difficult. Any help?

[EFerrari]

11. bleever: do you have a link to where you have looked there?

I'll go look, too. It slipped my mind this a.m.

[bleever]

14. Here ya go:

Main site: http://www.sos.nh.gov/

If you click the main link for primary results, you go here: http://www.sos.nh.gov/presprim2008/index.htm

Maybe I'm just missing it, but I can't find results sorted by counting method. Nor have I found any source using official NH info for all factors of the equation.

Yeah, everybody's all over this story, but they're all over the surface of it. My two cents: if there was any goofiness, it was either a) residual machine malfunction, b) residual machine malprogramming, or c) willful interference by parties outside the Democratic Party looking to make trouble.

[EFerrari]

16. I vote for c. Someone may have been hoping to set Democrats

against each other.

If that's true, let's disappoint them.

I'll go see if I can find anything. Thank you!

[Maven]

17. Well, here's a list of NH municipalities that use machines

http://www.sos.nh.gov/voting%20machines2006.htm

All others not on that list use hand counts.

So I suppose theoretically you could match up the machine list with the vote counts and get totals in each category. If you have about 10 hours to waste.

I would also like to see where these numbers are coming from. Seems like they're shuffling around the web without any attribution.

[bleever]

19. Thanks!

Why I read thirty or forty different reports and never saw that link referenced, I don't know.

I want to see the math done rigorously. And I fully support a recount, so that we can be sure that NOBODY can mess with the votes.

[EFerrari]

18. Wait: What we need is a list of the municipalities or counties

that use paper ballots.

I bet that's how those numbers were generated. Ya think?

[bleever]

20. See Harvey's post just above. nt

[EFerrari]

23. I saw the same list at the gov site but

they don't list paper counting venues.

So, the idea is to find a list of those, then do the math. Check the link I posted above to checkthevotes. It looks like that's what they did there.

[bleever]

26. At the bottom of the list in that link it says

that all other places use paper, and count by hand.

I'll look at the checkthevotes link. Thanks.

[EFerrari]

22. At this page, about 4 screens down, municipalities by method:

http://checkthevotes.com/#Compare%20All%20Machine%20vs%20Hand%20Counts

[Maven]

25. That's for Republicans

Here's the breakdown for Democrats:

http://checkthevotes.com/index.php?party=DEMOCRATS

I admit if this is true it's a very strange pattern but I still don't know how you even COULD deliberately key the machine results to the hand counts, since the machines aren't linked in NH and the hand-counted municipalities were among the last results to come in.

[EFerrari]

28. Oh -- I was only looking for a list of paper counting venues.

You're right. The data is for Republicans. But it gives a list of the venues one would need to put a list together for Democrats -- which as you saw, the SoS doesn't.

As to the other, I don't know enough to even guess.

[Maven]

30. OP misuses the word "flipped"

In '00/'04, the problem was that the machine data within certain municipalities reflected almost the exact opposite of the exit polls for the same municipality. Hence, there was appreciable evidence that votes were flipped from one candidate to another.

Here, that's not what's being alleged. All they're showing is that the ratios within each type of vote counting mirror one another, which is a strange pattern indeed but does not show the same type of flipping we saw in '00/'04. My point was that if you actually wanted to make this result come about you'd have to A) be psychic, since the hand-counted results were among the last to come in, and B) somehow manage to manipulate individual, unlinked tabulators in each municipality to get the numbers to match.

[EFerrari]

32. I do understand your point and don't know enough to speculate

about possible scenarios.

Did the handcounted venues match the pre election polling? In order to come up with a formula to attempt to produce the result implied in the OP, you'd have to have some kind of reliable estimate for the late returning handcounted municipalities, wouldn't you?

[Maven]

37. "Did the handcounted venues match the pre-election polling?"

Not really.

In fact, Clinton won decisively in several hand-counted municipalities, just not as many as Obama. Likewise, Obama won decisively (sometimes by a percentage ratio of 2:1) in many machine-counted municipalities.

I gave some examples here:

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=132x4009174#4012627

The problem is, even if you had estimates you're talking about getting the vote count right to within a couple hundred votes, without leaving evidence of overvoting (of which there was none). It just doesn't seem remotely likely given the setup that NH uses and the order in which different venues rendered results. Nor would it even be necessary (or even desirable) to accomplish such a neat pattern to throw the election.

[EFerrari]

43. I count 136 municipalities that used paper ballots and you gave

several examples of the counter case where Clinton won in those. Is your sample small or did you find "many"?

I'm not trying to be rude but can't figure out a way to go look for myself.

[Maven]

51. I would have to look at the CTV link and count

As I recall, the sample was not small--around 40 or so that I counted right after the election. The examples I gave in that link were some of the more dramatic ones. I'd have to count again to give you a total.

But there is something else interesting here--if you look at the CTV data you'll notice that in large towns that were hand-counted, Clinton still won decisively overall. The hand-counted ratio of Clinton/Obama in those large towns (mostly southern NH, where Clinton got most of her votes) is very similar to the machine-counted ratio.

Likewise, in small towns Clinton did slightly better in machine-counted votes, whereas Obama only did slightly better overall in hand-counted votes.

[gateley]

12. I haven't been paying attention to this, but I DID hear on Air America a few

days ago that Ron Paul supporters complained because they voted, and the votes weren't included in the totals. These were hand counted ballots, too.

They gave an example that let's say 3 people voted for RP, but the official tally showed 0. And I believe it happened more than once, and not just with RP votes.

As long as there are some questioning it, it wouldn't hurt to do a re-count, would it?

[creeksneakers2]

33. They only have an anomaly in one precinct

Ron Paul got 31 votes. The tally sheets recorded zero for him. Since its just one precinct, the likeliest explanation is human error or dishonesty by a single poll worker, not widespread fraud.

[gateley]

54. Thanks for the clarification. nt

[whatchamacallit]

15. You've gotta hand it to the Junta

HAVA is to 2000 Election Crisis as Big Wooden Horse is to Troy

[Maven]

21. EDIT: Attribution for numbers in the OP has been found

So I suppose it's a matter of ratios--the ratio of votes for one candidate to another in each category (machine/hand-counted) is the same. Still, I'm not sure how you could deliberately bring about this result or why you'd even want to if you were planning to fix the primary.

[EFerrari]

24. Go see my #22. That's where the numbers seem to come from. n/t

[Maven]

27. I saw that.

See my response.

[EFerrari]

35. Delete.

[creeksneakers2]

29. Here is why as a long time Clinton observer I'm against a recount

The audit won't just be a recount. It will turn into an investigation and the investigation will turn into a witch hunt. This happens every time the Clintons are accused of something. At the beginning, the accusers alway ask, "What harm could taking a second look do?" Then it all starts.

So far we have widely circulated falsehoods like the claim the Chris Matthews said that exit polls showed an 8% lead for Obama. A chart is circulating says Clinton only was projected to get 35% of the vote in the original exit poll but that number comes from one line in a British newspaper that didn't even have direct access to the data. We have invalid apples to oranges comparisons like hand counted to machine counted districts. Those comparison have been debunked, but the debunking is ignored.

Its already getting worse. People are raising the issue of ballot security. They believe that Obama ballots will be removed by insiders who want to cover up the fix. A recount won't satisfy a conspiracy theorist. Nothing ever does. When no evidence is found for the original accusation, the theorist just says it was covered up.

Soon we'll have revelations like somebody who worked at the software contractor made a donation to Bill Clinton in 1996. They'll find a Clinton supporter who had access to the ballots after the election. TV will pick it all up and they'll report it with phrases like "Hillary's ballot rigging scandal." It will go on and on until people get bored with it and it stops being covered on cable.

The proponents of the recount accuse those who are against it of not supporting clean elections. That is not the case with me. I'd love to see audits of our elections. But those audits should be scheduled BEFORE the election and not brought up afterward. A post election decision to audit casts suspicion on the winner. And if you are a Clinton you have the Clinton curse - every accusation must be followed by a witch hunt.

[Maven]

31. I agree, and I would add

I believe a recount will show nothing amiss and the result will be that commentators will say, "See? All those nuts alleging fraud in 2000 and 2004 were just sore losers. This shows voting machines are accurate."

[EFerrari]

36. Sure and someone in NH who knows what they're doing can

take it right back to them. "You don't even know where these ballots have been, so you can't know if the count is right".

Of course, the Noise Machine has the volume advantage.

[Maven]

41. That they do

Let them go ahead with the recount. I wish EVERY contest was hand-counted and have worked to keep DRE machines out of my state. So far, it's been a losing battle. I just hope they do this quietly unless and until they find the numbers were off enough to change the outcome.

[EFerrari]

34. This has nothing to do with the Clintons. n/t

[creeksneakers2]

39. It does now.

Somebody somewhere will report something like they saw a suspicious guy near a counting machine and the vast right wing conspiracy will want an investigation to find out if the unnamed person could be associated with the Clintons. The left will pick up on it.

I've seen all this many many times before. It always turns out the same. It has already started this time. The Clintons get singled out for scrutiny.

They are saying that this is the first time a result came in that didn't match pre-election polls. Here's the report from the results of gubernatorial election in Minnesota that elected Jesse Ventura:

http://www.salon.com/news/1998/11/06newsb.html

Its always a conspiracy theory when it comes to the Clintons. Its their own unique curse.

[EFerrari]

40. Are you a Hil supporter? Because the people I see pushing this

around here are HRC supporters for the most part. Those of us on DU that are pursuing this aren't interested in Hillary Clinton. We're interested in election integrity.

I've been asking and asking for someone to show me the Obama or the WhoEver supporters making claims of fraud. I've asked about five times now and no one has yet pointed me to one thread. Can you? Because it's getting weird already -- like I can't see something everyone else is seeing.

[creeksneakers2]

44. I'm a regular Hillary defender but may not be a supporter

I don't like her negative attacks on Obama. I think they are unfair. I also doubt her political judgment because those kinds of attacks backfire. So I'm thinking about switching to Obama.

Lots of people on DU are interested in election integrity and even put substantial personal effort into it but there was no avalanche of calls for a New Hampshire audit BEFORE Hillary won.

Suspicions are being raised everywhere here about potential fraud. I don't see many outright accusations. But suspicions become accusations. Most aren't blaming Hillary either but that will come along. Who is the most likely suspect for election fraud? The winner.

[EFerrari]

47. I've been in this forum since November 22, 2004. This forum

is why I came to DU.

Had Obama won, or my granny for that matter, and under these circumstances, I promise you this forum would be raising exactly the same questions. I don't know about the rest of DU.

And then, as althecat and autorank did, they'd post their stuff to GD so more people could read it.


Since Hillary didn't count the vote, it's silly beyond belief to blame her. Some wacko somewhere might do that. That's what wackos do. :)

[yodermon]

42. Here's the key piece of info alleging "vote flipping"

From http://www.opednews.com/articles/opedne_bruce_o__080110_obama_clinton_3a_remar.htm%20

Clinton Optical scan 91,717 52.95%
Obama Optical scan 81,495 47.05%

Clinton Hand-counted 20,889 47.05%
Obama Hand-counted 23,509 52.95%

Note how the percentages are identical down to the hundredths. e.g. Clinton Optical == Obama Hand-counted.

This result would in fact occur if Clinton & Obama had been flipped in the machine tally.

However, this data is not evidence that this has actually occurred.

It simply means that a manual recount of the op-scan ballots is a good idea.

[EFerrari]

45. Right and right. We're were just trying to figure out how

this could be an intended result.

[K Gardner]

46. It may truely NOT have been intended. It could have been a malfunction, or one of those

one in a million freak happenstance occurences that, legitimately happen. But we won't know until we look, is all we're saying.

[EFerrari]

49. Understood. And if we had a mandated audit in place like a sane

electorate would, we wouldn't have to trust or guess. We'd have verified while we could.

[K Gardner]

50. What can we do, or is anything even in the works, to ensure that? You know, SC is ALL touch screen..

not evena paper ballot trail to TRY to follow.

[EFerrari]

52. I don't know if anything is in the works. That would be a great question

in an OP.

If I were the Queen of the ER movement, I'd be mounting challenges in every state that uses these POS with my millions of volunteers and billions of dollars. :)

But, more seriously, someone does have an answer to this. Maybe Kelvin Mace, maybe emlev or another ER regular. That's why I'd post it in an OP.

[NobleCynic]

75. Considering how foolishly obvious this is if it is a case of fraud

It could be a disgruntled employee

[creeksneakers2]

48. I don't see this as evidence of anything at all.

Hillary got 100% of her own votes. Start with that.

Suppose Hillary got 65% from machine counts. Then she must have received 35% of her vote from hand counted. That would account for 100% of her vote.

If Hillary got 65% form machine counted, then Obama had to get 35% from machine counted. He therefore had to get 65% of hand counted.

Try it with any number you want. It will always turn out this way. It will "magically flip." The person who came up with this logic didn't think it through.

