The New York Times and CBS News recently announced that they will begin adding online survey panels from YouGov to traditional phone surveys as part of their election coverage, arguing that "[a] deluge of cheap partisan polls has swamped a shrinking number of high-quality, nonpartisan surveys, making it hard to know who is really ahead in many political campaigns. The solution? More nonpartisan surveys."  YouGov is a  U.K.-based research firm founded in 2000 that is non-partisan and uses online survey panels.  The panel the New York Times and CBS are using counts more than 100,000 members.

This decision by the NYT and CBS has been controversial.  Until recently, public polling used random-digit dialing, as this method is considered the gold standard for producing probability--random--samples.  As the American Association for Public Opinion Research explains, "in a probability sample, everyone in the population of interest (e.g., all registered voters in a political poll) has a chance of being selected for an interview. Knowing those chances is critical to creating valid statistical estimates" and knowing the "margin of error."



But as Nate Cohn of the NYT points out, telephone surveys are not without their problems, the most important of which is declining response rates ("Only 9 percent of sampled households responded to traditional telephone polls in 2012, down from 21 percent in 2006 and 36 percent in 1997, according to the Pew Research Center"), especially among young and nonwhite voters who are "least likely to own a landline and least likely to respond to telephone pollsters."  The difficulty in reaching and interviewing people casts doubts on the ability of telephone polls to produce accurate, representative samples and do a good job of predicting elections, particularly as the people who are least likely to participate in telephone polls (young voters) become an ever greater share of the population.

YouGov does not use probability sampling.  Instead, the research firm "attempts to build a large, diverse panel and then match its panelists to demographically similar respondents from the American Community Survey, an extremely rigorous probability survey conducted by the Census Bureau. This step is intended to mimic probability sampling. But it can require significant assumptions about the composition of the electorate, including partisanship. These assumptions are contestable and based on varying amounts of evidence."  In addition, it is worth noting that the panels may not be nationally representative because not everyone is online, and non-internet users tend to be older, poorer, less educated, and Hispanic.

Nevertheless, it appears that YouGov produces results that are as good or even better than those derived from traditional survey methods, which explains why media organizations like the New York Times and CBS news have decided to incorporate it into their "diverse suite of surveys employing diverse methodologies, with the knowledge that none are perfect in an increasingly challenging era for public-opinion research."

Read more:
http://www.nytimes.com/2014/07/28/upshot/explaining-online-panels-and-the-2014-midterms.html
http://www.people-press.org/2012/05/15/assessing-the-representativeness-of-public-opinion-surveys/
http://www.aapor.org/AM/Template.cfm?Section=Reports1&Template=/CM/ContentDisplay.cfm&ContentID=6210

TeachingwithData.org resources:
Introduction to Statistical Methods in Economics (http://www.teachingwithdata.org/resource/3078)
Dancing statistics: explaining the statistical concept of sampling & standard error through dance (http://www.teachingwithdata.org/resource/3910)
Evaluating Polling Methods and Results (http://www.teachingwithdata.org/resource/3927)
Voting Behavior: The 2012 Election (http://www.teachingwithdata.org/resource/3920)