A tale of two polls and two polling methodologies
A new poll by ICM for the Sunday Telegraph this morning should ease some of the jitters within the Lib Dems. After Friday’s YouGov survey putting their share down at 13% the ICM numbers out today are CON 37: LAB 36: LD 18. The only change on Tuesday’s Guardian ICM survey is a one point drop for the Lib Dems.
The timing of the polls meant that all the interviewing by ICM took place after the Simon Hughes admission in the Sun while only about a third of the YouGov respondees would have aware of that development.
So how is it possible that the two pollsters most watched by political gamblers should be giving such a different view of the Liberal Democrat position? This gap does not make sense. I think it can be put down to three key factors.
The weightings that are used. In order to ensure that their samples are representative ICM asks how people voted last time and then builds in an allowance to deal with “misremembering”. In the last poll for which there is data ICM’s past vote weighting was; LAB 38.5%: CON 32.2% LD 21.9%. YouGov uses what it describes as a “political party identifier” to weight its samples on the following basis: LAB 47.1%: CON 34.7%: LD 13.9%. By comparison the May 2005 General Election result was LAB 36.2%: CON 33.2%: LD 22.7%. Also YouGov does not weight by the likelihood that people will vote.
“Spiral of silence” adjustment. A top-line adjustment is made by ICM to deal with those who say they do not know. The pollster assumes that 50% of donâ€™t knows vote the same way they did last time. In last week’s Guardian poll this adjustment added one point to the Lib Dem share. This correction was originally introduced to deal with “shy Tories” who, it was thought, were reluctant to admit their allegiance to an interviewer.
The actual question that is asked. The YouGov internet panel members is presented with the following on his/her PC screen: “If there were a general election tomorrow, which party would you vote for? Conservative, Labour, Liberal Democrat, Scottish Nationalist/Plaid Cymru, some other party, would not vote, donâ€™t know”. ICM has a much tougher approach that seeks to weed out non-voters by stating initially: “Some people have said they would not vote in a new General Election, while others would go and vote at their polling station. I would like to know how certain it is that you would actually go and vote in a general election?” Then the interviewer asks: “The Conservatives, Labour, the Liberal Democrats and other parties would fight a new election in your area. If there were a General Election tomorrow which party do you think you would vote for?”
Each of the three factors would, I suggest, explain the differences between the two pollsters. I have never been able to understand YouGov’s political identifier weightings and if Peter Kellner is reading this it would be great if he could add a comment. So which pollster is right? Who knows but what seems to be happening at the moment is that many more Lib Dems are now saying that the do not know.
The reasons I press the pollsters for their detailed data is that you cannot judge a survey by its headline figures alone.