A week and a day after his appointment with the polls, the polls indicate that the Colombians willing to vote will mostly do so for Gustavo Petro.
However, it does not seem that they will be enough to give him a victory in the first round.
The quality-weighted average of the latest polls leaves him around 42%, the level he obtained in the second round of 2018. The eight points that remain until half plus one needed to close the victory on the next 29 seem like a difficult gap to cover, especially given its slow growth in recent weeks.
Thus, the question that everyone is asking is who will be second, earning the position to dispute the presidency in the ballot on June 19.
Until a week ago, that position seemed assured for
Fico
Gutiérrez, drawing a scenario of left-right polarization that today is much less clear due to the sudden rise of Rodolfo Hernández.
The former mayor of Bucaramanga has launched his candidacy based on a speech against Fico and Petro, but not against polarization, but establishing a third anti-establishment pole in which he pretends to be alone.
This is how he has snatched the space of centrality from the one who traditionally held it, Sergio Fajardo, who has sunk in all the polls with no sign of improvement.
The opportunity to do so was given by the traffic jam of Gutiérrez's campaign.
After a dizzying ascent once it became known in mid-March that he would be the right-wing choice to take on Petro, unlike his predecessor (outgoing president Iván Duque) he has been unable to convert the opportunity offered by being a lesser-known person in effective supports, even ending a certain decline in his trend in recent weeks.
Two factors could have played against him: his leaning to the right, which without being the most extreme on the Colombian scene (where there are politicians with declared admiration for Trump, Bolsonaro or the Spaniards from Vox), is undeniable;
and its association with old elites connected to the poorly valued outgoing presidency at a time when the demand for change seems to be the majority in the country.
Thus, the former mayor of Medellin has been uncompetitive in the two key axes (left-right and new-old).
Even so, the central scenario right now is not that Rodolfo passes, but that Fico does.
That happens in six out of seven worlds according to the application of an open source probabilistic model on the average made by EL PAÍS.
This model serves to better calibrate the uncertainty around the scenarios drawn by the surveys.
It does this by repeatedly simulating supposed choices with different outcomes but anchoring their probability to the average.
From this approximate exercise it turns out that Petro is first in 90% of the simulations, and Fico beats Rodolfo to get into the second round in three out of four.
But 22% of the time the one that happens is the one from Bucaramanga.
A not insignificant probability.
All of this is due to the basic fact captured by the polls that Gutierrez holds an advantage over Hernandez.
But the size of the advantage varies greatly from one poll to another: the technical tie produced by CELAG (with a very slight advantage in favor of Rodolfo) or CNC (in favor of Fico) up to the more than ten points anticipated by Massive Caller draw two scenarios very different.
This variation is especially important as the spotlights that illuminate the electoral preferences of Colombians will go out this weekend.
The electoral law prohibits the publication of any type of survey in the week prior to the vote.
This always makes it difficult for undecided or hesitant voters to make decisions, which in a changing dispute such as the one apparently between Fico and Rodolfo can be crucial.
The vote is always a combination of two opposing forces: the desire to choose the person or platform closest to one's own convictions, and the aspiration that the vote not be lost but rather serve someone who, thanks to it, can reach the can.
The polls are the mirror that guides the coordination of efforts that candidates in dispute need to convince those who hesitate between expression and usefulness.
It already happened in 2018, when Sergio Fajardo was on the verge of going to the second round under the darkness of the poll ban.
It seems that this year history could repeat itself.
Methodology and sources
The average number of polls is carried out on the total number of likely voters, both valid and undecided, including the last ones published by each polling house registered with the National Electoral Council up to the date of publication of the article, from March 14, 2022 (the day after the inter-party consultations that ended up defining the candidates for the presidency).
The compilation is collected from the platform-recipes-electorales.com
,
and is contrasted with the publication in various media.
Each survey has a slightly different weight in the average that depends on the numerical evaluation carried out by the
semáforo de encuestadoras de lasillavacia.com
, which qualifies each of the polling houses according to a series of technical, methodological and accuracy parameters. in the end result.
To avoid excessively skewing the average, the assigned weight only varies by 20%, so that the highest rated pollster's survey receives 100% and the lowest rated receives 80% weight over the final average.
When a pollster is registered with the CNE but is not valued by
lasillavacia.com
, the average assigns it an intermediate point between both extremes.
This average, unlike the previous ones, eliminates the undecided from the calculation base.
This is due to two reasons: first, there is now at least one survey (The 3 Musketeers) that does not specify the number of undecided, which makes it impossible to reproduce an average with undecided;
Second, the number of undecided voters has been reduced as we get closer to the elections, to the point that normalization with their elimination from the base now is less risky methodologically than at the beginning of the campaign.
In fact, the two pollsters with the best rating prioritize the dissemination of their estimates without including the undecided.
The table shows the estimates corresponding to each survey carried out and included in the average once the undecided are eliminated from the calculation base.
The probabilistic model applied to the average described above to obtain an approximation of scenarios is based on the one
carried out by recipes-electorales.com
, but fed with its own normalized average.
As is the case with the data used as food for the average from the aforementioned platforms, the code used for this analysis, as well as its raw results today, which serve as the basis for the graphs produced, is available to anyone in
this open repository
.
Subscribe here
to the EL PAÍS newsletter on Colombia and receive all the key information on the country's current affairs.