Sunday, September 21, 2008
The value of opinion polls
From the point of view of a politician I tend to have a mixed view of opinion polls. It is always nice of course to see one that predicts the Liberal Democrats doing well and even reinforcing one's perceptions of public opinion. On the other hand they can record support going down as well as up and it is at these times that one tries to distance oneself from them. Given that we cannot have our cake and eat it then as a general rule I maintain a healthy scepticism of any poll, no matter how favourable it is.
It is on this basis that I approach the mega poll being touted by Politics Home, which reportedly sampled almost 35,000 people over 238 marginal constituencies. This is bigger than any poll I have ever seen and for that reason people are giving its conclusions some credence. The question is though whether the information is reliable enough to accurately predict the state of play in individual constituencies, not least those which are currently represented by Liberal Democrats, who tend to have a strong personal vote and record of campaigning, and those in which we are challenging.
Mark Pack on Liberal Democrat Voice rightly cautions against drawing conclusions on the basis of uniform swings in polls and I believe that this applies to this mega poll in the same way as it would to any other, even though the sample is larger in this case and that YouGov has deliberately targeted the marginals. Forty four Liberal Democrat seats are certainly better than others have predicted but I believe that this poll indicates that we can do even better than that.
I say this for two reasons. Firstly, the margin of error in any poll does not statistically improve much above a sample of around 1,200, no matter how many people are asked. Secondly, this is not a series of individual constituency polls taking local circumstances into accounts rather it is a national poll based on a specific grouping of constituencies and predicts no more than general voting trends across those diverse areas. After all on average the polling company only approached 147 people in each constituency, a figure which will not be representative and does not have an acceptable margin of error built into it to determine local results.
Clwyd West MP, David Jones is nevertheless very excited about the poll. He points us to the section on Wales which predicts that the Tories could win anything between 15 and 18 seats at the next election, an improvement on their previous high tide mark under Margaret Thatcher. However there is no individual polling evidence in any of the seats concerned that statistically stacks up to support this.
As an example in Brecon and Radnor the idea that the Tories may take the seat back should energise Labour voters in the south of the constituency to come out and vote tactically as they have done for many years. Equally, the notion that Newport East might go Conservative when it is the Welsh Liberal Democrats who are the main challengers is absurd, and why no mention of Swansea West, which on this sort of swing against Labour would also fall into Welsh Liberal Democrat hands? Ceredigion has bucked trends before and I am confident that it will do so again and return Mark Williams to Westminster.
At the end of the day this poll is just a snapshot like any other. It is nice to look and and discuss with others but the poll that really counts is on election day and there is still some time to go before we reach there.
It is on this basis that I approach the mega poll being touted by Politics Home, which reportedly sampled almost 35,000 people over 238 marginal constituencies. This is bigger than any poll I have ever seen and for that reason people are giving its conclusions some credence. The question is though whether the information is reliable enough to accurately predict the state of play in individual constituencies, not least those which are currently represented by Liberal Democrats, who tend to have a strong personal vote and record of campaigning, and those in which we are challenging.
Mark Pack on Liberal Democrat Voice rightly cautions against drawing conclusions on the basis of uniform swings in polls and I believe that this applies to this mega poll in the same way as it would to any other, even though the sample is larger in this case and that YouGov has deliberately targeted the marginals. Forty four Liberal Democrat seats are certainly better than others have predicted but I believe that this poll indicates that we can do even better than that.
I say this for two reasons. Firstly, the margin of error in any poll does not statistically improve much above a sample of around 1,200, no matter how many people are asked. Secondly, this is not a series of individual constituency polls taking local circumstances into accounts rather it is a national poll based on a specific grouping of constituencies and predicts no more than general voting trends across those diverse areas. After all on average the polling company only approached 147 people in each constituency, a figure which will not be representative and does not have an acceptable margin of error built into it to determine local results.
Clwyd West MP, David Jones is nevertheless very excited about the poll. He points us to the section on Wales which predicts that the Tories could win anything between 15 and 18 seats at the next election, an improvement on their previous high tide mark under Margaret Thatcher. However there is no individual polling evidence in any of the seats concerned that statistically stacks up to support this.
As an example in Brecon and Radnor the idea that the Tories may take the seat back should energise Labour voters in the south of the constituency to come out and vote tactically as they have done for many years. Equally, the notion that Newport East might go Conservative when it is the Welsh Liberal Democrats who are the main challengers is absurd, and why no mention of Swansea West, which on this sort of swing against Labour would also fall into Welsh Liberal Democrat hands? Ceredigion has bucked trends before and I am confident that it will do so again and return Mark Williams to Westminster.
At the end of the day this poll is just a snapshot like any other. It is nice to look and and discuss with others but the poll that really counts is on election day and there is still some time to go before we reach there.
Comments:
<< Home
Whichever way you look at it, however, it is a very good poll for the Tories, awful for Labour, not much better for the Lib Dems and only so-so for Plaid.
I'm afraid, Peter, I can't see Mark Williams holding Ceredigion. Plaid will re-take it.
I'm afraid, Peter, I can't see Mark Williams holding Ceredigion. Plaid will re-take it.
# posted by 
: 11:43 AM
: 11:43 AM
The other point about this poll of course was that it was taken over two weeks at the end of July and the beginning of August. Recent polls are much better for the Liberal Democrats.
As for Ceredigion, we will see. Most people said we would not win it in the first place.
As for Ceredigion, we will see. Most people said we would not win it in the first place.
I'm afraid Plaid is being tarred with the same brush via it's association with Labour in the Assembly.
Post a Comment
# posted by : 7:02 PM
: 7:02 PM<< Home
