Another 2008? The Tactical Vote the Left Is Waiting For Has Already Happened, and It Is Not Enough

In 2008 Zapatero won without almost anyone crossing blocs: IU lost a quarter of its vote, ERC more than half, and the Socialists collected enough to add five seats. Some hope that fear of a PP–Vox government will repeat that concentration. The data say it has already happened: of every hundred Sumar voters from 2023, fifteen have gone to the PSOE and the space to its left has shrunk to little more than half. What is left to concentrate gives the PSOE between 5 and 14 seats, takes between 2 and 5 from PP and Vox, and never touches the majority. D'Hondt pays four times less for a vote that moves within a bloc than for one that crosses it.

25 October 2026 · Miguel · Leer en español

Every campaign has its favourite precedents. On the left, 2008: a worn-out Socialist government, a crisis that was coming and that the government denied, and a victory by vote concentration. On the right, 2011, which we cover in a second article: the collapse of that same government, with the crisis already upon it. The two share a feature that is usually forgotten: in neither was there a large transfer between blocs. What changed in 2008 and in 2011 was what happened inside each bloc and, above all, who stayed home.

That is the premise we take seriously here. The blocs of 2026 are very watertight: movement happens inside them, not between them. If so, how much could a 2008, meaning a concentration of the left’s vote in the PSOE (the Socialists), change the result on November 29? We calculated it on today’s forecast, moving votes within each bloc in each of 20,000 simulations and rerunning the seat allocation.

What 2008 was

The PSOE went from 11.03 to 11.29 million votes between 2004 and 2008, from 42.6% to 43.9%, and from 164 to 169 seats. The conservative People’s Party (PP) also rose: from 9.76 to 10.28 million, from 37.7% to 39.9%, and from 148 to 154 seats. Both big parties grew at once because the vote concentrated. Izquierda Unida (IU, United Left) fell from 1.28 to 0.97 million, from five seats to two. ERC (Catalan republican left) fell from 652,000 to 298,000 votes, from eight seats to three, and the Catalan Socialists (PSC) rose from 21 to 25 seats: a good part of the Socialist gain was in Catalonia. Turnout fell almost two points, from 75.7% to 73.9%.

It was the campaign of the “useful vote”. The financial crisis had begun in the summer of 2007; the government maintained that Spain was safe, and the campaign became a two-man duel, Zapatero or Rajoy, that drained the small parties. The post-election studies by the CIS (the public polling institute) confirm it: the direct transfer between PSOE and PP was one or two points in each direction. The PSOE did not convince PP voters. It convinced IU and ERC voters that the ballot that counted was its own.

What has already happened in 2026

Translated to today, 2008 would be a transfer from Frente Amplio (Broad Front, the coalition to the left of the Socialists that succeeds Sumar) and Podemos to the PSOE, and from the nationalist left to the PSOE, to stop a PP–Vox government. The CIS layer of our model, which crosses voting intention for 2026 with recalled vote in 2023, says that movement is already largely done.

Of every hundred Sumar voters from 2023, the model keeps only 30 in Frente Amplio and sends 32 to Podemos, 15 to the PSOE and 7 to abstention. Sumar had 3.04 million votes and 31 seats; today’s forecast gives Frente Amplio 1.87 million and 11 seats, and Podemos 0.55 million and one. The space to the left of the PSOE has lost almost a million votes in three years, and the split between two brands wastes much of what is left: in a seven-seat province, two lists with 5% each get nothing; one with 10% wins a seat.

The PSOE, for its part, keeps 67 of every hundred of its 2023 voters. If it still stands at 7.15 million votes it is because it collects by other routes: 1.35 million from people who abstained in 2023, 440,000 from Sumar, 270,000 from new voters and 170,000 from people naturalised since then. It is a different PSOE, with a third of its voters replaced, but the same size.

What is left to concentrate

The question is what would happen if, between now and November 29, the useful vote went further. We moved to the PSOE 15%, 25% or 40% of the vote the forecast gives today to Frente Amplio and Podemos, equally in every province. The 25% is the order of magnitude of what IU lost in 2008. The 40% is an extreme. We then tried the same adding ERC, EH Bildu and BNG, which would give up half that fraction, as ERC did in 2008.

With 25% of Frente Amplio and Podemos moving to the PSOE, about 600,000 votes, the PSOE gains 9 seats on average, Frente Amplio loses 5 and PP and Vox lose 3. With 40%, close to a million votes, the PSOE gains 14 and the right loses 5. If ERC, EH Bildu and BNG also give up 20% of their vote to the PSOE, the right loses 5 seats and the nationalist left another 5 that go to the PSOE. In none of the nine cases does the probability of a PP–Vox absolute majority fall below 96%: they start from 199 seats and concentration takes at most five from them.

