#Matchettgate: The story of how unionist newspaper Scotland on Sunday attempted to deceive the nation with a fake poll

So now that the fog has cleared a bit, let's recap on what we know about the stunt that Scotland on Sunday and its self-styled "award-winning reporter" Conor Matchett attempted to pull last night. The datasets from Savanta ComRes polls on independence typically present various different versions of the Yes/No figures, and one of those versions does not use turnout weighting - in other words, unlike the headline numbers that are normally used for publication, it does not give greater weight to the people who are most likely to actually cast a vote. Sometimes the unweighted numbers are extremely similar to the published numbers, but on other occasions there's a significant disparity. In two polls so far this year, the published ComRes numbers have been dramatically better for Yes than the unweighted numbers... 

January ComRes poll before turnout weighting: 

Yes 47% 

No 43% 

Undecided 10% 

January ComRes poll after turnout weighting, published numbers: 

Yes 51% 

No 38% 

Undecided 10% 

February ComRes poll before turnout weighting: 

Yes 43% 

No 46% 

Undecided 11% 

February ComRes poll after turnout weighting, published numbers: 

Yes 47% 

No 42% 

Undecided 10% 

So as you can see, there's no automatic significance to the unweighted numbers showing a small No lead - it has already happened this year, and is perfectly consistent with the possibility of a Yes lead on the headline numbers. And yet, last night, Scotland on Sunday published the unweighted numbers from a new ComRes poll and presented them as if they were headline numbers. They tried to make out that the No lead was a significant departure from previous polling and had been caused by the Sturgeon-Salmond controversy, when in fact the numbers were literally identical to the February poll that had Yes ahead by 53% to 47% once undecideds were excluded. That doesn't necessarily mean, incidentally, that Yes must be ahead on the weighted numbers, but it does mean that we simply don't know - there could be a Yes lead, a No lead, or a dead heat. 

So what the hell are Scotland on Sunday playing at? The short answer is "I've no idea", but here are a few logical possibilities... 

* The poll contained both weighted and unweighted numbers, as per usual, but Scotland on Sunday disliked the weighted numbers and ran the unweighted ones instead. That would be, not to put too fine a point on it, utterly outrageous. 

* The poll was half-finished and the weighted numbers weren't available yet, so to meet a publication deadline Scotland on Sunday decided to just run the unweighted numbers. That would be downright amateurish. 

* The poll was never intended to be a Yes/No independence poll, and therefore ComRes didn't collect enough information to produce turnout weighted numbers. But when the raw data suggested a No lead, Scotland on Sunday couldn't resist retrospectively turning it into a Yes/No poll to produce damaging headlines for the SNP, even though they'd been warned the numbers were unreliable. That would be pretty cynical. 

Wherever the truth lies, it's hard to think of any explanation that doesn't involve a heavy dose of shoddy journalism.