A few weeks ago, I wrote an article about how Google Trends data was going against all opinion polls by suggesting that more people are searching for ‘vote Trump’ than ‘vote Clinton’, and that the former search was moving further ahead of the latter as the US presidential vote drew nearer.
As it turned out, Donald Trump’s victory in yesterday’s vote suggests that internet searches may be a more accurate indicator of what’s about to happen than any sort of survey or opinion poll. The same was true for the UK’s referendum on EU membership, where polls and bookmakers expected a Remain vote, but more people were Googling ‘vote Leave’, and that was what more people went on to do.
So, does internet search engine activity offer a window into the true behaviour of humans? Is what people are doing in front of their computers and smartphones an insight into the realities of what we will do rather than what we say we’ll do? And if so, just what feats of prediction is Google Trends capable of?
To test its psychic powers a little further, I’ve put three of this coming weekend’s events into Trends in an effort to see whether what people are Googling will reflect what really happens.
The X Factor
ITV’s Saturday night staple is now down to its last seven contestants, but which of them is getting the most Google activity?
The answer is the controversial Honey G, by a country mile. In fact, she’s so far ahead that you can barely see anyone else’s graph when plotted against hers. Some way behind her in second place is Matt Terry, who the Engage Web X Factor boffins tell me is all but certain not to be eliminated.
Trends only allows you to plot five different entities on one graph, so let’s assume that those two contestants are through and it comes down to one from the other five. Here is their activity from the last 24 hours:
As you can see, it’s pretty close between the five of them, but Sam Lavery just has the lowest average vote. Second to last are 5 After Midnight, who I’m told are one of the favourites so perhaps we have to consider variant spellings here, while rank outsider Saara Aalto is performing the best of the five.
We do have to consider variables, such as somebody putting in an exceptionally good or bad performance on the night, but today’s stats suggest that Sam is generating the fewest searches and, in a reality TV world where hype often converts to votes, I’ll take this as an indication of her getting the least approval on Saturday. Particularly if she ends up in a sing-off against 5 After Midnight, this puts her in major danger of elimination.
England vs. Scotland
Old enemies England and Scotland will go head to head on the pitch at Wembley Stadium this Friday in what will be their first competitive meeting since 1999, with both nations at pretty low ebbs in footballing terms, but what outcome are people Googling?
With sports, we have to remember that it comes down to action on the pitch rather than a public vote, but these results likely reflect betting patterns, which in turn often reflect the likelihood of an outcome. Therefore, an England win here seems to be the most prominent prediction.
Brazilian Grand Prix
With the Grand Prix in Sao Paulo wrapping up on Sunday, I’ve taken the five favourites to win the race and looked at the worldwide search volumes rather than just the UK’s, where Lewis Hamilton is always likely to be a Google hit.
It looks like young Belgian/Dutch racer Max Verstappen has been in every motorsport fan’s Google search box this week. I thought this may have been down to his outspoken views on F1 that were widely reported on Monday, but the peak of his traffic came in bursts a few days before that. Is Google willing him to victory?
To summarise, as of this morning, Google Trends thinks:
– Sam will be eliminated from X-Factor this weekend
– England will beat Scotland in football
– Max Verstappen will win the Brazilian Grand Prix
If Google were to place a £10 treble bet on these three outcomes, it would be over £500 richer on Monday morning if they all came off.
It will be interesting to see how well these crude Google statistics reflect what’s actually going to happen this weekend. This may well turn out to be a total flop, but in any case, I’ll review the predictions next week and analyse the ups and downs of using Google Trends to forecast real world events.