Who's Gonna Win?

[silverscreenselect]

This is an interesting web site that keeps a running update on the chances of each presidential candidate in the general election, if it were held today. The web site uses an average of recent polls, state by state, and then does a Monte Carlo analysis by running thousands of simulated elections and sees which candidate amasses 270 electoral votes more often. Based on their current projections, John McCain has an 80% chance of beating Obama, while Hillary has an 88% chance of beating McCain.

One big caveat is that state by state analysis in a number of states is based on a small number of polls, some of them quite old, and should be taken with a grain of salt. Even this fall, you're not going to see a lot of polling done in places like Wyoming and Utah. The site should be interesting to follow to see how the landscape chances over time however.

http://hominidviews.com

[starfish1113]

Some of these numbers are funky as all get out.

In the Obama-McCain matchup, Obama's % chance of winning Iowa is 99.9% and McCain's % chance of winning Missouri is 99.5%. No way that these two states are that far out of play.

Likewise, in the Clinton-McCain matchup, Clinton has a 99.9% chance in Florida and McCain is given a 95% chance in, of all places, Arkansas.

I guess this has to do with, as SSS said, the small number of polls taken thus far, but those numbers will have to be modified significantly before I take this seriously. It is fun to look at, though.

[TheCalvinator24]

I believe McCain would beat Hillary Clinton in Arkansas.

[starfish1113]

I believe McCain would beat Hillary Clinton in Arkansas.

I think so too, but I don't give him a 99% chance of doing so!

[slam]

Probably an interesting idea to do this, but the results have to be seriously flawed. I didn't look at it in any detail but there is absolutely no way that a major candidate has 80%+ over the other major candidate in this current political environment. Moreover, the swing from Democratic to Republican can't be that much (88% for Clinton over McCain vs. 80% for McCain over Obama) in this election. There just isn't that much difference between Hilary and Barak.

My suspician is that this site is paid for by someone who wants Hilary to look better against McCain than Obama does. Maybe some Republican supporters who would rather face her or some Clinton supporters who want their candidate to win.

Either this site is seriously politically influenced or the methodology is just incompetent.

[NellyLunatic1980]

http://www.electoral-vote.com is a much more reliable (and non-partisan) website.

[slam]

http://www.electoral-vote.com is a much more reliable (and non-partisan) website.

Not sure this site is showing the true story either, but it's certainly not as blatant as the first one.

[hf_jai]

Probably an interesting idea to do this, but the results have to be seriously flawed. I didn't look at it in any detail...

Then how on earth do you feel qualified to voice this opinion?

... but there is absolutely no way that a major candidate has 80%+ over the other major candidate in this current political environment. Moreover, the swing from Democratic to Republican can't be that much (88% for Clinton over McCain vs. 80% for McCain over Obama) in this election. There just isn't that much difference between Hilary and Barak.

It's based on polling data and probabilities. I think whoever did the analysis made some incorrect assumptions made, but I wouldn't call the results "seriously flawed." Hillary and Barack are not that far apart on policy, but they appeal to some very different concerns among the voters, as well as demographic differences in critical battleground states.

My suspician is that this site is paid for by someone who wants Hilary to look better against McCain than Obama does. Maybe some Republican supporters who would rather face her or some Clinton supporters who want their candidate to win.

Either this site is seriously politically influenced or the methodology is just incompetent.

You may be right about the site being paid for by someone wanting Hillary to come out on top, but since you "didn't look at it in any detail" you don't have much to base that opinion on. Also you shouldn't make the assumption that it's someone from the GOP. Obama supporters don't like to admit it, but there's a whole lotta Republicans who would very much rather have McCain run against him than Hillary.

http://www.electoral-vote.com is a much more reliable (and non-partisan) website.

Edited: Nevermind. I see you have to hover your mouse over a state to see results of what I guess are some sort of polling. I still don't know the source(s) so I am skeptical.

[wbtravis007]

This is an interesting web site that keeps a running update on the chances of each presidential candidate in the general election, if it were held today. The web site uses an average of recent polls, state by state, and then does a Monte Carlo analysis by running thousands of simulated elections and sees which candidate amasses 270 electoral votes more often. Based on their current projections, John McCain has an 80% chance of beating Obama, while Hillary has an 88% chance of beating McCain.

One big caveat is that state by state analysis in a number of states is based on a small number of polls, some of them quite old, and should be taken with a grain of salt. Even this fall, you're not going to see a lot of polling done in places like Wyoming and Utah. The site should be interesting to follow to see how the landscape chances over time however.

http://hominidviews.com

This is the silly. Anybody who puts any stock in this whatsoever knows nothing about politics.

