A month ago I posted about how I was testing an intuition that researchers might be better off asking participants to predict the behaviour of a friend, rather than their own behaviour. I’ve used voting intention to do this because it is to the point, involves asking a simple question that can be easily repeated, and – crucially – because it can be readily validated against the forthcoming UK general election outcome.

I used Google Surveys to independently field the two questions A) and B) shown below between the 29th April and 1st May, at the start of electoral campaigning. I then re-fielded them, unchanged, between the 2nd and 4th June, just ahead of this Thursday’s election. That gave four data sets in total, two for each question.

A) How will you vote in the UK general election on the 8th June?
Single-choice, answer list comprising: Conservative, Labour, Liberal Democrat, SNP (in Scotland) / Plaid Cymru (in Wales), Other (e.g. UKIP, Green, Independent), I haven’t decided yet / I don’t know, I probably (or definitely) won’t vote.¹

B) Please think about someone you know well and regard as a friend. How do you think THEY will vote in the UK general election on the 8th June?
Single-choice, answer list comprising: Conservative, Labour, Liberal Democrat, SNP (in Scotland) / Plaid Cymru (in Wales), Other (e.g. UKIP, Green, Independent), I honestly don’t know how they will vote, I don’t think they will vote at all¹

The aim – I stress – was not to predict the election outcome but to test the power of this potential framing effect by comparing relative findings across the four surveys. Good job too, because one thing I discovered in the process is that Google Surveys isn’t at all representative of UK political opinion! If Labour go on to win an outright majority then I’ll have to eat those words, but despite their surge in the polls such an outcome still seems mightily unlikely. The reason for the bias, I’m sure, lies in Google’s sampling, which is mostly drawn from online news websites. By definition, those visiting such sites are more likely to be interested in politics, and it’s a fact that such people are attracted more to the political Left than Right. Whilst Google do handily auto-weight by age, gender, and region, any attitudinal bias inherent in the sample sourcing remains.

Fortunately, there was much interest in the findings despite this Labour bias, as follows:

If my intuition that ‘asking about a friend’ proves more accurate than asking participants about their own intentions, then not only should we expect a relatively high turnout amongst younger voters, especially those 18-24, we should also see a LOWER than expected turnout amongst older (65+) voters…

…Many in the media talk about whether the younger vote will turn out this time, and the potentially positive impact on Labour’s chances if it does, but I have seen nothing, until now, that suggests that older voters may stay at home. The blue bars in the charts below confirm that a month ago younger votes were less likely to declare that they would vote, but now say they are much more likely to do so (for example, 24% of 18-24 year olds thought they probably or definitely wouldn’t vote when asked a month ago, but that has fallen to 18% now). The nature and size of this effect in this age group is similar when we ask about a friend (the orange bars). Yet, whilst self-declared intention to vote hasn’t changed much over the same period amongst those aged 45 to 64 (hovering around the 8% mark on average), those aged 65+ answering in relation to what they think a friend will do predict that almost 15% will avoid the polls on Thursday. As you can see, this change has materialised over the course of the last four weeks.

I probably (or definitely) won’t vote (self declared) / I don’t think they will vote at all (friend prediction)

Another observation is the slightly strange reading for 35-44 year olds, which perhaps suggests that they are the ones feeling most disaffected with how campaigning has unfolded so far? However, of most interest to me is whether the orange bars or the blue bars tell the most accurate story, and I wait for Thursday’s result and the subsequent turnout figures to find out!

People declare themselves more certain about HOW they will vote now compared to a month ago, but when asked in relation to a friend levels haven’t changed much over time…

…The findings suggest that, as individuals, our feeling of certainty about how we’ll vote increases as campaigning unfolds and we approach election day. However, when looking at the “friend prediction” data it seems that we knew this all along – participants in those surveys were just as confident about how a friend would vote a month ago as they are today. In other words, we at least claim to know the minds of others better than we know our own! This is going to be harder to validate, because the blue and orange bars are close now (3.5 points between them, on average), but at the very least it says something interesting about how our own perception differs from the impression we give others and/or about our own mind-reading powers!

I haven’t decided yet / I don’t know (self declared) / I honestly don’t know how they will vote (friend prediction)

There was lots more of interest that was more political than methodological, and which I’ll omit for that reason! That said, our data confirmed the dynamics that polls have reported over the same period – in particular the improvement in Labour support, and higher voting intentions amongst younger voters, which gives confidence that the samples, despite the bias mentioned, are credible and consistent enough for comparative analysis across the four data sets. I’ve focused on differences by age bracket, but I could have said things about gender too. Most notably that men are more likely than women to claim to know the mind of their friend. Hmmm… less sure about that one.

Can’t wait for Thursday night!

¹ Google Surveys limits answer lists to seven items only, hence the SNP / Plaid Cymru and Other groupings.