November 6, 2012 is quickly approaching. Gov. Mitt Romney has already shown he is a much more tenacious opponent than Senator McCain was in 2008 and therefore has expanded his path to the White House. Nevertheless, the road is still going to prove a challenge.
Roughly every two weeks until Election Night, I will be posting The Pragmatic Conservative’s “Electoral College Watch.” This is our way to wade through all of the data to look at how the Electoral College is shaping up. As many note, election forecasting can be a futile effort. My attempt definitely has its issues, but I do attempt to look at the election from a holistic standpoint.
This post will be long (and I apologize for that) in order to 1) lay out my “model’s” methods, 2) show my baseline Electoral College map, and 3) comment briefly on what our map means for Obama and Romney.
*Note: Models/predictions can attempt to control for every possible variable, but they can never anticipate national/international events nor capture campaign tactics/strategy, both of which can easily shift electoral chances.*
1) The “Electoral College Watch” methods:
In an attempt to determine which candidate will reach the magic 270 number, the model takes into consideration three major components: 1) Polling, 2) Historical Election Results, and 3) Demographics. Most states are forgone conclusions as to who their Electoral College votes will support, so this analysis only examines 15 states. I call them “states of interest” as “toss-up” and “swing states” both can be misleading terms.
My “states of interest” are OR, NH, PA, NC, VA, FL, OH, WI, MI, IA, NV, NM, CO, AZ, and MO. Mainly for historical factors or current demographic/candidate’s strengths and weaknesses, these 15 states are theoretically winnable for either candidate and will determine who wins the general election.
Polling data includes all polls conducted since March 1 (roughly when Romney sealed the GOP nomination). As the summer goes on, the Spring polls will be slowly dropped as they will become too old to be meaningful.
Data pulled includes the raw matchup, Obama’s approval rating, the candidate’s Independent vote, party enthusiasm (Obama’s Democratic vote less Romney’s Democratic vote and vice versa), and the Democratic, Republican, and Independent share of the undecided voters.
The raw matchup was then adjusted based on the “house effects” of the polling agency and whether the poll was of Registered Voters or Likely Voters (some agencies have built in advantages toward a particular party and Registered Voters polls tend to understate the Republican candidate’s vote). For instance, PPP has a historic Democratic “house effect” of around 3 points and its polls are of Registered Voters, which on average understates the Republican candidate by 1.6 points; therefore, Romney’s share of the vote would be adjusted upward by 4.6 points. In order to maintain the poll’s represented vote total, the “house effect” adjustment in split between Obama and Romney (i.e. Obama reduced by 2.3 and Romney increased by 2.3). I then took a weighted average of the polling results, with the more recent polls weighted more heavily.
Historical Election Results
No election year is like another, but by examining a decade of election years you can smooth out the results.
The model incorporates the results of the Presidential, Senate, and House elections in each of the 15 “states of interest” between 2000 and 2010. This incorporates the ultra-close election of 2000, the wave elections of 2006 and 2010, the blowout election of 2008, and the rise of the Republicans in 2002 and 2004.
I then compare each party’s share of the vote in each state for each category to the national results of the party in each category to determine if each party either under- or over-performed. For instance, Kerry performed 3 points above his national popular vote percentage in Oregon in 2004, while the 2008 Oregon Republican Senate candidate performed 0.6 points better than the national Republican Senatorial candidates’ popular vote.
After all the election years are averaged (using a basic mean), I took two additional differences – 1) the difference of the averaged Democrat and Republican candidates results and 2) the difference of the averaged results for each party to their respective party’s national average vote share. For instance, in Oregon, the Democratic Presidential candidate, on average, performs 7 points better than the Republican candidate and the Democratic Presidential candidate performs 5.2 points, on average, better than his national popular vote percentage.
Next for each “state of interest,” I examine a series of demographic information based on the 2010 Census. Demographics of interest are the percentage of the population that is female, over 65, black, Hispanic, and college educated. I then look at the 2008 unemployment rate and the current unemployment rate.
