Consumer Trends & Insight
Corporate Reputation
Digital Trends
Employee Engagement
General Election
Government Affairs
Life At Edelman
Women In The World
Influencer Marketing
Integrated Marketing
Digital Design
Brand Marketing
Film Production
Community Management
Media Relations
Corporate Communications & Advisory
Brand Strategy
Data & Research
Financial Services


1 June 2017

If you think polling is hard, try interpreting it

Written by: Harry Spencer, Account Manager at Edelman

General Election

Edelman’s public affairs polling analyst Harry Spencer provides his latest (and not altogether totally serious) take on the polling

We have a lot of new polls, including ones of Wales and London, which are basically consistent with the national picture. Making the challenge of interpreting things harder is that fact most of these are from Yougov, which is finding a consistently better situation for Labour than others. Here’s an imaginary Q&A.

Who’s ahead in the polls right now?

The Conservatives. That’s about the only thing which is clear.

How far ahead of they?

I have no idea. Generally, the best approach is to take the average of all the different polling firms. That would give you a 9% Conservative lead. However, some firms haven’t done a new poll in a while – if you limited it to the past week, you would get a Conservative lead of 7%, from roughly 44% to 37%.

What’s the degree of uncertainty?

Very high. There is a wide spread of results in the polls, partly as a result of the different approaches they have taken to handling the 2015 polling disaster. This explains the range in Conservative leads from 3% to 14%. Somewhere in the middle is usually the best bet, but it’s highly uncertain.

What’s the short-term trend in the polls?

By this I mean over the past week. The answer is, I really have no idea and we need more data. There are not enough data points to really judge, but I think the Conservative lead may have slipped a bit again this week – that’s what Panelbase and Yougov show, although no dramatic shifts – the really big shift was earlier.

What’s the medium-term trend in the polls?

Over the last few weeks the contest has been completely transformed – the Conservative lead of 20% has been cut in more than half, and the Prime Minister’s lead in personal ratings has also dropped substantially. Most of this has been Labour gaining support rather than the Conservatives losing support.

What’s changed?

In technical terms there are a few different changes going on compared to at the start of the campaign:

  • Labour are no longer losing voters to the Lib Dems – and may be gaining a few instead
  • Labour are no longer losing voters to the Conservatives directly
  • Labour are picking up more former UKIP voters – roughly 2% of the electorate has switched directly from UKIP to Labour, whereas before this was non existent.
  • The Conservatives have lost support among 2015 Lib Dem voters
  • There has been a surge in young voters and previous non-voters saying they are going to vote, and vote Labour

Of course, whether this is the case on the day is impossible to know.

What hasn’t changed?

Several things have not changed:

  • Half of the 2015 UKIP vote has gone to the Conservatives. That is still, in all polls, absolutely the case.
  • The Conservatives are also keeping virtually all of their 2015 voters – their retention rate is still higher than Labour’s.
  • The over-65s back the Conservatives by huge margins – in the order of 70% to 20%.

Why has this all changed?

One thing that is definitely going on is it is becoming a clearly two-horse race – support for smaller parties is at its lowest in decades. The squeeze is driving Labour support up and keeping Conservative support very high. While it has been fashionable to trash the Conservative manifesto (and it’s fair), probably just as important was the complete washout of the Lib Dem campaign, which has left Labour as the only alternative. Undoubtedly though the Conservatives have not had the campaign they would have wanted, partly because they seem to have been caught by surprise themselves, and partly because fighting an election on your Brexit strategy when said strategy does not exist was always risky.

I’d also say that Theresa May’s pledge to consider bringing back fox hunting and the failure to deliver a ban on the ivory trade were really bad ideas – they alienated Labour waverers considering the Conservatives by reminding them what they don’t like about the party. And put simply, people love animals and dislike needless cruelty – editorialising much?

Meanwhile Labour has had a good campaign, and a spell of (albeit artificial) unity has served them well.

Are the polls wrong?

Definitely. Some of them must be. The fundamental divide in why the polls differ is in how they handle the surge in support for Labour among young people. Firms which show smaller Conservative leads take this more or less at face value – they believe all the 20 somethings who say they are going to vote, and vote Labour.

Firms like ComRes which produce big Conservative leads, to put it colloquially, do not believe the 20-somethings answering their surveys. They assume, on the basis of a strong historical track record, that in practice lots of these voters won’t turn up.

It is fundamentally a question of what the electorate looks like – if it’s what it says on the tin that’s better for Labour – if, as is traditional, older voters turn up and younger ones don’t, this will be a pretty comfortable win for the Conservatives.

Personally, I expect the latter, because the list of politicians relying on younger voters and getting beaten is very, very long. Some of the polls with Labour only 3% behind have 18-24 year olds turning out at a higher rate than the over-65s – I will purchase, prepare and eat a hat if that happens.

What would it mean in seats?

The magic number is 6.5%. That was the Conservative vote lead in 2015. The only thing you need to know is that, if its higher than that, they’ll make net gains, and if its lower, they are virtually certain to make losses.

If the Conservative lead is less than 5 it will very likely be a hung Parliament, and if it’s bigger than 8 they’ll have a pretty comfy majority. If it’s in double figures, than a majority of 100 or so is still perfectly possible. A few percent around 6.5 either way will make an enormous difference in seat outcomes.

That said, there are signs that both parties might be picking up lots of “wasted votes”. For Labour, the danger with picking up loads of young voters is that they are not especially efficiently distributed – there is no value in increasing Jeremy Corbyn’s majority in Islington North from 20k to 30k.

For the Conservatives, a lot of the Labour-held seats which have larger UKIP votes to eat into have large majorities. This means, it’s at least possible they will reduce their deficit in lots of seats from say 10k to 2k, but still lose the seat.

What would keep me up late at night if I were May?

I would be most worried about the erosion in my best PM ratings. And very concerned about the considerable possibility of winning with a barely increased majority. Such a result would technically leave me the winner, but with my internal party authority shot to pieces. The fact this will happen just before the small matter of negotiating Brexit, which will necessitate disappointing some parts of my coalition of support, would be especially concerning. The fact that such a situation is a very real possibility (if not likely) would concern me a lot.

What would keep me up late at night if I were Corbyn?

I’m going to generously assume he wants to win the election.

The most concerning thing for Labour is the UKIP to Conservative swing alone is more than enough to cost dozens of MPs their seats, and nothing has changed on that front. If the Conservatives keep their voters and the UKIP voters who switched, defeat is almost certain.

I’d also be worried about the softness of the vote in general. Reliance on young voters is usually a good way to lose. And Labour’s poll share is very dependent on that.

Please update your browser.

This website requires Chrome, Firefox, Safari or Internet Explorer 9+