Skip to main content

Argentina’s Primaries: Did The Polls Get it Right This Time?

By | [email protected] | August 15, 2017 1:28pm


Polls aren’t doing so great in the world these days. Brexit; Trump’s victory in the 2016 US Presidential elections; the Colombian peace referendum and, on a local scale, the 2015 presidential elections. Most polls predicted one result and then we learned it was actually the opposite.

In the last couple of years, we have grown used to pollsters getting their predictions terribly wrong, increasing the shocking effect of unexpected results that we never thought could possibly happen. And Argentina’s primary elections from Sunday were not the exception.

In the weeks previous to August 13, several media outlets such as Cronista, Clarín and Página 12 published — sometimes selectively — the predictions that pollsters released daily, all of them pretty much suggesting a similar result.

However, much like the greater part of the media landscape, many pollsters failed to escape the strong political division in Argentina, and there results were increasingly reflected a certain bias.

True, some did predict the results accurately (one, in fact), but most conducted polls at the request of the political parties with which they have close ties with and pretty much ended up telling them just what they wanted to hear. This both altered the political leaders’ perception of where they stood and caused people and the media to misread the real outcome of Sunday’s primary elections.

Clarín analyzed how each pollster performed before the elections and how accurate their predictions were in the Buenos Aires province. Very few did well enough. Most of the others, not so much.



Opinaia is the only online pollster (the others conduct phone call-based polls) out there. Its last analysis was released the Wednesday previous to the election and it pretty much said that Cambiemos would win by less than half a point (it predicted Cambiemos would win by 0.4 percent, instead of the final 0.08); missed the candidates’ percentages by very little, and well within the margin of error: Cambiemos’ Esteban Bullrich by 0.29 percent, former President Cristina Fernández de Kirchner by 0.61, Cumplir’s Florencio Randazzo by 0.7 and FIT’s Néstor Pitrola by 0.7. The only exception was Sergio Massa’s 1País, for which it predicted he would get an 18.5 percent of the vote, almost three percent more than what he ended up getting.



This one nailed the order in which the candidates would perform but failed to predict the “virtual tie” between the two first candidates: it predicted that Bullrich would beat Fernández by two points. True, it’s within the usual margins of error of two or three percentage points, but in this case they make a fairly important difference.


This one accurately predicted Bullrich’s surge in the days prior to the election, and the votes that Massa and Randazzo would get. However, it also said that the former President would come out on top on Sunday by one point (so it was within the margin of error.)



It accurately predicted the candidates’ order. Their percentages, not so much. The polling company said that Massa would get 11 percent of the vote, four points less than what he actually got. It did say that Bullrich would win the race by a not-so-large margin, but gave him 3.5 more points than he obtained and one and 1.5 more to the former President. It also missed the mark with Randazzo and the Worker Party’s Nestor Pitrola’s numbers.


This one mistakenly predicted that Fernández would win the election, but by less than two points — within the margin of error — and were pretty close to the percentages that the candidates ended up getting. It argued that the former President would get 34.8 percent, 0.7 more than what she got on Sunday, Bullrich would get 33.2, a point less that he actually did, and Massa 17.7, two more than what he managed to get.


The larger portion of pollsters fall into this category, as most predicted Fernández would win by at least three or four points. However, for the sake of being concise, we’ll select two.


This pollster is run by Gustavo Marangoni, a former government official of the Daniel Scioli administration that ran the Buenos Aires Province between 2007 and 2015. His last poll predicted that, as a result of a constant growth in the intention of vote, the former President would defeat Cambiemos’ candidate by seven points — 38.6 to 31.8. It did almost get it right with Massa, but they really missed the mark with the others.


Although it adjusted its numbers by the end, it assured during a great part of the campaign that Massa had actual chances of disputing Cambiemos’ second place, meaning that not only it misread the electorate but also grossly miscalculated the chances of 1País’ leader: it ended up giving him nine more points than what he finally got — 24.7 percent against the final 15.5 —  and gave Cambiemos four points less. The only silver lining: it missed Fernández’s percentage by 0.7 percent.