[silverscreenselect]

Here's how he does his simulations. Assume that the most recent poll in a state is Obama 50, McCain 40, Undecided 10 and that they surveyed 500 voters.

He calculates that Obama should receive 250 of those 500 voters and McCain should receive 200 of them. He ignores the undecideds. Based on that, he calculates that Obama should receive 55.6% of the decided voters and McCain 44.4%.

He then tallies 450 "votes" by picking 450 random numbers between 0 and 1. If the random number is between 0 and .556, it counts as a "vote" for Obama. If the number is between .556 and 1, it counts as a "vote" for McCain. Whoever has the largest number of votes in that state "wins" the state. He repeats the process for all 50 states plus DC and tallies up the total electoral votes and that determines who "wins" that simulation.

He then repeats each simulation 10000 times to find out who "wins" the election the most times. So if in his simulations, Obama wins 6000 elections and McCain 4000, then the odds of Obama winning the national election is 60%. In a particular state, if Obama wins 7000 times and McCain 3000 times, then the odds of Obama winning that state are 70%.

He takes all reliable polls in each state that are less than one month old and averages them out to get his final numbers. If there are no polls in the last month, he uses the most recent one he has.

This particular methodology is based on the idea that a fairly small difference in raw numbers can translate into a significant difference in overall results. The odds in favor of the house at double zero roulette are about 5%. However, if a player plays hundreds of times, the chances of his coming out a winner are much less than 47 1/2%, which is why there are a lot of 99%+ results here.

One obvious flaw in his simulation is that each state's results are independent of others and it doesn't take demographic trends into effect. That's not right. If Obama does very well in Nevada, he is far more likely to do well in states like Colorado and New Mexico with similar demographics. If McCain does very well in Ohio, he's far more likely to do well in Pennsylvania.

[tlynn78]

slam wrote:
Probably an interesting idea to do this, but the results have to be seriously flawed. I didn't look at it in any detail...

hf_jai wrote:
Then how on earth do you feel qualified to voice this opinion?

Wow. You really have been away for too long.



t.

[TheConfessor]

Statistically, there's probably a better than 0.1% chance of McCain dying before election day, so any analysis that says he has a 99.9% chance of winning a state seems bogus.

[slam]

Probably an interesting idea to do this, but the results have to be seriously flawed. I didn't look at it in any detail...

Then how on earth do you feel qualified to voice this opinion?

I'm qualified to voice this opinion because the results are obviously very flawed. I know enough about the current status of the presidential election to be very confident in this conclusion.

How about this deal:

The site states that Hillary is 88% to win vs. McCain. If she gets the nomination, we can bet on the outcome. The site says that 7 to 1 are the fair odds if I bet on McCain. To sweeten the deal, I'll let you only give me 6 to 1. I'll bet up to $1,000 on McCain with you (but we'll make sure our money is held elsewhere to insure payment).

Similarly, the site says that McCain is 80% to win over Obama which translates into 4 to 1 odds. If Obama gets the nomination, I'll bet on him and let you only give me 3 to 1 odds. Again, I'll bet up to $1,000.

The fact that I view these bets as very enticing (I could lay them off in Vegas for an instant signficant profit), tells me that something is wrong with the site - either its politics or its methodology.

[slam]

Statistically, there's probably a better than 0.1% chance of McCain dying before election day, so any analysis that says he has a 99.9% chance of winning a state seems bogus.

Actually, an average 71-year old is more like 4% to die in the next year. Since the election is 6 months away, that gives an average 71-year old a 2% of dying before the election. I'll grant that McCain may be less likely than average to die during the next year (look at his mother's age, for example), but I'd guess he has to have more than a .5% chance of dying before the election.

And before hf_jai (whoever the **** he is) questions my qualifications, I'll mention that I have several professional designations from the Society of Actuaries that attest to my qualification on this matter.

[Tocqueville3]

And before hf_jai (whoever the **** he is) questions my qualifications, I'll mention that I have several professional designations from the Society of Actuaries that attest to my qualification on this matter.

Ummm...he is a she.

And although we may not agree on much politically she knows a lot more than you think she knows.

As for your professional designations from the Society of Actuaries that attest to your qualification on this matter--we're all REEEEEAL impressed.

[TheCalvinator24]

And before hf_jai (whoever the **** he is)

Match to hf_jai.

[slam]

And before hf_jai (whoever the **** he is) questions my qualifications, I'll mention that I have several professional designations from the Society of Actuaries that attest to my qualification on this matter.

Ummm...he is a she.

And although we may not agree on much politically she knows a lot more than you think she knows.

