Of all the companies polling regularly during the campaign, none has hitherto been more stable in its estimates of referendum vote intentions than TNS BMRB. Up to now, six of the seven polls it had conducted to date this year had put Yes on 41% (and No on 59%) once the Don’t Knows were excluded, while the other put the Yes side on 40%. So it comes as something of a surprise that today TNS BMRB have put Yes on as high as 44%, easily the highest Yes tally yet (once Don’t Knows are excluded) in any TNS BMRB referendum poll.
The detailed numbers are Yes 32%, No 41%, while Don’t Knows are put on 27%. That represents a two point increase in the Yes vote compared with TNS BMRB’s last regular monthly poll conducted at the end of May, while No are down one and Don’t Knows are also down one.
However, TNS BMRB also conducted a more recent poll in the middle of June for the website scotland18september.com. That suggested the Don’t Knows had fallen to 22%, and that thus perhaps people were finally beginning to make up their minds. However, now the figure is back up to 27%, so it seems the Don’t Knows are not in fact coming to a decision quite so quickly after all. Otherwise the comparison with that more recent poll shows the Yes vote to be unchanged, while No are down by as much as five points.
Inevitably, this poll will lead to speculation that the Yes side have made a further important advance. But a word of warning; one poll a summer doth not make. The increase in Yes support could well be the result of the sampling variation to which all polls are subject. We should bear in mind that three other polls, by YouGov, Survation, and ICM, that were conducted at some point during the extended interviewing period for this poll (a fortnight from 25 June to 9 July) all either reported no change or a drop in the Yes vote.
Indeed, the paradoxical effect of the publication of today’s poll is to reduce the Yes vote in our latest poll of polls from 43% to 42%, as it means that the ICM monthly poll that was conducted in June and which had a relatively high Yes vote now drops out of the series. (The poll of polls is still dated 11 July as interviewing for last Sunday’s ICM poll concluded before that for today’s poll.)
The publication of today’s poll gives us the opportunity to follow-up our blog earlier this week on the controversy between YouGov and Survation about why those two companies, amongst others, are producing different estimates of Yes strength. One part of that controversy concerns how many 2011 SNP voters that will vote No are being identified by the different polls. TNS BMRB’s surveys add an extra dimension to this debate because until today at least, they typically have reported a relatively low Yes vote even though, unlike YouGov but like Survation, they have to ask their respondents afresh in each survey how they voted in 2011. In contrast to Survation (and YouGov), however, TNS BMRB’s polls are conducted face to face rather than over the internet.
In the four most recent polls the company had produced prior to today, it found on average that 21% of 2011 SNP voters said they would vote No (after Don’t Knows were excluded) – exactly the same as in Survation’s polls. That suggests that the difference between the polls in their estimate of the Yes vote does not simply lie in their estimate of how likely 2011 SNP voters are to say they will vote No. What TNS BMRB usually do share with YouGov is a lower estimate of how many 2011 Labour supporters say they will vote Yes – at 21%, their average figure before today was in fact the same as YouGov’s.
In today’s poll, though, that latter figure is notably higher, at 28%, while the proportion of 2011 SNP voters who say they will vote No has slipped to 15%. Both changes must be at least part of the reason why the Yes vote has increased in this poll, and the fact that both figures have changed is consistent with the suggestion that the differences between the polls is not just a question of how many No voting SNP supporters they find. However, we should note too that TNS BMRB’s weighting scheme (which like every other company’s typically results in an increase in its estimate of the Yes vote) has had a particularly marked impact in this poll, raising a (not unusual for TNS BMRB) unweighted Yes vote of 40% to one of 44%. That perhaps is reason for further caution before jumping to too many conclusions about the implication of today’s poll for the state of the referendum race.
TNS BMRB also make another attempt in today’s poll to ascertain how committed people are to their current voting choice. In line with what the company found in April, it appears that 55% are utterly committed to voting Yes or No (with the two sets of supporters showing almost identical levels of commitment). In contrast less than one in ten voters (9%) say that they are only ‘quite likely’ to vote the way they currently think they will do or that they are ‘not sure’ that they will. Not too many gains to be made in the next two months by trying to win over voters from the other side it seems. The focus of both campaigns between now and 18 September will have to be on the one on five voters who tell TNS BMRB that that still have no idea at all which way their preference will fall. It seems that it is in their hands that the future of Scotland lies.