On top of the new Harris this morning, there is also a new MORI poll CON 38%(-1), LAB 21%(-4), LDEM 19%(nc). Changes are from the MORI poll for Unison a week ago. Others remain at 22%, and as usual the largest chunk is going to UKIP on 8%, followed by the Greens and BNP, both on 5%.

Perhaps more interestingly, significantly more people (41%) said they thought the economy would get better over the next twelve months, with only 30% saying it would get better – the highest figure MORI have recorded in 12 years. I am conscious that I’ve promised to do an updated post looking more broadly at the latest economic optimism figures and haven’t delivered yet, but these figures certainly suggest that the public think the economy is bottoming out and due to start recovering.

This morning’s Metro holds a surprise – there is a voting intention poll from Harris. Once a regular pollster in the UK, Harris dropped out of political polling and became an online company. They produced one poll prior to the last general election (which compared very well indeed with the actual result getting the Conservatives bang on, Labour within 2 points and the Lib Dems within 1) and carries out a regular international poll for the FT, but these are the first voting intention figures we’ve seen from them since 2005.

The topline voting intention figures stand at CON 35%, LAB 20%, LDEM 16%. Others register at a rather remarkable 29%, including 10% for UKIP, 6% for Green and 4% for the BNP. We’ve seen inflated figures for minor parties from all the pollsters in the run-up to, and since, the European elections, but this is something else and something I’m rather sceptical of.

Full tables are on Harris’s website here. For those interested in methodology, Harris are on online company with their own panel, like YouGov. Their polls are weighted by age, gender, educational achievement, region and internet usage, but not it would seem by past vote or party ID (in fact, the recalled past vote in the tables looks very odd indeed – in their unweighted sample Harris found more people claiming to have voted Tory in 2005 than Labour, so hopefully the weighting corrected that!). Instead Harris use something they call “propensity score weighting”, a proprietory weighting they say corrects for behavioural and attitudinal biases from different peoples likelihood to be online. Exaclty how it does so, we don’t know.

While Harris did ask how likely people would be to vote, they do not appear to have weighted or filtered by it.


New ComRes poll

ComRes have a new poll in tomorrow’s Independent on Sunday. The topline figures, with changes from their last poll a fortnight ago, are CON 39%(+1), LAB 22%(nc), LDEM 18%(-2). Comparatively little change here, and the “other” vote would appear to still be up, the maths suggests they are probably at around 21%.

I’ve no news yet on whether there are or aren’t any other polls in the works tonight.

New MORI Poll

Ipsos Mori have released a new poll conducted on behalf of Unison. Like ICM’s poll last night, it’s something of a return to normality after the extremes of last month. The topline figures, with changes from MORI’s last poll, are CON 39% (-1), LAB 25% (+7), LDEM 19% (+1) (the previous poll, you may remember, showed Labour way down at only 18%)

The poll included a booster sample of public sector workers allowing MORI to better gauge public sector voting intentions – there support stood at CON 32%, LAB 29%, LDEM 19%. An idea I often see expressed is the public sector employees constitute a “payroll vote” who monolithically vote Labour. It’s not true – public sector employees are more likely to vote Labour, but as this shows, it’s not some sort of guaranteed block vote – at present the Conservatives even have a narrow lead.

With the Iranian election still disputed, some support for President Ahmadinejad has come from what seems to be the only solid poll during the campaign. Iranian election polls are notoriously dodgy – there is a list of those that emerged during the campaign on this wikipedia page, they show wildly contrasting figures, report only partial results, cover only major cities (or often only Tehran itself), or have sample sizes that suggest they are straw polls, not controlled samples. It’s not a particular surprise, given that opinion pollsters in Iran have an unfortunate history of ending up in prison.

One poll that does seem to have a solid methodological basis is this one by Terror Free Tomorrow, a US non-for-profit company. Amongst those expressing a voting intention this showed Ahmadinejad on 67%, Moussavi on 27%, Karroubi on 4% and Rezai on 1%. The actual results were Ahmadinejad 63%, Moussavi 34%, Rezai 2% and Karroubi 1%.

The authors, Ken Ballen and Patrick Doherty, wrote an article in the Washington Post yesterday saying that, despite what seems to be the Western media’s keeness to assume that the election was fixed and Ahmadinejad couldn’t possibly have won fairly, their poll suggests that the Iranian public did support Ahmadinejad, and the result is likely to be genuine.

This itself has caused much debate – including Gary Langer here, Mark Blumenthal here and Nate Silver here.

So, does the poll show that Ahmadinejad really won after all? Well, the poll itself looks entirely trustworthy. It was a random telephone poll, conducted in Farsi but from outside Iran for the safety of the interviewers, sampling was using random digits on randomly chosen telephone exchanges. There was a good response rate – of 2,364 contact attempts, 42% resulted in interviews (non-contact was 26% and response rate 58%) and the sample was weighted by province, rural/urban, age and gender.

Obviously a phone poll is going to exclude people without a telephone (though phone penetration in Iran is pretty high – even the worst provinces have a penetration above 75%) and the proportion of respondents who classed themselves as middle class seems far too high. However, both these skews would likely favour reformists, not President Ahmadinejad, so if anything it should have underestimated his lead.

Criticism of the poll has largely focused upon the high level of don’t knows and refusals. The figures I gave above exclude these, but in the original sample 27% said don’t know and 15% refused to answer. These don’t seem as outlandishly high as some people have suggested – while it was only three or so weeks out from the election, it was only a couple of days after the main contender had announced his candidacy and before the final candidate list was announced.

That said, people have rightly pondered whether this big chunk of “don’t knows” and refusals were “shy reformers”. Might these figures be people in a police state who didn’t want to admit to a total stranger over the phone they might vote for reformers? We don’t have full cross breaks for the poll, but the authors indicate that the don’t knows were disproportionately (60% to 40%) reformers. If they broke 60-40 in favour of Moussavi the result would certainly be closer, but Ahmadinejad would still be comfortably ahead.

There’s also some comment that the rest of the answers don’t show a populace particularly enamoured by a hardline Conservative President. 60% for example, supported unconditional talks with the USA, 77% favoured normal trade, and 89% supported peaceful US help with nuclear power. 84% supported a free press and 77% think the Supreme Leader should be elected.

However, the Iranian public also expressed some views very much in line with those of President Ahmadinejad. 62% opposed any peace deal with Israel and favoured “all Muslims fighting until there is no State of Israel”, 32% had a negative opinion of Jews. The two biggest threats to Iran were almost universally seen as Israel and the US. 46% thought Ahmadinejad’s policies had succeeded in reducing unemployment and inflation.

My problem with the poll isn’t anything to do with those don’t knows or Iranian people’s policy views – it’s the fieldwork dates, May 11th to May 20th. It took place over three weeks before the election. International coverage of the campaign was suggesting a walkover until the last week or two, when the Moussavi campaign was suddenly being reported as coming to life.

In other words, this poll strongly suggests that had the Iranian election been held on the 20th May, President Ahmadinejad would have won in a landslide. It doesn’t necessarily follow that Iranian public opinion didn’t shift away from him in the three weeks between then and polling day. Equally, it gives no particular reason to think it did – it’s just too early to tell us, though frankly it would be one mighty turnaround and on the balance of probabilities, one has to conclude that the polling evidence (I make no claim on anything else) is in favour of Ahmadinejad genuinely having topped the poll.

That said, it’s also worth echoing a point Nate Silver made: if people were scared to tell a pollster their real intentions, they might well also have been scared to vote that way, so even if polling figures match results, it doesn’t necessarily follow that it was a free and fair election. There’s more to real democracy than just counting the papers correctly.