Monday, October 29, 2012

All polls are not created equal

By Richard K. Barry

Rasmussen is not calling.

Those of us who anguish over the polls day in and day out tend to have conversations about polling methodologies. It sounds more exciting than it is. Anyway, some of it is well beyond me as it involves the kind of math I abandoned in my second year of university. But some of it pertains to something as simple as whether or not a given polling company calls cell phone numbers in addition to land-line numbers.

You may not be surprised that a lot of people, younger people in particular, rely exclusively on cell phones. If you don't include cell phones in your sample, you are going to miss accounting for these people. Guess which candidate cell phone users are most likely to vote for?

Pollster and political strategist Stan Greenberg at Democracy Corp makes the case, which I'll post in detail because it's important: 

The main issue is cell phones and the changing America that most are under-representing. Our likely voter sample includes 30 percent reached on cell-phones from a cell-phone sample conducted in parallel with our random-digit phone sample. Some other surveys have moved to that level and methodology, but most have not. They are missing the new America, and we’re not sure we are keeping up either.

In the real America, most Americans are now cell-phone only or cell-phone mostly users. With no one really sure what is the right proportion for the likely electorate, everyone has been cautious but that may be the riskier option.

Pay attention to this. In the last half of 2011, 32 percent of adults were cell-phone only according the Center for Disease Control that is the official source on these issues; 16 percent were cell phone mostly. But the proportion cell-phone only has jumped about 2.5 points every six months since 2008 – and is probably near 37 percent now. And pay attention to these numbers for the 2011 adult population.

  • More than 40 percent of Hispanic adults are cell phone only (43 percent).
  • A disproportionate 37 percent of African Americans are cell only.
  • Not surprisingly, almost half of those 18 to 24 years are cell only (49 percent), but an astonishing 60 percent of those 25 to 29 years old only use cell phones.
  • But it does not stop there: of those 30 to 34 years, 51 percent are cell only.

That's a lot of potential Obama supporters that some pollsters are managing to avoid.

Rasmussen, for example, a Republican-friendly polling company, doesn't call cell phone numbers and consistently find greater support for GOP candidates.

I don't have data in front of me that goes through which polling companies call cell phone numbers and at what percentages, but it seems like a pretty important question.

Back in September, Nate Silver covered the same territory finding that companies that call cell phone numbers show a better result for President Obama.

I'm just saying what we all know, which is that the development of technology is moving very quickly and those who aren't keeping up are going to get it wrong. 

All polls are not created equal, though we do seem to treat them like they are.

(Cross-posted at Lippmann's Ghost.)

Labels: , ,

Bookmark and Share


  • In Rasmussen's defense, they do a very good job of molding their interview audience to the audience they are polling. Having said that, Rasmussen has a fairly bad track record. This is why Nate Silver weights their polls lowly, even when they have a large sample. It isn't the cell phone issue directly; it is that they have not predicted well in the past.

    PPP however, is very good. They are maximally weighted in the 538 model--at least when they have a large enough sample size. And when PPP says we're losing, I worry.

    But what do you mean by, "It sounds more exciting than it is"? This stuff is great! And it so impresses the ladies! "Come on over here babe; let me regress your trend line..."

    By Anonymous Frankly Curious, at 7:50 PM  

  • Hey, don't be fresh.

    By Blogger Richard K. Barry, at 7:54 AM  

Post a Comment

<< Home