Read The Shares, Not The Gap

Two polls, both by NOP, are the first out of the gate in the general election. We can draw some lessons from how they are reported.

Two polls, both by NOP, are the first out of the gate in the general election. We can draw some lessons from how they are reported.

The first, for Channel 4's Powerhouse Programme, has the Tories up five points from the one broadcast the previous week, from 29% to 34%, while the other, published this morning in the Daily Express, has the Tories on 31%, with no change from the last poll on the NOP website, the one done for the Sunday Times in March (sic).

Fieldwork for the first was 'over the past weekend', according to PA News and '5-6 May' according to Powerhouse, while the poll for the Daily Express was 4-7 May, according to the Daily Express.

Both, done over the same period of time, with the one projecting to 253 seat majority (Express says 250 on their 'splash') while the other to only (only?) 207, a 46 seat discrepancy. Both can't be right.

Or can they? Split the difference and you get 50% for Labour, 32.5% for the Tories, and 12% for the Liberal Democrats. And look at the results of the two polls: The first, 49%, 34%, 11%; the second, 51%, 31%, 13%. Both within, well within, sampling tolerance. Indeed, within plus or minus two percent for the share of each party, never mind the plus or minus three usually quoted.

So the lesson from this? Read the shares, not the gap, which doubles the sampling tolerance, watch the Tory share especially, for if they don't get up into the 40s, they're dead.

Watch the comparisons, for to contrast to the outcome of the last election, Labour's 44.4% (not the 43.2% reported in this morning's Daily Express, which includes the votes for Northern Ireland), the Tories' 31.2%, and the Lib Dem's 18.0%, and then calculate the swing (the change between the two leading parties, divided by 2), to get the best guide both to the changing shares of the parties, and by extrapolation, the standing of the parties.

One final caution for today (more tomorrow, when we'll see the MORI poll in the Times and Gallup in the Telegraph), is that if you use the model on the ICM web site to calculate the state of the parties in terms of seats, you'll be surprised to find that the Labour Party didn't have a majority of 179 at the last election as we all thought, but about 60 fewer.

More insights about Public Sector

Society