Ashcroft poll puts Reform ahead in Runcorn. Is it right?

Labour launches its by-election campaign trailing Reform in the first poll for the seat. But there is reason to doubt these Ashcroft numbers

We now have two sets of numbers to consider for the Runcorn and Helsby by-election. One - and by default the superior of the two - a poll of actual voters on the ground. And the other a model forecast.

The poll comes from Lord Ashcroft, who had form for producing a slew of constituency surveys during the 2010-15 parliament. They weren’t toe-to-toe accurate in aggregate. They could have been better. But they weren’t far off in telling you the direction of travel.

The numbers from Ashcroft’s own poll for Runcorn gives Reform a comfortable lead of nine percentage points.

In by-election terms, that’s Reform winning the seat from Labour by more than a few thousand votes.

It suggests one quarter of voters are either undecided or refusing to say how they would vote. Ten per cent of those who backed Mike Amesbury would move Reform. Almost four-in-ten Tory supporters would, likewise, move Reform.

These are eye-opening movements. But they fly in the face of what my own Britain Predicts model is saying, which is that Labour are currently ahead by 3pts - the equivalent to a lead of a few hundred votes, just south of a thousand.

So who’s right?

Constituency polling is a tough nut. An effective capacity to sample a few hundred - or a thousand - representative voters within a small geographical zone is the reserve of the few, not the many. And it’s at this first hurdle that I fear the Lord Ashcroft poll falls flat.

The cross-breaks show a significant under-sampling of wards outside of Runcorn. As my own write-ups have pointed out, it is in Runcorn where Reform’s support will be most concentrated.

When you pull up the electorates for the Runcorn and Helsby constituency by ward, you can work out the following. Two-thirds of the seat is Runcorn. The rest are the commuter-belt towns and villages stretching south-west to Chester and Ellesmere Port.

The Ashcroft poll breaks the seat down into four cross-breaks. The most commuter and rural of which - the least friendly to Reform according to Britain Predicts - makes up 22 per cent of the seat.

But only… one per cent of the poll.

I say again. One per cent.

The most Reform-friendly part of the seat, Bridgewater, Grange, and Halton, meanwhile, make up 50 per cent of the poll.

And that’s as a weighted sample.

But this part of the constituency accounts for less than a third of the total electorate.

It’s a 20pt oversample in Runcorn central, and a 21pt undersample in Commuter County Cheshire.

Britain Predicts knows the vote share breakdown by ward in Runcorn and Helsby. It can be wrong. But it has Reform in third, behind the Tories, in the Gowy villages and Helsby, while agreeing with Ashcroft in that Reform is ahead of Labour in Runcorn.

This may not sound like much to the untrained eye. But had Ashcroft’s team properly sampled Runcorn and Helsby, it would likely be enough to cut Reform’s lead down from the 9pts it published, to within the margin of error of the Britain Predicts numbers.

My calculations suggest it would instead be Labour 32 per cent, and Reform 35 per cent.

That is, of course, assuming the gradient is the same, and that Reform are polling poorly in the Cheshire villages.

The poll doesn’t give us clarity on that. But the headline that four-in-ten Tories are moving Reform here should give one pause for though. That is a development, more than enough to, in isolation, put Reform within striking distance of Labour.

And that’s one thing both the Ashcroft poll - were it properly sampled - and Britain Predicts agree on: this is a tight fight. A Lab lead of 3. Or a Ref lead of 3.

Happy campaigning.

The summary is this: the Ashcroft poll over-samples in Reform heavy areas. And under-samples in areas weak to Reform. When corrected, it suggests a Reform lead not of 9pts, but 3pts.