GE2015 showed that betting based on single seat polls is a recipe to lose money

At the end of last week I drove through the Hartlepool constituency on the way back home from my holiday in North Northumberland in the hope of seeing something that I could snap to illustrate the by-election. Alas nothing was obvious on the A19 but maybe I was concentrating on the traffic!

Like David Herdson on Saturday I am far from convinced that the Tories can take the seat and would rate both LAB and CON chances at about 50%

But surely the big polling lesson from GE2015 was not to take single seat polls too seriously. A huge number of the more than 100 seats surveyed proved to be far from predictive. At the time I wrote that Cameron’s victory in that election was the death knell for such surveys and it has been noticeable how very few have been commissioned since.

If you look at the Hartlepool betting trend the big driver has been the Survation poll that was carried out three weeks ago and a lot could have changed since.

The huge challenge with constituency surveys for the pollster is to get any sort of sample that is big and balanced enough to be of significance. Inevitably these have to be done on the phone because it is hard achieving a sample response of 500 or more online. The difficulty of phone polling is the declining proportion who use their home phones and we have all becoming increasingly sceptical of mobile calls from sources we do not recognise.

This makes it harder to get a balanced sample so that the views of those in hard to find segments get their responses scaled up. Thus in the Survation Hartlepool poll only 36 of the target number of 75 likely voters in the youngest age group were found so effectively they were given a more than double weighting.

My second concern about backing the Tories is that this is taking place on the local election day making it much harder for the parties to attract activists to help in this fight. In the council elections for Hartlepool the Tories have only managed to field 11 for the 33 council seats at stake which suggests fewer activists on the ground.

While the Tories remain the 60% betting favourite LAB is the value option.

I should add that making a bet is not a prediction of an outcome – but a suggestion that the odds on offer are better for a contender than might appear.