Venerating past vote recall
Many a navel has been gazed into, and many a hand wrung, about the impact of past vote recall on political polls. The common practice is of course for voting intention polls to be weighted by a formula that includes past vote recall to ensure that the sample is politically representative. For voting intention polls this is entirely justified.
However, some people incorrectly believe that without being weighted by past vote recall, readers of other types of political surveys will be misled.
One obvious objection to this view is that measuring voting intention is unlike other sorts of political poll, because voting intention seeks to model actual voting behaviour. For such polls, not only is past vote recall important, but other factors such as likelihood to vote are also vital for accuracy. And yet those who are most critical of other (non-voting intention) political polls donâ€™t suggest that such polls should be weighted by factors such as likelihood to vote.
However, the main contention of those who venerate past vote recall is that without it, you will always get a pro-Labour bias. But that is simply not borne out by experience â€“ weighting a sample by past vote recall makes no discernable difference to non-voting intention polls.
To keep things contemporary we compared two sets of data from the most recent ComRes poll for the Independent on Sunday, published on 19th October. We asked respondents whether they agreed or disagreed with four statements: three political and one about Christmas spending. These two tables can be downloaded in full at www.comres.co.uk , but in summary they show the following:
â€œIt is right that taxpayersâ€™ money should be used to bail out banksâ€
NatRep weighted: Agree 37%, Disagree 58%
Past vote weighted: Agree 37%, Disagree 58%
â€œI will scale back my Christmas spending plans to save moneyâ€
NatRep weighted: Agree 62%, Disagree 35%
Past vote weighted :Agree 62%, Disagree 36%
â€œGordon Brownâ€™s decisive handling of the bank crisis means that Labour has a good chance of winning the next electionâ€
NatRep weighted: Agree 37%, Disagree 53%
Past vote weighted Agree 37%, Disagree 54%
â€œIf David Cameron had been prime minister, he would have handled the bank crisis better than Gordon Brownâ€
NatRep weighted: Agree 25%, Disagree 56%
Past vote weighted Agree 25%, Disagree 56%
There are two statements where past vote weighting makes a difference – of 1% (well within the margin of error) â€“ and one of these statements is not political at all.
The results speak for themselves.
Andrew Hawkins – CEO of ComRes