Spain General election29 Nov 2026in 8 days PP + Vox majority 99% 2023 investiture bloc 1% Calendar Track record Research
Research · Spain
EN ES

Reading Spanish polls: the correction test and the CIS's raw figures

Abstract. The core methods paper for Spain. Taking off half the average past miss, with each election predicted from the others, was further from the result at most of the eight general elections since 2004, so Spanish polls are taken as published. The CIS's direct vote, rescaled to those naming a party, was further out than its own estimate at five of seven elections, so we use the estimates. The paper also summarises the seat calculation, which reproduces the 2023 Congreso exactly.

Polling Forecast · updated 9 November 2026 · 8 min read · Spain, method, leave-one-out, correction, CIS, voto directo, estimación de voto, D'Hondt

Summary

This paper sets out how Polling Forecast reads Spanish opinion polls for the next general election, and why, unlike in Britain, it does not correct them. Final polls at the eight general elections since 2004 missed by about a point per party on average, and by more at two of them, but the direction changed from one election to the next, and a correction built from past misses would have made the forecast worse at 5 of the eight. The CIS publishes what its respondents said before its own adjustments, and that raw figure was further from the result than the CIS's published estimate at five of seven elections. The seats instead depend on the D'Hondt count in 52 provinces, most of them small. We model that count exactly: given the official 2023 votes, our calculation reproduces all 350 seats.

What has changed since this paper was written

Updated 9 November 2026. The rest of the paper is as first published, except where a sentence below says it was corrected.

  • An early election (5 October). Pedro Sánchez dissolved the Cortes; the general election is on Sunday 29 November 2026. The forecast, the election-day model and the days-to-go figures use that date, and polls may not be published from 00:00 on 24 November (Madrid time) until the polls close.
  • Only elections from 2000 shape the forecast (28 September). Earlier elections stay in the long view and on the history pages but no longer feed any test, correction or error size. In Spain nothing before 2000 was used, so nothing changed.
  • Our figures beside the polls (26 October). The correction is off in Spain, so our projected shares are the poll average, and the pages say so rather than repeating the figures.

The electoral system

Each voter chooses one closed list in their province. The Congreso de los Diputados has 350 seats, and the rules are in articles 162 and 163 of the LOREG (Ley Orgánica 5/1985 del Régimen Electoral General).

  1. Seats per province. Ceuta and Melilla elect one deputy each. Every other province starts with two seats, and the remaining 248 are shared among the 50 provinces in proportion to population by largest remainders. The apportionment is fixed in the decree that calls each election; for 2023 it was Real Decreto 400/2023.
  2. Threshold. A list needs 3% of the valid vote in the province, blank votes included, to share in its seats.
  3. D'Hondt. Within each province the seats go to the lists above the threshold by the D'Hondt method: seats are handed out one at a time to the list with the largest quotient \(v_p / (s_p + 1)\), where \(s_p\) is the number of seats it has so far. In Ceuta and Melilla the single seat goes to the list with the most votes, which is what D'Hondt does with one seat.

A majority is 176. A Prime Minister needs 176 votes at the first investiture ballot, or more yes than no at the second (Constitution, art. 99), so a government can be formed with fewer deputies if enough others abstain.

The data

Results by province for every general election since 2004 come from the Junta Electoral Central's official summaries in the Boletín Oficial del Estado. Province boundaries come from the Instituto Geográfico Nacional, and statistics from the INE. Polls come from the CIS's own studies, from pollsters' and newspapers' own releases where the figures could be read, and otherwise from Wikipedia's tables, graded C and marked on the site. The final-poll record has 140 final polls at eight elections from 2004 to 2023. The Spanish data: results, maps, statistics, deputies and polls gives every source and licence.

