December 16, 2019

Article at Gizmodo UK

View original

Why the UK's Voting Intentions Were Safe From Google

The day before the election, I had a look to see if Google Trends data provided any opportunities to beat the bookies… or at least skip the need to stay awake past the exit poll. While certain things were quite telling – search on the parties and leaders was much lower than 2017 for one – in all, the conclusion was a resounding: ¯\_(ツ)_/¯.

Curiously, that wasn’t the public’s reaction when it actually went to the polls and overwhelmingly voted for Boris Johnson and the Conservative Party, condemning Labour to its worst performance since 1935 and its fourth spell in opposition. By the time of the next election, Labour will only have had one election-winning Prime Minister in half a century.

So why couldn’t Google Trends data see this coming? Google may know a lot about you as a person, but for the country as a whole it’s more complicated. Here’s why.

Interest picked up right at the end

So that anyone would read the damned piece, I had to file before election day itself. This meant my data was incomplete. For thoroughness’ sake, here’s what the search data looked like from the beginning of the campaign to election day:

Conservative Party vs Labour Party vs Liberal Democrats

The Leaders one is even more interesting, because the last three days of the campaign was the point where Boris Johnon got a clear and consistent lead in search interest.

Boris Johnson vs Jeremy Corbyn vs Jo Swinson

So what does that mean? Well we know that there was an awful lot of apathy about the unpalatable choice the electorate was facing, and that manifested itself in a lot of “don’t knows” in the polls. Perhaps they really were making their mind up at the last second and it broke for Johnson.

Though to be fair, even with these charts, I doubt I’d have been confident to change my verdict of “¯\_(ツ)_/¯”, so why isn’t it more clear cut?

Google can’t see intent behind search

As anybody who has ever typed Katie Hopkins into a search bar will know, not every Google enquiry is out of respect, admiration and love. So yes, we can see that more people were Googling the Labour Party, but we don’t know *why* they were Googling them. It might be to scope out the opposition, win an online argument or just to make themselves really cross in the time-honoured internet fashion.

By the same token, people might Google Boris Johnson to dig into his, uh, ‘fluid’ political opinions, or his colourful use of language while at The Spectator or being paid “chicken feed” by The Telegraph.

In other words, Google might have a pretty good idea of how you’re voting from the kind of links you click and terms you use, but when everything is bundled into a Labour/Tory/Lib Dem basket, it all gets muddled.

...and even those with an open mind might not have been swayed

Then there are people who are just assessing their options. What with the interest in tactical voting this election, lifelong Lib Dems might have been looking at the Labour manifesto to see if they could stomach it, and vice versa.

Equally, Labour had lots of eye-catching policies that people might want to read about – free broadband, for example – but being open to something isn’t the same as ultimately voting for it.

...or at least Google and political parties think so

Google may make all kinds of things from smartphones to drone delivery, but the open secret is that the real money keeping it going is from context-sensitive ads. In other words, when you search “buy socks” on Google, the top results will be sponsored by big-money sock merchants.

It’s the same with politics, only considerably more grubby. And Google has a whole library of every grubby paid-for ad – of which eight Tory ones were banned for being misleading, and another (unbanned) one linked to a site designed to look like the Labour manifesto. Isn’t democracy marvelous?

Anyway, the point is that people typing in these search terms don’t always land on friendly sites. And that could have an impact too.

Our system remains deeply weird

Despite having the chance to change our voting system back in 2011, the British people in their wisdom decided that First Past the Post was going great.

Thanks to this tremendous foresight, tactical voting was deemed a must for this election, but the advice wasn’t always clear. In Kensington, say, Labour won in 2017 by 20 votes, but at the European elections, the Liberal Democrats won with around three times as many votes. So who should you tactically back if you want a pro-EU candidate?

Labour, it turned out was the answer. The Conservatives beat Labour’s Emma Dent Coad by 150 votes as Sam Gyimah took 9,312 for the Liberal Democrats.

This only covers engaged voters

The thing about searching on Google is that it’s an active choice. You need to engage with it and type in a keyword, and that means you need a certain baseline of interest – something we’ve already established just wasn’t there this time around for the majority of voters.

For everyone else, research may come more passively – most likely via absent minded Facebook and Instagram scrolling, which may or may not eventually materialise in Google searches.

In the 2017 election, the most shared content was about the Tories’ love of hunting foxes and ambivalence towards the sale of ivory. They didn’t leave that giant ‘kick me’ sign in the manifesto this time, and both were explicitly reversed.

Mind the age gap

Measuring Google searches tells us something, but in the same way you’d get different results if you canvased at the bingo hall or at a night club, there’s a certain about of sample bias here. Older demographics use the internet comparatively little, if at all. Some poor suckers don’t even know how to change the default search engine from Bing, who would also be lost in the cracks here.

In the 2019 General Election, the early analysis is telling us that – as per usual – younger generations backed Labour, while older voters stuck with the Conservatives. The point at which that crossover happens is a good indicator as to the general mood of the country.

But this time the divide was truly astonishing:

In short, we can see that (relatively speaking) lots of people were looking up Labour on Google during the campaign. What we can’t see is how many pensioners were throwing Labour leaflets in the bin or nodding approvingly while Boris Johnson said “Get Brexit Done” five times per day on local radio.

Each seat is different

It’s a cliche, but a UK general election is really 650 individual fights, and although they tend to follow overall trends (the ‘national swing’) each one has a slightly different flavour. This is most obvious when you look at Northern Irish seats where the parties are completely different to the rest of the UK, but it’s also true in towns with strong Lib Dem traditions or those that have an MP whose staff dislike him so much that they hijack his Twitter account to resign in style. For example.

This time was weirder than most, because the Brexit Party was standing in less than half the seats it said it would, and there were by my count* no fewer than 15 ex Labour and Conservative MPs standing as independents, for Change UK or for the Liberal Democrats. In the end, not one of these wild cards was elected, but they certainly muddied the waters and drew votes away from others.

So what lessons do still stand from my original post? The main one is probably that despite the excitement on Twitter, most people simply weren’t getting election fever. Even compared to last time round, this election brought out an apathetic sigh rather than a roar for change.

Is this exercise worth repeating in 2024? I think so – for one thing, more people demographically will be able to vote who routinely use Google, assuming it’s still a thing in five years’ time. Whether Gizmodo agrees is another matter, of course...

* Change UK (3): Anna Soubry, Mike Gapes, Chris Leslie; Liberal Democrats (6): Philip Lee, Sam Gyimah, Luciana Berger, Chuka Umunna, Sarah Wollaston, Antoinette Sandbach; Independent (6): Chris Williamson, Gavin Shuker, Dominic Grieve, David Gauke, Anne Milton, Frank Fielding. Have I missed anyone?