Politics: Electoral College Project November 201

If the state polling aggregate numbers at RCP are correct:

What if the state polls are under-reporting GOP strength by 1% and undecideds break 2-to-1 for the challenger?

Method:
1. Using the RCP state poll averages for data (O, R)
2. R1 = R*1.01 adjusts for tendency of polls to under-sample Republican voters - this is a debated assumption about systemic error in polling
3. und = O - R1 this is the percent undecided voters
4. O1 = O + (und*0.3333) and R2 = R1 + (und*0.6667) accounts for the debated assumption that undecided voters break 2-to-1 for the challenger
5. O1 + R2 to check to see if the numbers add up to 100%
6. O1 - R2 yields the final vote in that state; a negative value means Gov. Romney wins that state while a positive value means President Obama wins that state
7. EV is the electoral vote value of the state
8. O-EV and R-EV gives the various electoral college totals for the candidates if they hold/lose that state.

Result: President Obama 281 Gov. Romney 257

In order for the state polling data to show a Romney victory, the existing data needs to be under-estimating Republican support by 6.4% resulting in the table below which resembles Electionprojection.com's 2008/2004 average turnout model.

Poll pros who really do this seriously:
538 (315-223 Obama)
electoral-vote.com (294-220 Obama with 24 too close to call)
electionprojection.com (303-235 Obama)

I'm just a "blogger in pajamas" but why should they have all the fun?

The Romney people are either actually confident of their chances or they are putting on a very brave face. Geraghty offers up two items today that explain their confidence (actual or spin).

So as it stands:
1) If the state polling is accurate, the President is headed toward around 300 electoral votes and easy win.
2) If the national polling is accurate, the race is very close (on par with 1976, 2000, 2004) and the winner will probably get something in between 270-290 electoral votes.
3) It is theoretically possible that a candidate could just barely win the popular vote and collect a sizable electoral college win consistent with both sets of polling data.

As an extreme hypothetical, if a candidate won every state by 0.5%, there could be a 538-0 electoral college shutout with only a 0.5% popular vote differential. However, in practice, that is unlikely.

I went to the databank and plotted the data of the electoral votes of the winning candidate (y-axis) and the popular vote differential (x-axis). I went back only to 1964 because that was the first time the electoral college totaled 538. I dropped the 1968 election data point because there were some electors that cast their ballots for a 3rd party candidate (46 for George Wallace) which was the last time that occurred in numbers beyond the occasional lone protesting faithless elector.

As you can see, there is a somewhat linear relationship between margin and total electoral votes. There were three obvious exceptions where the electoral vote margin was disproportionate to the popular vote margin (1980, 1984, 1988). You can also see that we have very little data points on the left side of the graph. The three closest races occurred in 1976, 2000 and 2004. 2012 looks to be on that side of the graph.

Carter had a 2% popular vote margin in 1976 yielding 297 electoral votes, thus, a 1% win by either Obama or Romney would likely yield electoral votes less than 297. However, the design of the electoral college permits a narrow popular vote win to be magnified because a candidate gained support from a larger collection of states. This benefit would be lost in a popular vote only system which Democrats generally support because of their 2000 experience. Ironically, if indeed Obama gets ~ 300 EVs but only wins the popular vote by 0.5 to 1%, his political standing is enhanced by the "quirky" electoral college because he can claim his support was over a larger number of states.