The BBC commissioned ComRes to recruit a representative sample of the British electorate for the TV debate in Cambridge on 31 May 2017.
ComRes recruited 155 people for the final debate audience of 134. Recruitment was conducted between 15 and 30 May 2017. The recruitment process was stringent, securing a balanced audience using recognised, high quality and best-practice recruitment techniques as outlined below.
ComRes has a strong track record of recruiting for similar events, having done so in the recent past including for Channel Four's EU Referendum Debate in 2016 and the BBC Scottish Referendum Debate in 2014.
ComRes used a two-phase online and telephone approach to recruitment, ensuring that all respondents were screened thoroughly to meet the agreed criteria. Audience members were selected using a detailed recruitment questionnaire (including various questions to ensure consent and whether audience members wanted to ask a question to the panel).
The recruitment questionnaire was designed by ComRes and approved by the BBC, and included a series of questions to identify demographics and political attitudes. Minimum and maximum quotas were set to ensure a good spread of the voting public as below.
Demographic questions were designed to ensure a balance on age, gender, ethnicity, socio-economic status and working status.
Adults who had campaigned politically at any election from 2014 onwards were automatically filtered out and not recruited, so as to avoid recruiting 'activists'.
Respondents were asked questions on their awareness of the forthcoming General Election and their likelihood to vote, to ensure they were politically engaged.
In addition to using this detailed set of screening and profiling criteria, ComRes ensured that the audience was representative of the result of the 2015 General Election while also taking into account current voting intention trends.
The Conservative and Labour - as the UK's largest parties - had the largest share of supporters in the room, roughly equivalent to each other. Smaller parties were supported by an appropriately lower number.
The audience also included undecided voters, including people who voted Conservative, Labour or for other parties at the last Election but who are now considering changing their vote.
In addition, ComRes also ensured a 50-50 split of Remain and Leave voters at the EU Referendum, including a small number of people who did not vote.
While the above criteria were not intended to produce a nationally representative sample in the strictest sense, they do ensure a balance of views in the audience including people who are undecided.
QUALITY CONTROL AND CHECKS
On recruitment, participants' email addresses and telephone numbers are recorded and then in the week before the debate, written confirmation invitations and information packs were sent via email, asking participants to bring identification with them, along with a print-out of the invitation email/mobile version.
Closer to the day, additional screening was conducted among those recruited and regular reminders sent.
Audience members were contacted by telephone and rescreened on several key questions to ensure consistency between those and their initial answers. This enabled us to identify inconsistencies and potentially to exclude those individuals from the audience.
The ComRes team checked audience members in before the debate and confirmed yet again the screening criteria.
Andrew Hawkins is the chairman of ComRes
This blog first appeared on the ComRes site, and can be read here