Lib Dem Voice got the presidential race badly wrong, and it raises questions about their surveys

Your ads will be inserted here by

Easy Plugin for AdSense.

Please go to the plugin admin page to
Paste your ad code OR
Suppress this ad slot.

deweySo, Lib Dem Voice took the route of the Literary Digest as their survey got the result of the presidential election completely wrong. They predicted that the first round results would be 52% for Daisy Cooper, 30% for Sal Brinton and 18% for Liz Lynne, with the actual result being 47% for Brinton, 27% for Cooper and 26% for Lynne. That’s one candidate given almost double the votes she actually got, while the other two are underestimated by about 50% each. Basically, as a prediction of the result, it’s not much better than a random number generator would have been.

So, we’ll have a quick pause for a ‘told you so‘ because that prediction felt wrong to me for the reasons I set out there – the LDV surveys come from a skewed sample that isn’t a balanced representation of Liberal Democrat members. Yes, I know they like to put various disclaimers on them, but those disclaimers always come after a headline that says ‘Lib Dem members think‘ (or something similar) which means the first impression is that this poll represents all members. Indeed, if you just look at the headline – and that’s all you get on the LDV Twitter feed and on other social media – you don’t get any disclaimers, and just get told ‘what Lib Dem members think’.

Now, we often get the claim that these surveys have shown similar results to other surveys of Lib Dem members undertaken by polling companies, so I went looking for the evidence on that. As far as I can see, this is based on a few questions from a few years ago (and Mark Pack’s FAQ on it that people point to is over two years old too), so hasn’t been done on a significant scale or recently. Pointing out that something was vaguely accurate a few years ago does not magically make it accurate now – especially when there’s a very big piece of evidence (the Presidential survey) that says it’s not.

This matters because the LDV surveys and their results are taken seriously by many people, and they could well be giving a wrong impression about what party members think. As it stands, people are being told that Lib Dem members overwhelmingly continue to support the coalition and think the party is on the right track, but what if they don’t? If the people being surveyed aren’t representative of the wider party membership, why are their views being presented as if they are? The most recent piece of comparable data suggests that using the LDV poll as a guideline to what members think isn’t accurate, and it’ll take a lot more than pointing at something from a few years ago to change my mind.

5 thoughts on “Lib Dem Voice got the presidential race badly wrong, and it raises questions about their surveys”

  1. That poll came with a much stronger set of qualifications than other polls due to the finding that 91% of people intended to vote.

  2. The LDV survey suggested the party establishment candidate would do rather worse than she did, suggesting that the coalition and leadership are more popular with the membership than the LDV surveys would imply.

  3. I think the big problem with this sort of polling is the Presidential election is going to have a skewed voting base. The people voting in that sort of survey are those paying more attention to activist base stuff than the typical party member. On “general opinions” I suspect that their polling isn’t far off, and it’s certainly going to be more accurate than anything attempted by the mainstream polling organisations.

    But for this sort of thing, Sal had name recognition amongst the wider membership, whereas before the campaign Daisy was almost completely unknown except within the conference attending/clued in activist/turns up to LDV regular types. Jennie had to tell me I had in fact met her, still not 100% sure she’s right on that, to most of my local party, she’d be a complete unknown.

    It is, of course, impossible to know, and unless someone wants to commission a polling company (or indeed an internal calling group with access to party membership data like, for example, the Bradford Phone Bank) to do a full members survey they remain the best we’ve got.

    Hmm, wonder how much it would cost to get the phone bank to call 1000(ish) members and weight the sampling properly?

    1. It has been done by YouGov as they have asked me on occasion if I’m a member of a political party, and there was a poll done by them in 2008 (I think it was a test to see how popular Lembit was amongst members and to see if Ros could beat him) that was specifically for members. I know they’ve also assisted some academic research into political party members as well (I did a post about some a couple of months ago) on top of the other research questions they’ll occasionally drop into their political surveys.

Comments are closed.