Posts Tagged ‘polling’

4th December
written by Sean Noble








There are many theories, excuses, and explanations about what went wrong for Republicans on Election Day.  Some are fairly accurate, others wildly off-base, and most are premature.  I think some fact-based analysis is in order, and I’ll include my opinion as well at the end. Much of the data comes from either exit polls or post election polls I’ve read.

Of ballots currently counted (33 states have certified the result), Barack Obama leads Mitt Romney 65,362,092 to 60,715,413 – a lead of 4.6 million votes or 3.6% (50.94% to 47.32%).

In 2008, Obama defeated John McCain by roughly 9.5 million votes or 7.3% – 69,456,897 to 59,934,814.  Romney exceeded McCain’s vote total, but Obama is about 4 million votes shy of his 2008 total. 2012 turnout was less than that of 2008, both on a raw vote basis and when measured as a percentage of the eligible electorate.

With that said, turnout was closer to 2008 levels in the key battleground states than in non-competitive states.  In the 12 most competitive states, the current cumulative raw vote is more than the total from 2008, whereas the balance of the country is down about 5.7% from 2008.

A few key takeaways:


  1. Support for Obama was weaker in 2012 than 2008.

Obama is the first President since FDR to win reelection with fewer votes than his initial election.  His share of the vote fell in every one of the 12 most competitive states, including as much as 3.7% in Wisconsin.


  1. Romney was unable to convert millions of voters who abandoned Obama.

If we were to look at this simplistically, we could argue that Obama didn’t win the election. Rather, Romney lost the election by failing to motivate the millions of voters who couldn’t vote for Obama to vote for him instead.  Tellingly, a clear majority (52%) of voters said the country was on the wrong track, but Romney only won 84% of these voters.  Obama won 93% of those who said the country was going in the right direction.


  1. Romney won independents by 5 points on Election Day – a 13-point shift from 2008 when Obama won independents by 8 points, and Romney still lost.



Republican pollsters, pundits, strategists and campaigns got it wrong, seriously wrong.  I was one of them.  Going into Election Day I completely believed Romney was going to win.

We relied on data that turned out to be inaccurate.  We were convinced, based on polling results and historical comparisons that Romney was going to win.  That we were shocked at the outcome demonstrates that we relied on serious miscalculations.

There has been significant discussion about how badly Republican and center-right polling got it wrong.  That is true.  The polling that Romney relied on, the party committees relied on, and that Senate candidates relied on was all faulty because it had a common problem – a severe underestimation of Democrat turnout.

Early indications are that this error was driven by use of overly restrictive likely voter screens and an expectation that youth and minority voters would plateau at 2008 levels, rather than continue to grow.  It wasn’t the methodology per se – that is, the error wasn’t necessarily because some pollsters used random digit dialing and others used specific lists of registered voters.  Ultimately, it appears a combination of factors resulted in “weight” being added to Republican and independent voters.

Republican and conservative pollsters tended to use a likely voter model that split the difference between 2004 and 2008, a model that leaned Democrat narrowly and made winning independents the critical piece of a winning coalition for Romney and other Republican candidates.

By contrast, Democratic and news organization polling estimated a turnout model more like 2008, but with continued growth in minority voting as a share of the electorate.  This model was closer to reality, and therefore their polling was closer to the mark.




Just as there was never a silver bullet for defeating Obama, there is no single cause for this loss.  It was a culmination of a number of weaknesses and missed opportunities that, in combination, proved fatal.  Below are several of these factors in a list that is by no means exhaustive:


  • Romney’s decision to not rebut the attacks by Obama on Bain, tax returns, and offshore accounts in the summer proved fatal.  Romney’s delay in defining himself was a major strategic error. Given the amount of money that was spent attacking Obama by the outside, any money spent attacking Obama by Romney was wasted. He should have been promoting himself.  In the exit polls, voters split somewhat evenly into four camps when asked what was the most important quality in a Presidential candidate: “Shares my values” (27%), “Strong leader” (18%), “Cares about people like me” (21%), and “Vision for the future” (29%).  Romney won all but “Cares about people like me” by strong majorities – between 54% and 61% of voters in each group.  However, he lost by 63 points(!) on “Cares about people like me,” which Obama won 81% to 18%.


  • In a post-election poll, swing voters were asked to describe both Romney and Obama in a word or phrase. For Romney, the most prevalent word was “liar” and for Obama it was “ineffective.”  Since elections are a matter of whom do you trust, even “ineffective” trumps “liar.”  Tellingly, this was among swing voters.


  • The much-discussed “gender gap” is better understood as a “minority gender gap.”  Women made up 53% of voters.  Overall, Obama won women by 11 (55-44) and Romney won men by 7 (52-45).  However, Romney actually won white women (38% of voters) by 14 points (56-42).  The problem is that he lost African American women (8% of voters) 96-3, and Hispanic women (6% of voters) 76-23.  By comparison, Romney “only” lost Hispanic men (5% of voters) 65-33 – 21 points closer (net) than Hispanic women.  Put another way, if the Hispanic gender gap had been the same as the white gender gap, Obama’s lead would be cut by nearly 20%.  In 2004, the gender gap was nearly identical for all races and ethnicities.


  • Late-breaking undecided voters defied history and broke to the incumbent President.  Obama won those who decided on Election Day or the few days prior 50-44.  Past elections, including 2004 when Kerry won this group 53-44, found the opposite to be true.  Conventional wisdom has long held that challenger candidates benefit from late deciders, who, by not already having joined the incumbent’s column, have in some sense already decided to consider an alternative.  This “incumbent rule” is clearly no longer a rule – at least in presidential elections.  As Election Day neared and the campaign saw Romney and Obama tied or Obama narrowly ahead but at 47% or below in key states (albeit using admittedly problematic turnout models), they expected Romney to carry a strong majority of the remaining voters.  Most campaign strategists and pollsters expected this effect to help turn the tide in a number of states that, in reality, Romney ended up losing by 5-6 points (e.g., Iowa, Wisconsin, Colorado, etc.) and to firm up what should have been his victories in Florida and Virginia.


  • Urban turnout was higher than expected.  Employing a long-term ground game and using constituencies of identity politics proved to be effective for Obama in these communities.  For example, in Wisconsin, Obama won by nearly identical margins in 2008 and 2012 in Milwaukee and Madison combined – roughly 300,000 votes.  In the balance of the state, Obama won by 111,000 votes in 2008, but lost by 96,000 votes in 2012.


  • Superstorm Sandy slowed Romney’s momentum and allowed Obama to look Presidential in the days leading up to the election.  Perhaps this is an explanation for the late-deciders breaking to Obama, but it also changed the narrative and preempted Romney’s message that Obama was campaigning on “small ball” issues rather than an overarching vision.






  • Republicans need to get real about grassroots campaigning – and engage in their own style of “community organizing.”  This should be an ongoing effort that does not require rebuilding from scratch each cycle.


  • Candidates matter, and Republicans need to engage in a robust candidate recruitment and training effort.  They need to avoid situations where massive investments in TV ads, grassroots efforts and mail are squandered because of a candidate who says something stupid.  While “47 percent” and “legitimate rape” were certainly not the only causes of the 2012 Republican collapse, they did play a role and had effects beyond the outcomes in the Presidential and Missouri Senate races.


  • Republicans need to get serious about data – collecting it, understanding it, and using it. Obama’s campaign had a group they called “the scientists” to plow through the huge amount of data they collected from polling, the field, public sources, the voter file, fundraising, and many other sources.


  • Finally, Republicans need to recognize that the demographics are changing and they need to do something about it – most notably with the growing Latino community.  I’m not arguing for amnesty (although, I would point out that Reagan campaigned on a platform of amnesty against the union-backed Walter Mondale in 1984), but we need to do something that reaches out to Hispanics in a credible and sincere way.
22nd October
written by Sean Noble

The headline of a London Telegraph story reads: Barack Obama sees worst poll rating drop in 50 years.

That’s pretty ominous and fatalistic sounding, but there are a few details that make it somewhat misleading.  Here is the bottom line of the story:

Gallup recorded an average daily approval rating of 53 per cent for Mr Obama for the third quarter of the year, a sharp drop from the 62 per cent he recorded from April.

His current approval rating – hovering just above the level that would make re-election an uphill struggle – is close to the bottom for newly-elected president. Mr Obama entered the White House with a soaring 78 per cent approval rating.

 Yes, he is taking on water, no question, but when there is an artificially high expectation, the grind of Washington, D.C. is going to take a heavy toll.  Obama’s election was a watershed moment in U.S. history: we elected the first African-American in our history.  That’s a big deal, and with his campaign of “hope and change” most people’s expectations were very high.  Even many voters who cast a ballot against Obama had high expectations.  Think about it, his approval was at 78 percent on inauguration day.  That is a remarkably high number for a President when there isn’t a national crisis that rallies the nation (e.g. Bush 41’s approval in the 90’s after Desert Storm and Bush 43’s approval in the 90’s after 9-11)

So, of course Obama’s numbers were going to come down.  Yes, he certainly had something to do with his own slide in popularity. The American people are very uneasy with the rush of government involvement in huge swaths of our economy (stimulus, auto bailout, bank bailout and now the march toward a government takeover of health care) and the incredible amount of money that is being shoved out the door to pay for all this.  Remember when Clinton was President and we (read Republicans) were screaming bloody murder over a budget deficit of $300 billion?  That’s chump change.  This year the U.S. budget deficit is an eye-popping $1.75 TRILLION.  That’s almost SIX TIMES larger than Clinton’s deficits.

So, yes, Obama has seen huge fallout in his numbers, one, because he started so high, and two because he is trying to too much too fast and Democrats in Congress are not helping him.

23rd September
written by Sean Noble

“The only poll that counts is the one taken on Election Day.”  It’s a line used thousands of times a year, and I’m pretty sure it was first used by the late Stephen Shadegg – campaign extraordinaire and father of Arizona Congressman John Shadegg.

So the latest poll to pop up, showing Goddard wiping the floor against any Republican in next year’s Governor’s race, should be taken with a grain of salt… or better yet, with a salt lick.

Here is the totality of what you need to know about the efficacy of the poll:

“The pollster is known as a Democratic firm, and the survey was automated and done via telephone.” (emphasis added)

So this was an autodial, robo-call?  It is a known fact that Republicans have less tolerance for automated phone calls than Democrats (ok, maybe not a “known fact” but anecdotally, it takes more calls to Republican households than Democrat households to get the same number of respondents on a robo-call survey – trust me, I’ve seen it.)

Given that this was an autodialed robo-call, (which would cost all of about $500 to conduct) I find it a little surprising that Arizona Republic reporter Matt Benson would write on it – and wait until the 11th graph of the story to indicate that it was a robo-call survey. 

I’ll make Matt a deal – I’ll commission a robo-call survey of 600 Arizonan’s and I’ll give him the exclusive to write the story – which I expect to be as long as this one.

You’ve got my number Matt.