After the Alba Party was launched a couple of weeks ago, I ran a Twitter poll asking how people were planning to vote on the list. “Nothing scientific,” I stressed, “just curious to see how my own followers are planning to vote”. The poll was reasonably popular, attracting more than 2000 votes, and a couple of people suggested that I should repeat the exercise every week or two to see if anything changes.
So that’s what I did yesterday, and I assumed that people were intelligent enough to understand that the “non-scientific” disclaimer still applied without me having to reiterate it every single time – but no. Instead I had to wade through a sea of drivel all evening along the lines of “what sort of polling ‘expert’ thinks that a self-selecting poll is meaningful, you’ve lost the plot James, but whatever keeps you busy, ho ho ho”. Some of these were people who really should have known better. Lesson of the day: never expect an intelligent reaction, and always insert footnotes for numpties.
The poll initially followed a very similar path to the first poll, with Alba moving into a double-digit lead over the SNP and the Greens trailing far behind in third place. But no sooner had I provisionally declared Alba the winner before a sudden influx of thousands of votes pushed the SNP into a modest lead over Alba. The most likely explanation is that the second poll attracted a lot more retweets from SNP supporters than the first poll did.
So with around twice as many overall votes as the first poll, the results are not directly comparable, but what’s interesting in both cases is the absolute numbers. Unlike many self-selecting polls, Twitter polls are restricted to one vote per account. OK, some people have multiple accounts or fake accounts, but it’s reasonable to assume that most votes come from real, unique people. At time of writing, there have been 4656 votes, of which 32.6% were for Alba – so that’s approximately 1518 Alba votes. It might surprise you to know that on a typical turnout, only around 80,000 real votes for Alba would be required to win eight seats (one in each region). So even my little Twitter poll has picked up roughly 1 in 50 of the voters that Alba would actually need.
None of this is to imply that the other 98% of the required votes are necessarily out there – but it’s a useful reminder to look at absolute numbers, and not just percentages, before dismissing an exercise completely out of hand.
A number of people have mentioned being interviewed for a Panelbase poll with Wings-type questions, so we should have another real, scientific poll quite soon. It remains to be seen whether the full poll is for Wings or whether he asked for a few questions to be tacked on to someone else’s poll, or to an omnibus poll.