[Maven]

53. This is not "flipping"

Please see my posts above.

However I agree that manual counts are always better.

[creeksneakers2]

55. I read your post before about Ohio and I agree with you

But I think we are talking about two different meanings of flipping.

There is flipping meaning a machine subtracts votes from one candidate and gives them to another.

Then there is the flipping of proportions of vote totals that yoderman points out. Yoderman claims evidence of fraud because the results of proportions of vote totals for the two candidates on the two type of machines exactly mirror each other. Where Obama get 60% Hillary get 40%. Where Hillary gets 60% Obama get 40%. Yoderman believes this is evidence of fraud. What is really is is the result of a math problem that will only turn out this way.

To set up Yoderman's calculations, they used the following laws.

A = Machine votes for Hillary.
B = hand counted votes for Hillary
C = Machine counted votes for Obama
D = hand counted votes for Obama.

A+B = 100%
A+C = 100%
B+D = 100%
C+D = 100%

So we have a matrix
Hillary - A / B
Obama - C / D
machine counted / hand counted

If I know the results for any one of the squares I can predict all four squares.

If A = 30% then B must = 70%
If A = 30% then C must = 70%
If C = 70% then D must = 30%

So if you take the machine count for for Obama against his hand count numbers you'll get 70% / 30%
Do the same for Hillary you get 30 / 70%

Yoderman sees this as evidence of fraud because the proportions flipped in such exact numbers.

[OnTheOtherHand]

57. no, I don't think it's quite that bad

It's perfectly possible that Clinton could have, say, 51% of the two-way vote in hand-count places and 53% in op-scan places. There's nothing that requires those two percentages to total 100%.

(Row percentages, column percentages....)

It's more like weird numerology.

[creeksneakers2]

58. She could have 51% of the total vote in hand counts

and 53% in scans. She could not have 51% of her own vote in hand counts and 53% of her own vote in scans. Her own vote has to add up to 100%. Its hard to explain but I tracked down where this started. See how they did it:

http://www.electiondefensealliance.org/

New Hampshire 2008 Primary Analysis
This page is an evolving compilation of outstanding citizen investigation of the highly suspicious New Hampshire primary voting results. We are borrowing and synthesizing from many sources cited and credited here.

Thursday 1/10: Bruce O'Dell writes:

Theron Horton and I have confirmed that based on the official results on the New Hampshire Secretary of State web site, there is a remarkable relationship between Obama and Clinton votes, when you look at votes tabulated by op-scan versus votes tabulated by hand:

Clinton Optical scan 91,717 52.95%
Obama Optical scan 81,495 47.05%

Clinton Hand-counted 20,889 47.05%
Obama Hand-counted 23,509 52.95%

The percentages appear to be swapped. This seems highly unusual.

[eomer]

60. You're using the wrong denominator.

She could not have 51% of her own vote in hand counts and 53% of her own vote in scans.

None of the percentages in the allegedly weird swap have the total votes for a candidate as the denominator. The denominator for the first and second percentages is the total opscan votes and the denominator for the third and fourth percentages is the total handcount votes. Total Clinton votes and total Obama votes don't enter in anywhere in these percentages.

[creeksneakers2]

62. Yes, they do.

Add across both rows and up and down both columns. They all add up to 100%.

The calculations were made by adding the number of votes for Obama and Clinton in each category and designating that total as 100%. Then the vote totals were cross multiplied to come up with the ratios of one vote to another that made up the 100%.

They did the same thing for machine counts versus hand counts. Took the entire total, designated that as 100%. Took the ratios and expressed them as percentages that added up to 100%.

They said they were going to probe the data at the center. Think I should send them an E-mail?

[eomer]

65. No, they don't.

Clinton Optical scan: 91,717 / (91,717 + 81,495) = 52.95%
Obama Optical scan: 81,495 / (91,717 + 81,495) = 47.05%

Clinton Hand-counted: 20,889 / (20,889 + 23,509) = 47.05%
Obama Hand-counted: 23,509 / (20,889 + 23,509) = 52.95%

The total Clinton votes of (91,717 + 20,889) doesn't appear as a denominator in any of the four percentage formulas.

The total Obama votes of (81,495 + 23,509) doesn't appear as a denominator in any of the four percentage formulas.

Please look at the hypothetical example in my post downstream and you can see that your formulas are not correct.

