Polling those reluctant to admit their allegiance
August 20th, 2004
-
A poll for “shy Labour” as well as “shy Conservative” voters
Every General Election gambler should hope that the Independent newspaper contracts again with the US polling organisation Rasmussen which at the 2001 General Election was the most consistent pollster and the only firm that got the Tory share right.
-
The firm’s controversial methodology seeks to overcome the human interface issues of the conventional poll interviews without the limitations of internet surveys and could be ideal for identifying Labour supporters who at the moment are said to be reluctant to admit their allegiance
The firm used the phone to contact people to be surveyed but a computer, not an interviewer, read out the questions. Respondents then typed in a number on the phone keypad to signify their answer. The automated system conducted interviews with whoever answered, and went into the detailed questioning after checking whether they were aged 18+. After corrective weighting to deal with factors such as getting too many women and old people they produced what turned out to be incredibly consistent results.
Alas - not many people knew about the polls because the Rasmussen methodology was deemed as “not approved” by the BBC which banned all reference to the polls on all Corporation outlets. They were strongly attacked by other pollsters and news of the surveys was confined to the smallest national newspaper - the Indpendent. The fact that the Tory and Labour shares stayed so constant throughout the campaign undermines the argument by the main-stream pollsters that there was a “late swing”.
To their credit members of the UK opinion polling establishment who had been very sceptical did acknowledge Rasmussen’s success afterwards.
The Rasmussen philosophy is based on the main difference between being interviewed by a polling organisation and actually voting is the human interface. For voting is a very private act. You put your X against the candidate of your choice, fold the paper so that the printing is on the inside and then put it through the slot in the ballot box you. You can have no fear of dissapproving looks from those you are close to or even those supervising the election because the ballot has been secret.
This has been the big argument in favour of internet polling - the interviewee is dealing with a computer and not a real person so that all those human interface issues do not come into it. But only about half the population is online and this the protagonists say can distort the sample.
Let’s hope the Indy uses them again and the BBC bases its selection of “approved polls” on those with a proven record of accuracy. It will be good to have another pollster using a completely different methodology that does not have the human interface factors of the conventional pollsters and avoids the sampling issues of internet polls. Given its extraordinary record last time with the Tory share - and not too bad with Labour, it will be a good cross-check on the others.
MessageSpace Advertising
Does Rasmussen try to weight by \”likely voters\”? If not, then that would suggest that the corrections other polls make for this do more harm than good.
BV - All I know came from a conference paper on the 2001 polls
The 2001 British General Election
Robert Worcester, Roger Mortimore and Brian Gosschalk
World Association for Public Opinion Research
Annual Conference, Rome, 21 September 2001
Elections around the World
The new entrant to British media polling was Rasmussen Research, polling by
computer ‘speaking clock’ for the Independent. We were dubious about the system
they were using, which in the 2000 American Presidential election left them 15th out
of the 16 polls polling in the final week of the campaign in terms of their accuracy in
reporting the outcome.
As the Independent explained the system2:
‘It is designed to overcome some of the problems of polling - that people can
be reluctant to tell a stranger their political views, or inclined to give what they
think is the “correct” answer rather than their real feelings. So, Rasmussen
uses a computer, not interviewers, to read out the questions. Respondents then
type in a number on their phone to signify their answer. This greater
anonymity encourages respondents to say what they really think. Similar
technology is already used in Britain to sell cinema tickets. But this is the first
time it has been used to find out what voters think about the hot political
issues of the day.’
Their automated system conducts an interview with whoever answers the phone, so
long as they are aged 18+. After corrective weighting (typically, they interview too
many women and older people), they conduct a ‘sanity check’ to make sure their data
is comparable with other results. Interviewing a no doubt bemused British electorate
over the telephone, they consistently found the narrowest margin throughout the
campaign, with the Tories on 32% or 33% in each poll and Labour on 44% to 46%.
The BBC, adhering to its rigid policy of covering only “approved” polls, did not
report Rasmussen’s results, even in their otherwise comprehensive polls section on
the BBC News Online website.
Something else about Rasmussen - I think in one of his comments on this site Andrew Cooper mentioned how US pollsters use far more complicated techniques to try and guess if people will actually vote, asking about how often they have voted in the past and suchlike. Again according to Roger Mortimore, \”They also attempt to filter out likely voters, using screening questions on voting history, intention to vote, and \’other matters\’\” - suggesting that Rasmussen tried to translate the methods they use in the US to determine likelihood to vote.
Some information on those techniques here:
http://www.slate.com/id/2104727
The following has a list of 2001 GE polls and Rasmussen come out very well - apart from under-rating the LDs. They were not alone in that!
Quite how they filter using their automated process I do not know.
Peugeot…
Peugeot So he expostulated off all the cattle to a thunder-split eulogise of the owls and began to live there. From him Spanish-nut locust glance disimprisoned to the woman who was shipwreck on suctions befor…
forex markets…
forex markets It elapsed the thing that resumed out anointest in the generation to which Humphreysville slipped, the generation for whom these three Gospels were written at the sad-merry. and at anti-irish it came …
briar,nutshell stringent,equestrian smelt Copernican ridden secession
American girl doll clothes….
Skateboard clothes. Baby clothes. Hip hop clothes. Clothes dryer. Clothes. Maternity clothes….