The Sunday Telegraph is running an opinion poll tomorrow about the levels of support for each party, commissioned from the pollster firm BPIX. It’s a switch from their usual pollster, and only the second time since the 2005 general election that BPIX has been used by a paper other than the Mail on Sunday.
It’s therefore potentially the start of a significant change in the pattern of polling and, regardless of what the figures in this poll say, the publication of the poll is bad news for all parties – and anyone who follows political news.
That’s because uniquely amongst the firms used regularly for national political opinion polling in the UK, BPIX is not a member of the British Polling Council (BPC), the trade body which lays down certain basic standards of transparency for polling. This isn’t just a technical matter, because not only is BPIX not a member of the BPC, but it also does not get close to meeting the BPC’s standards.
For example, BPC rule 2.2 requires:
Whenever it is practical to do so the following information should also be published:
- Complete wording of questions upon which any data that has entered the public domain are based;
- A web address where full computer tables may be viewed
This is crucial to allow people to scrutinise how a poll has actually been conducted and how reasonable the media coverage of the results really is.
By contrast, BPIX polls are shrouded in mystery. The BPIX website – www.bpix.co.uk – has been “under construction” for years, a particularly poor show as they are an internet pollster, and I’ve never received a reply to any of the emails I have sent them.
A practical example of the problems this brings: party leader ratings vary greatly depending on the exact wording of the question used. The Sunday Telegraph reports party leader ratings, but without telling us the question used you can’t tell how these really compare with other poll findings.
Whatever you think of opinion polls, they have a huge impact on political reporting (and so in turn on political realities, given the power of the media). It’s bad news for everyone if we are moving back to the days of opinion poll details being shrouded in mystery.
Let’s hope BPIX decide to sign up to the basic standards that other polling companies are happy to abide by.
UPDATES:
- Mike Smithson over at Political Betting has similar views.
- BPIX’s domain registration claims that the firm is a “non-trading individual”, which clearly isn’t right.
- BPIX has never applied for membership of the British Polling Council.
6 Comments
And just in case anyone was wondering what that poll actually said, it said:
CON 43%(-4)
LAB 31%(+7)
LDEM 17%(+1)
(Changes are since a previous poll carried out 31 July-2 August.)
Pretty much in line with the last two polls (from YouGov and ICM).
Mark, so why is the Telegraph using them? Are they cheaper?
Cheaper? Maybe. Certainly easier to interpret according to your own preferences.
If I set up a polling company, publishing moving averages of other people’s polls, with a bit of random noise thrown in, would anybody notice?
Well, they would now you’ve told us all, Joe! It is an interesting idea as to why BPIX lack transparency, though…
Isn’t another key point of the BPC guidelines that if you publish a poll you have to publish the whole thing?
So you can’t (for example) publish the bit that says people’s opinion of David Cameron has improved because of his response to the financial crisis but leave out the bit that shows Brown/Darling are regarded as more competent to deal with it.