Slashdot Mirror
Statistics, Elections, Frustration
The word is that the Florida recount will be completed later today (before 5 EST). In the meantime, a couple of interesting bits
related to math (which seems much more appropriate to Slashdot ;) The big one is of course the 'Voting Irregularity' in Palm Beach where supposedly thousands of seniors voted for Buchanan due to a badly designed ballot. this report (unfortunately, its a PDF) breaks down the returns on various counties and pretty much proves that something was wrong. Any math folks out there interested in doing their own take on the numbers?
bwoodard sent in
a mathematical argument for the electoral college written by
MIT Prof, Alan Natapof. Hopefully we'll have more word later today. Update: 11/09 01:55 PM EDT by C :For those of you interested in seeing why there is such controversy over the Palm Beach County ballot, you can take a look at the ballot to see for yourself if it might be a bit unintuitive. If you'd like more food for thought, you can check out this article which talks a bit about the usability issues behind the ballot's design.
Does anyone else find it incredibly ironic that
the brothers Daley would be on national media crying about election irregularities? Just for the record, their father stole the 1960 presidential election for Kennedy away from Nixon.
So long and thanks for all the fish . . . !!!
if GWB upholds his principles, he looses the election, and thus deserves to win.
if GWB betrays his pricinples, he wins the election, and thus deserves to loose.
[as seen on a bulletin board in Fermi Lab]
"Never apply a Star Trek Solution to a Babylon 5 problem"
Looks like a Babylon 5 problem to me ....
"It is a greater offense to steal men's labor, than their clothes"
Shows a scatter graph of Buchanan votes vs. Bush votes, by county. Assuming that Buchanan should get a fairly consistent percentage of the Bush votes by county (and this graph does seem to bear out that assumption), the Palm Beach results stick out like a sore thumb.
It has also been reported that 19,000 ballots from palm beach were invalidated because 2 holes had been punched for the presidential candidate. I wonder which two they were...?
It's important to emphasize that this does NOT mean that there was any sort of fraud going on. It was most likely an honest mistake, the official who designed the ballot said she was trying to find a way to use large fonts to help the elderly voters.
If it's clear that they were confused, though, it seems that the only fair thing to do is have a re-vote in Palm Beach, open only to those who voted the first time. (They all signed their names, right?)
---
According to Orvetti, Buchanan(-hole) has just conceded that the 3400 votes are "properly Gore's".
OK. I still have a problem with it (in the Constitutional sense), but if Buchanan's willing to relinquish his claim on those votes, now I can see at least some argument that might persuade a judge to grant a revote.
In any case, the recount results are skewing heavily to Gore, so my whole point may be moot. Let's hope so.
[Comic Store Owner]: Worst ballot ever !
"Sweet creeping zombie Jesus!"
-The Professor, Futurama
My credentials: MS in math, 4 years experience as a quality control statistician.
Here are some "flaws" in the analysis at http://web.mit.edu/norstadt/Public/election.pdf
1. The data at best supports the proposition that the Bush/Buchanan voting ratio for Palm Beach is significantly different than other Florida counties (but maybe not, see 5 below), but not that it is different from the "desired voting" ratio for Palm Beach.
2. Using Bush/Buchanan ratio is very weird - should use Buchanan/All ratio. Trying to introduce outside knowledge inherent in focusing on how "conservative" voters split introduces preconceptions that are non-statistical in nature. Why not Buchanan/Nader?
3. Normalization set chosen to be Florida. Why? Why not all US? Why not all of the South?
4. Expected distribution of Buchanan support rates by county is not known and cannot be determined by choosing a model. You must provide a non-statistical argument for why you believe the model holds.
5. In particular, even if the counties are normally distributed (doubtful), the maximum value should be chosen from an extreme value distribution, not a normal distribution. Extreme value distributions have very heavy tails. The fact that the choice of Palm Beach was made post hoc changes how the analysis should be done.
6. The basic premise is flawed because the vote rate might actually be from a different distribution than other counties. For example consider the distribution of voters born in Palm Beach. You cannot look at a model to infer that the emperical distribution is "wrong". In other words, maybe Palm Beach really does have more Buchanan supporters.
7. If the hypothesis is "the butterfly ballot causes confusion between candidate 2 and candidate 3", then it would be proper to test that hypothesis on ALL races using that ballot. In particular the Senate race on the same ballot did not display any anomoly despite having 3 candidates and comparable total votes.
I am not saying that the result is not anamolous, just that it is very easy to conclude that it is for the wrong reasons.
So you say. Expert testimony says otherwise. Jakob Nielsen:
Nielsen doesn't go so far as to say that this is specifcally what cost Gore the election, but with 19K incorrectly filled out ballots in two counties, I'd say it's a pretty fair guess.
Additionally, from Dan Bricklin:
Nineteen thousand. People with poor vision, people who received incorrect sample ballots. It's obvious that the statistical anomoly is there, especially when graphed. So rather than grousing about how dumb people are, why not design a ballot that doesn't skew the result?Ehm yes...thank you, that made it even more clear to me that the Electoral College system is totally unfair, I would even go as far as saying that it's NOT democratic at all. People should choose the president, not states.
:)
That's right, it's not democratic. It's not supposed to be. We are not a democracy, we are a "republic". The rational behind the EC was to have the states elect the president, with the states using the popular vote to decide how they voted. This also prevented the "mob rule" problems inherent in true democratic governments.
I find it humerous that the democrats that were on TV last week expousing the virtues of an electorial college system (when it was assumed that Gore would win the EC, but lose the popular vote) are now calling for it's removal and complaining how unfair it is
Finkployd
I would caution, however, that I've not run a T test or a chi-squared test or any other measure to see if this is really an outlier. It looks like one to me. However, with so few data points, it is hard, statistically speaking, to know for sure if it is an outlier, or just an unlikely, but statitically insignficant event. Mean and standard deviation do not tell the whole store.
I know I have a bias here. My personal solution to all the problems would be to apportion the invalid votes. If there's a vote for Bush and Buchanan, each gets 1/2 a vote. Add up the numbers, round down to the nearest whole vote and you are golden.
I've also seen references to Florida law that specifically states that ballots have an "X" on the right side of the name of candidates only. Don't know how true that is, but that would make these ballots illegal. If they are illegal, then the courts would have to deside what the most appropriate legal remedy would be. I don't envy them that task.
It would also would like to see a statistical analysis of the delta in the votes as a percentage of voters done. It seems on its surface that the 700-odd votes that were missing in the original count seems high for a county the size of Palm Beach. Almost all the other country results seem to be about what I'd expect. But all the counties aren't in yet, so it is hard to know for sure.
I'd also compare this year's results with prior years that Buchanan ran to see if there are any statistical differences between the two.
There are Lies, Damn Lies and Statistics :-)
I don't see a problem if there was 100% voter turnout. Isn't that what this country is about?
I also have a big problem with people dismissing the elderly or anybody else that got confused. They aren't so feeble minded to not know how they wanted to vote. In an ideal world they would vote in a way that made it clear that were little or no room for errors. Having helped the elderly on many occasions, I know they need special attentions that younger folks don't need. On the other hand, younger folks tend to need other things the elderly don't so the accomidations made to the elderly aren't that special all things considered.
Another mathematical fact. Right now the electoral college gives smaller states more power because they have proportionally more votes than their large bretheren. If we subtract out this bias by subtracting 2*states one from each candidates total, one see that Gore wins with 220 votes (219 needed in this hypothetical situation).
Finally, damn this is a barn burner.
Point of fact - the alleged confusion is supposed to arise due to the placement of the Democrat and Reform party punches, not the Republican one (which was pretty clearly located first on the ballot). Since the ballots in question were used in the West Palm Beach area, it might be useful to check how the numbers broke down there. Acc ording to CNN:
Since the original margin between the Bush and Gore was only 1784 votes, I'd call that signifigant.
Additional evidence that the ballots may have caused widespread confusion, from the same article:
So, another comparison, this time of the two-selection error rate on two different parts of the ballot. More than 19,000 voters selected two presidential candidates, which is more than five times as many as made the same mistake for senators, and more than ten times as many as Bush's lead statewide.
Say what you want to about Daley, Jackson, Mfume, and whomever else you feel is a little too leftward leaning for your tastes. But the numbers do tell a story here.
Look, the first problem is not with the pages alone. Its how the pages lined up with the holes. Check out the mock image I threw together. The black dots are where holes might have been if there was play in the ballot book. The light grey dots is where they should've been (and where the one news photo I've seen shows them).
The stories from people just don't match that perfectly aligned news photo. But if you consider that the holes could've been shifted (really, the book would've been what shifted, but for all intensive purposes...) you can see how it could be confusing. One lady said she saw a dot RIGHT NEXT to the word Democratic, not the separating line above -- that turned out to be the Buchanan hole.
But there is another problem people have cited. If you go in there thinking only of two candidates, Bush and Gore, you look only for those two. In English, we read left->right, top->bottom, left_page->right_page. You could easily say "First one is Bush, so that's teh first hole... Second one is Gore so that must be the second hole!". Why read further? Why examine the right page at all when your candidate isn't over there? Remember, your hindsight is 20/20, so don't be so quick to label these people as "stupid" or you might be so yourself!
The fact is, the ballot was redsigned on to two pages so they could use bigger fonts to make it easier to read. That was a nice idea. But, they never evaluated what the other ramifications were.
Furthermore, the sample ballots sent out to the voters, etc. were paper books! There were no holes in there, no gap, etc. It was not what the ballot actually was. No one could've guessed.
1) Print out the alleged bad ballot.
2) Now place it horizontally about chest to collarbone level and at half an arms length away.
The arrows seem a lot closer together!
If it is like my poll the holes are physically under the ballot book by nearly the distance the holes are separated from each other. This means that unless you can look at it from above, which is not possible if you are short and the ballot and book are in fixed positions, again like at my poll, then the arrow appears to point between two holes.
In other paper ballot elections I attended, the ballots and books were designed so that unneeded holes were covered and there was never any doubt which holes to punch.
The palm beach ballot design my seem ok at first glance, but in use it would be quite ambiguous. This also explains why it was initially approved by the dems.
My ballot had similar problems but only on pages that were for the reelection of judges. Some other issues had several places between them. I noticed when i got to the right half of the page and the punch was already punched. This is a bigger problem for me since i am short.
--- If you don't want to know the answer, don't ask the question.
that made it even more clear to me that the Electoral College system is totally unfair
Unfair in the sense that 4 populous states (CA, FL, NY, and TX) can't impose their will on the rest of the country, yes. You have to have AT LEAST a dozen states to win under an electoral system, but under a popular system you can basically promise every federal dollar to those four states and the rest fo the country can go to hell...
---------------------------------------------
Recursive: Adj. See Recursive.
THIS ELECTION MUST BE RESOLVED WITH A TAG TEAM STEEL CAGE MATCHUP OF TITANIC PROPORTIONS!
Gore vs. Bush in a steel cage. Cheney and Lieberman waiting for the tag. Both houses of Congress on the sidelines threatening to turn the match into a full-on bicameral brawl!
There has never an opportunity like this before, and there probably won't be one again.
What are you waiting for?
Bruce
Bruce
You are the real Bruce Perens.
In the Republican primary in that exact same area, thanks to support from a relative, Buchanan got up to 8,000 votes in 1996. In a primary, where a very small number of republicans actually turn out compared to election days. Of course, that was way before he left the party, but it is more than reasonable that he still has between 1/3rd of the supporters that he did back then. Saying "this is wrong because every other country voted differently." is complete bull. And, as most of us who's visited know, you never visit Palm Beach without seeing something odd. Anyway....
I would like to remind everyone that the electoral college works. Just because New York and California really really want Gore to win doesn't mean that the rest of the country wants what Gore represents. Imagine if the EU existed during Hitler's rise to power, and Nazi Germany dominated the popular vote for the elections to president of the EU. This is all hypothetical, but I'm simply afraid that a lot of people don't understand the power balancing that the electoral college brings.
Most of the Democrats who were in favor of the electoral system last week are still in favor of it (for example, this guy named Gore who you may have heard of -- still in favor of the electoral college, as well as all his publicly speaking advisors).
---------------------------------------------
Recursive: Adj. See Recursive.
http://www.howstuffworks.com/ ele ctoral-college.htm
--
Wooden armaments to battle your imaginary foes!
Ok, what have we learned from Campaign 2000 - First, the media is irresponsible. The early projection of Florida was idiocy at its finest. This election was and is still too close to call, yet the media egos want to be the first one to call the winner, as if anyone cares who called it first. I think the media needs to examine the use of exit polls, and also needs to get back to reporting raw numbers.
So what you're saying is that the media called "First post!" and then got moderated down?
Think of it as a UI design problem.
I had a similar problem with a UI a did for a Palm Pilot application that had to do with spacing of labels in closely spaced checkboxes (you don't have lots of room on a pilot). The users complained that it caused lots of problems, even though if you looked carefully at it, it was clear which checkbox belonged to each label and the users worked with the application every single day.
Naturally, I had to change it and split the form across multiple pages.
It's also important to note that many voters who made a mistake were forced to submit their ballot once punched. The poll workers did not allow them to destroy their ballot and get a new one.
So, you have a bad user interface that gives you one chance to get it right.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
What's slightly more disturbing is that the article goes on to say:
Is it just me, or does anybody else see this as a significant problem? Especially with the outcome of this election hinging on Florida's vote.Anyways... The ballots in Miami-Dade county should stand by law because all challenges to the ballot format must be made before the first vote is cast and ALL voters and campaigns in that county received an instructional copy of the ballot weeks before the election. That leaves outright fraud( forged absentee ballots, dead people voting, etc ) and the suppression of republican voters in the west(due to the early miscall of Florida going to Gore) as the most likely challenges.
I think a court ABSOLUTELY must not allow a revote. It would be unfair for one county or one state to have the power to elect the president knowing the status of the rest of the vote before walking into the booth. It also wouldn't be fair for a court to statistically modify the vote of Miami-Dade county based on demographic or other information as the votes were cast and if people were truly unsure of their vote, they could have asked for help from the officials at the polling center.
If there isn't a decisive winner, I think the ONLY fair thing to do, and this may rise to the level of a decision by the Supreme Court, would be to throw out the ENTIRE 25 electoral votes in Florida as if they had gone to a different candidate. Because neither man received a majority of the electoral votes(270), the Constitution says it is up to the newly elected House of Representatives to select the President. It seems the only way to ensure the rule of law is obeyed.
Don't leave your mind so open that your brain falls out. Don't close it so much that you cut off the blood.