h1

YouGov versus ICM — Will June 10 provide the answer?

May 26th, 2004

box

The May ICM poll showing Labour 4% ahead has further reinforced the gap with YouGov which, like Populus, has the Tories 4% ahead.Which pollster should gamblers believe?

This week’s YouGov poll showing that UKIP is heading to beat the Lib Dems for third place in next month’s Euro Election – an outcome that would surely mean the end for Charles Kennedy – should enable us to test the pollster.For if on June 10 YouGov can get both the low turnout Euro and the London Mayoral Elections correct then the doubts about internet polling will be ended once and for for all.

The harsh reality for all opinion polls is that on election days real people put real ballot papers into real ballot boxes and a real result is announced. On those days all the theories and all the differing approaches to the methodology are put to the harshest examination of all. General Election and other bets are decided.

Unfortunately onlyYouGov has carried out specific June 10 polls. So we will we will have to wait for a real test against actual numbers the new Populus strategy of compensating for “Shy Labour supporters”, and the ICM approach to dealing with those who are not “absolutely certain to vote” .

But what will be on the line will be the YouGov approach of polling solely on the internet amongst a group of people who have registered with the company and who get paid for their opinions.

    What concerns many is that to be part of a YouGov survey you first have to register with the company and you have to have access to the internet – factors that could skew the sample.

YouGov supporters retort that the sample surveyed in conventional telephone polls is restricted to people with a land-line who happen to be in when the interviewer calls – both factors that could distort the sample. With YouGov you fill in the questionnaire in your own time without there being any personal interaction with an interviewer.

The person being surveyed does not have to provide basic personal data because that has already been stored on YouGov’s files. Thus the “interview” takes place at a time of the person’s choosing and is focussed solely on the issues being examined.

A feature of a YouGov survey is that questions can be asked in a number of different ways.

The real test is in the results. At the last General Election the most accurate opinion polls were those of YouGov and ICM – which are still 8% apart.

A factor that might help YouGov’s is that those questioned can be less consistent in their answers to a computer than to a live interviewer where people might feel constrained by what they have said before.

Thus a little probing can produce a more revealing result. This was from a London Mayoral Election poll just before Christmas when Ken Livingstone’s return to Labour was being discussed. Note the way the anti-Livingstone element increases with each question.

Q1 “How would you vote with Ken as an Independent?”
37 Livingstone, 19 Norris, 12 Hughes, 4 Gavron (Lab) 3 Others
ANTI-Ken Livingstone component – 38%

Q2 “How would you vote with Ken as official Labour candidate?”
33 Livingstone, 20 Norris, 15 Hughes, 5 Others
ANTI-Ken Livingstone component – 40%

Q3 “Would you be more likely or less likely to vote for
Mr Livingstone if he stood as Labour’s official candidate?”

9 More 19 Less likely, 26 FOR anyway, 35 AGAINST anyway
ANTI-Ken Livingstone component – 54

These findings have played a major part in our call on the London election. Hopefully YouGov will be commissioned for a final week’s poll. If they get this one right then their credibility will be enormously enhanced.’






Comments are closed.