The VOTER (Views of the Electorate Research) Survey was conducted by the survey firm YouGov. In total, 8,000 adults (age 18+) with internet access took the survey online between November 29 and December 29, 2016. The estimated margin of error is plus or minus 2.2 percent. YouGov also supplied measures of primary voting behavior from the end of the primary period (July 2016), when these respondents had been contacted as part of a different survey project.
These respondents were originally interviewed by YouGov in 2011 to 2012 as part of the 2012 Cooperative Campaign Analysis Project (CCAP). In that survey, 45,000 respondents were first interviewed in December 2011 and were interviewed a second time in one of the 45 weekly surveys between January 1 and November 8, 2012. After the November election, 35,408 respondents were interviewed a third time. We invited 11,168 panelists from the 2012 CCAP. Of those invited, 8,637 (77 percent) completed the 2016 VOTER Survey.
The 2012 CCAP was constructed using YouGov’s sample matching procedure. A stratified sample is drawn from YouGov’s panel, which consists of people who have agreed to take occasional surveys. The strata are defined by the combination of age, gender, race, and education, and each stratum is sampled in proportion to its size in the U.S. population. Then, each element of this sample is matched to a synthetic sampling frame that is constructed from the U.S. Census Bureau’s American Community Survey, the Current Population Survey Voting and Registration Supplement, and other databases. The matching procedure finds the observation in the sample from YouGov’s panel that most closely matches each observation in the synthetic sampling frame on a set of demographic characteristics. The resulting sample is then weighted by a set of demographic and non-demographic variables.
On average, YouGov’s methodology produces less bias across a series of benchmarks than do other methodologies using online samples. For more information see: