Strange day for polls. Despite the average consistently showing the PP above the PSOE, the two surveys known today have made some eyebrows raise by placing for the first time those of Feijóo. The margin of error, therefore, is important when putting the magnifying glass on, because if they continue to be biased—something natural that happens with all polling companies—just as they were in the 23J, we could enter unknown territory.
There is only one survey, the one by SW Demoscopia for La Séptima among the nearly 20 polling firms in the market, which gives victory to Sánchez and is left out of all the calculations in this piece. The question now starts from two specific polls published this Sunday, 8 November: IMOP for El Confidencial and Sondaxe for La Voz de Galicia. Both place the PP below 30% and both give it an advantage over the PSOE. The possible tie appears when submitting those figures to a historical scenario, not by reading their original results.
There is another clarification from the start: an electoral error from 2023 does not demonstrate a permanent bias in 2026. Companies can change their samples, weighting, or estimates of voter turnout. In addition, between an interview and the vote, voters’ decisions can change. We do not know what part of those deviations persists today. Of course, the rest of the dozens of polling firms also still need to be seen, which have to keep extracting data to see the consistency of today’s figures.
Two polls with the PP below 30%: what deserves attention
IMOP places the conservatives at 29.8% and Sondaxe at 29.7%. Their estimates for the PSOE are 25.9% and 27.3%, respectively. That is the starting point: two PP victories, with different gaps and with Vox consolidated as the third force.
The 30% is not an electoral threshold that automatically grants or removes seats. Its interest here is comparative: the PP obtained 33.1% in the 23J, rounded to one decimal. The two publications place it more than three points below that result.
IMOP: the PP recovers ground and the PSOE falls back compared with October
The latest IMOP offers a favorable evolution for the PP compared with the delivery of 23 October. The conservatives move from 28.5% to 29.8%, while the PSOE falls back from 27.1% to 25.9%. Vox barely changes: from 20% to 19.9%.
The distance between PP and PSOE goes from 1.4 to 3.9 points. In the deputies published by the company, the PP rises from 123 to 126; the PSOE falls from 113 to 106; and Vox goes from 72 to 70.
So, that the PP keeps below 30% does not mean that it is falling in this series: it improves compared to its previous measurement.
IMOP on 23J: almost nailed the PP, but fell short with the PSOE
The antecedent of IMOP requires distinguishing by party. Its last publication before the 2023 blackout, of July 17, estimated 32,9% for the PP. The final result was 33,1% rounded: a difference of just two tenths.
The mismatch was greater in the PSOE, to which it granted 28,2% and which ended at 31,7%: 3,5 points more. It attributed 14,5% to Vox, versus 12,4% obtained; to Sumar, 12,8%, versus the final 12,3%.
In those four parties, the average absolute deviation was approximately 1,6 points, calculated with the rounded percentages.
The seats tell the same story with nuances: the PP’s 137 fell within the range of 134 to 142, and Sumar’s 31 within that of 31 to 34. The PSOE, with 121, exceeded the maximum of 108 expected; Vox, with 33, fell below the minimum of 38. IMOP got the PP much better than the PSOE right.
Applying today those differences in full would leave the PP at 30%, the PSOE at 29,4% and Vox at 17,8%. The popular lead would narrow to six tenths.
With the same operation on the October IMOP, the PSOE was 1.9 points ahead: also in the adjusted scenario, the PP’s relative position has improved.
Sondaxe: PP and PSOE rise, Vox falls
Sondaxe draws another evolution. Between its publication of November 1 and that of this Sunday, the PP goes from 29.2% to 29.7%; the PSOE, from 26.9% to 27.3%; and Vox, from 19.2% to 18.6%.
The popular advantage hardly changes: from 2.3 to 2.4 points. In contrast to the Socialist drop that IMOP records, Sondaxe registers a rise of four tenths. The two series match in the order of the parties, but not in the direction of the PSOE’s movement.
Sondaxe’s fieldwork was carried out from October 26 to 28 and from November 2 to 4, with 1,200 interviews in each wave. It is important to keep those dates: today’s publication collects responses obtained days earlier.
Sondaxe got the sum of PP and Vox right on the 23J, but underestimated the PSOE
The historical comparison by Sondaxe also uses its latest publication prior to the campaign blackout, from July 17, 2023. It is not mixed with the poll released on election night. It attributed 34.3% to the PP, 28.7% to the PSOE, 12.5% to Vox, and 14% to Sumar.
Compared with the result, it overestimated the PP by 1.2 points and Sumar by 1.7; it underestimated the PSOE by three points and was one tenth of a point from Vox. Its average absolute deviation across those four parties was 1.5 points, using percentages to one decimal place.
In the seats there is a detail that prevents speaking of an across-the-board failure: Sondaxe projected 141 for the PP and 29 for Vox. They added up to 170, exactly what both ultimately obtained: 137 and 33. It got the total for that block right, although not its internal breakdown.
It also calculated 109 deputies for the PSOE, which ended up with 121, and 37 for Sumar, which obtained 31. The socialist deviation was significant; the complete balance, however, is more nuanced than saying that the survey was wrong in everything.