The seat calculation

Our implementation follows articles 162 and 163 of the LOREG: in each province, D'Hondt among lists with at least 3% of the valid vote, with the number of seats per province of 2023. Run on the official 2023 votes, it reproduces the official result exactly, province by province:

PartyOur calculationOfficial
PP137137
PSOE121121
Vox3333
Sumar3131
ERC77
Junts77
EH Bildu66
PNV55
BNG11
UPN11
CC11
Total350350
Provinces where every seat matches5252
Live The seat law run on the official 2023 votes, against the official result Computed from the database, polls to 19 November 2026.

A separate check in our data pipeline runs the same rule on every general election since 2004 and reproduces all 350 seats at each of the eight. Two things are not modelled: exact ties, which the law settles by lot and which have not arisen, and a change in the number of seats per province, which is re-apportioned by population before each election (From votes to seats: D'Hondt in 52 provinces).

How far Spanish polls have missed

Final-poll average minus result, in points, party by party:

Party20042008201120152016April 2019November 20192023Average
PP+4.8−0.3+1.7−0.8−2.9+3.6−0.1+1.0+0.9
PSOE−5.2−1.7+2.5−1.5−1.4+0.6−0.4−3.3−1.3
Sumar (and predecessors)+1.2+1.2−1.0−3.1+3.7−1.8+0.9+1.0+0.3
ERC−0.5+0.80.0−0.3–−0.7+0.2+0.20.0
Junts+0.1−0.2−0.9−0.2–−0.5−0.6+0.3−0.3
PNV+0.1+0.3−0.10.0–−0.2−0.1+0.10.0
EH Bildu––−0.3+0.3–−0.1–−0.1−0.1
Cs–––+5.5+1.2−1.1+2.5–+2.0
CC+0.1–0.0––−0.2––0.0
Vox–––––+0.6−2.6+0.7−0.4
BNG+0.4–+0.1–––––+0.2
Live Congreso: final-poll average minus result, by party (points) Computed from the database, polls to 19 November 2026.

Averaged over the eight elections the table suggests a pattern, with the PSOE understated by about a point and the PP overstated by a little less. But much of the PSOE's figure comes from 2004, when the vote followed the Madrid train bombings of 11 March by three days and the last legal polls had been taken before them, and the sign changes from election to election for every large party. The left of the PSOE is measured as one lineage (IU, Podemos and its allies, Unidas Podemos, Más País and Sumar), because its parties have merged and split at almost every election (Measuring Spanish polling error, 2004–2023).

The test

Each election was predicted with the correction learned from the other seven: half the mean error, for parties measured at two or more elections, followed by a rescale so the shares keep the same total. The parties scored are the PP, the PSOE, Vox, the left lineage, ERC, Junts, EH Bildu, the PNV, the BNG and Coalición Canaria. Ciudadanos, which did not stand in 2023 and is not in today's model, is left out of the test.

ElectionPolls' average missCorrected, leave one outVerdict
20041.551.50closer
20080.740.75level
20110.740.78further out
20150.870.90further out
20162.652.72further out
April 20190.931.02further out
November 20190.690.76further out
20230.850.77closer
Average1.131.15
Live Correcting Spanish polls, each election left out in turn Computed from the database, polls to 19 November 2026.

The correction was closer at 2 elections and further out at 5, and worse on average. Our Spanish projection therefore takes the poll average as it is. The code checks this each time the forecast is built, so if a future election changed the verdict the correction would switch on without any change to the method.

Sumar and Podemos in the correction

Today Sumar and Podemos are polled apart, but the history measures them as one lineage. Any correction would therefore be worked out for the lineage and shared between the two in proportion to their shares, and the same folding and splitting is used wherever the lineage matters (the replays in If we're wrong: replaying past Spanish polling misses on today's polls and the election-day projection in From today to polling day in Spain). With the correction switched off, this changes nothing in today's projection.

Raw answers and the CIS's estimate

In Britain, starting from what respondents say before the pollster adjusts it beats the published headline (Why we start from British pollsters' raw answers rather than the published headline). In Spain one pollster publishes both figures for every study: the CIS, a public research centre, prints the "voto directo" (the party a respondent names, as a share of everyone asked) beside its "estimación de voto" (its estimate of the result). Private pollsters mostly publish only their estimate.

