~~Washington Dispatch Presidential Election Poll Model: Final Results & Predictions~~

[Will]

Washington Dispatch 2012 Presidential Election Poll Model: Final Results and Predictions

Mitt Romney (48.7%) leads Barack Obama (47.4%) by 1.3% nationally in the final national Washington Dispatch Poll of Polls, and leads the electoral college by 273 electoral votes to 265.





Mr. Romney holds a 10% lead among self-identified independents heading into election day, with nearly another 10% of independent voters undecided or planning on voting for third party candidates.





Mr. Obama retakes a very slim lead in Ohio in the final Washington Dispatch Ohio Poll of Polls, reducing Mr. Romney's lead in the model to 8 electoral votes.



Mitt Romney retains his leads in the states of Colorado, Florida, Iowa, Nevada, New Hampshire, and Virginia.













In addition to regaining a small lead in Ohio, President Obama maintains his shrinking leads in Pennsylvania and Wisconsin, and also leads in Minnesota.







The final results of The Washington Dispatch National and State poll aggregation models show a very different result from those shown by most individual pollsters, as well as popular poll aggregation models like those found at Five Thirty Eight, Real Clear Politics, or the Princeton Election Consortium, all of which project an Obama victory.

If the Washington Dispatch Models' results proved accurate, we could be in for a long, roller-coaster ride of an evening tonight as the polls close and election results come in. Most pundits, and the campaigns themselves, feel that Ohio could be a "must-win" for Mitt Romney. If the President were to take Ohio, many of his supporters would likely begin to celebrate an election victory. States like Iowa and Nevada are thought to be even more favorable to Mr. Obama than Ohio. However, our model shows Mr. Romney carrying Iowa and Nevada, and if that were the case, such celebration would prove premature as Nevada's electoral votes late in the game would push Mr. Romney over the top in spite of an Ohio loss.

We could leave things here, based solely on the numbers and the models, but where is the fun in that? This author humbly submits his own final prediction for the election, based in part on the above models and in part on gut feeling, a Romney victory with over 300 electoral votes and somewhere around 52% of the national popular vote.



Mr. Obama's slim lead of 0.2% in Ohio will not hold up if turnout today meets Mr. Romney's expectations and he cruises to a national victory of 4% or more. Ohio has traditionally leaned to the right of the national electorate in recent years, including 2008 when Mr. Obama carried Ohio by a margin 2% or so smaller than his national margin of victory. The polls this cycle have shown the opposite trend however, with Ohio trending 2% or so more favorable to the President compared to national polls, but even this would not be enough for the President to carry Ohio if Mr. Romney wins with 52% or so of the national popular vote.

The President's leads in Wisconsin and Pennsylvania have been consistently shrinking every week over the final month of the campaign. Despite Mr. Romney's gains in and late emphasis on the state of Pennsylvania, the electorate there is simply too favorable to Mr. Obama, who even in the worst case-scenario based on 2004 turnout, appears to need only to wrap up the final undecided Democratic voters in order to carry Pennsylvania's 20 electoral votes. Wisconsin is slightly less favorable to the President, but still the biggest reach in this prediction. Republicans have crafted an impressive ground game in Wisconsin over the past two years, managed to produce even more favorable turnout in the contentious Scott Walker recall election than they did in the wave election of 2010, and have a ticket that includes a popular Wisconsin representative--Paul Ryan. If Mitt Romney is going to pull off any surprise state level victory in 2012, it will be Wisconsin. Matching their recall election turnout advantage, or at least something in between that and their 2010 midterm turnout performance, would be enough to propel Romney to victory. This will of course be a tall order, given the national and state-level strength of the Obama Organizing for America turnout machine, which will in all likelihood outperform Democratic turnout in those previous two Wisconsin elections.

