• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

So how come no one ever executes a vote audit?

Status
Not open for further replies.

Dilbert

Member
Something lost in all of the election coverage has been the fact that all of the exit polls were VERY wrong throughout the night -- 2-4%, which is quite significant given the eventual margins of victory in most key states. There are a couple of possibilities:

1) The mathematical models used to match exit polling results to final returns are broken. If so, then why has this election been so different from past elections?

2) There was a deliberate strategy executed to game exit pollsters to influence the election. That would be an interesting theory, but not very likely -- how would you communicate that strategy to the large number of people required to execute it without compromising the secrecy that would be needed?

3) There was some systematic failure in translating the intent of voters to the actual balloting. As seen in the 2000 recount, looking over those ballots revealed many which were physically flawed in some way and might have been miscounted by machine, but which evidenced intent. (aka "hanging chads") There were also stories from 2000 about voter intimidation and fraud.

So, all of this leads to a much larger question: How come no one ever audits our election results and processes? We seem to hold the contradictory views that allegedly "every vote matters," yet knowingly allow a flawed voting system without any oversight or consistency.

Where is the analysis of the new voting methods used in this election? In particular, electronic voting has known insecurities, and VERY few places provided a paper copy of the transaction to the voter. Yet, strangely, no one has even THOUGHT about whether the system was used successfully or whether all of those votes were valid. Doesn't that bother anyone?

There were also various locations which moved away from punchcard voting to other methods, and I haven't heard a word about whether the new systems were any more or less error-prone. There were also the new additions of provisional balloting and the highly controversial "challengers" in certain areas. Again, who is looking at these issues?
 
what's it matter? at least you have the money to move to canada. I DON'T.

and i guess the reason that there are no audits is because there might be a chance that.......you know what? it doesnt matter, as long as there is an electoral college we will always be at the whims of disembodied entities who vote the way they wanna.....
 

ManaByte

Gold Member
-jinx- said:
Something lost in all of the election coverage has been the fact that all of the exit polls were VERY wrong throughout the night -- 2-4%, which is quite significant given the eventual margins of victory in most key states. There are a couple of possibilities:

CNN explained the exit polls very well last night around 12:30pm PST. They went into extreme detail on how exit polls are a sample of the vote and no where near 100% of the vote. They explained that the best use of exit polls is to determine the mindset of voters and why they were voting as opposed to determining a winner.
 

Dilbert

Member
ManaByte said:
CNN explained the exit polls very well last night around 12:30pm PST. They went into extreme detail on how exit polls are a sample of the vote and no where near 100% of the vote. They explained that the best use of exit polls is to determine the mindset of voters and why they were voting as opposed to determining a winner.
Statistics 101: A large enough random sample ought to represent the entire population. The fact that almost ALL of the exit polls were strongly for Kerry in swing states is not probable, unless there was another bias operating behind the scenes.
 

Phoenix

Member
Because the margins are large enough that it would be a pointless endeavor and just waste time, tax payer money, and harm the economy in the process.
 

ManaByte

Gold Member
-jinx- said:
Statistics 101: A large enough random sample ought to represent the entire population.

That goes against the "every vote counts" stuff people were screaming in the locked election thread last night.
 

Dilbert

Member
Phoenix said:
Because the margins are large enough that it would be a pointless endeavor and just waste time, tax payer money, and harm the economy in the process.
How can you say that the margins are "large enough" when you don't know the uncertainty associated with the voting process?
 

Phoenix

Member
-jinx- said:
How can you say that the margins are "large enough" when you don't know the uncertainty associated with the voting process?

I've taken enough probability and statistics to know that uncertainty is in accordance to probability. The probability that anything would change is marginal.

Go ahead and recount them. If you think that the results will change by 150K votes, you're nuts.
 

Doc Holliday

SPOILER: Columbus finds America
I would love to see how bush did in places where electronic voting was used versus old fashioned booths.
 

fart

Savant
we assume the margins are large enough that nothing would change. we trust our information sources to an amazing and wholly unwarranted extent.
 

Dilbert

Member
Phoenix said:
I've taken enough probability and statistics to know that uncertainty is in accordance to probability. The probability that anything would change is marginal.

Go ahead and recount them. If you think that the results will change by 150K votes, you're nuts.
You're entirely missing my point.

I have long maintained that the election process in this country does not have the precision to support an "every vote matters" philosophy. There are THREE outcomes to an election: win, lose, or statistical tie.

If you are measuring something with a ruler which is marked by the inch, and you tell someone that a caterpillar is "3.476 inches long," you've made a mistake of overprecision, given your measurement technique. The best you can probably do in that situation is to say, "It's approximately 3.4 inches," or, "It's somewhere between 3.3 and 3.5 inches," or, "It's somewhere between 3 and 4 inches."

