What is Digg?58 Comments
- lougoose, on 06/22/2009, -1/+34Next time I rig an election, I have to remember to use a computer to generate random numbers. Not my feeble mind.
- Mitchum, on 06/22/2009, -1/+34Finally, a statistical analysis exercise that I actually enjoyed reading through.
- nshady, on 06/22/2009, -1/+33Science. It works, bitches!
- darknecross, on 06/22/2009, -0/+14Good article, however I'd like it more if they included the numbers they used (or linked to a source), as well as the methods used to arrive at conclusions -- why is one statistic four in one hundred--, or links to the cognitive studies they mentioned.
Why do "News Articles" tend to omit source information? These aren't blogs. - SoulGrub, on 06/22/2009, -2/+13Clever buggers at Columbia U.
- onederwall, on 06/22/2009, -0/+10FTA- "The probability that a fair election would produce both too few non-adjacent digits and the suspicious deviations in last-digit frequencies described earlier is less than .005. In other words, a bet that the numbers are clean is a one in two-hundred long shot."
- j01101010, on 06/22/2009, -0/+8I was expecting them to mention Benford's Law at some point but they never did. Benford's Law:
http://www.mathpages.com/HOME/kmath302/kmath302.ht ...
Basically, some digits are more likely to appear as the leading digit (but also in trailing digits) in any statistical data table. - Barackalypse, on 06/22/2009, -1/+7Sadly, Ahmadinejad is popular in the only places that matter in Iran, with the Revolutionary Guard and with the Ayatollah.
- Aremihco, on 06/22/2009, -2/+8Wow... what a wonderful way to discover a "smoking gun".
- Petestreet, on 06/22/2009, -1/+7This is statistical probability.
The article states that there is a one in 200 chance that these numbers are clean.
They could be clean, but probably not. - inactive, on 06/22/2009, -1/+5IRS auditors use some of the same methods for detecting fraud.
- CaptainRant, on 06/22/2009, -0/+4The key to winning any election is being in charge of counting the votes.
- inactive, on 06/22/2009, -0/+4Well it's not like the US isn't rife with election fraud either.
- rb1248, on 06/22/2009, -0/+3I'm a Democrat, and yes Bush didn't win the popular vote in 2000, but he did in 2004 so he was definitely elected then by all definitions of the word.
- kanojo1969, on 06/22/2009, -0/+3I think this article is more editorial or opinion than article. It's not written by a journalist, but by a scientist. Of course, you'd think that he would be more likely to cite sources, just out of habit, but it doesn't really matter - I bet you could find the data easily enough from the information in the article.
- lolwutpear, on 06/22/2009, -1/+4and in 2008, when the Republicans won again!
wait, what? - SoulGrub, on 06/22/2009, -2/+5Mahmoud?
- gogog0, on 06/23/2009, -0/+2you're an idiot. i say that as an iranian.
- consonance, on 06/22/2009, -0/+2This isn't a research journal; it's a newspaper, so you don't have to cite your sources.
- quamb, on 06/22/2009, -0/+2The Devil is in the Details.
http://www.youtube.com/watch?v=RGbYyV0UMgM - inactive, on 06/22/2009, -0/+2I would also like to see this analysis done on the last five US presidential elections as well as the last several Iranian elections.
- ghatid, on 06/22/2009, -1/+3I'm not, by any means, saying that I don't think it was rigged, but I'm just saying statistics won't hold up in a court of law.
- knubblet, on 06/22/2009, -1/+3but the article points out several irregularities in the vote counts. denying a recount when there is evidence of fraud is denial of democracy.
- Petestreet, on 06/22/2009, -1/+3Well done statistical analysis.
I am impressed. - JBizness, on 06/22/2009, -2/+4Statistics are crazy. If you haven't read Freakonomics, I highly recommend it. 91% of readers liked it. But seriously, they use some interesting statistical tools to analyze data. The only thing I can remember off the top of my head is that you can use Benford's Law to help determine if numbers are being faked. While the article talks about how you would expect the last digit to have an equal chance of being any number 1-9, Benford's Law shows that for many (but not all) lists of random numbers, the leading digit will have an unequal distribution in terms of usage. Specifically, 1 is used nearly 30% of the time, while 9 is used less than 5%. It has limitations of course. For example, if you applied it to batting averages, 9 would essentially be 0 since no one bats that well. Or if you applied it to the age of people who consume alcohol regularly, we would expect the % chance of 1 being the first digit to be lower. any case, when applying such tools, you have to take the outcome with a grain of salt. Since statistics are variable by nature, not every "law" may be applicable. However, they can still be a strong indication that something just ain't right.
http://en.wikipedia.org/wiki/Benford%27s_law - ifruit, on 06/22/2009, -1/+3Organik is no doubt a 9-11 Inside job type of guy.
- j01101010, on 06/24/2009, -0/+1I don't see why not... People have used Benford's law to detect fraud before, why can't it be used here?
- MacroDaemon, on 06/22/2009, -2/+3He was running against Kerry. If there ever has been a democratic candidate that would have been hard to elect, he was it.
- JasonJones99, on 06/22/2009, -1/+2Using some crazy data to tell us something we already know, but still cool at the same time.
- SeanFergusonLF, on 06/23/2009, -0/+1Not just any computer. Make sure you generate "true" random numbers using atmospheric noise like http://random.org/
- inactive, on 06/22/2009, -0/+1200 to 1 isn't really too much of a long shot - but as another suspicious point in a long list of suspicious points, it certainly doesn't do anything to dissuade the likelihood that the election wasn't exactly 'free and fair.'
- SeanFergusonLF, on 06/23/2009, -0/+1Statistics are not crazy. The people who interpret them are.
- SeanFergusonLF, on 06/23/2009, -0/+1I concur. I say that as an American.
- Absorber, on 06/22/2009, -0/+1Never tell me the odds!
- SeanFergusonLF, on 06/24/2009, -0/+1While we're on the topic of faulty cognitive heuristics...
Make a bet with one of your friends. Have them flip a coin just ten times. If a heads or tails comes up three times in a row, you win. If not, they do. - SeanFergusonLF, on 06/23/2009, -0/+1I learned long ago to always look at the source when statistics are published in the news. Although the authors are Ph.D. candidates, political science is nothing like psychology nor statistics and it would be nice to be able to check their numbers.
Even if their math is flawless, p <. 005 is still a possible, albeit quite unlikely outcome. My intuition tells me that since the authors were working with such a large sample size that this probability should be lower.
On a personal note, I'm pretty convinced the election was a sham. It would still be nice to be able to check for myself. - surfacewound, on 06/22/2009, -0/+1Or you could just, you know, not manipulate the last few digits.
- heresy_fnord, on 06/22/2009, -1/+2They need to learn how to rig elections better. They could have learned from our 2000 election split at 49/51 instead of going all out with a high 70/30%. Make it look less obvious...
/s - SpruceCaboose, on 06/22/2009, -0/+1Whether or not he won is not the question. The question is "Was the election rigged?" That is the much more important question, as that one impacts much more than just who will lead the country. It affects the country as a whole, its very foundation that it was trying to solidify.
- darkciti2, on 06/22/2009, -1/+2I'd love to see this analysis done for the 2004 USA general election results.
- SeanFergusonLF, on 06/24/2009, -1/+1Because the numbers being dealt with were very large. Modifying the first digit could easily add or subtract 100,000 people from a province. It's likely the the first digit in the modified numbers were generally unaltered.
For example, let's say 100,000 people voted in a province. The real vote was 45% for candidate A, 45% for candidate B, and 10% for other.
Now let's say we to rig the results to be ~ 60/30/10, but no more than 70/20/10. This means we have only two reasonable choice for our first digits (5 or 6)/(2 or 3)/(1 or 0). So far our vote counts look like:
A: 6_,___ or 5_,___
B: 3_,___ or 2_,___
O: 1_,___ or _,___
As you can see, at this stage the degree to which we can reasonably manipulate the numbers is very small. Interestingly, the logarithmic curve illustrated by Benford's Law has a lot to do with this.
When we choose the second digit, the human element increases. However, since we're trying to approximate a 60/30/10 split, we can't stray too far.
A: 58,___ or 59,___ or 60,___ or 61,___ or 62,___
B: 28,___ or 29,___ or 30,___ or 31,___ or 32,___
O: 8,___ or 9___ or 10,___ or 11,___ or 12,___
With each progressive digit, the number of reasonable options, human decisions, and thus our ability to detect manipulation increases. - SeanFergusonLF, on 06/23/2009, -1/+1Benford's Law wasn't really relevant in this analysis.
- Vulveeta, on 06/24/2009, -0/+0Yes, that is the clue -- forensic study of reputedly accurate numbers will show the interference of a human hand, if such a hand were present in creating those numbers. This is when Political Science does indeed become a science. The study of probabilities, and the understanding of random numbers are brilliant advances in Political Science.
- kanojo1969, on 06/22/2009, -1/+1Congratulations, not only have you failed to understand the article, but you've also produced complete gibberish as an analogy. Each of those has a 1 in 20 probability of occurring but to get both together it's a real long shot; 1 in 123.
Consider yourself a genuine rarity. - stuwanker, on 06/22/2009, -1/+1A lot of the numbers are calculated from straightforward discrete probability. If you'd like, I'd be happy to go into detail.
- JBizness, on 06/22/2009, -1/+1Aaaaaaaaaaaaactually. Read Freakonomics. Statistics can be a very powerful tool. In one example, they were able to use statistics to find teachers who were cheating and giving their students answers to standardized testing. Retests under controlled circumstances supported many of the allegations and those teachers were fired.
http://www.wikisummaries.org/Freakonomics:_A_Rogue ...
It's just a summary, so it doesn't have all the data, but they at least mention some of the things they were able to discover by simply analyzing certain sets of data. And while statistics themselves may not necessarily hold up in the court of law, they can certainly cast enough doubt that a closer look would be justified. - sexyloser, on 06/25/2009, -0/+0Dugg for Star Wars reference
-
Show 51 - 62 of 62 discussions
© Digg Inc. 2009
— Content created and posted by Digg users is