Mexican Election opinion polls: Tools for propaganda?
Enrique Peña Nieto is the frontrunner in Mexico’s elections. But polls are divided over the size of his lead.
By MANUEL RUEDA
The most contentious topic of the Mexican presidential campaign three weeks from Election Day isn’t necessarily one of the candidates.
Since May, Mexico’s public opinion polls have produced wildly different results from one another, triggering widespread scrutiny, leading people here to question if polls are being used for propaganda purposes, and if they can actually change the outcome of the July 1st elections.
At the end of May for example, Reforma newspaper published a poll that put leftist candidate Andrés Manuel López Obrador (AMLO) just 4 points behind the once-dominant front-runner Enrique Peña Nieto.
Obrador’s supporters celebrated the findings, claiming them as a sign that the momentum is turning in their way. Some political analysts also interpreted Peña Nieto’s modest performance in the poll as a sign that the recent wave of student protests against the PRI candidate poisoned public opinion of the candidate.
But that same week, Milenio newspaper published a poll that gave Peña Nieto a comfortable 21-point lead over AMLO.
With less than twenty days left until the Mexican elections, we analyzed the most recent polls and asked experts in Mexico and the U.S. how such polling results could vary so widely. Here are some things to take in consideration about polls in Mexico and elsewhere.
1. Polls must be analyzed over time
At first sight, it might seem shocking to see how polls could contain such dissimilar results. Take a look at this table of national polls conducted between May 24 and May 27 compiled by the website ADN Politico.
We have four polling companies in this table, Mitofsky, Parametria, GEA-ISA, and Grupo Reforma, which has its own newspaper and polling department.
As you can see in the far right column, Mitofsky gave Peña Nieto a lead of 17.5 percent at the end of May, while GEA/ISA put the PRI candidate 21 points ahead of Lopez Obrador.
The difference between both candidates was 18 points in the Parametria survey and just four points in the Reforma poll.
However, if you compare these polls to previous polls carried out by the same companies a pattern begins to emerge. Here are some of the most recent polls conducted by these companies.
As you can see, Peña Nieto’s lead over Obrador became 7 points smaller in just 12 days in the Mitofsky poll. Peña Nieto’s lead in the GEA/ISA survey went down from 21.4 to 16 percent in about two weeks.
Reforma only conducts polls once a month, so we do not have numbers that are more recent than its end of May poll, which puts AMLO just 4 points behind the PRI candidate. However a poll conducted by that company at the end of April had AMLO 11 points behind Peña Nieto.
Their numbers are different, but all these polls are telling a similar story: Lopez Obrador has been catching up to Peña Nieto recently.
2. Poll results can be influenced by a variety of subtle factors
We may have a pattern emerging across the different polls. But that still does not explain why results are so varied. To find an explanation for this trend, we spoke to Ricardo de la Peña, polling director for GEA-ISA.
De la Peña’s survey has consistently given Peña Nieto the widest lead amongst all major polls. But this researcher is no loose cannon. In the 2006 elections, which President Felipe Calderón won narrowly with 35.9 percent of the vote, de la Peña’s estimates were within two percentage points of the final result.
Opinion polls in Mexico share some similarities. Any poll that is taken seriously, de la Peña said, interviews a sample group of at least 1,000 registered voters and conducts these interviews in person.
In order to make sure that these 1,000 people represent all of Mexico’s regions and different socio-economic groups, researchers typically visit 100 to 150 electoral precincts per survey, interviewing 8 to 10 people in each precinct.
Precincts are selected at random with the help of a computer program. This ensures that precincts with characteristics that are more common in Mexican society — for example, precints where the majority of residents only have a high school degree — are more likely to get picked than precincts with less common characteristics, such as those where a majority of residents are college graduates.
But after precinct selection, polling methods begin to differ, and there are several subtle factors that can influence how some people answer questions on their electoral preferences, such as the order in which questions are asked, the type of clothing the interviewer is wearing, and whether the interviewer identifies himself as a member of a company that the interviewee already associates with a certain political tendency.
“Interviewees tend to look for a minimum level of empathy with the interviewer,” de la Peña said. “So if they assume that the interviewer represents a political tendency, their answers can sometimes lean in that direction.”
De la Peña says that his company, GEA-ISA, takes several steps aimed at controlling bias. Interviewers are not allowed to wear colors that allude to Mexico’s political parties. They are not given uniforms, because that might make some interviewees feel less comfortable around them, and therefore less willing to share their true opinions.
GEA-ISA interviewers also identify themselves by mentioning the formal and little known name of the polling company they represent but not by mentioning the newspaper the poll is published in, which in this case is Milenio, a paper that some leftist activists accuse of favoring the PRI.
