Election 2012 Polling Lesson: Internet In, Telephone Out?
Online polling is ready for prime time but Internet voting isn't, says Survey Monkey's chief methodologist.
Social Studies: Obama vs. Romney
(click image for larger view and for slideshow)
Did Gallop, Rasmussen Reports, or another telephone pollster call you in the past few months to see who you were voting for in the U.S. presidential election? More importantly, did you answer their questions?
If you did answer, you're part of a dwindling breed. According to a recent study by Pew Research, the response rate of telephone surveys -- the percentage of households sampled that actually yield an interview -- has plummeted from 36% in 1997 to just 9% in 2012.
"These challenges have led many to question whether surveys are still providing accurate and unbiased information. Although response rates have decreased in landline surveys, the inclusion of cell phones -- necessitated by the rapid rise of households with cell phones but no landline -- has further contributed to the overall decline in response rates for telephone surveys," the Pew report states.
Well, if mobile and landline phone users are increasingly difficult to reach, what's the alternative? Online polling may be the answer.
To be fair, telephone surveys proved very accurate in the 2012 U.S. presidential race. For instance, most polls showed President Obama with a persistently narrow lead among Ohio voters; in Florida, the surveys suggested a dead heat leading up to Election Day. Both predictions turned out to be spot on.
"We know the telephone works, but before it's extinct, we need to consider some of the Internet options," said Dr. Phil Garland, chief methodologist at online survey company SurveyMonkey, in a phone interview with InformationWeek.
In the weeks leading up to Election Day, SurveyMonkey conducted online polls of its users and found that its model projections came very close to the final election results. The company's three predictive models (raw and weighted) forecast 48 of 50 states correctly the night before the November 6 vote, the company says. (Two models missed Florida and Ohio; a third missed Florida and North Carolina.)
SurveyMonkey's online polls have a 30% response rate, which is much higher than the 9% average of phone polls.
SurveyMonkey users can create online polls on whatever topic strikes their fancy. There are several ways to get people to answer a poll, such as sending an email invitation or posting the survey to your Facebook wall. After completing a poll, respondents see a screen that invites them to take additional SurveyMonkey surveys.
In the 11 weeks leading up to the U.S. presidential election, SurveyMonkey invited respondents to help "predict the results" of the vote. Those who participated were then asked 11 questions pertaining to the presidential race. The SurveyMonkey questions mimicked the questionnaires of the phone polls, which typically inquire about respondents' party affiliation, age, gender, zip code, and whether they're likely voters, said Garland.
Telephone surveys usually include 1,000 to 1,500 likely voters, but SurveyMonkey's sample groups were much larger, as many as 20,000 per day, Monday through Thursday.
But do SurveyMonkey's online polls accurately represent the demographic mix of the U.S. voting population? Yes, the company says, but maybe with an exception or two. "We have slightly more people with post-grad degrees, and we have fewer people with less than a high school education. But people notoriously over-report these things," said Garland, who added that the less invasive nature of online polling may embolden respondents to pad their resumes a bit.
If the trend of declining voter participation in telephone polling continues at its current rate, the next presidential election could see a 5% to 6% response rate. "Political scientists like Nate Silver rely on historical models and telephone polling data, so if the bottom falls out of telephone polling, a lot of people are going to need to get their data elsewhere," said Garland.
This doesn't mean, however, that Internet voting is also just around the corner. "I hope we're a long ways away," said Garland, citing the potential for election fraud. "It's hard enough to track fraud as it is, and it clearly occurs sometimes in some places. Maybe it's not enough to tip elections, and maybe it's not widespread, but we'd be naïve to think that every single vote, or non-vote, is valid," he said. "The question is, if we move to an online format, how much would that increase the probability that those kinds of things could occur?"
InformationWeek is conducting our annual Outlook Survey to explore how IT leaders are planning their priorities and budgets for 2013. The results of the survey will appear in an upcoming issue as well as in an in-depth report. Take our InformationWeek 2013 Outlook Survey now and enter to win one Samsung Series 5 Chromebook. Survey ends Nov. 19.
6 Tools to Protect Big DataMost IT teams have their conventional databases covered in terms of security and business continuity. But as we enter the era of big data, Hadoop, and NoSQL, protection schemes need to evolve. In fact, big data could drive the next big security strategy shift.
Big Data Brings Big Security ProblemsWhy should big data be more difficult to secure? In a word, variety. But the business won’t wait to use it to predict customer behavior, find correlations across disparate data sources, predict fraud or financial risk, and more.
Join us for a roundup of the top stories on InformationWeek.com for the week of December 7, 2014. Be here for the show and for the incredible Friday Afternoon Conversation that runs beside the program!