| If the election were held today and... | Obama | McCain |
| ...all McCain <5% went to Obama | 406 | 132 |
| ...all Obama <5% went to McCain | 306 | 232 |
| Shift since yesterday: | +10 | +11 |
| Shift this week: | +20 | -20 |
Today's Numbers: | Yesterday's Numbers: |
The chart at the left shows the distribution of electoral votes based on statewide polls. The colors represent the gap in the poll. The darker the color, the larger the gap. 50% +1 or 270 Electoral votes are needed to win, out of a possible 538.
This site is really designed to be used along with Electoral-Vote.com. I provide one useful way to interpret the data, but without the information that is provided at that site, you won't get as much use from this chart.
For a complete explanation of the methodology used, see the Electoral-Vote.com polling FAQ. The raw data used to generate the chart is also available there. Please note that other than using the data that they make available, this site is not affiliated with Electoral-vote.com
12 comments:
Three recent polls are out for AZ, showing McCain +2, +4, and +5. You take the lowest, +2, perhaps hoping for a turnout gap, cell-phone gap, reverse Bradley, or just the best case scenario. Electoral-vote.com will average the three. I'd throw out the two outliers and go with the +4 (still hoping for a better turnout. 538 processes the hell out of the numbers and decides that it represents a snapshot of McCain +6.4 with a projection of McCain +7.9.
Sometimes I doubt whether Nate's method is really working. In a few cases he has anticipated the change, but other changes have not shown up at all in his model.
It will be interesting to see whose predictions and assumptions were correct on election night (if the election outcome itself holds little suspense).
Voter suppression in Indiana
Actually, I took the McCain +2 number for AZ because that's what EV.com listed as of yesterday. Today they show it as McCain +6. Unless otherwise noted, all my numbers are the margins that EV shows. I may not always agree with them, but they make the data available, so that's who I use.
Sorry- I thought maybe you had picked a newer number off 538 or elsewhere before it got to electoral-vote.com. Sometimes I'm not clear on which polls electoral-vote.com uses. (They didn't use the "New West" number?)
It seems like Nate was right about "regression to the mean" in AZ. But if AZ was getting close, some other states may be closer than expected as well. Maybe NV is safe?
Yep, EV.com can be a bit confusing. Andy does make things easier than I previously thought... Each day you can download the current polling data snapshot at http://www.electoral-vote.com/evp2008/Pres/Excel/today.csv
I wish I had found that earlier, since it would have allowed a couple of cool things if I had time to implement them. With less than a week to the election, I can't do much with it other than update the running 50% tally.
Note, If you know spreadsheets, you could use this to make your running tally list easily enough. There's probably an easier way, but worst case you could do a bunch of sum() functions.
A couple of weeks ago I downloaded a list of states ranked by their popular vote margin in favor of Bush from 2004. I can't find it now (on another computer), but I thought it was from electoral-vote.com. Of course, the positions of states have changed in this election. IA, VA, NC will be in different positions for sure based on demographic changes and the individual differences in regional appeal of these candidates.
Lots of info there. http://electoral-vote.com/evp2008/Pres/pres_polls.html
I was trying to think of how something like your graph could somehow also convey state information but I never really worked out a good idea. The mini-map works to fill in some information. I divided states into 5 groups of ten or 10 groups of 5 (+DC) based on vote margins and was wondering how it would look to superimpose the state abbrevs in a range from top to bottom but that might not look real good, especially if the date lines are there, too. But my idea is it would be good to see which states these solid red states ARE, which states these solid blue states ARE, and swing states ARE, somehow within the main chart, even if it were somewhat decorative. Another way would be to actually have 50 lines of text with UT…AZ on the vertical axis at the margin (like, today's state ranking) but that would be serious (and kinda microscopic) clutter.
One problem is that Andy uses too many polls with differing leans or tilts, which produces zigzag state graphs, although you get that even with a single pollster. The regression lines smooth that out a little on his newer graphs. Nate tries to overcome that by adjusting for each pollster, which may work. Sam Wang represents the line on his EV estimator not as a line, but as a zone. I think using the median (like Colley rankings) is the best, if you have enough polls in a short time range.
I was wondering if you are happy drawing data from electoral-vote.com or would rather DIY next time if there is a next time?
I suppose that if I have a whole bunch more time on my hands next time 'round I might do it myself, but I doubt it. I'll never be as statistically savvy as Nate or Andy, and I like being able to blame someone else for any bias in my numbers. :-) But who knows. Four years is a long time, so I have a while before I need to think about these issues (though I expect that Sarah Palin will announce her canidacy sometime around noon on 11/5, so maybe not so long after all).
It seems plausible this week, but in the long run I don't think Palin will go far. Her positions are ones the general voting public strongly disagrees with, and I don't think she would get through a rough primary season in the top 2 or 3. The much wittier Huckabee already occupies her niche, and even if she did well with the Christo-fascist base, she would fail in the general. Also, she is a socialist, redistributing the wealth: $3000 from the hard-working oil companies to every Alaskan family, and she associates with terrorists, the Alaskan separatists such as her husband. She's not really "American" like the rest of us are. Thus, Rovean politics requires that she make exactly these accusations of her opponent to cover the chink in her armor.
The next Republican would have to be more centrist IMO unless they are exceptionally charming such as a Reagan. Romney, Huckabee, another governor might work. But I think you may be right in that the woman factor may be huge next time. I'd be surprised if one or both tickets didn't have a woman on the ticket next time, but someone with broader crossover appeal than Palin.
I suspect Hillary will see her path to the presidency closed off on the Democratic side, and will team up with Joe Lieberman to run as Republicans in '12. That's my prediction, and I'm stickin' with it.
Oh, don't get me wrong... I don't think Palin will WIN, just announce. Hell, I don't even think she'll be out able to stay out of prison until 2012. Course, if she was President, she could pardon herself, and the base doesn't seem to mind corruption, so long as it's done by a Republican.
Dividing states into "Centiles" (if that is a word)
Obama->McCain
DC
HI NY VT IL CA
MD NJ CT DE MI
WA ME NM MA OR
IA CO WI VA PA
MN NH IN OH NV
NC MO FL RI GA
ND MT WV SC TX
KT SD KS LA AR
AK MS TN AZ NE
OK WY AL ID UT
McCain->Obama
UT ID AL WY OK
NE AZ TN MS AK
AR LA KS SD KT
TX SC WV MT ND
GA RI FL MO NC
NV OH IN NH MN
PA VA WI CO IA
OR MA NM ME WA
MI DE CT NJ MD
CA IL VT NY HI DC
(That would look better in a fixed-width font like Courier.)
There's probably a mistake in there somewhere.
Or it could be 5 groups of 10.
51 is also divisible by 3 -- but I don't think there are really 17 swing states (and 17 red states and 17 blue states), are there? That would break the Blue/White btwn IA and CO and the White/Red btwn WV and SC.
Looking at it this way, the battle seems to be in the row with GA, FL, MO, and NC, rather than the rows with IA, OH, or FL, but this is your e-v.c table from Sunday ranked by Obama-ness, not net difference. RI is stranded in an odd place.
Check out the table called "Probability of Win by State" about 25% of the way down the page at election-projection.net. States are ranked on the left hand margin by how likely they are to vote Democratic. California is first due to its size making it more predictable than little DC, for example. It tells the number of EVs for each state and gives a RUNNING TOTAL as you proceed down the page. Just what I wanted to see, only better. This must be the one electoral-projection site McCain reads because the 269/270 line runs right through PennsylvaniA! There is, however, only a 4% chance that the polls are off so far that McCain can win it.
This also shows that if can just stake out PA, and somehow get outrageously lucky in CO NH NV OH FL NC and ND, seven states that are more likely to go to Obama, then MO IN MT GA AZ MS SD WV SC may just fall into place.
That really makes it clear to me how desperate McCain's situation is. Four states where he has about a 1/20 chance, followed by a state with a 15% and one with a 25% chance, plus win about 4 or 5 coin tosses and then some safer Russian-roulette ones.
He's toast! and I think i just found a new site to watch for the next… 4 days.
Wow, great site. I particularly like their interactive simulator. It lets us all be our own Nate Silvers. You can even test all the worst case scenarios. Want to know what happens if McCain steals VA, PA, OH and FL? Just set those three sliders to zero% and hit calculate He still wins less than 25% of the time (depending on the position of the "State-to-state outcome correlation" slider, between 8 and 24%, though the lower side of that is more realistic since state-to-state correlation is not all that high).
Post a Comment