Comments on FiveThirtyEight: Politics Done Right: 1-in-10,000,000

酒店經紀人, 菲梵酒店經紀, 酒店經紀, 禮服酒店上班, 酒店小姐兼職, 便服酒店經紀, ...

2009-10-03T04:34:45.921-04:00

酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店經紀,
酒店打工經紀,
制服酒店工作,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
酒店經紀,
菲
梵,

情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,情趣用品,...

2009-01-17T05:07:00.000-05:00

Surely Proportional Representation is the answer t...

2008-11-06T10:39:00.000-05:00

Surely Proportional Representation is the answer to low turnout?

I estimated the smallest possible share of the popular vote the winner of the US election could have, makes interesting reading:

http://roguemovement.weebly.com/2/post/2008/11/so-why-didnt-you-vote.html

In Australia you HAVE to go to vote, however you D...

2008-10-29T09:22:00.000-04:00

In Australia you HAVE to go to vote, however you DON"T HAVE to cast a vote.
In federal elections you turn up ID yourself, then are given two ballot papers. Then it's up to you to vote or not.
Informal voting is rare and voter fraud is (almost) non-existant.
The only person you register with is the Electoral Commision, proving where you live (when you move).
The fine for NOT voting is around US$400. -ms
My word is 'ditymp' which, of course, I read as "DIRTY Member of Parliament (MP)"

Actually, your individual vote is highly likely to...

2008-10-28T23:23:00.000-04:00

Actually, your individual vote is highly likely to be critical. Here is the scenario:

Imagine that in a particular state 1,000,000 votes are cast, and Obama wins 550,000 to 450,000, a margin of 100,000 votes. He would have won anyway with 450,001 vote, so 99,999 of his votes were totally irrelevant. However, without them, every single one of the 450,001 votes was essential to his victory. Therefore, Obama got 450,001 absolutely essential votes and 99,999 votes with no effect.

Since you can't know which group you were in, all you know is that there was a 78% chance that your individual vote was absolutely critical to the Obama victory, and a 22% chance that it was meaningless. And that would be in a state with a 10% spread; most swing states will be a lot closer than that.

I defy anyone to refute my logic. So go out and vote! The entire election depends on it.

@Larry AmpYou beat me to it - this graph is a perf...

2008-10-28T22:58:00.000-04:00

@Larry Amp

You beat me to it - this graph is a perfect candidate for a logarithmic scale. (though perhaps they thought that would make it less emphatic...some readers might not understand how to read one)

I would argue that even those who live in states t...

2008-10-28T17:38:00.000-04:00

I would argue that even those who live in states that are definintely are going to go one way or another (ie. NY or UT (where I live)) should still vote. Although there is a zero percent chance that my vote (already cast) will determine the election, it will help to give moral weight to Obama if the electoral college is closer than predicted. Further, national races are not the only races, and I would argue that school board, mayor, and state races impact your life more than national races do.

Additionally, local races are how state politics undergoes evolutionary change. Although I live in the reddest state in the union, my mayor, my congressman, and my city council are now all democrats (not true when I moved here 5 years ago).

Utah has the lowest voting participation of ANY state, simply because I think most people feel that the result is pre-determined. It is in this sort of environment that we have room for change.

We also have the best abbreviation for any municipality. SL, UT.

I don't buy the Chris Matthews theory about the In...

2008-10-28T15:39:00.000-04:00

I don't buy the Chris Matthews theory about the Indiana Governor's race. Jill Long Thompson hasn't been doing nearly as well in polls there as Obama has, and neither has Montagano in that 3rd district race against Souder. Perhaps narrower margins of defeat for these two candidates will be correlated with an Obama victory next Tuesday, but I think Obama stands a far greater chance of winning Indiana than Jill has of unseating Mitch Daniels.

Gelman assumes that the probability of casting a d...

2008-10-28T15:19:00.000-04:00

Gelman assumes that the probability of casting a decisive vote is proportional to 1/n where n is the number of voters. He bases this on some supposed historical data, but this is quite controversial, and I do not subscribe to his view. See footnote 3 of his paper. if you do not agree with his assumption, the odds of casting a decisive vote or miniscule.

