Getting It Right: YouGov and Online Survey Research in Britain

Abstract

A BSTRACT The past two decades have witnessed significant changes in how survey research is conducted in Britain. One of the most important innovations is the use of national internet surveys. Internet surveys are now used by the national media and the British Election Study to provide information on party support and the dynamics of public opinion on a wide variety of topics. The survey house YouGov has played a pioneering role in these developments. YouGov's track record of "getting it right", i.e., of providing accurate forecasts of the results of several major elections, has convinced many -not all -observers that online surveys will have a major role to play in future studies of voting and elections. This paper describes the historical development and current status of internet polling in Britain, focusing on the survey methods employed by YouGov. The paper concludes by discussing future innovations in online survey research.

Journal of Applied Communications, 2017, vol. 101, no. 1 Full Issue

“…Public opinion research often uses non-probability samples as an effective way to gain insight on population estimates (Baker et al, 2013). Non-probability samples require corrections for nonrandom selection and nonresponse, but multiple studies have shown that non-probability samples have the potential to generate results of equal or greater rigor than probability-based samples (Abate, 1998;Twyman, 2008;Vavreck & Rivers, 2008).…”

Journal of Applied Communications 100(3) Full Issue

“…Non-probability samples are commonly used in public opinion research to make population estimates and in this case was the best way to reach the population of interest: high water users. Although there are limitations in being able to generalize non-probability samples, they have been shown to yield results as good as, or even better than, probabilitybased samples (Abate, 2008;Twyman, 2008;Vavreck & Rivers, 2008). The researchers fully acknowledge the limitations of opt-in panels and the lack of coverage associated with on-line survey designs.…”

“…Non-probability samples are often used in public opinion research to make population estimates . While non-probability samples require adjustments for nonrandom selection and nonresponse, previous literature has shown non-probability samples have yielded results that are as good as or even better than probability-based samples (Abate, 1998;Twyman, 2008;Vavreck & Rivers, 2008).…”

“…Weighting was conducted post hoc using post-stratification methods (Kalton & Flores-Cervantes, 2003) based on the 2010 Florida census data to ensure the sample reflected the adult Florida population and to provide results intended to approximate the population of interest. When using non-probability opt-in samples post-stratification weighting methods have been found to yield results that are as good as those obtained using probability-based samples (Abate, 1998;Twyman, 2008). Data was analyzed using SPSS statistical software including descriptive statistics and ANOVAs.…”

Change-Point Analysis of the Public Mood in UK Twitter during the Brexit Referendum

“…The outcome of the referendum came as a surprise to many, with the final YouGov poll before the result incorrectly giving Remain a four point lead [22], bookmakers predicting an 86.29% chance of a Remain victory just hours before the polls closed [23], and Nigel Farage, a leading proponent of the UK leaving the EU, conceding as the polling stations closed that it "looks like Remain will edge it" [24].…”