This is a shipment from the Kiko Llaneras newsletter, an exclusive newsletter for EL PAÍS subscribers. You can also listen to his new podcast and follow the polls.
A month ago Pedro Sánchez called elections. Have the polls moved since then? Little, but yes. The PP continues as the first force in our average, with 31% of the votes, followed by the PSOE (27%), Vox (19%) and the Frente Amplio (7%). But the small changes we see almost all point in the same direction: the PP's advantage narrows.
The clearest movement of the month is the popular pullback. In our average, the PP has fallen from 32.7% to 31.1%, while the socialists stayed almost where they were (from 26.4% to 26.6%) and Vox gained half a point.
It has been an eventful month. The first, the call itself. The second, the difficulties of the left in putting together a truly broad front. The third event was the tragic accident in Urda, on October 28. The first two events boosted the PSOE, but with the train crash the starting hypothesis was a punishment for the Government. Is it seen in the surveys? It is not clear. The five pollsters who asked before and after the accident see the PSOE going down, but only slightly: between two tenths and a point and a half. Our average, adjusted for the differences between pollsters, tells the same story: if Urda has moved votes, so far they have been few.
It wouldn't be so strange. I have a shortcut for thinking about campaigns: almost nothing matters. It is easy to exaggerate the latest that has happened, and a lot of votes have been decided. This month has put that rule to the test. An accident with 69 deaths, which has cost the Minister of Transport his job, is one of the most intense things that can happen during a campaign. And yet, the polls have not turned around. The campaign, for now, has perhaps been motivated more by the mobilization of the left than by the train tragedy.
The counterpoint of the week was put by SW Demoscopia in a survey for La Séptima. According to his estimate, the PSOE would be the first force with 30.9%, ahead of the PP (28.8%), and Vox would fall to 17%. She is a pollster with little public history and that requires caution. But the scenario is plausible. The distance between PP and PSOE has also narrowed in our average, from more than six points a month ago to four and a half now.
What options does the PSOE have to be the most voted? Our model gives it one in five: it is ahead of the PP in votes in 21% of the simulations. When the elections were called, that probability was 16%; It rose to 24% the day before Urda, and has now returned to 21%. It's a real option for two reasons: four-point errors are not uncommon, and there are three weeks left before voting.
Who will govern?
The graph shows our prediction from the surveys and 20,000 simulations (see methodology at the end). The central result leaves the PP with 129 seats, eight less than in 2023. They are followed by the PSOE (106; 15 less), Vox (68; more than double its 33), the Frente Amplio (15) and Podemos (2), far from the 31 they achieved together with Sumar. The rest of the parties share another 31 seats. My usual note applies here: the ranges are wide because the surveys are approximate. There is a 10% probability that the PP will remain below 108 seats, and another 10% that it will exceed 151 seats.
That is to say, the central result is that the PP remains far from the absolute majority, but that it easily joins Vox. They have around 197 seats, 21 above the 176 necessary to win an investiture.
How firm is that sum? Our model allows us to estimate the probability of each majority, which is the key question. The PP only reaches absolute power in 1% of the simulations, and in another 1% it could do without Vox with the support of PNV and the Canarian Coalition. These are the options for a right-wing government without the ultra party: barely 2%. A month ago they were triple that, but the fall of the PP has reduced them. A PP Government with Vox adds another 89% of options. It is the most likely scenario. But probable is not certain. In 2023, four days before voting, this same model gave PP and Vox a 55% chance of joining, and then they did not join.
Those who have reacted to Urda are the betting and prediction markets. In Polymarket, Alberto Núñez Feijóo has a 90% chance of being the next president, compared to Sánchez's 9%. The day before the accident it was around 82%, and before the call, it was 79%.
Can the left come back?