To determine whether a state is heavily weighted toward a demographic, I look at the difference between its demographic percentage and the national percentage. To determine whether the voters in that state are “better off today than four years ago,” I take the difference in the 2008 unemployment rate and the current one.
Adjustments to Obama’s Spread
The model examines Obama’s spread in each state with a negative spread indicating Romney leads. Therefore, using the adjusted raw matchup spread as the baseline, I adjust the spread based on the polling details, the historical elections results, and the demographics. From this point on, the only the polling data will change – unless I decide to add additional economic indicators; the other metrics are the state’s fundamentals.
I take each of the metrics within each component and either add to or subtract from Obama’s spread in each state. Adding to Obama’s spread indicates the metrics are working in his favor (i.e. an approval rating of +5 or greater or a share of the black population that greatly exceeds the national percentage). However, if the metrics work against Obama, points are subtracted from his spread (i.e. states with higher than normal 65+ population or a significant share of the undecided voters are Republicans).
2) “Electoral College Watch” Baseline
The “Electoral College Watch” breaks the states into 5 categories: Likely Obama, Likely Romney, Lean Obama, Lean Romney, and Toss-Up. The 15 “states of interest” will fall into the Lean categories or Toss-Up. All other states are Likely states.
A state is Lean Obama if the final adjusted Obama spread (FAOS) is +5 or greater and Lean Romney represents states with the FAOS of -5 or below. Toss-ups are the states that fall between -5 and +5 FAOS.
Obama begins with 189 Likely Electoral College Votes and Romney starts off with 170 Likely Electoral College Votes. The 15 “states of interest” represent 179 Electoral College Votes.
The “Electoral College Watch” baseline has New Hampshire (+5.5), Pennsylvania (+6.5), Wisconsin (+7.8), Michigan (+6.3), New Mexico (+18.3), and Colorado (+5.0) as Lean Obama. Lean Romney states are North Carolina (-7.3), Arizona (-15.3), and Missouri (-9.3). This puts the baseline Electoral College Vote at 253 for Obama versus 206 for Romney with 79 Toss-Ups.
Ohio and Virginia both currently have negative spreads suggesting they are within reach for Romney; while Oregon, Nevada, Iowa, and Florida have positive spreads. However, Iowa, Florida, and Virginia are shaping up to be as close to true Toss-Ups as you can get with spreads of only +1, +0.5, and -0.3 respectively.
Yet, if you give Romney Ohio (-2.3) and give Obama Oregon and Nevada (+4.5 and +4, respectively), then Obama would have 266 Electoral College Votes (only 4 short of victory), while Romney would only have 224 (46 short). Obama would only need to win one of the three remaining Toss-Ups, while Romney would need all three to win.
3) Romney’s path is clear, but challenging
Obama has a series of built in advantages going into the summer – the power of the incumbency, general likability, a strong voting coalition, and fundraising prowess. For these and other reasons, Obama begins with a small lead. However, Obama’s campaign has been sloppy thus far and he has no coherent message – not to mention the economy remains the #1 issue for voters.
Romney, on the other hand, is in a much better position than many would have thought. He has fewer paths, but has opened up a lot of states many did not consider in play (such as Wisconsin, Michigan, and I think, Oregon). Demographics work against him in some states, but Romney has so far proven to be a dedicated and skilled campaigner who has a consistent message that could resonate more with voters as they start to pay attention. Romney also has substantial fundraising abilities and a strong campaign infrastructure, something the Obama campaign did not have to deal with in 2008 and an aspect I don’t think they were anticipating.
Obama has to start playing defensive in states his campaign wasn’t originally planning on, which will pull resources from other states. Romney needs to start playing aggressively in the “states of interest” to begin shoring up support and pulling undecided voters his way (particularly Independent voters). He has quickly shored up Republican support, but additional party enthusiasm would help (something he can accomplish, potentially, with his VP pick). Expect both to ramp up their state operations to bolster the GOTV campaigns. The election may very well be won on the efficiency of the campaign’s GOTV effort.
All in all, 2012 is going to be an interesting Presidential campaign to watch.