[eomer]

61. Weird numerology.

Yep. Anytime you get a whole bunch of numbers together, you can always find something somewhere that looks weird.

Of course, sometimes something that looks weird actually is weird, but I'm doubtful this in one of those cases -- I think it's one that just looks weird.

[OnTheOtherHand]

63. it's too weird

I was looking for similar weirdness, and noticed that per Leip, Bush's share in Arkansas matched Kerry's share in California, and in Connecticut, all to the nearest hundredth. The Inference Is Obvious!

More thoughts:
http://www.dailykos.com/storyonly/2008/1/11/91642/8304/344/434942

[eomer]

59. A hypothetical example shows your formulas are wrong.

The results could have been:

Clinton Optical scan 10,000 10.00%
Obama Optical scan 90,000 90.00%

Clinton Hand-counted 30,000 60.00%
Obama Hand-counted 20,000 40.00%

You can see that your formulas are wrong because 10% + 60% does not equal 100% as your formulas say that (a + c) should.

[creeksneakers2]

64. There are only two methods of counting votes

If Clinton had 10% of her vote optical scan then she would need to have 90% hand counted. Your figures totaling scanned and hand counted for Clinton only add to 70%. Where did the other 30% of her votes come from?

Your vote totals for Obama hand counted and Obama scanned add up to 130%. How could that be?

You can see at my post that when the calculations were made that led to this anomaly they were using the 100% of total figure to determine ratios of one thing to another.

[eomer]

66. [Add] up to 130%. How could that be?

Because you are mixing percentages that have different denominators. It is only when looking at a list of percentages that all have the same denominator that they will add up to 100%. (Edit: or, I should say, that they must add up to 100%. It is always possible that they may accidentally add up to 100%.)

For example, there is no reason that the percentage of U.S. residents who are male, plus the percentage of U.S. residents who are non-Hispanic should add up to 100%, right? The percentage who are male plus the percentage who are female should add to 100% and the percentage who are Hispanic plus the percentage who are non-Hispanic should add to 100% but the percentage who are male plus the percentage who are non-Hispanic will be something more than 100% (could be 130%). This last percentage is probably not useful for anything since we added two percentages that are unrelated to each other and therefore are not likely to end up with anything meaningful.

If you want to look at percentages with a denominator of total votes for Clinton or at percentages with a denominator of total votes for Obama, then using the same example would give these numbers:

Clinton Optical scan 10,000 25.00%
Clinton Hand-counted 30,000 75.00%

Obama Hand-counted 20,000 18.18%
Obama Optical scan 90,000 82.82%

They don't add up to 100% when combined with any of the percentages we were talking about before, because they are using a different denominator.

[creeksneakers2]

68. The denominator is a percentage is always one hundred.

Lets look down one row. Only scanned ballots for Hillary and scanned ballots for Obama are counted. Together, the total percentage of Hillary scanned ballots and Obama's percentage of scanned ballots must equal 100% or all the scanned ballots.

Your hypothetical example has the total of scanned ballots 25% + 82.18% which = 107.8%. It has to equal 100%. That's all the ballots and no more.
The conspiracy theory calculation only has two categories of scanned votes - scanned votes for Obama and scanned votes for Hillary. So the percentages must add up to 100%, not 107.8%.

[OnTheOtherHand]

72. there is No Way that you can win this argument

And you should not want to, because it really has no bearing on the debate about New Hampshire.

Please reread eomer's original hypothetical until the perfect parallelism with the OP sinks in. (He was trying to add the point that I tried to add -- that you can look at ROW percentages or COLUMN percentages, but not simultaneously. Clearly it didn't quite come across.)

ETA: Or maybe look really close at #65. eomer is going out of his way to get on the same page with you. Meet him.

[eomer]

73. Let me clarify.

By denominator I mean the total number of votes that is defined to represent 100% in each case (for a particular set of percentages). The denominator will be different when you are calculating percentages for a row than it is when you are calculating percentages for a column or when you are calculating percentages for the entire matrix.