Socialists will say, rightly, that nothing is written. And they have a pattern in their favor: in the last three general elections, the left finished better than the polls said. In April 2019 he surpassed them by about four points; in November of that year, just barely; and in 2023, by almost four again. The most remembered case is the last one: the right started with an eight-point advantage when the elections were called and ended up winning by only one and a half. It is true that this pattern has been reversed in regional and European elections - since 2023, the right has surpassed the polls 15 out of 20 times - but it makes sense to only look at the general elections: they are the most important elections, in which more people vote, and they can have their own dynamics.
The brake on socialist optimism comes from two places. First, from the distance: the right now starts with about 15 points of advantage, almost double that in 2023. And second, from the microdata of the polls, where a consolidated rightward movement is seen. According to the October 40dB barometer, 10% of socialist voters in 2023 now say they will vote for a right-wing party.
The summary is the same one I offered a month ago: is a left-wing comeback likely? No. Is it impossible? Neither. That is why PP and Vox are left without a majority in one in ten simulations.
Methodology. Our seat calculation results from a four-step process. 1) We start from an average of surveys that weights each survey by its size, its date and the historical accuracy of its pollster, according to our comparator; each day use only the surveys published until then. 2) We estimate the vote in each province from that average, using as a reference the results of the 2023 general elections and, for new parties or those that now compete separately, the provincial data from the 2024 European elections and the 2026 Andalusian elections. 3) We incorporate the uncertainty of the surveys taking into account their historical precision and the distance to the voting date. 4) We simulate 20,000 elections, varying the votes in each one and distributing the seats according to D'Hondt with a 3% barrier per constituency. More details can be read in the description of the 2023 general election model.
Other stories
🎙️ 1. Podcast premiere, sign up!
I have something new that I'm very excited about: I'm launching a new podcast. It's called Between Data, with Kiko Llaneras and it appears every Thursday in EL PAÍS. I do it with José Juan Morales and Nicolás Tsabertidis.
My challenge is to make it sound like this newsletter. It even has a section that we have called the random data. In the first episode we talk about the 29N surveys, Urda and writing with AI. You can listen to it here:
Can I ask you two favors? The first is that you subscribe on Spotify, YouTube, Apple Podcasts or iVoox, so that it arrives to you every week. The second, if you like it, share it and give it five stars: it is what helps us the most to reach more people. For me this is a learning experience, so any clue or idea is welcome, from themes, to software to make reels or sponsorship proposals. Just send me an email.
😴 2. They gave us an hour and we slept less?
On Sunday, October 25, we turned the clocks back and gained an hour. Do we use it to sleep? No. This is what the data from DuckDik, a Spanish health bracelet company, says, with information from hundreds of people. Its users slept 7 hours and 27 minutes that night, almost half an hour less than normal on a Saturday. Those who did not reach seven hours rose from 27% to 39%.
If we do the math, we see what happened. Users went to bed at their usual time, around 00:34, but got up about 20 minutes earlier than normal, in real time, and almost an hour and a half earlier according to the updated clock. In addition, they slept worse: they woke up more times and spent more time awake. Because? The first part has an explanation: the body does not know that we have changed the clocks. That's why we don't win the hour. But that we got up even earlier is a mystery to me. Do we get up to take advantage of the extra hour? If you can think of a better explanation, tell me.
🧬 3. The dictionary changes the definition of “life”
Until now, if you looked for “life” in the RAE dictionary, the first meaning said: “Force or essential activity through which the being that possesses it acts.” It didn't sound very scientific. The new edition just presented defines it as the “property of organisms by which they are capable of exchanging matter and energy, maintaining an internal organization through their own metabolism and reproducing thanks to a heritable molecule, DNA.”
The definition fascinates me.
The change was proposed two years ago by Carlos Briones, a biochemist and astrobiologist at the CSIC. As Antonio Martínez Ron tells it in El Diario, Briones believed that none of the old meanings had a scientific basis and wanted one that could be used to search for extraterrestrial life.
But the definition ties life to DNA. What if life appears on another planet that does not use that molecule? José María Bermúdez de Castro, one of the academics who defended the change, explains that they discussed it: “If that happens, the definition is changed, because the dictionary has to evolve.” Language is alive and evolves, although it does not use DNA.