Likewise, in any real-world election process, it is probably unavoidable that errors will be made in measuring the intent of the electorate. The best that we can hope to do is to a) remove all potential sources of systematic error and b) characterize the amount of random error so that we know when we have a statistical tie. Systematic error can come into play maliciously with hackable systems like Diebold or voter challenge campaigns, or accidentally by not synchronizing polling hours or methods. Random error will continue to exist...but how big IS it? Is a lead of 500 votes sufficiently large to say that it's outside the margin of error? 1000 votes? 5000 votes? We don't know until SOMEONE actually does the analysis.

Do I think this election is reversible? Probably not, but it would be nice to know that SOMEONE is looking into possible issues prior to the inauguration. The bigger issue is how we fix all the known problems with how we vote, and how to get on some kind of stable mathematical ground for the results. For a country which supposedly prides itself on having the best voting system in the world, we seem peculiarly happy settling with known problems...which is flat out dumb.
 

Phoenix

Member
There is an election committee in the United States that has measured the variability of using various methods to count ballots and has approved the process because it falls within a certain tolerance. This work HAS been done before. Its not like this is the nations first election of first challenge of it. When automated vote counting machines were introduced they had to be accurate within a certain tolerance and that had to be proven. Diebold didn't just roll out new machines and people said 'thank you, new machines - we'll use them'. The system is tested and tested and retested every election every campaign every vote. The tolerances are understood which is one of the reasons that statisticians are part of all the election news coverage teams now.
 

Dilbert

Member
Phoenix said:
There is an election committee in the United States that has measured the variability of using various methods to count ballots and has approved the process because it falls within a certain tolerance. This work HAS been done before. Its not like this is the nations first election of first challenge of it. When automated vote counting machines were introduced they had to be accurate within a certain tolerance and that had to be proven. Diebold didn't just roll out new machines and people said 'thank you, new machines - we'll use them'. The system is tested and tested and retested every election every campaign every vote. The tolerances are understood which is one of the reasons that statisticians are part of all the election news coverage teams now.
So what is the margin of error in our elections? I'd love to read the source materials.
 

Phoenix

Member
-jinx- said:
So what is the margin of error in our elections? I'd love to read the source materials.

The margin of error depends on the type of machine use. The benchmarks list 2% as being acceptable with anything above that being unacceptable.
 
From what I've heard of the electronic devices, they are far too potentially shady to be trusted. They can create "dummy" subsets of votes, and apparently many could be hacked via telephony or potentially through swapped disks that go from the offline voter box to the online compiler. Not to mention that the chief software engineer hired by one of the manufacturers was fresh out of prison off fraud charges. Go figure.

I agree that its extremely odd that the exit polls so contradicted the results.
 

Tenguman

Member
the last 2 elections showed that with elections as close as this, the percent error can range way too high...higher than projected. has a lot to do with polling methods, who's doing the polling, who's reporting the polling, and above all who are being polled. With an election with high turn out like this, it's just too much up in the air.

WHICH IS WHY POLLS SUCK

I didn't even look at them last night. I just kept track of the counted votes.
 

AirBrian

Member
Sincere question:

Wouldn't the early exit polls indirectly favor the stay-at-home moms/dads, unemployed, retired, and students? In other words, because of the very nature of the timing of certain demographic's voting habits, wouldn't that skew the results?

Much like how typically the rural vote counts come in first and then the urban, giving a sometimes "false" lead for Republicans.
 

Phoenix

Member
Ned Flanders said:
F

I agree that its extremely odd that the exit polls so contradicted the results.

Exit polls are a very small set of people in a wade variety of places and not as evenly distributed as the voting populace. There are only roughly 14K exit poll responses as opposed to some 111 million actual voters. Since your exit polls generally don't come from places all over rural America you will find skewing in the exit poll figures. That's why they are just guestimates when it comes to calling states and why some stations (like CNN) took forever to call those states because the polls weren't diverse enough to make a decision. Heck the only reason they called Ohio was when Kerry said he would conceed.
 

CaptainABAB

Member
Phoenix said:

Jinx - that report Phoenix links to should answer your question....

(Remember, this is based on the 2000 election)

"Government cannot prohibit news organizations from irresponsible
political reporting. It cannot bar the exit polls on which networks
largely rely for their early calls of a projected winner. The Commission
notes the body of evidence that has mounted since November documenting
the unreliability of exit polls. The networks now know, from
their internal investigations and from studies by their paid consultants
that exit polling is seriously flawed.
The dirty little secret of the last
campaign was that exit polls conflicted with the actual final results in
many states—and in five specific instances by as much as seven to sixteen percent.

Network officials acknowledge that these exit polls have become more fallible
over the years as more and more voters have refused to participate in them.

The Commission was shocked by reports that network interviewers at polling
precincts have offered tawdry inducements, such as small sums of money or
cigarettes, as enticements to citizens to participate in exit polling. Such conduct
cheapens journalism and creates an unhealthy polling place environment.
The Commission strongly encourages citizens not to participate in exit polling.
If candidates, political parties and election officials actively encouraged voters not to
participate in the exit polling game, it could further erode the credibility of exit
polls and network reliance on them."
 
Status
Not open for further replies.
Top Bottom