When calculating percentages by row, then the denominator for each row is the total number of votes in that row:

              Clinton              Obama                Total
Opscan    91,717  (52.95%)     81,495  (47.05%)    173,212 (100.00%)
Hand      20,889  (47.05%)     23,509  (52.95%)     44,398 (100.00%)
Total    112,606  (51.75%)    105,004  (48.25%)    217,610 (100.00%)

When calculating percentages by column, then the denominator for each column is the total number of votes in that column:

              Clinton              Obama                Total
Opscan    91,717  (81.45%)     81,495  (77.61%)    173,212  (79.60%)
Hand      20,889  (18.55%)     23,509  (22.39%)     44,398  (20.40%)
Total    112,606 (100.00%)    105,004 (100.00%)    217,610 (100.00%)

When calculating percentages by column and row, then the denominator for the entire matrix is the total number of votes in the entire matrix:

              Clinton              Obama                Total
Opscan    91,717  (42.15%)     81,495  (37.45%)    173,212  (79.60%)
Hand      20,889   (9.60%)     23,509  (10.80%)     44,398  (20.40%)
Total    112,606  (51.75%)    105,004  (48.25%)    217,610 (100.00%)

In the first case (by row), the denominator for the first row is the total number of votes in that row, or 173,212 votes. So the two percentages in that row are the ratio that the votes in each cell in that row bear to 173,212. In other words, 91,717 / 173,212 and 81,495 / 173,212. The denominator for the second row is the total number of votes for that row, or 44,498 votes. The denominator for the third row (the total row) is the total number of votes in that row, or 217,610 votes. In each case, the denominator will have (100.00%) next to it because it is, by definition, the number of votes that represents 100% for that percentage calculation.

So, getting back to the question at hand, whenever you have calculated percentages by row, you can't normally add the percentages in a column and get 100%. Similarly, whenever you have calculated percentages by column, you can't normally add the percentages in a row and get 100%. But the weird thing that Bruce O'Dell noticed is that in this case you can. When we calculate percentages by row, in this case, we actually can add the percentages for a column and get 100%. But that is just an accident.

To prove that it is just an accident I'm going to put together the above three ways of calculating percentages for the same hypothetical example I gave before.

Percentages by row:

              Clinton              Obama                Total
Opscan    10,000  (33.33%)     20,000  (66.67%)     30,000 (100.00%)
Hand      30,000  (25.00%)     90,000  (75.00%)    120,000 (100.00%)
Total     40,000  (26.67%)    110,000  (73.33%)    150,000 (100.00%)

Percentages by column:

              Clinton              Obama                Total
Opscan    10,000  (25.00%)     20,000  (18.18%)     30,000  (20.00%)
Hand      30,000  (75.00%)     90,000  (81.82%)    120,000  (80.00%)
Total     40,000 (100.00%)    110,000 (100.00%)    150,000 (100.00%)

Percentages by column and row:

              Clinton              Obama                Total
Opscan    10,000   (6.67%)     20,000  (13.33%)     30,000  (20.00%)
Hand      30,000  (20.00%)     90,000  (60.00%)    120,000  (80.00%)
Total     40,000  (26.67%)    110,000  (73.33%)    150,000 (100.00%)

In this hypothetical example, you can't take the first case where percentages were calculated by row and then add the percentages in a column and get 100%. So the fact that you can do it in the real numbers at the top of this post is just a fluke and that is what everyone thinks is too weird. You are getting messed up by thinking that the numbers had to add to 100% but in fact they didn't have to -- as my example proves -- they just accidentally did (or else did because of some fraud mechanism that made it turn out that way).

[OnTheOtherHand]

74. you win a sportsmanship prize for this one! n/t

[Maven]

69. That's what I thought at first too

However, that's not the issue they're focusing on.

The issue is actually the ratio of one candidate's votes to the other in each category.

In other words, take Clinton's machine counted votes and divide it by Obama's machine counted votes, and you get something like 1.125.

Now take Obama's hand counted votes and divide by Clinton's hand counted votes, and you also get 1.125 or something close.

They've just chosen to express this proportion as a percentage, by dividing each candidate's vote counts into the total Clinton+Obama in each category. That's my gripe with this OP--it's a deceptive way to present the data, to make it look like the same kind of flipping we saw in Ohio, when in fact that's not what those percentages represent at all.

[creeksneakers2]

70. You are still using a ratio

And one must be the inverse of the other, because there are only two variables compared to two varibles.

[Joanne98]

56. Thank you for posting this!

[HooptieWagon]

71. Math Problem!!!!!

The OP is incorrectly assuming that 100% of Clinton's votes = 100% of Obama's votes. This is incorrect, unless by chance both candidates recieved the same number of votes (tie). 100% of A may or may not equal 100% of B. The poster who added the actual vote tally (not the percentage) has it correct.