By Gregory B. Hladky
1:00 PM EST, November 13, 2012
When you look at the final numbers, pollsters like Quinnipiac University's Douglas Schwartz came out of this election looking pretty good.
Predictions about Obama's victory were nearly on the money. Republican candidate Linda McMahon's U.S. Senate debacle was nailed. And key swing state voting trends were well plotted.
So why is this guy so damned worried about cellphones?
"For us, it's more and more about cellphones," a weary Schwartz commented a few days after this year's polling-and-voting frenzy ended. "That's the biggest challenge by far."
The reason for all that concern was detailed in some of the numbers that emerged from exit polling of voters.
About one-third of all people who cast ballots this year don't have landlines at all. And those cellphone-only folks "tend to be disproportionately young people," says Schwartz.
People ages 18-29 represented about 19 percent of everyone who voted — almost the same as four years ago. Democrat Barack Obama did much better among young people than Republican Mitt Romney (though not as well as Obama did in 2008 against John McCain), and it was one of the reasons the Democratic incumbent won.
So more young people are voting. They could decide lots of future elections. Those same younger voters are increasingly only using cellphones. Which means pollsters like Schwartz are constantly worrying about whether they're reaching enough of those young cellphone-only types to get accurate readings of their political temperature.
Schwartz says it's tougher for pollsters to reach cellphone users for several reasons.
Each number must be individually dialed rather than having it done automatically by computer. The person on the cell could be crossing the street or driving or anywhere that's not "a comfortable or safe place for an interview," he explains, requiring a callback. There's also the possibility that the cell user has moved out of state but retained his or her old area code, which means the call is no good for a state poll. And lots and lots of kids not old enough to vote have cells, and unlike a landline they usually aren't going to hand the phone over to a voting adult.
All of which leads to the interesting question of robo-polls.
The Quinnipiac University Polling Institute and other traditional public opinion survey operations use live people making telephone calls and doing interviews. Robo-polls use computers to dial up potential voters and ask the people who pick up to answer prerecorded questions by pushing a button on the phone.
Robo-polls are less expensive than traditional polls because you don't have to pay all those live interviewers. They're often quicker, because you don't have to spend time asking all those questions. That combination of cheap and fast has led to an explosion in the number of robo-polls being conducted.
"It appears there were more robo-polls [this election cycle] than in the past," says Schwartz, and it's a trend that worries him.
The problem is that robo-poll operations can't call cellphones, and Schwartz believes that's a serious and potentially fatal flaw. "If you can't poll cell users, then you're going to be off in your predictions," he insists.
So far, robo-pollsters like Public Policy Polling and YouGov and Ipsos/Reuters have managed to compensate for the lack of cell-using voters in their surveys by adjusting their formulas.
In fact, those three automated polling operations were rated as tops for accuracy in this election by a Fordham University study that came out last week.
(The Quinnipiac Poll wasn't included in the Fordham study because it focused only on national surveys. Schwartz says the Q Poll concentrated its efforts in the closing weeks of the campaign on surveys in key swing states.)
Schwartz isn't convinced that robo-polls will be able to remain accurate for much longer. Cell-only households are continuing to expand as a percentage of the electorate.
"At some point, there will be a breaking point," Schwartz predicts. "They (the robo-pollers) won't be able to do it just using landlines."
The Q Poll has become Connecticut's premier polling operation in the 24 years since it got going. Schwartz became the poll's director in 1994. It began as a Connecticut-only survey, then expanded to polling in New York and New Jersey as well. In recent years it has begun to garner a national reputation.
One major factor in the rise of the Q Poll has been the size of its voter samples. Schwartz says his interviewers always seek to have opinions from at least 1,000 likely voters — a pool that cuts way down on a poll's margin of error.
The other major opinion survey in this state is the Hartford Courant/UConn poll, which often samples opinions of just 500 voters. The Courant/UConn operation did fairly well in that Fordham study, coming in 13th, well ahead of better known pollsters like Gallup and Rasmussen.
Schwartz says the Q Poll will continue to focus on the most critical battleground states in elections rather than compete with pollsters that try to gauge the nation as a whole.
In this election, the Quinnipiac University survey targeted major swing states like Florida, Pennsylvania and Ohio, and Schwartz says those forecasts came within a couple of percentage points of the actual votes — which were key for Obama.
"That's where the election was going to be decided," Schwartz says. "That's become our niche, if you will… and we continue to believe that's what really matters."
Copyright © 2013, WTXX-TV