My interests this cycle include the question about what percent of the vote will be white versus non-white and party ID in the exit poll and the margins among Independents.
The attached document is an insider’s look at these issues across seven surveys released yesterday and today.
A few comments:
- There is an * where we had to do an estimate as the data was not released in this format. While we have done so carefully, we could be off a point or two.
- Other than the bizarre CNN party ID result, this certainly should quiet all the folks complaining in late September that the polls were all biased because party ID had to be in a comparable range with past exit polls. It will be – and multiple polls have carefully measured the movement in that direction in October. (ABC/Post has consistently been in the -4% range over days of tracking, but Obama must have really had a big time Sunday on their track. Welcome to the impact of the NFL on the respondent profile!)
- Party ID – and therefore what is meant by an Independent – varies. Where it is available, I am showing party ID with leaners. Most polls (although not NBC/WSJ) usually show the ballot among Independents WITH the leaners included in the Independent number. In any case, while it looks like Romney’s margin with Independents has dropped compared to ten days or two weeks ago, you need to keep some perspective here: IF party ID is getting closer between GOP and DEMs, it is generally not the DEM number dropping, its INDs getting a bit smaller as overall folks flop a bit toward “Republican.” So, it has a modest impact on who is left as an “Independent.”
- This data – and the margins among non-white – shows the primacy of the fundamental question of the cycle to me which is: What’s the percent white versus non-white, and of course, the margins among both sub-groups.
- Some perspective: Kerry lost whites by 17 points and Bush won by 2.5 points (and did very well among Hispanics, which has changed for the worse for Republicans since then). In 2004, whites were 77% of the electorate. Today, if whites are 74%, and Romney won whites by the same17 points and loses non-whites by 59 (19%/78%), that’s now a 2.5 point Obama victory instead. It is a reminder that at 75% white IF Obama can carry a 59% margin with non-whites, nationally, Romney has to win by closer to 20% among whites to win the national vote.
A few other quick thoughts about the polls:
- In the last NBC/WSJ poll, Republicans had a 9 point margin in terms of interest in the election compared to Democrats. I have not seen a margin that high for a party in a presidential year (recognizing we did not develop the scale until 1992). It is my hope that the party ID on the exit poll closes further as that margin is higher enough to have an impact on turn-out at the margins. Specifically, my hope is people 65+ push to 19% or perhaps even as high as 21% of the exit poll.
- To be candid, with a 1% margin – polls have done what they have been designed to do and give an insight into how close the election is, but, at a 1 point margin, they can not be that predictive about who wins or loses. You are well within the margin where turn-out and even minor variations in terms of how well a candidate does with a sub-group can impact the result.
- National polls of course certainly can not be used to predict the Electoral College, but, for what it is worth, the winner’s margin in CO, IA, and OH have all been within two points of the winner’s margin nationally over the last two elections. My guess is these three states will also be within 2% of the winner’s margin this cycle – VA too.
- Our work looking as a county based segmentation project shows Romney movement in the types of rural counties that comprise a large percentage of the vote in PA. The Romney campaign made a tough and gutsy call to go into PA – and beyond a doubt, it was the right decision.
Fingers crossed – we’ll see.
Seven national polls show a net margin of 1.7%, only Democracy Corp has the margin about 3%, but across seven polls, the median result is Obama with a one point advantage – that’s about 1,340,000 million votes nationally – but, we’ll know soon enough what the outcome is …