"3. A crucial part of our analysis is the recognition that p is of order 1/n (see the Appendix
for discussion and references on this issue). For example, in his review of rational choice
models of voting, Dowding (2005) refers to models of social utilities – such as
considered here – as the ‘B-term solution’, but dismisses such models based on a mistaken
belief that the probability of a decisive vote is as low as 10^90. In fact, 10^7 or 10^8
are more reasonable values for US presidential elections (Gelman, King, and Boscardin
1998) with much higher probabilities for many congressional races (Mulligan and
Hunter 2002). Models for the probability of tied elections can get complicated (see
Gelman, Katz, and Tuerlinckx 2002) but the extremely low estimates cited by Dowding
seem implausible, given that there have been several very close presidential elections
in recent decades, as well as over 500 congressional elections in the past century that
were decided by less than 1,000 votes. The probability of a decisive vote is low but is
clearly distinct from zero, if multiplied by a benefit term that is proportional to the size
of the electorate."

Thank you for that paper nate. Finally a model for...

2008-10-28T15:01:00.001-04:00

Thank you for that paper nate. Finally a model for voter behavior that could accurately explain turnout.

@Troylots of them. Essentially, any of the cases w...

2008-10-28T15:01:00.000-04:00

@Troy

lots of them. Essentially, any of the cases where Obama gets Kerry+Iowa+New Mexico and just needs one more state - in that situation at least one of Ohio, Colorado, Virginia, Nevad, Florida, Indiana, North Carolina & Missouri (in no particular order) is virtually guaranteed to be tight.

A scenario that might be interesting is the number...

2008-10-28T14:36:00.000-04:00

A scenario that might be interesting is the number of simulations where there are enough states in doubt that neither candidate reaches 270.

I'm not saying "too close to call." I'm looking more at recounts and such. Is there a situation where there could be an OH, FL, NM (to take some example states from past elections) combination that will leave us hanging for a while?

I read that paper on rational social voters, and t...

2008-10-28T14:08:00.000-04:00

I read that paper on rational social voters, and there is actually a mathematical mistake. They say that your probability of being a pivotal voter is proportional to 1/n. This is not quite true. Say that the polls say that the election is exactly 50/50. There are then three sources of error on election day: sampling error in the poll, bias in the poll, (538ers are familiar with those two) and, last of all, sampling error in the election itself. This is the unavoidable randomness of an election - the chances that people just can't make it to the polls, the chances that people randomly saw this ad as opposed to that ad the night before, stuff like that. The chance of being pivotal is proportional to 1/(sum of 3 kinds of error). (Actually, the root-sum-square, not the sum.) And the sampling error in the election itself is not propotional to N, it is proportional to sqrt(n). This error is, generally, negligible compared to the other two, but not always.

The more that you believe that an election is truly tight, the more your chances of being pivotal start to look a bit like 1/sqrt(n) instead of 1/n. When you multiply by a social benefit proportional to N, it becomes not just marginally rational to vote in such an election, but MASSIVELY so - on the order of sqrt(n)*B where B is the average individual benefit. If N is in the millions, sqrt(n) is in the thousands. So if you know that you're heading for a Florida 2000-type situation, you would be perfectly rational to spend many thousands of dollars to ensure your vote was counted, or, equivalently, hours of effort to prevent a tiny 1% chance of your vote not counting.

"Professor Douglas Rae wrote a short book analyzin...

2008-10-28T14:03:00.001-04:00

"Professor Douglas Rae wrote a short book analyzing the electoral college and individual voting. I can't remember the title and it isn't listed on his short curriculum vita page at Yale, but it should be available with a bit more extensive research."

So then he can explain also why no one votes for mayor in New Haven, and why John Daniels lost? Great.

But seriously, people clearly don't vote because they think their vote matters, even if they tell themselves that. It's an expressive act, a way of participating in something larger than oneself, like going to a protest, a vigil, a church service — a communion with civil society.

The olgiest lacked lacked ungshest, perhaps.But wh...

2008-10-28T14:03:00.000-04:00

The olgiest lacked lacked ungshest, perhaps.

But what does unperyon mean?

Salutations, gentlefolk,Note that after the 2006 e...

2008-10-28T14:00:00.000-04:00

Salutations, gentlefolk,

Note that after the 2006 elections the Pennsylvania House of Representatives was split 102 D / 101 R.

The 'tipping point' race - in the district around West Chester, in exurban Philadelphia, was decided by about 25 votes.

'oligiest'
(my verifyword today - a greasy oligarch ?

typo: it should be weighting not waiting...ungshes...

2008-10-28T13:58:00.000-04:00

typo: it should be weighting not waiting...

ungshest.

reedchex,It seems that opinion pollsters are inclu...

