Lurking beneath much of the polling we see is an enourmous problem - cell phones. Let's start by defining the problem. In 2010 Pew wrote the following:
"The latest estimates of telephone coverage, released last week by the National Center for Health Statistics, found that 25% of households (and 23% of adults) in the second half of 2009 had no landline service and only cell phone service (just 2% of households had no telephone service of any type). For certain subgroups in the population, the numbers are considerably higher: 30% of Hispanics are cell-only, as are 49% of adults ages 25-29"
The problem for pollsters is that it is much more expensive for a variety of reasons to poll cell phones. So the question on the table is simple: Are pollsters using land line only surveys undercounting Democrats, since the young and African-Americans are more likely than others to have only cell phones?
The question became more immediate for me when I looked at the latest poll from Michigan this morning. At first blush the poll looks like trouble: it shows Obama leading by a narrow 46.9 to 45.5 margin. On the heals of a recent EPIC-MRA poll it suggests that Michigan might be a dead heat. But then, if you read down below, you get the following statement:
83% of the people that the pollster talked to were over 51!!. They talked to virtually no one under the age of 30. They had so few respondents under the age of 30 that even if they wanted to adjust for their increadible error, they couldn't because the sample size of the people that they did talk to is too small to be representative. This paragraph is a pollster literally telling you they DO NOT have a representative sample.
It was really an amazing paragraph to read. Lest you think this is just a problem with them, let me direct you to SurveyUSA's website. They have been on top of this issue for a while, but let's take their analysis of Marist polling. As the noted with respect to Marist's Colorado poll:
In Colorado, the latest NBC News / Marist College poll finds Barack Obama a nominal 1-point ahead of Mitt Romney 46% to 45%. But 1 of the big stories in this Marist Poll, as in Marist polls released in Nevada, Virginia, and Florida, is how differently home-phone respondents and cell-phone respondents vote. In Nevada, cell respondents are 18 points more Democratic than home-phone respondents.
Why does all of this matter so much? Here is a comparison of state polls conducted by different pollsters. I have been doing this since 1996, and I have never seen anything close to divergence between different pollsters. How you calculate the current state of the race depends entirely on whether you believe Rasmussen (in which case Romney has a small lead) or PPP (which has Obama up 8 in their latest national poll). While I am always suspicious of Rasmussen, I am starting to believe the the source of the difference between the pollsters is really about cell phones. And no, Rasmussen does
not poll cell phones.
There others suspicious as well. Charlie Cook in a tweet about 10 days ago cautioned against reading too much from pollsters who don't poll cell phones.
I am going to study this issue mroe deeply over the next week - but there is a real reason to think polling is understating the true level of support for Democrats.