Friday, December 20, 2013

Considering the state of play among young voters

It always used to baffle me that politicians and journalists alike seemed to take it as read that the female vote is more important than the male vote. I wondered if that mindset simply boiled down to political correctness, until someone patiently explained to me that women tend to be considerably less partisan than men, and are therefore more important in the sense that their votes are more likely to be up for grabs. But I'm still not so sure that the fetishisation of the youth vote has a similar rational basis. There are two ways of looking at it - on the one hand, young people are the least important section of the electorate in that they are the least likely to actually turn out to vote, but on the other hand they are the greatest prize of all because they have (on average!) a greater number of voting years ahead of them than anyone else. In normal circumstances the two factors might be thought to offset each other, but given that the Yes campaign are aiming to win the independence referendum outright (and given that practically no country has ever willingly surrendered its independence after winning it), securing the hearts and minds of the younger generation for the decades to come is not such an obvious consideration this time around. It's also worth bearing in mind that nobody of any age has ever voted in an independence referendum before (unless they did it in another country), so neither is it the case on this occasion that young people are unique in lacking hard-to-break voting habits established over a period of years.

In spite of all that, the youth vote is undoubtedly regarded as special, perhaps because many commentators are determined to believe that the SNP gave 16 and 17 year olds the vote for tactical rather than principled reasons, and would be all-too-delighted to have a little gloat about how the whole thing had 'backfired'. A few months back, it was practically being stated as fact that this had already happened, and that younger voters were breaking disproportionately against independence. But the evidence for that claim was never particularly strong, and has become weaker still after the latest batch of polls. We now have datasets from three polls that were wholly conducted after the publication of the White Paper, and although they present a decidedly mixed picture, none of the polls suggest that young people are the most anti-independence group, while one of the polls suggest that they are in fact the most pro-independence group. Let's take the three in turn -

YouGov unambiguously suggest that 18-24 year olds are the most pro-independence age group, with 42% of them planning to vote Yes and 50% planning to vote No. By contrast, the Yes vote in the three older age groups falls within a narrow band between 30% and 32%, with the No vote ranging between 50% and 55%. Incidentally, YouGov are still failing to poll under-18s, which is astonishingly bad practice in a referendum that has a minimum voting age of 16.

Ipsos-Mori suggest that 16-24 year olds are the second most favourable of the four age groups for the Yes campaign. Paradoxically, though, the Yes vote among young people is very slightly lower than among the overall sample - this apparent contradiction is brought about by the particularly heavy support for Yes among 35-54 year olds, which pulls the overall figure up.

TNS-BMRB are the only pollster that offer any real succour for those who believe that young people represent some kind of problem for the Yes campaign, but even here the picture is far from clear-cut. It's true that out of the six age groups that are listed, 16-24 year olds are less likely to be Yes voters than anyone other than the over-65s. But they're also the third least likely to be No voters, after 25-34 year olds and 35-44 year olds. In a nutshell, if TNS are right there are simply an awful lot of undecided young voters out there.

Probably the best way of summarising the totality of the available evidence is "we don't actually have a scooby whether young people are more likely than their elders to vote for independence or not", but for what it's worth my own gut feeling is that they will indeed ultimately break slightly more for Yes - and that includes the symbolically important 16-17 year old group.

* * *

Scottish Skier mentioned in a comment on Wings over Scotland that he felt that TNS-BMRB are severely underestimating the Yes vote due to shy voter syndrome - ie. since they are the only pollster that interviews people face-to-face, Yes voters are less willing to admit their true intentions to them. As evidence, he points out that TNS are reporting a much lower percentage for Yes among the 2011 SNP support than other pollsters are. This theory certainly has the ring of plausibility to it, but the problem is that TNS are similarly finding a lower No vote among 2011 Labour voters. So it could be that 'shy Yes voters' are one of two factors at play here, with the other more obvious one being the relatively new TNS practice of asking how people think they will vote on the actual referendum date, rather than in a hypothetical referendum 'tomorrow'. The latter factor probably lowers the Yes and No figures by a more or less equal amount.


  1. Do they do face to face at the doorstep or in the street. In this day and age many people won't open doors to unknowns, and in the street most people dodge interviewers. So how do they introduce randomness? I think TNS polls are innacurate for those among many reasons.
    But all polls have problems, eg, how do pollsters using the inet account for the many people who are unconnected or who are not very inet savvy and are not on their lists of approved people to interview. And how are those approved lists made up?

  2. TNS interview people in their own homes. I don't know whether the interviews are prearranged or if they just turn up unexpectedly on people's doorsteps and ask to be let in.

    On the point about internet polls, the theory is that as long you weight the raw data correctly, you can get a representative sample even from a relatively narrow online panel. For instance, if you don't have enough women over the age of 65 in your panel, you can just scale up the responses of the ones you do have.

    I seem to remember that in the early days of YouGov, Peter Kellner wrote a piece explaining that they had done testing, and found that the responses of a properly-weighted sample from an internet panel were more or less identical to the results of a conventional telephone poll - except on one specific topic, and on that topic internet panels always produced different results regardless of demographic weighting (probably because the very fact of being regular internet users was shaping those people's opinions). For the life of me I can't remember what that topic was.

    As for an 'approved list', pretty much anyone can join the panels, although occasionally people get banned for inappropriate behaviour. It always amuses me to recall that Mike Smithson has been banned from the YouGov panel for years!

  3. I'd say it's the question change. People are more likely to know how they would vote tomorrow than how they will vote in 9 months time.

  4. Thanks for your explanation.
    Were you being coy when you wrote " For the life of me I can't remember what that topic was." And do you mean when the topic is Scottish voting intentions.

  5. No, I wasn't being coy - I genuinely can't remember what it was. It might have been attitudes to Europe, it might have been the death penalty, it might have been a social issue like gay marriage. It was a long time ago when fewer people used the internet, so the problem may have resolved itself since then.

    But one intuitive concern about internet polling that hasn't gone away is that the sample is always composed of people who have been polled countless times before, which by definition makes them slightly different from 'normal' people. Telephone polling doesn't have that flaw (although it may have plenty of other flaws).