You might remember that a few days ago I pointed out a gross factual inaccuracy (and I use those words advisedly - there's absolutely no doubt whatever that it was an inaccuracy) in the reporting of the recent ComRes poll by both David Maddox of the Scotsman and Magnus Gardham of the Herald. Both claimed that the Scotland Votes calculator suggested that the poll would translate to a close result of SNP 30 seats, Labour 27 seats. This is categorically untrue. There is no conceivable way of projecting from the ComRes poll that would give the SNP fewer than 42 seats, or Labour more than 13. The most realistic projection would probably be something like SNP 44, Labour 13.
After I pointed out the inaccuracy, the commenter "yesindyref2" mentioned that he had raised the issue in a comment on Gardham's article at the Herald website. Instead of correcting Gardham's mistake, the moderators simply deleted the complaints about it. This is obviously deeply troubling - it's hard not to conclude that "inaccuracy without embarrassment" was being preferred to "accuracy with embarrassment". That's not my (or I would hope anyone else's) idea of what journalism should be about.
Unbeknown to me, another reader of this blog also sent a complaint directly to the editor of the Herald, Magnus Llewellin, in which he quoted in full my explanation of Mr Gardham's inaccuracy and how it came about. I've been sent a copy of both the complaint and Mr Llewellin's response, which is extremely courteous. However, this is a rare occasion when I am actually just as troubled by a civilised response as I would have been by the more familiar mainstream media reaction of "your complaint is noted, but we never make mistakes, so go away".
It looks very much like Mr Llewellin's instinct upon receiving the complaint was not to make an effort to understand it and then establish whether it was well-founded, but instead to send a warm letter to mollify an unhappy subscriber. In fact, what he does in the letter is go through the motions of defending himself against a criticism that no-one has actually made. He points out that Scotland Votes is a respected tool which is also used by the Scotsman. He notes that seat projections are not a precise science and each method will produce a different result. All of this is true, but none of it has even the slightest shred of relevance to Mr Gardham's inaccuracy or how it came about.
The Scotland Votes predictor is no different to any other predictor - if you put nonsense figures in at one end, you'll get nonsense figures out at the other end. The inaccuracy of the supposed "projection" you end up with is not the fault of Scotland Votes, but of the journalist who failed to use it correctly. Scotland Votes ONLY WORKS if you put in figures from national polls. The ComRes poll was not a national poll, but was restricted to Labour-held seats only. And yet Gardham failed to spot that distinction, and put the raw voting intention figures into Scotland Votes as if they were national figures. What he did was quite literally as ridiculous as taking the raw voting intention figures from a constituency poll in Dumfriesshire, pumping them into the predictor as if they were national figures, and "projecting" that the Tories are on course to win twenty Scottish seats.
Even more worryingly, Mr Llewellin claims that Gardham did not present the figures as anything other than what they were - "a certain website's calculation based on a national swing". In fact, Gardham did not present them in that way, but if he had done he would have been wrong. The calculation was NOT based on national swing, but rather on the false assumption that regional voting intention numbers can be treated as if they are national voting intention numbers.
Mr Llewellin's final remark is that he is satisfied that the use of the "projection" was justified because it illustrated that the poll was "significantly less bad" for Labour than previous surveys. No. It was an outright inaccuracy that helped to bolster an ENTIRELY FALSE IMPRESSION that the poll was significantly less bad for Labour.
I can't believe any of this is still in dispute several days on. Most of us spotted Gardham's schoolboy error within seconds. Readers have been seriously misled, and a correction should be issued. End of story.