We contacted Alejandro Moreno, the polling director at Grupo Reforma, to see if his methods differed greatly from those described by de la Peña. After all his poll put Lopez Obrador just 4 points behind Pena Nieto in late May, while the GEA ISA poll put the leftist 21 points behind the frontrunner.
In an e-mail, Moreno said that his company, which also owns the Reforma newspaper, had a policy of not allowing employees to conduct interviews with other media outlets.
So we contacted political analysts and polling experts in the U.S., who presumably have an opinion on Mexican pollsters that is not affected by any political or economic interests. They described Moreno as a “well respected,” researcher.
“The Reforma poll is a reputable poll,” said Christohper Wilson, a Mexico expert at the Woodrow Wilson Center for International Affairs. “Within any poll there is a margin of error and my guess is that this poll’s margin fell in López Obrador’s favor.”
3. Polls don’t work out as well when voters don’t have marked preferences
Wilson explained that pollsters in Mexico have a tough time this year, because they are asking people about candidates for which their preferences are not very strong.
Mexican candidates posses similar policies on key issues, such as the best method to tackle drug violence, they are all shy on abortion, saying it should not be criminalized but adding they won’t push for legalization.
Mexico’s presidential hopefuls are even moderately similar on their economic views, with the three main candidates advocating for greater investment in social welfare programs and an economy that generates jobs by engaging in trade with other countries.
There are differences of course, in candidates’ background and in the reputations of the political parties they represent.
But differences in Mexico are not as great as they are in Venezuela, for example, where President Hugo Chávez, a socialist, faces an opponent who wants to steer the country back towards a free-market economy.
The lack of political polarization in Mexico is not an ideal scenario for pollsters according to Wilson.
“Polls are easy and most reliable in cases where people have very strong opinions.” Wilson said. “In that case it doesn’t matter how you ask the question, it doesn’t matter how you set it up.“
Mexicans’ lack of a strong preference in this election cycle can be found in the polling data. Most polls taken over the past month, suggest that 17 to 22 percent of Mexicans still haven’t decided who to vote for on July 1, just 19 days away.
4. Polls have a limited impact
Much has been said in Mexico about the possible use of polls for propaganda purposes. And candidates here regularly mention polls that favor them in order to try and convince voters to support their cause.
During an interview on TV giant Televisa last week, AMLO said that “his pollsters” had done a survey that put him 2 points ahead of Peña Nieto.
Meanwhile, the PRI recently aired a TV ad that shows graphics of the polls in which Peña Nieto is leading alongside footage of the candidate speaking at campaign rallies and shaking supporters’ hands in the background.
But can polls really influence how people vote?
Ricardo de la Peña from GEA-ISA argues that their influence on voter behavior is limited, as polls are just one of many other factors that help people to decide who to vote for.
“Polls allow the average voter to see what other voters are thinking,” de la Peña said, adding that other factors that influence the elections include political ads, candidate’s performances, media coverage, the opinions of friends and relatives and the opinions of journalists, politicians, columnists, and other members of the country’s elite.
One interesting effect of polls according to de la Peña is that they can influence the behavior of campaign donors, and other political operatives, who will sometimes lean towards supporting the candidate leading in the polls, regardless of his ideology, in order to advance their political and economic interests, should that candidate win the election.
De la Peña said that voters are most likely to think that polls that coincide with the opinions that they hear around them are the most credible ones.
But he explained that, because voters have a limited social circle that is often made up of people with similar political views, they may be surprised when they see a poll that contradicts the preferences of their friends and relatives.
James Bell, the international polling director at the Pew Research Center, told Univision News, that polls can be influential to the extent that they are portrayed as big news events, and generate a perception that a candidate in a race is the inevitable winner.
Bell mentioned for example, that the passage of Arizona’s SB 1070 immigration law significantly changed what Mexicans said about the U.S. in a 2010 Pew Center survey on Mexicans’ attitudes towards their northern neighbor.
Favorable opinions of the U.S. dropped from 69 percent to 56 percent in Mexico since 2009, but there are sharp differences between those interviewed before and after the Arizona measure was signed into law by Gov. Jan Brewer on April 23, 2010, according to the Pew Center report.
Sixty-two percent of those interviewed between April 14-20 had a positive view of the U.S., while that dropped to 44 percent among those interviewed between May 1 and 6.
Polls may not be as earth shattering news as an immigration law that allows police to racially profile immigrants.
But they still exert some influence on elections. And when poll results are close they can even encourage voter participation.
“It is important whether people or not believe that there is a clear leader or whether or not there is a clear second place candidate,” said Wilson. “If people just believe that Peña Nieto has the election locked up that would not be good for turnout.”
(Photo: Manuel Rueda)