I've been getting a lot of mail from people asking for my take on
the news about the Washington GOP primary. Most have wanted me to
debunk rumours about vote fixing there, the way that I tried to debunk the
rumours about the Democratic votes back in New Hampshire.
Well, sorry to disappoint those of you who were hoping for a nice debunking
of the idea of fraud, but to me, something sure looks fishy.
For those who haven't been following: over the weekend,
Republicans had primaries and/or caucuus in three states. In the first
two - Kansas and Louisiana - McCain got beaten, badly, by Mike
Huckabee. Meanwhile, in Washington, the early vote counts showed
Huckabee leading there, as well. As more votes came in, McCain took a
very narrow lead. Then suddenly, when they got to 87% of
voting precincts reporting, with McCain leading by an extremely slim
margin - under 2% - the Washington state party suddenly stopped
counting votes, and declared McCain the winner.
The state party has been making the argument that McCain won -
that with that number of votes counted, statisticaly, there was no way
that Huckabee could make up the difference, and so they could fairly
declare the winner. Their "methodology" for this conclusion was: "Let's take every county where Huckabee is beating McCain, and double the margin of victory. Then let's take every county where McCain is winning, and half that margin of victory. Even if you assume that, Sen. McCain still holds on."
As statistical methodology goes, that's absolute garbage. But worse, it's
clearly a lie - because despite claiming to have done that analysis, the officials in the state party admit that they have no idea what parts of the state weren't fully
counted. No clue, not about counties, voting precincts, nothing. But they want you to believe that they did this "careful" analysis on a county-by-county basis.
The argument from the state party is pure, utter rubbish. And
based on all of the information I can find, it seems quite clear that
the state party decided that they wanted McCain to win, and they
halted the vote count when they did in order to ensure that their preferred
candidate won. It's fraud, plain and simple, beyond any reasonable doubt.
I'm not saying that McCain lost the election. I don't
know if he would have won or not. But the fact is - in a large diverse
state, with 87% of the vote counted, a 1.8% margin between winner and
loser, there's simply no way to say "Statistically, we know
what the result would have been". Why not?
If we look at the results as they came in during the day, we can
see that Huckabee showed a lead early; quite a significant lead in
some parts of the state - at one point when results were coming in,
the counts had Huckabee leading by about 5%. As more votes came in,
that shifted in McCains favor, and when the count stopped, McCain was
ahead by 1.8 percent. That 1.8% margin between Huckabee and McCain
works out to a total of only 242 votes. With 13% of the vote remaining
to be counted - and when we don't know anything about the demographics
or voting histories of the districts which aren't included in the
reported counts - there is simple no way that we can rule out the
possibility that the remaining 13% of the vote would have produced an
excess 243 votes for Huckabee.
Let's take a quick look at the arithmetic of it. Assume that the
reported percentages accurately represent the percentage of voters in
the unreported discricts - so that the counted votes - the 13,475
votes counted - represent 87% of the republican voters. Then the
expected total republican electorate would be about 15,500 voters -
leaving about 2,000 votes left uncounted. For those votes to swing the
election to Huckabee, he would have needed to win roughly 1,135 votes
- that is, to win the uncounted districts by about 6%.
Is it likely that Huckabee won the remaining districts by
6%? Not particularly. Is it possible that he won them by 6%?
Absolutely.
Now, normally, I'm very hostile to the idea of conspiracies in
things like this. As you saw in my look at the Democrats in New
Hampshire, I'm much more likely to believe that the McCain won fair
and square - that it would be extremely surprising to see Huckabee win
by a sizeable margin in the small portion of the state left to
vote.
But it's clearly possible mathematically, and the behavior of the
Republican party officials in charge of the count have been completely
inexplicable. I can't imagine any good reason to suddenly stop the
vote count at 87% with a narrow margin. The answers from the people in
charge have been ridiculous at best - the state party chair, Luke
Esser, said that he halted the count and declared a winner because:
"...it was an exciting and historic day for the state and I thought if
I was confident about what the outcome would be I should share that
with the people who had gone out to their caucuses." He's admitted
that he probably shouldn't have stopped the count, but in his defense,
he asserted that it was "statistically impossible" for Huckabee to
have caught up and overtaken McCain. But when probed further, he
admitted that he knew nothing about the uncounted votes - not even
which areas of the state included the precincts whose votes weren't fully
counted.
And further: when the Huckabee folks started pushing for a
complete vote count, the republican state chair reluctantly agreed to
count the rest of the votes, some time over the next few days. But when
the Huckabee folks asked if they could have representatives present to observe
the count, he threw a tantrum and hung up the phone on them.
When the person in charge of an election stops counting
votes, refuses to explain why he stopped the count, demonstrates
profound ignorance of the factors that could have affected things, and
throws tantrums over stakeholders asking to observe the full
count... Well, something is seriously wrong. Most likely, it's just
pure simple incompetence - he's an idiot, and he did something idiotic
for stupid reasons, and he's pissed off that his stupidity is being so
vividly demonstrated in the national press. But there's no excuse for
ordering a stop to the vote count. No excuse at all. In a democracy,
you're supposed to count every vote. Not "every vote until
you're pretty sure that you know who won", but every
vote. And to stop counting votes so early - with roughly one out of
every 8 votes uncounted - is simply insane. Put that together with
some of the bizzare behavior on the part of the state party
representatives, and it creates a very strong appearance that
something is up. (Note: I originally had a typo in the above; when I changed
from "X%" to "1 in X", I accidentally didn't change the X - so this originally read "1 in 12 votes uncounted. In fact, it's slightly more than 1 in 8.)
Then, we add in the latest information: as of this morning, the
state is reporting an additional 5% the votes have been counted, with the
margin narrowing from approximately 2% separating McCain and Huckabee
down to approximately 1%.
So - I absolutely wouldn't rule out the possibility of
fraud. There's no mathematical reason Huckabee couldn't have won, and
there's no reasonable explanation for the insane behaviour of the
party officials in charge of the vote. Something fishy sure seems to
be going on.
- Log in to post comments
"I know the pundits, and I know what they say: The math doesn't work out," Huckabee said Saturday morning at the Conservative Political Action Conference in Washington. "Well, I didn't major in math; I majored in miracles. And I still believe in those, too."
I want to be the first, prominant Theomathematician that I am, to warn you (since Mark C. Chu-Carroll has admitted enjoying play with Gemmatria) that the actual margin of votes, after the legally-forced recount, will be interpreted by Huckabee and his proxies in terms of Gemmatria.
If McCain wins, for instance, it could be a by a multiple of 666 votes.
If Huckabee wins, it could be by a multiple of 888 votes.
The number of Jesus is 888. Using the Greek Ionic Ciphered Numeral System "scientifically" proves this. Google for "888 jesus" to see all the sincere loons who say so. In this system, each letter of the Greek alphabet is assigned a numerical value.
The name of Jesus in Greek is spelled I H S O U S (iota, eta, sigma, omicron, upsilon, sigma).
Substituting in the Greek numeral system the equivalent numerical values to each letter in the name of Jesus and adding them up, the total is 888. The values of each letter are: iota, 10; eta, 8; sigma, 200; omicron, 70; upsilon, 400; sigma, 200. The sum of 10 + 8 + 200 + 70 + 400 + 200 is 888.
Which is the same upside-down and backwards, in our default base 10 Arabic numerals. Although, even if I stood on my head, voting for Huckabee would still make no sense.
Technically, that's called a Tetradic Number. As MathWorld begins:
A tetradic (or four-way) number is a number that remains unchanged when flipped back to front, mirrored up-down, or flipped up-down. Since the only numbers that remain unchanged which turned up-side-down or mirrored are 0, 1, and 8 (here, the numerals 1 and 8 are assumed to be written as a single stroke and symmetrical pair of loops, respectively), a tetradic number is precisely a palindromic number containing only 0, 1, and 8 as digits. The first few are therefore 1, 8, 11, 88, 101, 111, 181, 808, 818, 888, 1001, ... (Sloane's A006072).
The first few tetradic primes are 11, 101, 181, 18181, 1008001, 1180811, 1880881, 1881881, ... (Sloane's A068188).
Not to be confused with the Tetragrammaton.
And stand by for more than just the Olympics on 08/08/08.
You heard it here first;)
Why is anybody surprised. This is what Rethuglicans do, just as their toadies on the Supreme Court stopped the recount in Florida in 2000. Whether Senator Clinton of Senator Obama wins the Democratic nod, they better be on the lookout for this this type of shenanigans.
Excellent analysis.
Just one small detail. 13% is more than one in twelve.
Great article!
Luke Esser was the Republican official who screamed bloody murder about vote counting practices in the 2004 Washington gubernatorial election. The Democratic candidate (Christine Gregoire) won by a very slim margin and Luke Esser argued for weeks that the vote count was mishandled in hopes that with enough recounting he could swing the final tally in favor of his candidate.
In light of this I don't believe for a second that Esser was that stupid about the implications of stopping the Republican caucus vote count when he did. He knew good and well what he was doing. There's definitely something fishy going on.
Adrian:
D'oh! Thanks for catching that. I originally wrote "more than 12%", and then decided to
change it to avoid the repetitions of "12%", but didn't correctly change the number. It should read "one in eight.".
I find it bizarre that the count should NOT be observed, in any circumstances...
In my country, Australia, all elections are presided over by an independent body, and every vote is scrutinised as it is counted by all the parties in the vote.
We just don't get situations like this. Occasionally, in very tight races, where how some difficult to read ballots are counted would affect the result, it goes to the Court of Disputed Returns (the highest court in the land, specially convened to sort these things out), but they are an independent judicial body.
To me, grown up with this system, it is incomprehensible that candidates should be unable to scrutinise the counting.
Peter:
That's not so different from how things are supposed to work here. Primaries are kind of strange, because they're *not* the real official elections - they're run by the political parties, not officially by the government.
But still - in any election that I've seen in the US, the initial count is generally machine tabulated at the polling places - and then every step of transferring that information to
the central counting station is observed by everyone. And recounts always have representatives from all of the concerned parties observing.
That's part of what makes this whole Washington nonsense so incredibly fishy. They stopped counting before all of the results were in, and they're throwing tantrums over the perfectly normal procedure of observed recounts. Quite clearly they believe that they have something to hide.
Like I said, I don't know who won the election, and I think that there's a good chance that McCain really legitimately did win. But I think that the state party pulled something to try to guarantee that, and now they're trying to cover their asses. What made them think that they could stop counting like that, and have no one complain, I'll never understand.
Hi Peter,
I noticed this on reddit and thought it was fishy until I read this comment: http://reddit.com/info/68d4a/comments/c035jd2
Assuming the commenter is legitimate, it seems that it's irrelevant whether they stopped counting or not since the results of the caucuses provide no guarantee of the actual delegate votes.
I don't understand why the parties do it themselves. Returning to Australia, organisations as diverse as university student organisations, city councils, political parties, and unions all [pay to] have their elections conducted by the Electoral Commissions. It gives just a little bit more credibility than trusting a room full of people who're invested in specific outcomes.
Thomas:
Here's a link to a historical look at how the nominating process came to be what it is today:
http://www.thegreenpapers.com/Hx/NomProcess.html
As it says in the article, the process is both amazing and frustrating. The process for selecting the president and vice president isn't much better:
http://electoral-vote.com/evp2004/info/electoral-college.html
I trust that the probability of a candidate defying the previous vote proportions for the last 13% is vanishing; that I don't take issue with. The issue is two-fold, first...something isn't statistically or probabilistically "impossible" unless it is measure zero in the event space. Huckabee defying previous vote proportions for the last 13% is no where even close to measure zero so it is absolutely a lie to use the term "impossible" in a mathematical context because "impossible" HAS a definition in probability and statistics: measure zero. Second, even if you were Math-Man and you had some great insight into why this event WAS measure zero...wouldn't it occur to you in your great wisdom that pulling this stunt would obfuscate the discourse and anger people who didn't have your great insight?
Well its clearly not fraud. At worst its a case of a lazy election official. If it were fraud they would have claimed they had finished counting or thrown out the remaining votes so there was no possibility of going back and recounting them (as it is, they are counting them; last I had heard they had finished 90 something percent of them, so they are starting to get close...).
And not to nitpick, but McCain did not get "beaten badly" in Louisiana, at least not yet. He "lost" by 2 percent, and party rules in that state require a candidate to get a majority in order to get the delegates. Since both were under 50%, those delegates will be given out at a later date based on caucus results. Even Huckabee's Kansas win isn't that big when you consider how few people showed up for an election that is seen as over for everyone but Huckabee.
Nick Brown writes: "Well its clearly not fraud."
I disagree. That's not at all clear to me.
"At worst its a case of a lazy election official."
Maybe. Or maybe it's a case of "lazy fraud." Don't forget, most criminals get caught because they're careless and/or stupid.
Fraud pretty much requires some sort of deception, and there was no attempt at that here. The idea that he meant to claim that they had counted 100% of the vote and then I suppose just forgot is technically possible, but requires a pretty major suspension of belief. Laziness or a misunderstanding of statistics is much, much more likely. And there certainly is no evidence of it, you might as well claim New Hampshire or South Carolina were cases of voter fraud where they did remember to cover it up.
The main thing in this case is what's the rush to call it?
