See Spanish Original.
|Electoral preferences according to surveys (January 2012) La Jornada|
- After learning the official figures their inaccuracies became clear
The bad publicity against Enrique Peña Nieto would be enough to send any candidate to the bottom of the polls, but not in Mexico, wrote Randal C. Archibald, correspondent for The New York Times, on June 12th. The writer was referring to the extramarital affairs of the aspiring presidential PRI candidate, the indications of corruption various PRI leaders were involved in, the PAN members that warned of the negotiations between the public and the organized crime were Peña Nieto to become President, and the marches and protests against him led by students.
It is an understatement to say these factors create a negative public opinion of the candidate from Mexico, that began before his campaign registration: on December 3rd of last year, when the pre-candidate suffered from a mistake at the International Book Fair in Guadalajara.
At the beginning of December of last year, the surveys were giving the candidate from Mexico 50 per cent of the votes, or more, although by then Humberto Moreira had already been dismissed as the national president of the PRI for his alleged responsibility in the state Coahuila that he governed for its indebtedness scandal. However, on January 10th of this year, BGC-Ulises Beltrán and Associates continued giving him half of the intended votes (compared to 26 and 24 per cent, Josefina Vázquez Mota and Andrés Manuel López Obrador respectively). While, at the end of this month Consulta Mitofsky reduced his standing by three points, to situate him in 50 per cent of voter preferences (28 for JVM and 22 for AMLO). Roy Campos acknowledged that the fall of the PRI could be a consequence of its own errors in December (Graph 1: surveys in January).
The fact that the surveys didn't reflect Peña's reputation in the phase prior to his presidential campaign, or if so, only minimally, had an immediate negative effect on the credibility of the pollsters themselves. Víctor Hernández broke down the Mitofsky data using figures supplied by Roy Campos [head of Mitofsky] himself on Joaquín López Dóriga's radio program: if the numbers he mentioned there on how many voters would choose the same party as in 2006 were true, then the tendencies 50 (EPN)-28 (JVM)-22 (AMLO) were clearly false. Based on the Mitofsky numbers, the percentages were the following: EPN: 30.7 per cent; AMLO, 28.42, and JVM, 23.82, which means in reality there is a technical draw between AMLO and EPN, since the margin of error on the survey is more or less (±) 3.1 per cent, observed Víctor M. Toledo. In addition, the pollster ascribed 100 per cent of the intentions to vote assuming they would act on their preferences (this is, the universe where people agree to answer for who they would vote for) and set aside the percentage of the surveyed who did not want to respond. If the latter were distributed in proportion to the trends of each candidate, a technical tie would occur again between AMLO and EPN.
Hundreds of cyberactivists elaborated graphics from Excel and embedded the videos to show such inconsistent methodology.
Then, Roy Campos, as if cured before ever being sick, (on December 24) said that a polling business that distorts results will not remain long in the market; that to spread a false survey is absurd; if it reports what is not present in the population, it will not appear only because the survey says so; that making a survey to generate false data throws away money, and that it is easier to say lies without using surveys.
In any other country, a foreign correspondent would say, the creation in the middle of May of the #YoSoy132 movement, which emerged precisely in repudiation of Peña Nieto, from a student sector that spread like wildfire, would have caused a collapse of his candidacy. This didn't happen in Mexico or in the polls. Except for GEA-ISA, which recorded a significant fall in the first half of this year (and an inexplicable recuperation from then on), BGC-Ulises Beltrán, Buendía, Parametría and Mitofsky reflected a minimum fall and then – it seems from the graph – the aspiring PRI candidate maintained control over voter’s intentions (Graph 2).
The controversy about the statistic practices and the criticism grew as the elections approached, but, except for isolated cases, the surveyors kept silent. In a special case the editorial director and communist of Milenio, Ciro Gómez Leyva, wrote several articles to vehemently defend the daily poll that was carried out in his publishinig house in alliance with Mitofsky Consulting.
On the eve of the election, apart from a modest and discreet decline (given the circumstances) of the aspiring PRI presidential candidate, the only relevant event for the surveys was the advance, unanimously recognized, of Andrés Manuel López Obrador, from third to second position (Graph 3).
Since the final hours of July 1st, according to the results of the Preliminary Electoral Results Program (PREP), it became evident that the official numbers did not have anything to do with, except for the distribution of positions between Peña Nieto, López Obrador and Vázquez Mota, with reflecting trends that were predicted by the surveys. Taking as an average a margin of error of ± 3 percent, according to the methodology used by the polling houses, almost no one had reached with certainty the range, not only for the votes each candidate obtained, but, above all, for the margins of difference between each other.
With respect to the vote given to the PRI, IPSOS-BIMSA, Reforma, Covarrubias, De las Heras and Berumen were within the range of allowable error (± 3) (Graph 4).
With regard to the voting given to AMLO, only BGC- Ulises Beltrán y Buendía Laredo came out with an acceptable range of error (Graph 5).
However, when one analizes what occurred with the margin of difference between the candidate from Mexico and that from Tabasaco, the surveys reveal their almost total ineffectiveness. On this point, the studies are not related to the official numbers of votes or to the efficiency or scientific expertise that is assumed by their authors. In this exercise, only De las Heras and Berumen carried out an accurate forecast. The rest had margins of error between 3.8 (Reforma) and 12.2 (GEA-ISA) (Graph 6).
Gómez Leyva did not mention the unique and striking uniformity of the surveys or the need to make an assessment of the daily polls carried out for its newspaper and for GEA-ISA. He spoke only of their failure: “Editorially – started his article on July 3rd – there is no worthwhile justification.”
The GEA-ISA owner, Jesús Reyes Heroles González, said that if his work were to be far from reality, it probably was due to people who reported that they would vote for the PRI candidate but did not go to the polls.