In order to allow the community to begin using the British Election Study data as soon as possible, we have released the data without some of the eventual features we intend to implement. These include:
- Full coding for occupations: the occupation question was asked in an open ended format and is currently being manually classified.
- Manual coding for the most important issue question: the MII question is open ended. The BES team intends to work on a unified schema for the MII question in the new data as well to some of the older surveys. In the meantime we have released automatic issue codings by using machine learning on the older surveys.
- Data linking: the British Election Study intends to merge in data from several other sources including candidate studies and census area statistics to allow for a fuller picture of an individual voter’s context.
- Constituency characteristics: marginality, past vote share, candidates (note that parliamentary constituency is already included as a variable in the current release)
- Question orderings: the order of questions and rows was sometimes randomized within the survey. This information will be made available as a supplemental file in future.
- Question timings: the amount of time a respondent spent on each page was also recorded by YouGov, we also intend to make this available as a supplemental file in future.
In addition, the British Election Study data is currently in an early version so users are encouraged to check their variables carefully and report any problems they find.