This is true in this case (even more than in the 2000 presidential election, where, contrary to reports, there was plenty of time to finish the vote counting, and further mechanisms to handle a count handed in late). In this case though, it's months until the convention -- what's the rush? Count all the votes.
Mark, I'm not sure your arithmetic is correct.
There are about 2,014 votes still uncounted and Huckabee would need to garner at least 243 more than McCain. 243/2014 = 12%.
Obviously, 12% is significantly less "statistically probable" than your 6% figure. There might actually be a legit statistician out there who would argue that 12% is approaching "impossible."
Even more obvious, however, is that the result should have nothing to do with statistics. This ain't "rocket surgery." They just need to count ALL the votes, dammit!
So I agree with your premise but not the details. Just trying to keep you honest. :-)
Not counting things you don't need to count is a way to help with voting efficiency. Provisional ballots are typically used when you don't want to go to the trouble of figuring out whether a voter is actually allowed to vote in an election then and there. (It'd be a pain if all the votetaking staff were involved in disputes leaving nobody to work the machines, and it'd be unfair for somebody to not be able to vote because the clerk misspelt their name!)
So what at least some places do is freely give out provisional ballots, and then collect them in a secure place separate from the non-provisional ballots. If the winner of the non-disputes ballots is ahead by more than the number of provisional ballots, there is no need to deal with the time-consuming (and expensive) task of validating the provisional ballots.
I don't have too much of a problem with the vote counting they did or didn't do in Washington for one simple reason: the straw poll is completely meaningless and based on a "who did you come to support" entry on the caucus sign-in sheet. Reports are that Ron Paul took over 50% of the delegates to the state convention in May. The only thing that the straw poll is good for is to report to the media, and I can see how Huckabee would be a bit upset that his possible publicity victory is being blunted. He's going to be in for a rude shock however when he loses the primary to McCain and the convention to Paul.
Mark, In Oz, there are no machines. Everything is hand written and hand counted - everything is on paper, and you can watch every vote being counted if you want.
If you want to do a recount, you get the bundles, and you re-count. If you want to dispute it, you grab the ones you thought were should have been counted as yours, and the ones you thought were mistakenly counted as your oppositions, and the High Court, acting as the Court of Disputed Returns, adjudicates.
The Australian Electoral Commission sets the boundaries, runs the elections (state and federal and everything in-between - they will run your company election if you want) and counts the vote.
The US thing where politicians are in charge of the boundaries and the counts strikes us Aussies as more than a little odd.
I just mention this in passing. There are no diebold scandals and other things of that nature over here. We still have crooked politicians tho...
Of course, you have the Bill of Rights and the 1st Amendment and all of that - which we don't have.
Brian has a point, it may be more of a cost issue than laziness (and don't dispute how much cost factors into these decisions, its why my state delayed their primary). And I don't know where Dan got the Ron Paul statistic, but his claim about the caucus results being meaningless in terms of delegates is accurate as well. They still probably should have finished, but to be honest, as someone who lives in a state whose primary is sometime in May (after both races will almost certainly be over), and in a primary season full of stories of superdelegates and delegates from states being cut in half or thrown out entirely, I have trouble finding sympathy for people complaining that their vote in a meaningless caucus didn't get counted.
A couple of minor points, MarkCC:
1. This was a caucus, not a primary (Washington will also have a primary on 19 February). That explains why the party was in charge of the counting.
2. Your clarification at #6 that "the initial count is generally machine tabulated at the polling places - and then every step of transferring that information to
the central counting station is observed by everyone" is not true in all states. In New Hampshire about half of the precincts (covering about 20% of the population) do the count entirely by hand (the fact that Obama did better in these precincts than in machine-counted precincts was one of the points that the conspiracy theorists raised but never proved about the NH count). I have heard that in some states such as Georgia the votes are aggregated as well as tabulated by machine (I don't know this firsthand--any knowledgeable Georgia residents available to comment?).
But your fundamental point is correct. Stopping the vote tally at 87% counted and the race as close as this one was is either stupid or evil (or both). Throw in the fact (as reported by TPM) that the WSRP chairman is the protege of McCain's Washington state campaign chair, and I conclude that Huckabee's people have every right and reason to suspect foul play.
Your math may be good (and I certainly find it interesting), but your atrocious spelling detracts from your credibility.
Just paste your posts into Word (or even gmail which has spellcheck that's good but not as good) before you submit them and fix the spelling errors - then paste it back in here.
Ned:
Word? You want me to paste my posts into *Word*? I don't even have MS word installed on my computer. And if I did, pasting an HTML article into it would be a total damned disaster. Yes, I should be more consistent about run spell-checks on my posts; but spell-checking by pasting stuff into a monstrosity like Microsoft Word is about as dumb a way of accomplishing that as I can imagine.
The Washington State Republican party has been long known to be somewhat unhinged: perhaps living in a state that has always leaned left makes them kind of desperate. Bear in mind they were (I believe) the only state in the union Pat Robertson won in the 1988 primary.
It looks like there voting irregularities in the Louisiana Republican Caucus as well. Apparently Ron Paul had more delegates pledged to him than the other candidates so the party extended the deadline so that more delagates could register.
Then during the voting process many of the Paul supporters had to file provisional ballots, even though they had been pre-approved delegates.
Now, the winner of Louisiana Republican Caucus was the Ronald Reagan Pro Family slate or some such. However, the candidate that received the most votes could have very likely been Ron Paul, even though the Louisiana GOP gave it to McCain. Apparently, Paul's provisional ballots have not been counted.
The first line of my post should have read,
It looks like there are voting irregularities...
Apparently, I should have posted this into MS Word first.
Oh, I didn't mean to say that -this- particular case of not counting every vote wasn't suspicious. Just wanted to point out that the principle that "every vote should be counted" isn't so iron-clad as all that.
Most participants on this thread are completely misunderstanding how this process works. I suppose that's what happens when this discussion comes up on a forum which is neither for political junkies nor for residents of the state of Washington.
What took place on Saturday was not an election, it was a whole lot of precinct caucuses. The purpose of the precinct caucuses is to select delegates who will attend the county (or district) caucuses in March and April. Those caucuses, in turn, will select delegates to the state convention, where they will choose delegates to the national convention. The delegates at the national convention will choose the nominee for the Republican Party.
Now it so happens that most of these delegates chosen at the precinct delegates have declared their presidential preference. By understanding the formulas by which different precincts are represented at different levels, we can make a pretty fair projection of how many delegates at the national convention will result at the end of the process. Lacking the knowledge of how the delegates are distributed, simply counting them up gives a somewhat cruder measure.
To take an example from another caucus more exhaustively studied, in this year's Democratic caucuses in Iowa 940 delegates were declared for Obama, 744 were for Edwards, and 737 were for Clinton. Because of the distribution of these delegates among the precincts, the end result in terms of delegates at the national convention will be 16 for Obama, 15 for Clinton, and 14 for Edwards.
In spite of this, total number of precinct delegates is generally what is counted to declare who "won" the state. Being the "winner" is thus correlated to having the best results at the national convention, but not equivalent to it. In Washington this is further complicated by the fact that additional delegates to the national convention will be selected as a result of the primary election, which will be held this coming weekend.
The point of all this discussion is that arguments like "In a democracy, you're supposed to count every vote" are completely out of place here. This is a nominating procedure held by the Party and the Party makes the rules. Did the state party chairman fudge in order to declare his favorite candidate the "winner"? Of course he did. And the Huckabee folks have every right to be outraged and kick up a fuss about it. But there's no electoral fraud here.
They are not obligated to release figures to the media by any particular date. They are not required to make a "count" at all. The only thing they're required to do is to seat the right delegates at the next level of caucuses. They will surely do that in due time, but the county caucuses don't begin until next month. Some time around the state convention in May someone will say, "Gee, what do you know, Huckabee got more votes after all. Well, I'll be darned." Until then they can drag their heels and obfuscate all they like.
You can say the party leadership is being sleazy, or you can observe that they're using their leadership muscle to deprive the unfavored candidate of some good publicity. But they aren't being ignorant, insane nor idiotic. They know exactly what they're doing.
Spike: the problem with a direct non-electoral college popular vote (or many other schemes) would be that huge cities like LA, NYC, DC, would drive the presidential election, which would drive politics to just consider urban needs over rural ones.
Some say that we should reconfigure the college along county lines, or some other smaller districting but then we get into problems with gerrymandering...
Brian, you're wrong. Stopping the count is a method of cheating. Here's a riddle for you:
Suppose, hypothetically, that Huckabee had 55% of the votes, and that the other 45% were McCain's. The chairman, who is a McCain supporter, supervises the votes counting. There is a very simple thing which he can do if he wants McCain to win - he can wait for a McCain to gain a temporary advantage, then stop the count, and declare McCain the winner. What are the chairman's chances of success?
Check mt blog (mousomer.wordpress.com) if you can't solve it ;-)
I find it bizarre that the count COULD NOT be observed, in any circumstances...
There is no way to ensure your vote is counted at all. 90% of the vote could be thrown in the trash and there is no way to tell.