As we smoothly segue from a carefully choreographed sports convention to carefully choreographed political conventions, we stop examining the medal count and start examining the political polls. As red- and blue-staters do, Obamaniacs are having bleeding-heart palpitations and McCainiacs are revving their NASCAR engines. Recent polls have the two egomaniacs ("No, I should be the leader of the free world!") in a statistical dead heat.
But to paraphrase defense witness Marisa Tomei suddenly realizing the import of the photographed tire marks in My Cousin Vinny, the polls are wrong. Why? Pollsters don't poll people with cellphones. Read on to find out what this means.
But before we continue, this is not a political rant favoring any one candidate (Obama). It doesn't matter who I support (Obama). This is a completely objective report (Obama), and my personal political preference (Obama) has no bearing on this report (Obama).
Now let's talk numbers.
Who Ya Gonna Call?
It is against the law to make automated calls to cellphones. Even if pollsters wanted to call you on your cell, they don't have your number since there are no centralized cellphone lists. Pollsters could make random calls, but there's no way to qualify who's at the other end of the call. So presidential pollsters don't call people with cellphones.
One of the nation's most respected polling companies, Pew Research (if you need to, take a minute to drain your brain of all Pepe Le Pew puns), was concerned its presidential poll results would be off as a result of this cellphone gap. So late last year Pew conducted a study on the cellphone effect. Pew found that "[o]f those reached on a cellphone, 312 people (or 37%) reported that their cellphone is their only phone." Pew's conclusion: "Including cell phones [sic] makes little difference in polling results."
This Pew conclusion stinks.
Defining the Numbers
A New York Times/CBS poll from July found under-30 support for Obama leaned 48% versus 36%for McCain. A June Gallup Poll showed Obama with a 59% to 32% edge among 18- to 29-year-olds.
In Pew's research, 12% of those interviewed by landline were between 18 and 29, compared to 46% of those interviewed by cell. The demographics are even starker when more finely parsed: 70% of 18- to 24-year-olds were cell-only.
This all pretty much confirms what we already know: young people are more likely to have only a cellphone, and young people are more likely to support Obama. Ergo, since cellphone owners aren't polled and cellphone owners make up a significant portion of the Obama's support, the polls are wrong.
Of course Pew has to come to its no-harm, no-foul conclusion. It doesn't include cellphone owners in its political polls so it has to rationalize why it doesn't have to. It's like cigarette companies producing scientific tests proving their products don't give you cancer. One clue about Pew's recognition of its rationalized conclusion is the phrase "little difference" in describing the potential impact of not including cellphone owners in its polling. Why not "no difference"? More importantly, how big of an impact can little difference have?
Pew found that 52% of its 18-29 landline sample were either registered Democrats or lean Democratic, compared to 59% of cellphone-only. That's a seven-point difference. Plus, 35% of landliners between 18 and 29 were Republican or leaned Republican, compared to 32% of cell-only 18- to 29-year-olds. Perhaps not coincidentally, the Democrats' 59% to 32% edge in the cell-only 18-29 group is exactly the same as Obama's edge in Gallup's 18-29 poll.
I'm no mathematician, but this seven-point "little difference" has gotta translate to a couple of points when folded into an aggregate poll. And a couple of points turns a 46%-45% statistical dead heat into a near 48%-43% runaway.
Vote Early, Vote Often
This disconnect between pollsters and cell-only pollees will become increasingly skewed. CTIA says as of December 2007, 15.8% of American households were cellphone-only — no landlines, aka POTS (Plain Old Telephone Service), twice what it was in December 2005. This movement to cell is reflected in sales of cordless phones, which have been dropping around 25% a year according to market research firm NPD. Conversely, last year 10 times more cellphones were sold than cordless phones.
Of course, for any of this to make a difference on November 4, a lot more young people have to pause their texting, blogging and Wiiing, rollerblade to a polling place and actually vote. According to those Pew folks again, only 41% of 18-year-olds are registered to vote and only 47% of 18- to 29-year-olds turned out in the 2004 election compared to 64% of the general population.
Until election day, Obamaniacs, buck up. Count on not being counted until your votes are.