"Opinium" poll in the Express shows 7 point Tory lead - UPDATED

Following new polls from TNS BMRB and a rare sighting of a Harris poll, we have another new entrant - this is shaping up to be an extremely heavily polled election! Anyway, tomorrow's Express has a new poll from Opinium - who they? Looking at their website here, they seem to be an online outfit being run by my old colleague Mark Hodson (meaning two of the new online pollsters conducting voting intention polls at this election, Opinium and Angus Reid, are being run by ex-YouGov staff).

The topline figures of the poll are CON 37%(-2), LAB 30%(+1), LDEM 16%(+1), with changes from a week ago. I don't recall a poll in the Express a week ago, so this may have been unpublished, or it may just be that the contents of the Express entirely bypass the blogosphere and no one at all noticed. Back in October Opinium said they were going to launch weekly UK voting intention polls and were running it already to collect back data, but this is the first results from it that I have noticed. One thing that immediately attracts attention is that figures of 37, 30 & 16 implies support for others at 17 points, a very high figure compared to most other companies.

All I know about Opinium at the moment is that the polls are conducted online using a panel, I've dropped Mark a line to get some more details. The Press Association refers to it having been carried out "using Opinium's research panel of 1,960 adults", which suggests a panel study re-contacting the same people, rather than different samples being drawn from a larger panel. Time, however, will tell.

UPDATE: I've heard back from Mark, so here is the methodology rundown for Opinium. Firstly, the fears about a panel of only 1,960 asked every week are thankfully unfounded - Opinium are drawing samples from their larger panel of 40,000 (so about the same size as Angus Reid's).

Weighting is by gender, age, region, working status and social class - there doesn't seem to be any political weighting, but the robustness of the samples has been tested on things like past vote and propensity to vote for the main parties, so presumably Opinium feel their samples are sufficiently politically representative to not require it.

There is a two stage voting intention question - "If there were a general election tomorrow, for which party would you vote?", with prompting for the three main parties or "some other party" and then a second question for those who say other asking which other party people would vote for. There is no squeeze question or re-allocation of don't knows - an aproach that is broadly similar to YouGov.

The topline voting intention figures include all those who say they would "definitely" or "probably" vote in a general election tomorrow (which sets Opinium aside from the other online pollsters, neither YouGov nor Angus Reid have a likelihood to vote filter).

For those who care about such things, their BPC membership is pending. As I mentioned yesterday, they have been quietly running voting intention polls internally for the last six months or so to build up back data, Mark has sent me over all those results and I'll put them into the database later on.