Lib Dems on 10% in Poll of Polls: Inaccurate by 2% which more accurate sums show
21 May 2012 7 Comments
Much is made of the polls and in many ways they are important as they are a barometer to how the public are feeling about the party at the time. We can dismiss them as irrelevant or inaccurate but the reality is that if they were higher we would be using them as evidence we were doing the right thing so the reverse must also be true; just look at how UKIP are using the polls, this is not dissimilar to how Clegg portrayed it when we beat Labour in the local elections a few years ago. One of the issues with polls is the discrepancy between polling companies which shows a 5% difference in Lib Dem support. Many people then look at the poll of polls to get an average and UKpollingreport keeps a running score on its site. This is currently 10% and he has recently come out to defend this despite the party securing higher in both local elections. But there has got to be a question around the methodology to create the average which shows a depressed Lib Dem score.
“on average Liberal Democrat support has been seven points higher in local elections than what they were polling at the time. Over the same period, polls have been largely accurate in predicting Liberal Democrat support at general elections, with the exception of 2010” UKPollingReport
This is more or less true for the general election results where we see that in 2001 the poll of polls was 1.3% lower than the result, in 2005 the poll of polls was 0.2% above the result and in 2010 it was 3.3% higher. However, in Scotland the polls in 2001 were 6.1% lower than the result and in 2005 5% lower. In local elections the poll of polls have indeed been lower but in the last 10 years the gap has been closer to 5% than 7% as UKPollingReport states (even if you look at his own chart):
So let’s assume that the poll of polls is more or less accurate for a general election result and 5% lower than for local elections, the poll of polls often reports the Lib Dems on 9-10%, even on UKPollingReport’s site (today it is 10% but it often moves to 9%) when according to the election result it should be reading 11-12%. Now this may be insignificant and may even be a change in pattern now we are in government but maybe the poll of polls is depressing the Lib Dem vote?
We have had more polls in the first 2 years of this government than we have ever had in a whole term. YouGov have been particularly prolific and often report a lower score for the Lib Dems than other polling companies. However, because they release a poll nearly every day when you come to make the average this actually decreases the Lib Dem average. However, it is not an average of the polls, it is an average of the results which includes the same result many times over. An average of the polls would be to take the same number of polls by different polling companies over the same period and make an average. If you do this, and I have (using the results from the UKPollingReport site), what you find is that the Lib Dem poll of polls is 11-12%. In fact there is another poll of polls run by Electoral Calculus which is more accurate than the UKPollingReport one, which shows exactly this result.
Not that this is anything to get excited about as it is still 10% lower than we are used to in recent years but maybe it’s a straw we can try and grasp at?