Nevada is perhaps the most peculiar result in the model itself and in this prediction. All signs seem to indicate an Obama victory there. The polls themselves show stronger performance in Nevada than in other swing states for Mr. Obama, the demographics of Nevada appear to be shifting more in the Democrats' favor every year, the 2010 wave election and Tea Party insurgence were not enough to overcome the Democratic machine and vote out Senate majority leader Harry Reid, and finally the campaigns themselves seem to have conceded an Obama victory in Nevada in the final week or so of campaigning. But the Washington Dispatch model tells a different story, with Romney leading heavily among independent voters in the state with the highest unemployment rate in America under President Obama's leadership. Democrats won the turnout battle in 2008 by 8% in 2008. Republicans cut that advantage to 2% in 2010. Holding the Democrats to a 5% turnout advantage in 2012 would be enough for a Republican victory.

Michigan is, frankly, fool's gold for the Romney/Ryan ticket. The Washington Dispatch model shows a 10% Obama lead there, and we stopped updating Michigan two weeks ago. Romney will perform much better in Pennsylvania and Minnesota than he will in Michigan. In the event of even this prediction proving to conservative on Mitt Romney's chances, and Romney winning in a land slide, it is far more likely in such a scenario that Mr. Romney would carry Pennsylvania and/or Minnesota than Michigan.

Finally, it is interesting to note the projected electorate by Party ID predicted by Gallup: a small republican majority. If that came to pass, Mr. Romney would almost certainly win the election handily. Even more interesting to note is the Rasmussen October 2012 Party ID update, as shown in the national poll model near the top of this article. Rasmussen reports has been tracking this constantly-running, monthly updated poll of America's partisan self-identification for nearly a decade. Despite using a model of Democrats +2 in his national election polling this year, Rasmussen's Party ID poll for October shows Republicans with an enormous advantage, their largest one recorded. In the 2004, 2006, and 2010 elections, actual Republican turnout outperformed the Rasmussen Party ID poll by around 2%. In 2008 the margin was roughly equal. Republicans have yet to under perform the survey in actual election day turnout. If this trend were to hold on election day 2012, it would signal a wave election for Republicans even more impressive than in 2010.

Most of the polls, and virtually all of the statistical experts like Five Thirty Eight's Nate Silver, who are admittedly much brighter than this author, seem to indicate the the Washington Dispatch Poll Models will not be accurate and that President Obama will win. The key factor driving the difference between our model and others is The Washington Dispatch Poll Aggregation Methodology, as described below. Other aggregation models simply use polls' top-line result of total support for either candidate. On average across the polls, sampling by party ID has reflected the demographic make-up of the 2008 electorate. As a result, sites like RCP and 538 produce numbers also based on turnout very similar to the 2008 election.

The Washington Dispatch makes the assumption that nationally, while turnout will not be as favorable to Republicans as in 2010, it also will not be nearly as favorable to Democrats as in 2008. Using the average from the 2004 through 2010 elections, the primary Washington Dispatch turnout model, on which the official numbers are based, shows a Democratic turnout advantage of 2.5%.

In order to maintain uniformity, as well as to remove the human element, The Washington Dispatch has used this same average turnout formula to produce the turnout model for each individual state in the Washington Dispatch Swing State/Electoral College Tracker.

Sometime after the election this week, The Washington Dispatch will examine the differences between actual results and our models, and attempt to decipher why the models were accurate or inaccurate. We fully admit that we may well end up with the proverbial egg on our face late tonight or early tomorrow morning, but we've thoroughly enjoyed this and hold out hope that our models will prove insightful, and even in the event of failure will prove salvageable for future elections with some corrective tweaking.

Methodology:

Rather than simply averaging the overall result of polls pitting President Obama against Mitt Romney (as is done on sites like Real Clear Politics), The Washington Dispatch digs into the cross-tabs of data behind polls (those that freely publish the necessary data breaking down responses by party identification), and averages each candidates' share of the vote among Republicans, Democrats, and Independents. Then, these shares are applied to the actual turnout by party identification (as measured by comprehensive exit polling of between 13,000 and 18,000 respondents for the 2004, 2006, 2008, and 2010 elections) for the past four elections. To calculate poll numbers for The Washington Dispatch's official head-to-head match-up between Barack Obama and Mitt Romney, each candidates' average share of the vote by party affiliation is applied to the average turnout by party affiliation over the past four elections.

http://washington-dispatch.com/post/...edictions.aspx


Went to the polls shortly after 6am this morning and cast my vote for Romney. Fingers crossed for the upset.

