“Two weeks before the holding of the general elections, the most probable result estimated by Ipsos grants an absolute majority to the sum of the two main right-wing parties, PP and Vox. With a participation of 70%, lower than the average of all the general elections held to date, the popular would obtain 145 seats with 35.3% of the votes. This would mean 14.5 points and 54 deputies more than those achieved in the 2019 elections. Vox, for its part, would obtain 36 deputies with 12.6% of the votes: 16 seats and 2.5 points less than four years ago. Between the two they would add 181 seats in the Congress of Deputies, that is, five above the absolute majority. This is the photograph that reveals the data from the Ipsos pre-election poll for La Vanguardia set before the start of the electoral campaign. One single certainty: the PP will most certainly be the party that obtains the greatest number of votes and seats on 23-J. From there, we enter the terrain of uncertainties. Some sharper than others. Today, they would do it with some ease. But we must not forget that some disputed seats represent a zero sum: what one block incorporates would have to be subtracted from the other, making the difference now estimated less than it may seem at first. Taking the worst ends of the forks of the two parties on the right would add up to 170 deputies, six below the absolute majority. Adding the two highest extremes would gather 186 seats. But it must be taken into account that there are seats that are disputed between them, so the two extremes could not occur simultaneously. In any case, the possibility that the PSOE and Sumar alone manage to reach 176 seats is not being considered. The best data for the Socialists together with the best data for those of Yolanda Díaz would result in 152 seats, 24 below that absolute majority.
This is the beginning of the Ipsos pre-election survey report for La Vanguardia that was published on Sunday, July 9, with data collected before the start of the electoral campaign. After having read it, could we affirm that the polls have been wrong?
This scenario envisioned an asymmetric 6-point turnout favoring the right versus the left. The hypothesis that we used, based on these data, is that the right would vote as in the elections of April 28, 2019 (when it obtained the support of 11.5 million voters) while the left would do so as in the electoral repetition of November 10 of that same year (when it obtained 10.5 million).
However, the results of 23-J have confirmed only part of the hypothesis. Indeed, the left (PSOE Sumar) has voted this past Sunday practically as in 10N (PSOE UP Compromís MP): 10.7 million votes. But the right (PP Vox UPN) has stayed at an intermediate point: 11.1 million ballots. How is it explained that participation was not finally so unequal between right and left? In the absence of the first post-electoral studies, we have an intuition: the demobilization of 400,000 center-right voters has taken place in the last week. Perhaps due to the expectation of a foreseeable victory for the right, perhaps due to the PP's post-election pacts with Vox in city councils and autonomous communities, perhaps due to the very expectation that Vox would enter the Government... In any case, in April four years ago, the PP together with Ciudadanos achieved 8.5 million votes. This past Sunday, the popular ones stayed at 8.1 million.
What happened to bring about these changes? Well, neither more nor less than an electoral campaign that has caused electoral realignments, especially during the last week. Could we have better estimated the vote of 23-J? Yes, but only really in the case of the PSOE. And we had data to do it. The polls did not fail, in any case (and only partially) the electoral kitchen did, that is, the decision on the correct model to estimate the vote.
Before these elections could be considered at least three interpretations based on the same raw data from the survey. Three models: (1) an estimation using vote memory from the last municipal elections on May 28, 2023 as a key variable; (2) an estimate using that of the last general elections on November 10, 2019; (3) an estimate adjusting the asymmetric demobilization to the estimated participation (71%), giving greater weight to the most recent election (28M). This last option was the one we used to make the estimate published by this newspaper. His mean total error for all games was less than 1 point. Its weakness: the underestimation of the PSOE (28.2% estimated compared to 31.7% obtained, that is, 3.5 points less). This was the one significant difference that kept alive the possibility of a right-wing majority. Today we know that a model more faithful to (1) would have been closer to the final result because we could have estimated the PSOE at between 31% and 32% (we insist, using the same raw data from the survey), which would have ruled out the possibility of a majority of PP and Vox. This was an analytical decision attributable only to the pollsters, not to the survey. In short, another lesson on the importance of the electoral kitchen.
