What's Behind The Dodgy Push Poll?

A dodgy push poll with fishy weightings has left New Labour all in a lather fuelling the spin the lamentable leader could make it back to No 10 after all. You can't make it up - but someone did. Behind the foggy figures and heady headlines lies a fishy poll and a telling Sunday story.

If it's too good to be true, it probably is. Even with a scant glance at the latest opinion poll, hacks would be hard pressed to come up with a headline: "Brown on course to win election", but not on The Sunday Times.

It seems a new Sunday Times poll reveals New Labour is just two points behind Tories. "The YouGov survey places David Cameron’s Conservatives on 37%, as against 35% for Labour — the closest gap between the parties in more than two years."

Five or six per cent maybe but two? Does anyone believe the Tory lead has shrunk to a mere two percent, as suggested by the poll? Not the bookies. But maybe the money markets waking up in a cold sweat after a night of Brown nightmares.

The rogue poll is clearly an outlier. The Orange Party is deeply sceptical about commissioned push polls, particularly when they show such sharp spin and take no account of the battleground of the key marginals.

Wise Wells at UKPR warns: ”Until we see some more polls that support or contradict the further narrowing of the polls – be wary.”

But the real question is - was it is real?

As Wells notes, the changes from the previous poll are well within the margin of error: "YouGov’s polls this week have been very consistent in showing a 6 point lead, and these figures are actually within the margins of error of a true position of CON 38%-39%, LAB 32%-33%."

Now that's more like it.

Push polls are only as good as dodgy weightings, skewed sample and questions asked - or not asked. The outcome relies on spurious uniform swings.

The actual tables show a different picture. The left column shows an unweighted Tory six percent lead, the bold one to the right, actual weighted numbers, has two percent.

So how are they "weighted"? The views of 'Labour IDs' on the poll sample will have to be scaled up in order to meet the 'quota' for each poll. Polling pundits point out this is something that they have not had to do in the past to the same extent. So why is it happening now?

A quick look at the method reveals YouGov "interviewed a sample of 1,436 adults online across Britain on Feb 25-26."

Do ordinary folk actually sign up for opinion poll panels? How much credence can be placed on a sample taken online rather than in face-to-face interviews? A tinyYouGov 'panel' made up of faceless online people is hardly representative of the whole country.

When asked the question "Do you think Gordon Brown is doing well or badly as PM", both the weighted and unweighted samples show the same result.

A staggering 58 percent in total reckon Brown is behaving Badly.

Very little differences in the view of a pathetic PM between weighted and unweighted. But miraculously a significant difference between the two for the headline-grabbing poll. Enough to send New Labour jumping up and down with glee. Now that's odd.

The Orange Party suspects a very different agenda at work and play here.

A Murdoch inspired wake up call to Dave. A poll which may tip the struggling Supreme Leader over the edge to name the day. A headline grabbing story after being eclipsed by the Observer and Rawnsley's Bullygate revelations.

The poll came on the day of Dave's big speech to the party faithful in Brighton at a time when he is under increasing pressure to come out of the closet with clear, realistic policies and stop wobbling around. "Now we'll see if the Tories have the mettle" thunders Murdoch's Sunday Times.

Speculation that Bunkered Brown may name the day any day soon is intense. But Bottling Brown has form. A poll showing victory is within his grasp may just be enough to swing the balance.

The two point lead is the smallest YouGov has recorded since Bottling Brown's election-that-never-was back in 2007. Then the Tories went on to peak in May 2008 with a 26 point lead. The message is in the Sunday Times bottle. Bottling Brown beware.

Meanwhile in the real world of fantasy school league tables and dumbed down exams...

Thousands of school pupils had their GCSE results secretly downgraded last summer to avoid the embarrassment of too big a jump in the numbers achieving top grades, reports The Sunday Times.

Exam chiefs ordered boards to raise the bar and move the goalposts to cut the number of pupils receiving A* to C grades in Mickey Mouse exams to avoid more egg on their faces over rigged grade inflation.

Push poll opinion polls and dumbed down school exams. You can't make it up. But some will have a damn good try, until they're caught out.

Top picture: Scarfe, Sunday Times. Mid graphics: YouGov/Sunday Times survey results (click images to enlarge)