Spain's CIS poll and D'Hondt arithmetic: why 169 seats may not be enough

Spain's CIS poll and D'Hondt arithmetic: why 169 seats may not be enough

A projection based on the microdata from the CIS barometers (Spain's official sociological research centre) puts the bloc its author calls "reactionary" at 169 seats in his central scenario. The figure has been circulating on Twitter and has sparked a storm of rebuttals: for some, it confirms that Spanish society has not shifted to the right; for others, it is an exercise in statistical sleight of hand. The number, however, hides more than it shows.

The first objection is elementary arithmetic. According to a calculation making the rounds, if the seats of right-wing parties that the author places in the "others" group are added to those 169 MPs, Congress would still be right-wing. The projection would not be measuring a bloc, but arranging the board to suit itself. The second objection is methodological: CIS microdata are not an electoral poll, they are an opinion barometer, and treating them as a survey has its limits.

What the D'Hondt law says that the headline doesn't

The key lies not in adding up votes, but in how they are distributed across provinces. In a five-seat constituency, with the PSOE at 38% and another left-wing party at 12%, a split of PP at 35% and Vox at 15% gives the right three seats. The same 50% of the vote, split 25% and 25%, yields only two. And the seat that is lost goes to the PSOE without it winning a single extra vote. The law rewards the most-voted party and punishes vote splitting.

Hence the strategy some attribute to the PSOE: fragmenting the vote to its right, with a strong Vox eating into votes and a PP leaning further right to fish in the centre-left. It is nothing new in electoral terms, but it explains why seat projections swing so much with small variations in vote share.

The author's track record: from the vaccine to the podcast

The discussion is not only about the method. It is also about who signs it. The author of the projection has a track record of predictions that failed to materialise, such as the forecast that no elbichito-19 vaccine would be achieved, and an economics podcast that some describe as "delusional". Some listen to it in order to invest against his forecasts, while others recall his time as a moderator on an economics forum, where he is accused of having censored critical messages.

That track record weighs on the conversation as much as the numbers. An analyst with that résumé can be right, but his signature shapes how the work is received. The projection is not judged on its own: it is judged alongside its author.

The scenario nobody rules out: the "multi-coup"

Among the critics there is a strand that flatly denies the usefulness of any calculation. They argue that the microdata are meaningless in a context they describe as a regime prepared to perpetrate an unpunished "multi-coup". It is the electoral version of "it doesn't matter what the polls say": if the result is going to be forced, arithmetic is decorative.

Against that, the defence of the microdata is clear: the surveys aren't worth a damn, but the CIS microdata are. The distinction makes sense: a barometer measures opinion, not voting intention, and its value lies in the trend, not in a snapshot of a single day. Even so, projecting seats from it requires assuming a translation from opinion to vote that no microdata can guarantee.

The result, as of this discussion, is a figure that serves to confirm the thesis of whoever reads it, not to settle it. With these materials, the only honest prediction is a hedged one: if the right goes into the election divided, the D'Hondt law will make it pay the toll in the small provinces. How much remains to be seen.