The direct vote is not comparable with a result as printed, because it includes the undecided, abstainers and those who do not answer. We rescale it to those who name a party or blank: party \(p\)'s direct vote divided by the sum of direct votes for all parties and blank. That is the Spanish equivalent of Forschungsgruppe Wahlen's "Politische Stimmung" in Germany. We then compare it and the estimate from the same study with the result, for the CIS's last study before each election.

ElectionPartiesDirect voteEstimateDirect vote closer
200481.651.936/8
200881.751.013/8
201190.690.662/9
2015111.121.155/11
201691.681.284/9
November 2019122.031.773/12
2023110.810.653/11
Average1.391.21
Live CIS: direct vote (share of those naming a party) against the estimate in the same study, average miss per party Computed from the database, polls to 19 November 2026.

The estimate was closer at five of the seven elections, and on average. The two exceptions are 2004, by about a quarter of a point, and 2015, by a few hundredths. The largest differences run the other way: in 2008 the direct vote had the PSOE at 47.8% and the PP at 32.6%, against a result of 43.9% and 39.4%, and the CIS's estimate (40.2% and 38.7%) was much nearer. The comparison covers seven of the eight elections, because the April 2019 study is not yet in it.

So the Spanish case is like the German one. The published figure beats the raw one, and the raw figure is available from one pollster only. A method that started from raw figures would rest on the CIS alone, with every house effect of the CIS inside the forecast, and would have been further out more often than not. We show the CIS's direct vote on the polls page beside every estimate it has published since 2023, and we use the estimates.

Where the forecast can still be wrong: the provinces

The seats do not follow the national shares in proportion. In 2023 the PP won 33.1% of the vote and 39.1% of the seats; Sumar won 12.3% of the vote and 8.9% of the seats. Most of that gap is D'Hondt applied in small provinces, where a party needs a quarter or a third of the vote to be sure of a seat. The last seat in each province is where a small national change moves seats, and the forecast is built to show it.

  • Average. Each pollster's latest poll within 30 days of the newest counts once.
  • Provincial shares. Each party's 2023 share in each province is scaled by its national change (proportional swing). Sumar's 2023 vote is split between Sumar and Podemos in their national ratio; SALF, which did not stand in 2023, takes its national share everywhere.
  • Seats. D'Hondt with the 3% threshold, province by province.
  • Simulation. In 2,000 simulations each party's national share is moved by a random error with the size of its typical final-poll miss, each party's share in each province by a further random factor, and the seats are counted again (Simulating Spanish polling error: where the chances come from).

The share of simulations in which a party wins a seat is its chance of being in the Congreso; the share in which a combination of parties holds 176 seats or more is its chance of a majority.

Limitations

  • Eight elections, two of them dominated by one shared miss (2004 and 2016), are a small sample.
  • Proportional swing from 2023 ignores local candidacies and pacts; if Sumar and Podemos stood together again, or a regional list changed, the provinces would move differently.
  • Seats per province are re-apportioned by population before each election; we use 2023's until the next decree is published.
  • Many polls this cycle reach us through Wikipedia's tables; each is marked.
  • The next election is due by 29 November 2026; these are chances for an election held today. From today to polling day in Spain widens them for polling day, and What we don't know yet: limitations of the Spanish forecast collects the rest.

References

Constitución Española (1978), article 99.

Jennings, W. and Wlezien, C. (2018). Election polling errors across time and space. Nature Human Behaviour, 2(4), 276–283.

Ley Orgánica 5/1985, de 19 de junio, del Régimen Electoral General (LOREG), articles 162 and 163.

Real Decreto 400/2023, de 29 de mayo, de disolución del Congreso de los Diputados y del Senado y de convocatoria de elecciones.

How to cite

Polling Forecast (2026). Reading Spanish polls: the correction test and the CIS's raw figures. Retrieved 21 November 2026 from https://www.pollingforecast.com/research/spain-method. Live figures change as new polls arrive; the date above is when the text was last revised.

You may also like