CRISIS FOR DOOMED ED.

Crisis for doomed Ed Miliband. The Guardian‘s ICM poll is in, and the news is grim. ICM reckons Labour is 3 points ahead of the Tories on voting intention. Since the last poll, the Tories have lost 4 points and Labour 1. That last poll was sensational indeed, as it showed Labour and the Tories on 36% each. Naturally, that was a crisis for doomed Ed Miliband too. Surely, the Tories giving back this hardwon triumph is newsworthy.

Wait…what was the headline again?

Guardian/ICM poll: public confidence in Tories’ economic competence surges

Yes! It did! From 28% to 40%, while Labour went from 19% to 24%! So it’s possible for both options to gain here – presumably the public has doubts about the economic competence of UKIP and the Lib Dems, which is fair enough.

But that’s not my point. My point is that the headline question – the point that Labour had apparently pulled back into the lead – has disappeared, replaced by this downticket question. Worse, the Guardian has form. They do this every time their favoured poll favours Labour.

Let’s rewind a post from February, 2013. Through 2011 and 2012, as Labour held a steady lead in the polls, the Guardian repeatedly referred to its own ICM poll to say that a) the lead was not enough and b) that the Tories did better on the question of who would manage the economy.

a) was defensible, up to a point. ICM’s methodology deals with don’t knows by assuming that they will mostly break the way they did last time, so someone who voted Lib Dem in 2010 and now doesn’t know gets counted as a Lib Dem, or more precisely a bit less than a Lib Dem after the likelihood to vote weighting. You can’t do polls without making assumptions like this.

That said, the Guardian, like most newspapers, persists in denying that any other polling by any other methodology exists. When ICM’s poll showed a Labour lead of 3, Ipsos-MORI’s showed a Labour lead of 10 with Labour on 40 per cent and Comres’s showed a Labour lead of 9 with the Tories under 30 per cent. YouGov’s showed Labour on 40 per cent with a lead of 7. If you were informed that four comparable polls gave a lead of 3, 7, 9, and 10, you would surely have a different impression – perhaps that the lead averaged 7.25 points, which ain’t any kind of “crisis for doomed Ed”, and more to the point hasn’t changed in four months.

b) is much less defensible. In October, when the ICM poll showed (shock!) a 7 point Labour lead, the Guardian discovered that the “who would be best on the economy” question had come in and the Tory lead on it was down to 4.

When the next big ICM poll showed up, with a Labour lead of 12 overall, Patrick Wintour of the Guardian managed to discover a way in which this was still a crisis for doomed Ed Miliband. This time out, they asked a question about who was responsible for the financial crisis, rather than who was best at managing the economy. More people thought Labour were than the Tories, probably because it happened on Labour’s watch. So, crisis, crisis, crisis.

Wintour didn’t bother to mention that this question had moved 5 points towards Labour since the last time they asked it.

Now, on this occasion, the headline voting-intention question has come in for Labour, and suddenly the economic competence one that was junked back in February is crucial. There is a word for this technique: misdirection. It’s how find-the-lady artists work the suckers. It is no kind of journalism.

It’s also very common. The Evening Oligarch commissions the Ipsos-MORI poll I mentioned above. On the day it showed a Labour lead of 10, what, pray, was the headline?

Sorry Ed, they’re just not that into you: six out of 10 polled don’t like Mr Miliband

Yet again, one question didn’t deliver the party line, and so another was swapped in. Find the lady. Step right up. You had to read four pars into the story to find out the actual result. Imagine if the football coverage headlined with the attendance or the number of throw-ins, and left you to work out who might have scored more goals.

If you read professional pollster and British blogging institution, Anthony Wells, you’d find out that the main conclusion from that question is that a majority of the British dislike all the party leaders, and who would say they were wrong?

So, anyway, as a result of this I refused to buy the Guardian for the first time ever; I had it in my hand, but I knew about the poll, and I just thought, I’m not going to pay for this shit.

I can see how it works, after all. The ICM poll is both low reading and infrequent, and all others are simply ignored. Because it is infrequent, each poll provides the context of thought at the paper for some time. All polling is subject to sampling-error, but if you re-poll every day like YouGov, a bad sample will be evident very quickly. If you don’t poll again for a month or more, you can get so much more out of a small investment of polling.

First of all, the hyper-Blairite political desk reads the poll and digs down until they find a question that suits. They show the draft story to their sources in Alan Johnson or Peter Mandelson’s entourage and derive a quote to the effect that nice cuts, Dave, moderniser reform friends of you know the rest. That makes a front page. Then, the ex-Trotskyite opinion desk reads its own paper, and discovers the crisis. They then unleash themselves in the opposite direction, convinced inwardly that social democracy is still social fascism and this is just more evidence for the revolutionary way out of the crisis, or at least for a big meeting at Conway Hall. That fills the leader, and with luck, kicks off a good row that can fill more space. The reader, presumed to be in either camp, is either cheered by the news and stimulated with outrage by the leader, or the other way around. Either way, they are misinformed, but who cares? Everyone is happy.

The really shocking discovery would be that, probably, the June ICM poll was an unusual sample and this one is just reversion to the mean. That would suggest that all the product generated from it was so much eyewash. And we know they know it; they find it necessary to hide it. Today, they mention someone else’s poll:

A range of opinion polls also suggest that Labour’s lead over the Conservatives is narrowing, with a new ComRes survey suggesting just 22% of voters rate him as a good leader compared with 28% giving David Cameron a favourable rating.

I wonder if that might be the ComRes poll that put Labour on a lead of 9?