If its vote errors were repeated in full, the current Sondaxe would stand at PP 28.5%, PSOE 30.3% and Vox 18.5%. The PSOE would move ahead of the PP by 1.8 points. In the adjusted previous wave, that distance was 1.9: it changes almost nothing.
If all the error were repeated, the average would bring PP and PSOE closer to a tie
The operation is simple: to the percentage published today, the difference is added between the result of the 23J and the estimate of that polling firm. The sign is preserved: it is added where it fell short and subtracted where it exceeded.
With equal weight for IMOP and Sondaxe, the average result of that exercise would be PP 29.3%, PSOE 29.9% and Vox 18.2%, rounded. The gap between the first two would be six-tenths in favor of the PSOE. It is an average of two hypothetical scenarios, not the general average of Demócrata polls.
Here is the basis for the headline: repeating the 2023 deviation could turn two published PP advantages into a much tighter contest. The calculation does not prove that that tie exists today, nor does it allow calling the difference “a technical tie” without additional statistical analysis.
The exercise also shows why it is worth keeping the two polling firms separate: complete repetition would first keep the PP ahead in IMOP and first put the PSOE ahead in Sondaxe. The average does not eliminate that discrepancy.
And what if the polling firms had already corrected half?
It is reasonable to consider that the companies have learned from 23J. What we cannot say is that IMOP or Sondaxe have corrected exactly 50%: we do not have a verification that measures that improvement.
To explore that possibility, we propose a second hypothesis: that only half of the historical deviation remains. It does not mean improving 50% of the entire quality of a poll, but adding or subtracting half of each vote difference observed in 2023.
In IMOP, the exercise adds 0.10 points to PP and 1.75 to PSOE, and subtracts 1.05 from Vox. In Sondaxe, it subtracts 0.60 from PP, adds 1.50 to PSOE, and subtracts 0.05 from Vox.
With that assumption, PP would remain first in both houses: by 2.25 points in IMOP and by 0.30 in Sondaxe. The average would be approximately PP 29.5%, PSOE 28.2%, and Vox 18.7%, with a popular advantage of 1.3 points.
The result changes the reading of the full repetition scenario: reducing the historical error by half returns leadership of the average to PP. It is a sensitivity test, not a demonstration of what percentage of the error continues to exist.
The official campaign has not started yet: there is no closed outcome
There are 21 days left until the elections on 29 November. The legal campaign begins on Friday 13 and ends on 27, according to the call decree. The fifteen official days of campaigning remain ahead, in addition to the remaining pre-campaign stretch.
The background from 2023 and today’s data are also not at the same distance from the ballot boxes. The historical publications that were compared appeared six days before 23J; the current ones are published three weeks before 29N. None of them should be treated as if they were interviewing citizens after voting.
The final difference between a poll and the count can combine measurement problems, estimates of turnout, and decisions made after the interviews. We have not isolated how much weight each component carried in these two houses during 23J. For that reason, it would also not be correct to attribute all of the deviation to their methods and then add a supposed campaign comeback: we could count twice part of the same phenomenon.
The campaign can mobilize, demobilize, or redistribute support. Its lagging effect has no known sign: it does not guarantee a recovery for the PSOE nor that the PP will widen its advantage.
A PP and a PSOE close to each other do not equal a majority for Sánchez
Vox remains decisive in both exercises. With a complete repetition of the error, its average would be around 18.2%; with repetition of half, 18.7%. Both levels are far above the 12.4% of 23J.
In Spain, closeness in national vote between the first two does not by itself determine the distribution of seats. The constituencies and the support each candidacy achieves within them matter.
As an additional illustration, when moving the midpoint-of-the-error scenario to a proprietary provincial model, IMOP would give 119 seats to the PP, 114 to the PSOE, and 67 to Vox; Sondaxe, 118, 117, and 68. PP and Vox would sum to 186 in both cases. These figures have not been published by the polling firms and have more uncertainty than the arithmetic operation on percentages.
The model uses the 29N seat distribution and the definitive provincial votes of 2023. It maintains the territorial weight from then, approximates the footprint of new candidacies, and does not incorporate a complete check of the 2026 coalitions. It serves to illustrate dependence on the assumptions, not to ensure a future majority.
The polls published today keep the PP ahead. The comparison with 23J explains how much that lead could be narrowed under specific errors; the campaign, turnout, and the next fieldwork will determine whether that possibility finds support in new data.
How the comparison was made
The two most recent publications by IMOP and Sondaxe identified here, and their public estimates from July 17, 2023, are compared. The reference outcome is the definitive one from the Central Electoral Board, including overseas voting and with PSOE and PSC grouped together. The adjustments use percentages rounded to one decimal to maintain consistency with the previous exercise.
The full scenario applies 100% of the historical difference; the second, 50%. That 50% is an illustrative choice, not a methodological correction accredited by the companies. The averages assign the same weight to each house. The hundredths in the tables allow the count to be checked and do not indicate that predictive precision.
In the seat simulation, the rest of the vote is adjusted proportionally to add up to 100%, and D'Hondt is applied by constituency. No win probabilities or confidence intervals are assigned.