Article

2006-11-06 Home Front: Politix

NY Times Editorial: Close Election? It's Do Over Time!!

Archived material is restricted to Rantburg regulars and members. If you need access email fred.pruitt=at=gmail.com with your nick to be added to the members list. There is no charge to join Rantburg as a member.

Posted by Swamp Blondie 2006-11-06 13:10|| || Front Page|| [2 views since 2007-05-07] Top

#1 I fully expect every close election to be sued to death by the dems.

Posted by DarthVader">DarthVader 2006-11-06 14:54|| 2006-11-06 14:54|| Front Page Top

#2 Interesting irony that the NYT mentions Bush/Gore and the Washington Gov's race in 2004. Neither race would have been a controversy were it not for the thousands of illegal alien votes, dead votes and felon votes that weren't thrown out. In Gregoire's case, investigators estimated that over 5000 votes were illegal or illegitimate. Get ready for another round of massive donk voter fraud tomorrow. Anyone who challenges or questions it will be tarred as a racist, a bigot, an extremist right-wing operative or a 'disenfranchiser'.

Posted by mcsegeek1 2006-11-06 15:14|| 2006-11-06 15:14|| Front Page Top

#3 After the circus in Florida the call for election reform should have centered IMO on a national standard voting process. One that used a mechanical counting device backed up by a paper trail. up until the last primary here we were still using the old lever machines. Easy to use. Reliable and unhackable. Tamper proof? Well I do suppose it could happen but enough observers are present when the tally is taken after the polls close the chances of rigging an election are slim to none. As to voter fraud. IDs should be required. It won't stop all fraud but should cut it down.

Posted by Cheaderhead 2006-11-06 15:27|| 2006-11-06 15:27|| Front Page Top

#4 DV, I agree and I would not be surprised, if it's close enough, to see the donks litigate in enough districts to prevent a Congress from being organized in January.

Posted by Nimble Spemble 2006-11-06 15:31|| 2006-11-06 15:31|| Front Page Top

#5 My mom pointed out that there will be *voting* tomorrow, but possibly very little *electing*, as so many of the close races will be decided after extended recounts and even court time.

Posted by Seafarious">Seafarious 2006-11-06 15:32|| 2006-11-06 15:32|| Front Page Top

#6 I remember reading someone predicting that we won't have closure for months after the election. I'm sure that will be true unles enough of us get out and vote so that the vote is not close enough to litigate.

VOTE. Remind your friends (who vote Republican) to vote. Vote like your life depends on it.

Posted by anon 2006-11-06 16:55|| 2006-11-06 16:55|| Front Page Top

#7 It's way past time that voter fraud be considered what it is, nothing less than full blown treason. It should be punished by death. After the entire democratic party gets executed for vote fraud, then maybe they'll create a real party.

Posted by Silentbrick">Silentbrick 2006-11-06 17:02|| 2006-11-06 17:02|| Front Page Top

#8 What does this mean in actuality? In terms of a two-candidate race in which each has attained around 50 percent of the vote, a 1 percent margin of error would be represented by 1.29 divided by the square root of the number of votes cast.

Huh? What!? Where the hell does this come from? It's not even close to statistical probability so far as I know? In statistical probabability, as I learned it, you take the total number of entries and divide by the number for each side to give you a result.

For example,

In a sample of 100, you have 53 for and 47 against. 53/100=53% for; 47/100=47% against. Obviously, the 53% for wins.

There's no adding 0.29 (and why 0.29?, why not 0.30 or 0.50 or whatever) and then dividing by the square root of the number of votes cast! None! That's an example of manipulating the numbers to give you a result you want.

This method won;t even give you a median statistical result (which is also easy to calculate). So, maybe he's trying to use a mean squares result - but that's also the wrong technique so far as I understand.

There are lies, damn lies, and then there are statistics.

Unless someone can tell me where I'm wrong then this whole calculation is a basis for massive fraud.

Posted by FOTSGreg">FOTSGreg 2006-11-06 18:52|| www.fire-on-the-suns.com]">[www.fire-on-the-suns.com] 2006-11-06 18:52|| Front Page Top

#9 Reminded me of nothing so much as a question my father used to pose, If a third of four is 2 and a half of three is eight, how long are two hatchet handles and a little bit?

Posted by Nimble Spemble 2006-11-06 18:58|| 2006-11-06 18:58|| Front Page Top

#10 Oh! And another thing, dimwit!

The US Presidential election is not decided by the popular vote! We use something called the Electoral College System which is a device to prevent majority one-party areas from exercising virtually dictatorial power over the rest of the nation (ie large population blue-state areas from deciding and controlling who gets to be President despite what the rest of the country says!).

But I guess you don't agree with the Founders on that particular apparatus anyway, right?

No, no. The tyranny of the majority is always the right way to go. Power to the people! Right, dude?

Posted by FOTSGreg">FOTSGreg 2006-11-06 18:58|| www.fire-on-the-suns.com]">[www.fire-on-the-suns.com] 2006-11-06 18:58|| Front Page Top

#11 The formula is correct, but the description is not.

For n sample data points randomly chosen from a normally distributed population, the margin of error at a 99% confidence level is approximately equal to 1.29 divided by the square root of n.

Key assumption: that the population is normally distributed, that it is a set of discrete points (probability distribution rather than a continuous probability density function) and that the sample is truly randomly drawn from the total underlying population.

The writer seems to have mistaken "99% confidence" for "1% margin of error". The formula actually says that the margin of error is going to be less than or equal to that formula, 99% of the time.

Posted by lotp 2006-11-06 19:04|| 2006-11-06 19:04|| Front Page Top

#12 Or, to put it another way, if our sample indicates that candidate A has 47% support, with a margin of error of 2% (99% confidence level), that means that the true level of support for A in the overall population will be somewhere in the range of 45-49%, 99% of the time.

The '99% of the time' is an odd thing to say about a one time event like an election -- the underlying statistics assumes we are drawing a simple random sample from the whole population and could do so again and again.

So the whole thing gets collapsed (somewhat misleadingly) into saying that we are 99% confident that A's overall support is in the 45-49% range. But what the numbers really mean is that in 99% of polls taken with these results in this population, A's true level of support will be between 45 and 49%.

Posted by lotp 2006-11-06 19:09|| 2006-11-06 19:09|| Front Page Top

#13 lotp

Thanks for the correction! Your explanation I understand, but what was being said had me completely confused (I'm not a statistician, but have done some statistical work).

Posted by FOTSGreg">FOTSGreg 2006-11-06 19:21|| www.fire-on-the-suns.com]">[www.fire-on-the-suns.com] 2006-11-06 19:21|| Front Page Top

#14 You're welcome. I'm not a statistician either, but I have to use both normal and Bayesian statistics in my research. It takes a while to wrap ones mind around this stuff -- our brains are wired for quick overally pattern matching, not these abstractions.

Add in a writer who's pretty clearly math challenged and the resulting article is a mess.

Posted by lotp 2006-11-06 19:33|| 2006-11-06 19:33|| Front Page Top

#15 I believe what a confidence interval of ±2% says is that if we sample the population with the same size sample repeatedly, 99% of the samples will return a result that include the true value witin the confidence interval and that 1% of the samples will return a result that does not include the true value within its confidence interval.

Posted by Nimble Spemble 2006-11-06 19:40|| 2006-11-06 19:40|| Front Page Top

#16 Yes, but that description while common is a bit misleading.

A confidence interval can never be specified without also associating a confidence level with that interval. It need not be 99% and usually isn't, in most analyses, since to achieve that level requires significant sample size and good reason to believe that the sample is truly randomly drawn.

It's the failure to specify parameters like that that, combined with other egregious omissions, that leads applied math people like my sig other to roll their eyes at reported poll results. ;-)

Posted by lotp 2006-11-06 19:48|| 2006-11-06 19:48|| Front Page Top

#17 Only on Rantburg, does a call for electoral cheating turn into a debate on statistical analysis.

The people here are scary...but that's okay. I can live with it.:) I want all the scary people on my side of things.

Posted by Silentbrick">Silentbrick 2006-11-06 22:03|| 2006-11-06 22:03|| Front Page Top

23:55 twobyfour
23:39 Zenster
23:30 elbud
23:27 Flomoter Ulolush5791
23:07 bunyip
23:05 Atomic Conspiracy
23:04 Barbara Skolaut
22:59 Barbara Skolaut
22:56 Barbara Skolaut
22:47 Seafarious
22:47 Atomic Conspiracy
22:41 Zenster
22:38 Zenster
22:34 Zenster
22:28 Zenster
22:28 Richard Aubrey
22:28 Remoteman
22:25 Zenster
22:21 anon
22:21 Alaska Paul
22:15 Zenster
22:06 Alaska Paul
22:03 Silentbrick
22:00 trailing wife