The numbers that will do whatever you want them to do

From the chart above it is possible to make just about any case you want over how David Cameron is being perceived. If you want to show that he’s moving up then just make your base point the YouGov July survey and compare with the latest numbers. If you want to show that he’s going down make your base survey February 2006.

The selective approach to all sorts of polling data is seen almost daily amongst so-called professional pundits. They have a feeling that something is happening so they look for the numbers that back their hunch. They might well be right – but they might be wrong.

So for gamblers and others trying to predict political outcomes you have to bring a bit more rigour into the process.

Firstly if you are trying to determine trends with any poll finding you have to look at the results from the same pollster to the same question over a period of time. Comparisons are only valid when you compare like with like.

Secondly it’s a good idea never to base a conclusion from one line of data from one poll from one pollster. So, as regular visitors might have noticed, I am now seeking to use charts a lot more so we can get a better medium-term view of how things are evolving using a range of firms.

With personal approval ratings only two pollsters regularly ask the same question – YouGov and Mori and with the former the internet pollster can vary the wording. So the YouGov numbers are based on the “well or badly” question while those for Mori are of findings from the “Are you satisfied or dissatisfied with the way Mr Cameron is doing his job as leader of the Conservative Party?” question.

The reason why Cameron appears to do better with YouGov is probably down to methodology. The Mori numbers are based on all respondents – not the “those certain to vote” who make up the group on which its headline voting figures are calculated. YouGov does not have a turnout filter but, unlike More, does weight on the basis of what people said they did at the last election.

What’s common to both pollsters is that there has been a clear increase in the number of negative responses as people have got to know the new Tory leader better and both a showing a decrease in the positive numbers compared with the leader’s first quarter in the job.

The positive figures with YouGov have moved down just a notch while there was a sharp decline when Mori last asked it. It will be interesting to see if this continues in the latest survey.

Mike Smithson