Random sampling and political opinion polling

Publication type

Conference Paper

Author

Luke Taylor

Publication date

Summary

The Inquiry into the 2015 general election opinion polls found that the primary cause of the polling miss was unrepresentative samples. Understanding Society includes questions on political support, and the large sample size of the study allows for results to be disaggregated by month (based on interview date). Understanding Society therefore provides us with a random probability sample which can be used to examine the change in political support over time and against which the performance of opinion polls can be compared. To ensure that the results are representative of the general population, each “monthly sample” has been post-stratified by demographic variables which correlate strongly with political support (using the survey cross sectional weight as a pre-weight). In the run-up to both the 2010 and 2015 elections, Labour support is consistently found to be above Conservative support (although these differences do not tend to be significant at the 95% level). For instance, in the month immediately preceding the 2015 election Conservative support is estimated at 34.8% (29.9 – 40% at the 95% level) whereas Labour support is estimated at 38.8% (33.5% – 44.5%). It is only after each general election that Conservative support as measured by UKHLS is consistently above Labour support.

Subjects

Politics, Public Opinion, Elections. Electoral Behaviour and Survey Methodology

Links