by Michael Handley (Updated Sat., Dec. 1, 2012 @ 9:38 a.m.)
The Romney campaign and GOP politicos, like Carl Rove, absolutely believed Romney was headed for victory on Election Day. Did Romney's polls skew toward the right by as much as an eight point spread because his internal pollsters fell into the wireless (cell phone) only polling gap trap?
A few journalists have reported that team Romney’s internal polling showed him comfortably moving into the lead in North Carolina, Florida, Virginia and a few other swing states. Add in Ohio, where Romney's internal polling had him in a statistical tie with Obama, and team Romney thought they were on track to lock in enough electoral votes to win the White House. GOP confidence in these numbers was such that Romney decided to not prepare a concession speech in advance.
An article by Noam Scheiber published today in The New Republic provides some details about the Romney Campaign's internal polling snafu.
The Internal Polls That Made Mitt Romney Think He'd WinWe know polls that included properly weighted cell phone only voters gave Obama the lead in all battle ground states through the summer and fall. In late September Nate Silver wrote that Obama has shown a clear lead in the 16 cell phone-inclusive polls of seven top battleground states taken since the convention. A national survey conducted by Pew Research Center for the People & the Press, in mid September, which include properly weighted cell phone only respondents, found that Obama had an eight-point lead over Mitt Romney among likely voters.
...In an exclusive to The New Republic, a Romney aide has provided the campaign’s final internal polling numbers for six key states, along with additional breakdowns of the data, which the aide obtained from the campaign’s chief pollster, Neil Newhouse. Newhouse himself then discussed the numbers with TNR.
...Newhouse and some of his colleagues have said that the biggest flaw in their polling was the failure to predict the demographic composition of the electorate. Broadly speaking, the people who showed up to vote on November 6 were younger and less white than Team Romney anticipated, and far more Democratic as a result. "The Colorado Latino vote was extraordinarily challenging," Newhouse told me. "As it was in Florida.
...One Romney aide told me that he ran into Tagg Romney, the candidate’s eldest son, as the results came in on election night. “He looked like he was in a complete state of shock,” the aide said. “[As if] these numbers cannot be real.”
Read the full story @ The New Republic.
The cell phone only trend is increasingly problematic for pollsters for several reasons.
The first major problem is that the fastest and cheapest polling research is done with automated Interactive-Voice Response (IVR) systems, sometimes called robopolls. Because they are fast and cheep, too many politicos still think robopolls are "good enough." Robopolls use automated dialers and a recorded voice to ask questions, with respondents indicating their answers by pushing a button on a touch-tone keypad or by giving their vocal response, which is interpreted by voice-recognition software. It is illegal under federal law to use an automated robopoll system to call cell phones, so robopolls sample only landline phone respondents. People who still answer landline phones turn out to be the most conservative segments of the electorate. Polls that include (progressive/liberal) cell phone only respondent "households" are more expensive and time consuming to conduct because live-interviewers must call cell phone respondents from prepared phone number lists - and those lists are expensive to buy.
A second problem for pollsters polling a particular state or district within a state is that cell phone owners are mobile. For example: A young person who gets their first cell phone number in Los Angeles (with an LA area code) and later moves to Dallas, then after a few years moves to San Antonio, typically always keeps their LA-based phone number. Preparing cell phone number call lists for a particular area code no longer guarantees that people who live in that area code can be contacted.
A fourth problem is that the way pollsters must obtain and select cell phone (only) numbers to live dial reduces the "randomness" of the people polled, which often skews results.
So, even when pollsters include cell phone (only) respondent "households" in their samples, with all the problems added together, their sample universe of completed polling surveys is much smaller and less random than it was in the late 1990's. This decreases the certainty and increases the variability of polling research results done by different pollsters at various intervals. (This is why Nate Silver's method of averaging many polls together allowed him to accurately forecast that Obama would win with 330 electoral votes.)
The number of Americans who have no landline phone service, who rely solely or mostly on cell phone service, exceeds 60 percent among all youth demographic groups, according to the latest Center for Disease Control (CDC) National Health Interview survey.
Broadly speaking, the people who live a cell phone only lifestyle are younger, less white and single - exactly the demographic groups that team Romney pollsters seem to have missed in their polling.
It is only the older whiter (increasingly more conservative GOP) age groups that can still be reached by land line phone for voice and automated robo-call survey and GOTV phone bank contact. The following pie charts depict the national cell phone only demographic patterns among all age groups of each group. Adding in the "cell phone mostly" numbers would increase the solid blue area by more than 16 percent for the minority demographic groups.
Analysis by Pew Research Center in 2010 shows support for Republican candidates is significantly higher in polls calling only or mostly landline respondents, than in dual frame (landline and cell phone respondent) surveys that combined properly weighted landline and cell phone respondent polling.
Pew Research found in 2010 up to a 10 point conservative bias gap between those who could be reached by landline and those who could be reached only by cell phone. Those who have gone cell phone only hold significantly more progressive / liberal political views than those who can still be polled by landline, live or automated robo calls, whether they also have a cell phone or not. This gap has likely widened over the last two years.
There is significant difference between the states in how the cell phone only and mostly verses landline numbers break out. In Texas, 39.9% of all households statewide were cell phone only at the end of 2011, with cell phone mostly households (19.2%) bumping the number of households across all age groups that rely primarily or exclusively on mobile phones to 59.1%.
Metro areas often have even higher cell-only adoption rates than the state as a whole. At the end of 2011, 53.3% of all Dallas County households were cell phone only with an additional 17.0% that were "cell phone mostly," according to a CDC survey published in October 2012.
Therefore, the most likely reason that team Romney thought they would comfortably win the battle ground states is they were running a 20th century polling operation for 21st century election.
This is why Carl Rove was so confused on election night and refused to accept Fox News' election call for Obama - watch the video.