What is one vote worth on November 29? Where turning out, or staying home, can decide a seat
Every vote is worth the same, but not every vote has the same chance of deciding a seat. Using the 50,000 simulations of our forecast, we calculated, province by province, the probability that a handful of votes changes the seat allocation: in Ceuta it is ten times higher than in Madrid. And in 2023, 10.4 million resident voters stayed home.
21 October 2026 · Miguel · Leer en español
On November 29, the 350 seats in Spain’s Congress will be allocated province by province under the D’Hondt rule. That has two consequences that are often confused. First, within a province every vote weighs exactly the same, because what counts is the provincial total, not the polling station or neighborhood where it is cast. Second, from one province to another, the probability that a few votes change a seat varies widely. It depends on how close the last seat is and on how many votes are in play.
We measured that second difference with the 50,000 simulations of the altnalysis forecast. It is one way of seeing the impact of voting, or of not voting.
The index: where a few votes can change a seat
| # | District | Seats | Index | Last seat decided by fewer than 1,000 votes | One vote decides it: 1 in | Contested by |
|---|---|---|---|---|---|---|
| 1 | Ceuta | 1 | 100.0 | 27% | 4,000 | PP and Vox |
| 2 | Teruel | 3 | 76.5 | 20% | 5,000 | Vox and PP |
| 3 | Ávila | 3 | 72.1 | 19% | 5,000 | Vox and PP |
| 4 | Segovia | 3 | 66.5 | 18% | 5,000 | PP and Vox |
| 5 | Lleida | 4 | 65.1 | 18% | 5,000 | PP and PSOE |
| 6 | Palencia | 3 | 53.3 | 15% | 7,000 | PP and Vox |
| 7 | Huesca | 3 | 49.4 | 13% | 7,000 | Vox and PP |
| 8 | Zamora | 3 | 46.4 | 13% | 8,000 | PP and Vox |
| 9 | Araba/Álava | 4 | 43.1 | 12% | 8,000 | EAJ-PNV and PSOE |
| 10 | Soria | 2 | 41.3 | 10% | 9,000 | PSOE and PP |
| 11 | Cuenca | 3 | 36.4 | 10% | 10,000 | Vox and PP |
| 12 | Rioja (La) | 4 | 34.0 | 10% | 10,000 | Vox and PP |
| 13 | Girona | 6 | 31.9 | 9% | 11,000 | Vox and PSOE |
| 14 | Navarra | 5 | 31.5 | 9% | 11,000 | PSOE and PP |
| 15 | Gipuzkoa | 6 | 29.8 | 9% | 12,000 | EH Bildu and EAJ-PNV |
- Ceuta tops the list: a single seat, closely fought between the conservative People’s Party (PP) and the hard-right Vox, and only about 38,257 expected valid votes. Its last, and only, seat is decided by fewer than 1,000 votes in 27% of simulations.
- Next come small provinces with the last seat up for grabs: Teruel, Ávila, Segovia and Lleida. In almost all of them, the PP and Vox are fighting over the last seat.
- At the bottom are Madrid (10.0), with many seats and more than four million votes, and provinces where the allocation is more settled, such as Melilla (6.8). A vote in Ceuta is about 10 times more likely to change a seat than one in Madrid.
In Madrid, with 38 seats, the last one is always decided by a small margin in proportional terms, but that margin is thousands of votes out of four million. In a three-seat province, the last seat can be decided by a few hundred.
Not voting counts too
In the 2023 general election, 10,396,976 voters living in Spain abstained. In the provinces at the top of the table, turnout gaps looked like this:
| District | 2023 abstention | Did not vote | Municipalities with the most abstainers |
|---|---|---|---|
| Ceuta | 44.3% | 26,750 | Ceuta (26,750) |
| Teruel | 24.8% | 25,131 | Teruel (7,136), Alcañiz (3,115), Andorra (1,710) |
| Ávila | 23.4% | 30,063 | Ávila (11,130), Arévalo (1,523), Arenas de San Pedro (1,401) |
| Segovia | 22.8% | 26,428 | Segovia (9,563), Cuéllar (1,731), El Espinar (1,710) |
| Lleida | 38.1% | 113,777 | Lleida (36,473), Tàrrega (4,518), Balaguer (4,484) |
| Palencia | 25.2% | 32,820 | Palencia (16,838), Guardo (1,515), Venta de Baños (1,385) |
| Huesca | 27.3% | 45,059 | Huesca (11,187), Monzón (3,618), Barbastro (3,424) |
| Zamora | 27.6% | 39,234 | Zamora (14,974), Benavente (4,101), Toro (2,127) |
In Ceuta, for example, 26,750 people did not vote in 2023, far more than the margin by which its seat is usually decided. That is the practical meaning of the index: where the last seat is tight, some of those who stayed home turning out can change the result, whoever they vote for.
Abstention is not spread evenly: in the city of Huesca it ranges from 18.8% to 51.5% depending on the section. But the key point stands: a vote cast in a high-abstention section weighs the same as one in a low-abstention section. What changes is how many votes are in play.
What this analysis does not say
- It does not say some votes are worth more than others. Every vote in a province counts the same; the index measures the probability that the provincial total is so close that a few votes decide a seat.
- It depends on the forecast’s uncertainty. With more precise polls the probabilities would change; the ranking of provinces is sturdier than the exact figures.
- It does not say who benefits from turnout. An extra vote decides the seat for whoever receives it; the index is the same for every party.
- The “1 in…” figures are approximate: they assume the vote goes to the party contesting the last seat and are rounded to thousands.
How we calculated it
- Last-seat margin. In each of the 50,000 simulations of the October 21 forecast, the D’Hondt allocation of each province is redone with the 3% threshold: the last seat belongs to the party with the smallest winning quotient, and the challenger is the party with the largest quotient left out. The margin is the number of votes the challenger would need.
- Probability per vote. The share of simulations with a margin below 0.25% of valid votes, divided by that window, gives the density of the margin at zero; divided by expected valid votes, the probability that one vote decides the seat. The index rescales it from 0 to 100.
- Abstention. Resident voters who did not vote in the 2023 general election, summed by census section and municipality from Interior Ministry polling-station data (excluding the overseas vote).