2008-10-28T13:57:00.000-04:00

reedchex,

It seems that opinion pollsters are including those who have already voted. In fact, given the Democratic tilt of those who have voted early, the polls are likely to go up, since this group will be given a higher waiting (the likelihood that they will vote, of course, is 1.00, as they have already done so).

My verification word (reflects badly on me): pitifuli

people are missing the big story from the Pew poll...

2008-10-28T13:55:00.001-04:00

people are missing the big story from the Pew poll

Obama's percent is similar to other polls but the Pew's results are different, in part, due to 3rd party candidates -- they now have a very significant 4% of the vote! (at McCain's expense) -- this result would seem to be very much in line with recent current events

i'm sure Pew's results being different are correlated to the fact that many other polls don't list 3rd party candidates as an option

But Pew's approach to the $64,000 questions seems to be much better than all others. They phrase it as close as possible to an actual ballot and they even adjust which 3rd party candidates you see based on your state (see below from questionnaire)

NOTE: BARR AND/OR NADER INCLUDED ONLY FOR RESPONDENTS IN STATES WHERE EACH
IS ON THE BALLOT.
Q.3 If the presidential election were being held TODAY, would you vote [READ AND ROTATE
REPUBLICAN AND DEMOCRATIC TICKETS FIRST AND ROTATE BARR AND NADER
TICKETS LAST] for the Republican ticket of John McCain and Sarah Palin, for the Democratic ticket of
Barack Obama and Joe Biden, for the Libertarian ticket of Bob Barr and Wayne Allyn Root, or for the
Independent ticket of Ralph Nader and Matt Gonzalez?

I'll admit Gallup concerns me slightly today. I r...

2008-10-28T13:55:00.000-04:00

I'll admit Gallup concerns me slightly today. I recognize a big Obama day from Friday polling dropped off. I'm also one to downplay virtually every tracker that appears to be an outlier. There could be some legit tightening though. We'll have to see how the next couple of days play out. I've watched the polls very, very closely over the last 90 days or so. Gallup is often times a leading indicator. We have two days of weekend polling in there for Mccain, but I'd still like to see no carry through on this potential momentum for McCain in gallup. It's actually a little scary. I'm not sure I have any idea who would be voting for McCain right now. to make it swing anywhere near even. The state polling and early voting is still overwhelmingly for Obama.

Unless I'm crazy, wouldn't it stand to reason that...

2008-10-28T13:50:00.001-04:00

Unless I'm crazy, wouldn't it stand to reason that since Obama has a sizeable lead in early voting, that the polls from here on out would tend to favor McCain? If tracking polls only poll likely voters who have not voted yet, and it appears lots more of Obama supporters have already voted than McCain, it seems to me that there would be disproportionately more McCain voters in the tracking polls these next few days.

Or am I wrong?

ps. the chances of a log version of the graph are ...

2008-10-28T13:50:00.000-04:00

ps. the chances of a log version of the graph are pretty slim. It would take a lot of paper just to write the zeros in the probability of a Utah vote being decisive. It's essentially the odds of 2/3 of the Republicans, and none of the Democrats, in Utah getting sick on the same day. Since a log graph needs space for each 0, it would not fit on this web page :).

Obama has 291 electoral votes locked up:Kerry stat...

2008-10-28T13:48:00.000-04:00

Obama has 291 electoral votes locked up:

Kerry states + NM, IA, NV, VA, CO

+ another 47 likely OH, FL

+ another 37 pure tossup NC, IN, MO

+ another 38 possible GA, MT, ND, WV, AZ, NE-2

He'll end up with between 291 and 413 electoral votes.

Scottking, you could vote for McCain in secret and...

2008-10-28T13:43:00.000-04:00

Scottking, you could vote for McCain in secret and still say you voted for Obama. After Watergate, no one could be found who admitted to voting for Nixon in his 1972 landslide win.

J - The following are the must-win states for McCa...

2008-10-28T13:30:00.000-04:00

J - The following are the must-win states for McCain*, with their associated poll-closing times. Within time groups, I have ranked them by predicted margin:

7:00 - VA, IN, GA
7:30 - OH, NC
8:00 - CO, FL, MO

You will have to decide for yourself whether VA is likely to be called before OH or CO; my own feeling is that it is likely.

* Note that NH (8:00) + NV (10:00) is also a unique path to defeat for McCain, as are a couple of other less likely combinations involving ND and MT.