Well, so much for the YouGov poll at the weekend being the last before polling day - we've had no fewer than three in the last few hours, although it does seem likely that this Survation poll will be the last one until the exit poll at 10pm tomorrow. It contains something of a curveball, because Survation have decided to headline the results of an additional voting intention question asked using a replica ballot paper, rather than the one asked in the normal way. Consequently, the headline numbers below are not directly comparable with anything that has gone before.
Scottish voting intentions for tomorrow's UK general election (Survation, 3rd-6th May) :
Liberal Democrats 7.1%
In line with recent Survation polls, the standard voting intention question produced a somewhat bigger SNP lead than that -
SNP 48.9% (-2.3)
Labour 24.8% (-0.8)
Conservatives 15.5% (+1.2)
Liberal Democrats 5.9% (+0.5)
Greens 2.4% (+1.0)
UKIP 2.0% (n/c)
In reality, only the results of the standard question are much use to us at the moment. There's no point trying to estimate exactly what the true SNP lead is, because different firms are producing very different numbers. What we can try to nail down is the trend. Two of the three polls today have suggested a slight narrowing of the SNP lead (on the standard question), while the third has suggested a slight increase. The odd one out is Panelbase, and it's noticeable that they started their fieldwork three days earlier than YouGov and two days earlier than Survation, so I suppose it's not impossible that they missed an ultra-late swing to Labour that the other pollsters managed to detect.
The most that can be said is that there's no firm evidence of any movement to Labour, but if it has happened it must be very small.
The Record, who commissioned the Survation poll, are making a song and dance about how the ballot paper version of the question points to a marginally less overwhelming landslide for the SNP. They suggest that this is partly because of people who would otherwise be SNP voters drifting off to 'local heroes' standing for the unionist parties, and partly because of anti-SNP tactical voting. The snag, though, is that the YouGov poll also asked an additional question in an attempt to see if local factors made any difference, and found that the SNP lead actually increased. Doubtless Survation would claim that their approach is superior because they name actual candidates, but nevertheless the evidence is obviously contradictory at this stage.
What does look increasingly likely is that the Liberal Democrat vote has strengthened as the campaign has drawn to a close. The YouGov poll at the weekend suggested that the Lib Dems were doing significantly better than at any point since the referendum, and today's poll from the same firm confirms that result wasn't a fluke. The Panelbase poll has the Lib Dems equalling their post-referendum record high, while the 6% in the standard question from Survation is not a record, but is still above average. This development may pose a problem for the SNP in one or two of the tougher Lib Dem-held seats.