Let me reiterate a prediction I have made previously: The 2012 election will be the last in which polling will rely on surveying landline phones.
Today, according to Pew Research, “fully a quarter of the U.S. adult population now [relies] solely on cell phone service.” Pew has also found that those most likely to fall into that group are young, African-American or Hispanic. Polling cell phone users is problematic for practical, economic and legal reasons.
At Zogby International we believe the future of polling lies with the Internet and establishing online panels both large and diverse enough to give reliable samples and results. We began interactive polling 12 years ago on an experimental basis and with due diligence began offering it to private clients and for our own uses to measure political attitudes and voter preferences.
Today, I can say with great confidence that our ability to use our online panel in national surveys is just as reliable as landline phone polling, and that it will soon emerge as more reliable.
Look at the numbers. Here are the averages for all polling firms as compiled by Pollster.com and Real Clear Politics (RCP) in early July for three of the standard measures of political opinion: presidential approval rating, generic congressional ballot and nation’s direction right or wrong. With these are results from the Zogby Interactive survey taken most closely to that time (June 28).
Presidential approval average: Pollster.com 45.2 percent, RCP 46.3 percent, Zogby Interactive 46.3 percent.
Congressional generic: Pollster.com, Republicans lead, 44.2 percent to 42.8 percent; RCP, Republicans lead, 43.8 percent to 42.8 percent; Zogby Interactive, Republicans lead, 42.5 percent to 41 percent.
National direction: Pollster, right 31.1 percent, wrong 60.9 percent; RCP, right 31.2 percent, wrong 62.2 percent; Zogby Interactive, right 28.9 percent, wrong 59.5 percent.
National horserace interactive polling we did in 2008 showed the same ability to match up with results from phone polling. Our last such poll in 2008 was on Oct. 4, and it showed Barack Obama with a 48 percent to 44 percent lead over John McCain, which was well within the margin of error of telephone polls taken at the same time. Interactive polling in states has also been successful, but not as much as in national polls. In
2006, our interactive polls accurately predicted the winner in 17 of 18 U.S. Senate races.
We went out on a limb in 2008 and ran interactive polls in a number of presidential battleground states, again concluding these several weeks before the election. For most, our results were within the margin of error of the averages from landline polls.
The demographics and size of some states make interactive polling more problematic. We do not have sufficient sample size within our panel in some states, and others have a population diversity that makes it more difficult to achieve a representative sample.
However, these are temporary problems. As other polling companies will learn as they enter the field of interactive polling, building a large and representative sample that can reliably poll states and other population sub-sets takes time and effort. Right now, reaching that goal is among our company’s highest priorities. We are confident it will be achieved.
As we have journeyed into the new world of interactive polling, our company and I have taken a number of potshots. The greatest misunderstanding has been about our methodology. As I have reported here previously, our samples are randomly drawn from a pool of several hundred thousand potential respondents. Survey respondents are invited to take a particular survey, very much like telephone poll respondents are called and asked to participate. A random probability sample of our representative panel receives email invitations to participate, with a link taking them to the survey on Zogby’s secure servers. Respondents never know when they are going to get an invitation to take part in a survey, and they never know what the subject matter is going to be until they click on the invitation link.
The move into interactive polling has been well worth it, despite the criticisms. Others will inevitably follow, and my fondest hope is that together as political and social scientists, we will carry on the valuable job of measuring public opinion.
John Zogby is president and CEO of the polling firm Zogby International. You can post comments on political topics in the Zogby Forums at Zogby.com.