The 7th most populous nation in Europe, the 6th economy if you rank them by GDP, the 10th by sovereign debt and a close second (over 21%) by unemployment rate (certainly the country with the best Jamón ibérico) has just two more days to form their opinion about how to cast their vote. Friends and contacts in Spain tell us there is a Poll embargo since December 14 and that the picture is mixed. That makes us a rare source of getting any kind of measurement on what is going on. We are certainly well aware that our method of using Google Trends has its limitations which cannot always be overcome without access to search microdata (currently out of reach for researchers). And using aggregate data of this sort is like trying to follow the score in a soccer match while being outside the stadium. If you are a soccer fan you certainly have a built in econometric model which allows you to tell the score in most cases.
The home team has mostly more fans present so you can identify which side screams or yells, a sigh of relief having just avoided to be scored upon, or a sigh of disappointment for a missed score are also well distinguished from a victorious roar when your own team scores or from the complaining screams when your team was just scored upon. If the team fans speak different languages then it makes it easier to disentangle who is who, booing the referee is also easy to detect and a number of other items fans can distinguish well but there is still so much uncertainty. Certainly there is a bit of guess work and a bit of intuition invloved but from 100% soccer fans more than the random 50% would be able to easily tell the final score if they had to, provided enough of these circumstantial “ifs” go their way.
With that being said and the precaution and caveats still holding we post this update today (we will post one tomorrow and one in the morning of Sunday) showing that Google search is certainly all about Podemos as we are approaching the finish line. We feel confident that we are identifying the search footprint of each party fairly well (certainly increasingly so as we are approaching the finish line) but not necessarily whether or not 1. this is a representative sample (Internet penetration in Spain is 75%, generational bias, etc) or 2. it correlates strong enough with voter intent (we know at least some cases when it doesn’t).
Nonetheless even our cleanest model with “programa <partyname>” which so far showed a close race between Podemos and C’s is starting to show Podemos pulling ahead. As always take it with a grain of salt but here are todays graphs: