The British Election Study is pleased to announce the release of the second wave of the Internet Panel. In total, 30,219 respondents took this survey, 26,870 of these also took wave 1. We are also pleased to release the first combined panel study file of Wave 1 and Wave 2 respondents. The overall wave on wave retention was 88.9%.
Wave 2 was conducted by YouGov between 22nd May 2014 and 25th June 2014.
DOWNLOAD the data here.
Weights and samples
We are releasing three new datasets: a new wave 1 dataset with some additional respondents (BES2015_W1_v3.0), the wave 2 dataset (BES2015_W2_v2.0) and the combined wave 1/wave 2 panel dataset (BES2015_W2_panel_v2.0).
Around 30,000 respondents were interviewed in each wave (referred to as the full sample). However, we also define a smaller core sample of around 21,000 respondents in each wave that constitute a cross-sectional group which is more representative than the full sample (i.e. the range of weights is smaller than in the full sample). For cross-sectional work, where you wish to say something about the proportion of the population giving a certain response, we recommend using this core sample.
The wave 2 core sample consists of all respondents from the wave 1 core sample who also took wave 2 and an additional 3,342 top-up respondents who were not in the wave1 core sample. These top-up respondents are chosen to rebalance the core sample demographically.
We provide a core cross-sectional weight for wave 1 (wt_core_W1) and for wave 2 (wt_core_W2) as well as a full cross-sectional weight for each (wt_full_W1 and wt_full_W2) that allows the use of the full sample.
We also provide a full sample weight for the panel analysis (wt_full_W1W2). This weight applies to every respondent who took both waves. 88% of the core wave 1 respondents responded to wave 2 of the British Election Study.
All of these weights are constructed at the regional level (i.e. Scotland, England, Wales and London are weighted separately to population values) and then adjusted to account for the oversampling of Scottish and Welsh voters.
These various overlaps are represented in the following Euler diagram:
The one set of respondents who are not covered by any weights are those who are under the age of 18. This is for two reasons: first, weighting respondents under this age is relatively difficult due to the lack of relative information to weight on (occupation, years of education, income etc). Second, the under 18s should be excluded in most use cases focusing on eligible voter populations. There are important uses of under 18s such as research on early socialisation or the Scottish referendum and we will release guidance on using responses from under 18s for these purposes as soon as possible.
The survey questionnaire includes many of the same items that were asked in wave 1 in order to allow within-person comparisons over time (e.g. has someone’s party preferences changed between wave 1 and wave 2). However, we have also included new questions and removed some questions that appeared in wave 1.
Wave 2 introduced a number of new questions:
- Local and European vote choice and turnout
- Method of voting (postal/in person and whether the voter went with anyone else)
- A battery of questions about the political attitudes and behaviour of people our respondents talk to about politics
- Left/right party placement questions
- Inequality perceptions (for comparison with US)
- EU approval (for comparison with the European Election Study)
- Willingness to help other EU member states financially (for comparison with the European Election Study)
- Committee on Standards in Public Life questions on expectation of fairness in upcoming elections
- A new battery of political knowledge questions (about international leaders)
- Electoral commission voter registration questions
- New user content (this will be explained in the upcoming codebook)
- Respondents’ union membership and subjective social class (i.e. working class or middle class identity)
Because many attitudes are generally taken to be relatively, they are not asked in every wave of the survey. This makes space for the additional content while still capturing longstanding attitudes and attributes of our respondents. The following items were asked in wave 1 but were not included in wave 2 except for top-up respondents (items given to topup respondents are marked with a *):
- Left/right and authoritarian/libertarianism item sets*
- Preferences for further equality for women, black people and gay people*
- How good Thatcher and Blair were for the country
- Social desirability items*
- Wave 1 political knowledge quiz*
- All wave 1 demographic items *
- Several of the wave 1 user items
Note: The new wave 1 dataset replaces earlier versions that have been released through the BES website (older versions are available on request).