[Loco Vette]

Originally Posted by TxAg

I hope you are correct Will.

[Will]

Originally Posted by TxAg

I hope you are correct Will.

Honestly I'm not very confident, but I'm seeing this thing through to the end.

My hope is that even if I'm wrong, the models will be somehow correctable/salvageable for future elections. I've really enjoyed this.

[VITE1]

Originally Posted by Will

Honestly I'm not very confident, but I'm seeing this thing through to the end.

My hope is that even if I'm wrong, the models will be somehow correctable/salvageable for future elections. I've really enjoyed this.

You have done amazing work and the detail you have Dug into is mind boggling.

I look forward to listening to you on the news in the near future. You have a great future and I hope a very profitable one.

[Fasglas]

Originally Posted by TxAg

I hope you are correct Will.

[jaxgator]

Originally Posted by TxAg

I hope you are correct Will.

Thanks for all of your hard work on this Will.

[Will]

Originally Posted by VITE1

You have a great future and I hope a very profitable one.

That would require someone paying me for my opinions. Who would want to do this is beyond me.

[Will]

Site is down, back-up location where you can leave comments, like, share, etc.:


Washington Dispatch 2012 Presidential Election Poll Model: Final Results and Predictions | RedState

[Broken Wind]

obama's going down. Not a doubt in my mind.

[Will]

Originally Posted by Broken Wind

obama's going down. Not a doubt in my mind.

I wish I could say the same.

Still skeered.

[VITE1]

Originally Posted by Will

That would require someone paying me for my opinions. Who would want to do this is beyond me.

The more often you are correct the more often wise people with money come looking of answers.

Get out and mingle with politicians and people in powerful positions. When you earn their respect then you can earn their money.

[Loco Vette]

Originally Posted by VITE1

The more often you are correct the more often wise people with money come looking of answers.

Get out and mingle with politicians and people in powerful positions. When you earn their respect then you can earn their money.

This. Will, you heard back form the bar exam yet?

[Will]

I'll eventually do a full write up on the site, but a quick breakdown for the moment for you guys and gals:

The model predicted Romney 48% to Obama 47% Nationally.

Romney got about 48%, Obama got about 50%. Undecideds broke heavily to Obama.

The model showed a dead heat in Ohio, 48% to 48%. We got a dead heat in Ohio, Obama leads 50% to 48% with 10% left to report.

The model showed a dead heat in Colorado, with Romney at 48%. Romney is at 47% with 10% left to report.

The model showed Romney at 50% in Florida. He's at 49% with 3% left to report.

The model showed Romney at 50% in Virginia. He's at 48%.

The model showed a dead heat in Iowa with Romney at 47%. He's at 46% with 4% left to report.

Etc., etc., etc.

Nationally the model showed Romney at 50% among independents. He won 50% of independents.

Everywhere I look, the model was incredibly accurate at predicting Romney's vote total, with true misses in New Hampshire and Nevada.

The model was incredibly INaccurate at predicting Obama's vote total. This is because: (a) Independents broke heavily to Obama at the end, and (b) the turnout model based on 2004-2010 used for the primary results was off, and the 2008 only or the 2008-2010 model turned out to be the better one, depending on the state.

Why? Turns out team Obama was right. 2008 was not a fluke demographically because of the wave election, and 2010 was merely a better year because of poor minority/democrat turnout.

Whites DID shrink even more as a percentage of the electorate, and as a result counteracted increased/decreased Republican/Democrat enthusiasm from 2008.

America is changing and Republicans are at a key cross-roads, which I have commented on here:

https://www.thevettebarn.com/forums/p...-25-years.html