AAPOR on the Election Polls

For what it's worth, AAPOR's statement about the 2012 election polls:

The following press release is being issued today. It was crafted by AAPOR’s current three presidents and our 2012 Election Rapid Response Team (Diane Colasanto, Mike Traugott, Rob Daves, Cliff Zukin, and Quin Monson). Considerable thanks goes to the Rapid Response Team for all they have done to help Council since last spring regarding 2012 election-related matters.  -- Paul J. Lavrakas, AAPOR President

AAPOR's Statement on 2012 Presidential Election Polling

During the past two months, journalists, partisans on many sides, and the public at large have focused a great deal of attention on the accuracy of the presidential pre-election polls.  At times considerable criticism was directed toward pollsters and their polling methods.

However, as was seen last Wednesday morning, the vast majority of the major pollsters were highly accurate in their final estimates for the presidential election, both at the national and state levels. The American Association for Public Opinion Research (AAPOR) would like to take this occasion to compliment pollsters who used established, objective scientific methods to conduct their polls, rather than subjective judgments about the electorate to make their forecasts.

“AAPOR is very pleased that the survey research profession has worked to respond to the increasing challenges facing public opinion polling by drawing on the best available scientific evidence, whether it is from scholars, government researchers, or political polling practitioners themselves,” said Paul J. Lavrakas, Ph.D., AAPOR’s current president.

Despite myriad challenges including the growing cell phone population, increasingly high levels of non-response, and even the effects of unanticipated events such as Hurricane Sandy, the final estimates of the 2012 election outcomes demonstrated that when pollsters remain committed to objective scientific methods, their pre-election polls are very likely to be an accurate forecast of the voting public’s behavior. 

“As importantly, to the extent that polls also are accurate in characterizing the attitudes, beliefs, and motivations of the electorate, we believe that pollsters, and the news media that use their poll findings, provide a great service to democracy by placing the opinions and preferences of the public in the forefront of the electoral process,” observed Lavrakas.

Which Polls Were the Best?

Nate Silver has a nice analysis of which polls did well and which polls did not so well in predicting the 2012 presidential election.  Well worth the read.  You'll see a more comprehensive article in Public Opinion Quarterly some time in the future, if they do as they normally do and do an election synopsis of poll accuracy.

Read his article.  Basically, live polls far outperform robo-dial polls.  No surprise there despite the pathetic efforts of some to defend such polling.  Of the bottom five polls in accuracy, three were robo-bullshit-polls (full list, any polls conducted).  The problem with the robo polls, other than being annoying as hell, is they cannot legally call cell (mobile) phones, which gives them among other things an age bias.  That makes them only slightly more accurate in predicting an election outcome as cutting open an animal and studying its innards.

Of the "big boy" polling firms, the ones who performed poorly (i.e., sucked) were Gallup, InsiderAdvantage, Mason-Dixon,  American Research Group, and Rasmussen (this of shops that conduced at least five polls).

Google's poll did well, leading Silver to suggest:

Perhaps it won’t be long before Google, not Gallup, is the most trusted name in polling.

Likely Studies of the 2012 Election

Every major event, and especially every presidential election, produces scores of academic studies.  2012 will be no different.  So what studies are we most likely to see eventually appear in the major academic journals?  Probably many of the same questions found in news accounts, but lemme take a stab at some likely themes.

• White Men Can't Vote.  Put in this category the studies that attempt to explain the Obama coalition and the role of the white versus non-white vote.  You'll find these mostly in political science journals, perhaps in sociology, trying to explain the dwindling role of white voters.

• Obama is Still Muslim, etc.  Here you'll find studies of why people did not vote for Obama, based mainly on racism as a factor but also belief in the various myths surrounding him (Muslim, born outside the U.S., is a space alien, and so on). These studies will be found everywhere, including masscomm (note, this is an area of interest to me as well).