Almost everything the PSOE gains comes out of Frente Amplio and the nationalists, not out of the right. The provinces where some seat changes under the 25% case are Madrid (in 98% of simulations), Barcelona (86%), Valencia (69%), Alicante (53%), Las Palmas (52%) and the Balearic Islands (39%). They are the large and medium districts where Frente Amplio has remainders it cannot convert and where the PSOE disputes the last seat with the PP or Vox. In districts with five seats or fewer, where Frente Amplio has between 2% and 7%, its votes are already lost today and transferring them barely changes the allocation, as we saw when calculating a withdrawal from the small provinces.

Why D’Hondt pays so little

A vote that moves from Frente Amplio to the PSOE only moves a seat when two things happen at once: Frente Amplio’s vote in that province was a remainder without a seat, and the PSOE’s quotient overtakes the last quotient of the PP or Vox. The first condition holds in many provinces. The second, in few: PP and PSOE are already the two big parties and their quotients are far apart in most allocations. The result is that the left’s concentration is worth about five PP and Vox seats per million votes moved. A vote that crosses blocs, from the PSOE to the PP, is worth eighteen seats per million: it subtracts from one and adds to the other in the same quotient.

The chart anticipates what the other article tells: moving votes within the left towards the PSOE is the least efficient channel of all. A PSOE voter staying home is worth almost twice as much as a Frente Amplio voter switching to the PSOE, and with the opposite sign.

And the right can concentrate too

2008 also had its face on the right: the PP rose by half a million votes and six seats without the left falling. In 2026 the equivalent would be a transfer from Vox to the PP, the “useful vote” to remove the government. We tested it with the same fractions. With 25% of Vox and SALF moving to the PP, 1.3 million votes, the PP gains 25 seats and Vox loses 25. The bloc gains 0.7. With 40%, 2.1 million votes, the bloc gains 1.5.

The right is already concentrated: with two parties adding up to 50% of the vote and few wasted remainders, D’Hondt is almost indifferent to how it is split between them. What is at stake in the PP–Vox transfer is who leads the bloc, not whether the bloc has a majority. Today that transfer is fairly balanced: 9 of every hundred 2023 PP voters go to Vox and 10 of every hundred Vox voters go to the PP. With 7.9 million on one side and 3.0 on the other, the net balance is about 400,000 votes towards Vox.

What 2008 had and 2026 does not

Two underlying differences make 2008 even less repeatable.

The first is the reserve of voters. In 2008 the PSOE could mobilise a 2004 electorate that was already its own. In 2026 the electoral roll has turned over by about 10%, and those coming in do not resemble those going out.

Of every hundred people who abstained in 2023, 58 intend to abstain again, 15 would vote for the left and 15 for the right: the pool of abstainers is not left-wing, it is split down the middle. Among the 1.6 million new voters, the right doubles the left: 37 to 22, with Vox as the first party. Only the naturalised, 600,000 people, lean left, and with 34% abstention. The PSOE of 2008 grew with the roll. The PSOE of 2026 swims against it.

The second is that in 2008 the IU or ERC voter who switched to the PSOE was voting for a party that was winning. The useful vote works towards whoever is ahead. Today the PSOE is second by six points, and the party that seems to be receiving the useful vote is the PP: 10 of every hundred 2023 Vox voters and 13 of every hundred voters of other parties.

Conclusion

The 2008 the left is waiting for has already taken place: the space to the left of the PSOE has lost a million votes since 2023 and the PSOE has collected part of it. What is left to concentrate, even at the pace of IU in 2008 or faster, gives the PSOE between 5 and 14 seats that come almost entirely from Frente Amplio, Podemos and the nationalists, and takes between 2 and 5 from PP and Vox. They need to lose 24 to lose the majority. If the blocs are as watertight as they seem, no possible concentration closes that gap. The only thing that closes it is part of the right staying home, and that is what the other half of this analysis is about.

How we calculated it

• Forecast. 20,000 simulations of the altnalysis forecast of October 25, with the same seed as the published one: votes of residents and of the overseas electorate by district and party.
• Scenarios. In each district, an equal fraction (15%, 25% or 40%) of the vote of Frente Amplio and of Podemos (only where it runs on its own; in Catalonia it runs within ERC and in Castellón, Valencia, Alicante and the Canaries within Frente Amplio) moves to the PSOE; in the full scenario, ERC, EH Bildu and BNG give up half that fraction. For the right, the same fraction of Vox and of SALF moves to the PP. D’Hondt is then rerun with the 3% threshold on valid votes.
• Efficiency. Mean change in PP and Vox seats divided by the mean votes moved, per million.
• 2023→2026 transfers. CIS layer of the model (module M4): direct intention from the latest barometers by 2023 recalled vote, with Bayesian recall correction and imputed undecideds. They are the same ones published on the forecast page.
• Historical data. Official results of 2004, 2008 and 2011 (BOE, the official gazette) and CIS post-election studies.
• Limits. The fractions are the same in every province; in 2008 the transfer was concentrated in Catalonia. We do not model the useful vote mobilising abstainers, which did not happen in 2008 either: turnout fell.
• Code. analisis_escenarios_historicos.py, in the altnalysis repository.