As for your professional designations from the Society of Actuaries that attest to your qualification on this matter--we're all REEEEEAL impressed.

I have no idea what she knows. All I do know is that she attacked a clearly correct opinion of mine (that the site in question had fallacious results) with clearly specious reasoning.

It seem that she is upset by what she thinks is an implied political opinion in my post. There was no political opinion intended other than to say I am convinced that the results from that site are nonsense and am willing to back up that opinion with money.

As far as an my designations go, asserting an opinion on an application of a mortality table is exactly what an actuary should be best qualified to do. Your eye roll reflects poorly on you, not me.

[Tocqueville3]

And before hf_jai (whoever the **** he is) questions my qualifications, I'll mention that I have several professional designations from the Society of Actuaries that attest to my qualification on this matter.

Ummm...he is a she.

And although we may not agree on much politically she knows a lot more than you think she knows.

As for your professional designations from the Society of Actuaries that attest to your qualification on this matter--we're all REEEEEAL impressed.

I have no idea what she knows. All I do know is that she attacked a clearly correct opinion of mine (that the site in question had fallacious results) with clearly specious reasoning.

It seem that she is upset by what she thinks is an implied political opinion in my post. There was no political opinion intended other than to say I am convinced that the results from that site are nonsense and am willing to back up that opinion with money.

As far as an my designations go, asserting an opinion on an application of a mortality table is exactly what an actuary should be best qualified to do. Your eye roll reflects poorly on you, not me.

Sweetiepie...my eye roll just means I think you think very highly of yourself.

Nobody likes pomposity.

[silverscreenselect]

The first point that the website tries to make is that, essentially, two candidates have to be fairly close in the poll (say 52-48 or closer) for there to be a decent chance of the trailing candidate to win a state. And even then, it's about 10-15% at best (the more independent polls that confirm the candidate is down that much, the worse the chances).

The second point is that the presidential election is not an election to determine who gets the most raw votes but who can get to 270 electoral votes.

The fact that Obama may lose Utah by ten points less than Hillary would or that he would win DC by ten points more than Hillary would does not change either of those. The fact that he has a realistic chance of winning Colorado and she doesn't and she has a realistic chance of winning Missouri and he doesn't do.

Obviously polling models in the primaries have often been badly off and just as obviously, preferences in mid-May do not always translate into November results. Carter led Ford by 30 points after the Democratic primary and Dukakis led Bush Sr. by 20, both of these results in July-August, not May. But by late October, this website should give you a pretty good idea of who is going to win.

[littlebeast13]

And before hf_jai (whoever the **** he is) questions my qualifications, I'll mention that I have several professional designations from the Society of Actuaries that attest to my qualification on this matter.

Ummm...he is a she.

Reminds me of the old days of commonly misgendering BB's! In fact, Jai was one of the BB's (I used to have a list in a notebook somewhere) who thought I was a girl....

But at least she didn't ask me out on a date like BiT did....

lb13

[MarleysGh0st]

And before hf_jai (whoever the **** he is) questions my qualifications, I'll mention that I have several professional designations from the Society of Actuaries that attest to my qualification on this matter.

Ummm...he is a she.

While we're clearing that up, I have a question.

Is Jai pronounced the same as Jay?

[littlebeast13]

And before hf_jai (whoever the **** he is) questions my qualifications, I'll mention that I have several professional designations from the Society of Actuaries that attest to my qualification on this matter.

Ummm...he is a she.

While we're clearing that up, I have a question.

Is Jai pronounced the same as Jay?

Yes.

Of course, I still pronounce it in my head as "Hi" because I'd thought of it that way so long before she finally corrected me...

On 1-70 between St. Louis and Kansas City, there is an exit for Highways H, F, and J. I've always thought of her everytime I've gone by it.... even if I don't pronounce the J part right....

lb13

[MarleysGh0st]

On 1-70 between St. Louis and Kansas City, there is an exit for Highways H, F, and J.

Does Missouri have something against numbers?

[littlebeast13]

On 1-70 between St. Louis and Kansas City, there is an exit for Highways H, F, and J.

Does Missouri have something against numbers?

They have numbered highways, but the majority of Missouri state roads are represented by letters (Or double letters).... and they must curve and angle all over the place, because you'll pass by exits for the same lettered highway numerous times.

J in particular has at least three exits on 70 between KC and StL that I can think of offhand (The aforementioned H-F-J, O-J, and J-DD)....

lb13

[NellyLunatic1980]

Hillary Clinton has been making all of these claims the past few weeks that only she can win the big states against McCain, only she can win the battleground states against McCain, only she can unite Democrats, independents, and Republicans against McCain, etc... but how much truth is there to those claims?

I've decided to take a different look at the polls and the electoral college map.

First, let's start with the 2004 electoral college:

Kerry states: California, Connecticut, Delaware, District of Columbia, Hawaii, Illinois, Maine, Maryland, Massachusetts, Michigan, Minnesota, New Hampshire, New Jersey, New York, Oregon, Pennsylvania, Rhode Island, Vermont, Washington, and Wisconsin (251 electoral votes)
Bush states: all others (286 EVs)

Now, going through all of the most recent polls for each state, this is a rundown of how blue or red each state currently is with either an Obama/McCain matchup or a Clinton/McCain matchup:

Kerry states - solidly blue (Dem ahead by 10% or more)
Both: DE, DC, IL, MA, NJ, NY, RI, VT
Obama only: CT, HI, ME, MD, MN, OR, WA
Clinton only: PA

Kerry states - leaning blue (Dem ahead by 6-9%)
Both: CA
Obama only: PA
Clinton only: MD, OR

Kerry states - barely blue (Dem ahead by 5% or less)
Both:
Obama only:
Clinton only: CT, HI, ME, MN, WA

Kerry states - tied
Both:
Obama only:
Clinton only: MI

Kerry states - barely red (Dem behind by 5% or less)
Both: WI
Obama only: MI
Clinton only: NH

Kerry states - leaning red (Dem behind by 6-9%)
Both:
Obama only: NH
Clinton only:

Bush states - solidly blue (McCain behind by 10% or more)
Both:
Obama only:
Clinton only: OH

Bush states - leaning blue (McCain behind by 6-9%)
Both:
Obama only: IA
Clinton only: FL

Bush states - barely blue (McCain behind by 5% or less)
Both:
Obama only: CO
Clinton only: IA, WV

Bush states - tied
Both:
Obama only: IN
Clinton only:

Bush states - barely red (McCain ahead by 5% or less)
Both: NM, NC
Obama only: AK, FL, MT, NE, NV, OH, SC, VA
Clinton only: IN, MO

Bush states - leaning red (McCain ahead by 6-9%)
Both:
Obama only: AZ, MO, ND, TX
Clinton only: AR, MS, OK, SC, VA

Bush states - solidly red (McCain ahead by 10% or more)
Both: AL, GA, ID, KS, KY, LA, SD, TN, UT, WY
Obama only: AR, MS, OK, WV
Clinton only: AK, AZ, CO, MT, NE, NV, ND, TX

If you take a look at just the solid blue states and solid red states:

Obama holds 14 Kerry states plus DC solidly for a total of 142 EVs while McCain holds 14 Bush states solidly for only 97 EVs.

Meantime, Clinton holds only 8 Kerry states plus DC solidly, but gains OH solidly, for 130 EVs, whereas McCain holds 18 Bush states solidly for 145 EVs.

This tells me three things:

1) Obama holds on to solid Democrats better than Clinton.
2) Clinton unites solid Republicans.
3) Clinton is gaining Ohio's 20 EVs at the expense of about half a dozen other states whose EVs total more than 20.

Now let's add in all of the "leaning" states, those states where a candidate's lead is greater than a typical 5% margin of error:

Obama holds 16 Kerry states plus DC and gains IA for a total of 225 EVs--just 45 short of the White House. McCain holds 18 Bush states and gains NH for 159 EVs.

Clinton holds only 11 Kerry states plus DC and gains OH and FL for 229 EVs--just 41 EVs short of 270. McCain holds 23 Bush states for 185 EVs.

What does this tell me?

1) There is no basis to Clinton's "big state" argument. Obama can and will win California, New York, Pennsylvania, New Jersey, and Massachusetts in November.
2) Even though Clinton can easily win FL and OH, she'll do it at the expense of eight states that should easily be in the Democratic wheelhouse: CT, HI, ME, MI, MN, NH, WA, and WI. She's risking the 67 EVs of those latter eight states to pick up the 47 EVs from FL and OH.
3) There are fewer states that Obama needs to defend than Clinton does. Obama needs to defend just three Kerry states (MI, NH, and WI) to Clinton's eight.
4) Obama puts more Bush states in play than Clinton does. Obama is competitive in 13 Bush states (possibly 15 cuz MO and TX are on that barely/leaning borderline), compared to Clinton's 8 (possibly 9 if you count her de facto home state of AR).

I've also discovered today that in the 1996 presidential election, Bob Dole won the white vote, the white male vote, and the blue-collar vote, but Clinton still won re-election.

So I don't know what argument(s) Clinton is going to deliver to the uncommitted superdelgates in June, but I've just shot holes into each of those arguments.