• Where'd The Christians Go?  By this, I obviously mean the conservative or evangelical vote and the small role it seemed to play in this presidential election compared to others.  You'll find these studies everywhere from political science to religion to masscomm, such as Journal of Media and Religion.

• The Roles of Polls.  In this, I expect to see a study or maybe several studies on the roles the polls play, from individual surveys to Nate Silver and the gang of geek/nerds who so correctly called the 2012 outcome.  Expect to see a lot of this stuff in Public Opinion Quarterly.

• Sandy.  Yes, there will be studies that attempt to explain much more fully than you see in the press about the role of Hurricane Sandy and the like.  Short answer: very little, but I'm sure we'll see analyses of it in the political science journals.

• Twitter and Social Media.  Yes, there will be a number of analyses of Twitter, from its role in the debates to whether it is a useful predictor of the election outcome.  A lot of this is basic content analysis, called by my major professor the "great intellectual cul-de-sac."  But analyzing big data is the future.  You'll find this mostly in the more geeky journals out of computer science.

Got any others?  Feel free to add some in the comments section below.  For the uninitiated out there, understand it can take as long as two years before an academic study sees publication.

Charles Darwin, Write-In Candidate

Charles Darwin, at least where I live, is a popular guy.

There's been a lot of attention about the write-in campaign using his name thanks to my U.S. Rep's rather shaky understanding of science (local stories here and here, but also found nationally).

Being the nerdgeek I am, I dumped the file of write-in ballots into a spreadsheet and tried to get an exact count of the variations of Charles Darwin.  Best I can tell, there were 3,908 write-in ballots for the mostly dead naturalist -- who if he had been elected, would have been 203 years old, roughly the same age as U.S. Sen. Robert Byrd when he died.

I may have missed some.  I show 6,907 different kinds of write-ins for the U.S. House race. I excluded write-ins for the other races. Some of the spellings of Darwin are creative, so I may have missed a few, but after some effort to transfer the PDF into a spreadsheet, this is what I found.

• Search for "darw" and I get 3,908 (not case sensitive, by the way).
• Search for "darwin" and I get 3,876.
• Search for "charles darwin" and 3,320 went with the full name.
• "darwinn" had two hits.

Having looked through the data, I'm comfortable with the "darw" search.  I can't find any other names that use those four letters together but who obviously meant someone else.

I should point out there were a lot of write-ins for "anyone else" or "any one else" or "any one but him" and so on and so on.  As many as for Darwin?  I don't think so.  Any use of the word "anyone," for example, gives only 248 hits, and any use of "anybody" came up with only 35.  Other popular write-ins?  Best I can tell:

• Big Bird got 23
• Bill Nye got 13
• Pete McCommons, local publisher, in one form or another (McCommunist) nailed a whopping, 170 votes.  Actually it's more than that, given all the various spellings or just "Pete Mc" and such.
• And finally, Darth Vader got two votes for U.S. House -- one of them from my son.

If you'd like to see the spreadsheet, you can download a simple column of the write-in names by clicking here.

And yes, I voted for Chuck Darwin as well.

Update (3:47 p.m. Friday)


Honey Boo Boo got 1 vote.  Just thought I'd throw that in.

It's widely reported already how many votes Jarvis Jones received (not nearly enough, but over 10).  John Doe did well with eight.  Jesus (4 votes) outscored Mohammed (1 vote).  Satan got 2 votes.

And finally, Jim Thompson got one vote.  Probably from himself.  Just saying...

Gays Elected Obama?

Gay people elected Obama.  How's that for a provocative lede?  You doubt me? Check out these fun stats, based on exit poll data.  Respondents were asked: Are you gay, lesbian or bisexual?

Yes   5%
No  95%

No surprise above in how many gay or non-gay voters were found in the electorate this week.  But check out the numbers below.

                           Yes, Gay    No, Not
Voted Obama           76%          22%
Voted Romney          49%         49%

In other words, among voters who said they weren't gay, it's a tie.  Among voters who said they were, Obama overwhelmed Romney.

Oh my, if Rush Limbaugh sees this, he'll have a fit.  Forget I mentioned it.

And I should point out that it's likely many of these votes were in states that were already solid Obama in the first place.

Obama and the Storm

It's a popular narrative among Republican types to blame Hurricane Sandy (and N.J. Gov. Chris Christie) for Obama's win this week.

Is it true?

First off, most models had Romney's momentum ending about Oct. 15, long before the storm hit.  Given the stunning accuracy of these models by Nate Silver, Sam Wang, and the rest of the nerdgeek squad, it's hard to argue against them.

And yet, and yet.

In exit polls of actual living breathing voters, 3 percent said they decided on Election Day who to vote for.  Another 6 percent said they decided "in the last few days."  This has the potential of supporting the Sandy Hypothesis.  After all, these late deciders broke for Obama, about 50-45 percent.  Not huge, but perhaps meaningful.  Voters were also asked how they would rate Obama's response to the storm in the role it played in their own vote.  Below, a breakdown:

Most Important Factor     15%
An Important Factor        27%
A Minor Factor                 22%
Not a Factor                    31%

These numbers suggest even more support for the Sandy Hypothesis.  After all, 42 percent gave Obama's response some kind of importance in their vote.  Among those who called it important, they cut for Obama over Romney by an almost 2-to-1 margin.

Yes, but here's what we don't know -- are the "important factor" folks also the late deciders?  And were they in key swing states?  After all, if a bunch of Illinois or California folks are in there, it really doesn't matter.  Those states were decided long ago.  I won't know more until I get my grubby little number-crunching hands on the raw data. 

My conclusion (so far): The Sandy Hypothesis remains untested.  The compelling forecast models (who were, after all, mostly right) argue the storm made no real difference despite what the conservative pundits (who were, after all, mostly wrong) say.  When I have the raw data I can crosstab it to death and examine whether late deciders were in key states and also voters who called the storm important in their decision.

My hunch?  Some did base their decision on the storm.  My other hunch?  Not enough did to explain the results seen Tuesday.  Still, a hypothesis deserves a fair test with some pretense of methodological rigor beyond the "feel it in their gut" bullshit you get from many pundits. 

And ultimately, what's wrong with basing your vote on how a president handles a recent crisis?  Nothing at all, best I can tell.  If Obama had blown his handling of Hurricane Sandy, you can be damn sure certain pundits of an ideological persuasion would've argued in favor of it playing a role in the vote. 

That's the problem with partisan pundits.  They're partisan.

Obama and the Catholics

In what will probably be a series of posts breaking down the exit poll data from the presidential election, I start today with religion.  Specifically, with Catholics. Why?  Mainly because I'm Catholic and this is my blog, but also because Catholics are viewed as a vital swing vote and they made up, in 2012, 25 percent of the electorate.  The numbers are kinda interesting.

• Obama won the Catholic vote overall, 50 percent to 48 percent.  That's a bit surprising given he was supposedly at war with the church.  Bishops, take note.

• Okay, but let's dig deeper.  Among white Catholics, Romney won 59-40 percent.  That's about the same racial breakdown for all voters, so not much to see here folks.  Let's move on.

• Among Catholics who attend church regularly, Romney won 57-42 percent.  Republicans usually win this category.

• Among Catholics who do not attend regularly, Obama won 56-42 percent.  These are mirror images of one another.  Fascinating stuff.  By the way, in the electorate there are more non-regular Catholics than regularly attending Catholics by a couple of percentage points.  In other words, Obama wins.

When the raw data are available I can break these down even further, but for the moment these rough crosstabs will have to do.  To sum up my blunt Catholic analysis, Romney wins big among Catholics who regularly go to Mass, Obama wins big among those who are less regular in their attendance.  But the kicker is there are more of the a latter than the former.

Stay tuned for more data crunching.