Only Mexican Voters Know Polls’ Validity

After six months on a high-tech roller-coaster ride unprecedented in Mexico’s political history, the final results have poured in--a blizzard of graphs, pie charts and percentage points from more than half a dozen major polls, all designed to predict which of the three leading candidates will win Mexico’s presidential election Sunday.

And the winner is . . .

Well, it all depends on how you look at it. And whether you should look at it at all.

Such, it seems, is the nature of the wild encuesta , Mexico’s first attempt to enlist a battery of experts to conduct independent, computerized, randomly sampled polls in advance of what promises to be the freest and fairest election since the Institutional Revolutionary Party (PRI) began 65 years of unbroken and sometimes authoritarian rule in 1929.

Among the experts were some of the most prestigious pollsters in the world: Washington-based Belden & Russonello, Louis Harris-Indemerc (the Mexican affiliate of New York-based Louis Harris International) and MORI of London.

In total, they interviewed, analyzed and tried to predict the responses of thousands of Mexicans based on the techniques they have used throughout the world to produce what have long been a staple of presidential politics in the West.

Perhaps not surprisingly, the results released Thursday and Friday--the deadline for publication of opinion polls--varied little in their findings:

Ernesto Zedillo, the candidate of the ruling PRI, would win by a large margin over his two major challengers with as much as 55% or as little as 38% of the vote.

But behind that prediction was an array of qualifications, controversies and uncertainties that say as much about the political climate in Mexico today as they do about the “science” of political polling in a virgin land.

At the heart of what came to be known here as “the war of the polls” is a basic question: Will potential voters even hint at the truth about their politics to a total stranger in a nation where a single party has had a stranglehold on power for more than six decades?

The mere fact that there even were encuestas, or polls, appeared as a dramatic illustration of President Carlos Salinas de Gortari’s reform policies, which permitted the importation of the high-tech hardware and software for independent experts to do their own polling for the first time. In the past, only the government was permitted to publish such polls.

Salinas also has sanctioned the sweeping electoral reforms, among them credible computerized voter-registration lists and identity cards, that have helped make this presidential race a real contest. The reforms have created one of the tightest electoral battles in Mexican history--and a vast market for the pollsters’ predictions.

As MORI President Robert Worcester put it at a news conference here last week: “What accounts for the increased interest (in the polls) is that for the first time in 60 years you’ve got a chance for an upset. . . . I think it’s a long shot, but you’ve got an electorate that is ready for change.”

The race has been a three-way affair among Zedillo and opposition leaders Diego Fernandez de Cevallos and Cuauhtemoc Cardenas--of the right and left, respectively--and percentages of support have changed almost daily. Also almost daily, it seemed, there was a new poll.

Their final predictions of a PRI victory notwithstanding, the pollsters were the first to congratulate the government for permitting the surveys in the first place. “The contribution of polls to emerging and continuing democracies is absolutely essential,” said Nancy Belden of Belden & Russonello.

But critics--chiefly the opposition parties and political analysts who support them--have asserted that the polls have been co-opted by the ruling party.

Most of the institutions that contracted with the polling firms are, in fact, supporters of the ruling party. Among them are U.S. and European banks with investments in Mexico, which depend on the stability of the status quo. Others are newspapers and broadcast media that openly endorse the PRI.

And the pollsters themselves represent companies with vested interests in the outcome of the election; most have had contracts with the government in the past.

“These are like turf battles in the Mafia,” Federico Estevez, director of the social sciences department at the Autonomous Technical Institute of Mexico, said of the competition among the pollsters. “They’re all fighting for future contracts.”

But Estevez, who has done his own “poll of the polls” since the campaign began, insisted that he finds their results credible, concluding last week that the presidential race is “more over than ever.”

The pollsters themselves were careful to say that they aren’t completely sure of the outcome. Most stressed that shifts still could occur before Election Day.

In releasing the poll that gave Zedillo and the PRI the largest margin of victory--as much as 44%, or even 55% if the undecided vote is factored in, over Fernandez’s 19%--Louis Harris President Humphrey Taylor said: “Is it certain that PRI will win? The answer, I think, is yes, unless something very dramatic happens.”

Belden, in releasing her poll giving Zedillo 46% and Fernandez 19%, noted: “At times, big events in the closing days of an election have a profound impact. However, the lead at this point is commanding, and it would take a serious turn of events to catapult another candidate into the lead.”

And MORI’s Worcester was even more guarded on his group’s poll, which gave Zedillo 38% to Fernandez’s 22%. He provided a list of statistical aberrations that could render his measurement--indeed, all the polls--simply wrong.

At the top of the list, Worcester said, is what he called “the spiral of silence,” a statistical “skew” that his Mexican partner, Miguel Basanez, says indicated that voters were more apt to tell the truth if interviewed in the anonymity of the streets than in their homes. Simply stated, Worcester said, the “spiral” is voter anxiety about stating any preference other than a ruling party that has been known to intimidate voters in the past.

MORI’s predictions, he said, try to factor in that skew through a process he called “massaging.” A competitor called it “propaganda.”

“It’s not propaganda,” Worcester said. “It’s psephology . . . the study of elections. And I think the failure to (factor in such skews) is stupid.”

Worcester said the hardest element to factor in is the large percentage of voters who either said they had not decided or refused to answer--a group that strategists with Fernandez’s National Action Party said almost certainly will vote for Fernandez or Cardenas and his Democratic Revolutionary Party.

And polls that tried to ask “attitude” questions in an effort to determine whether respondents were lying out of fear did raise questions about their findings.

In Belden’s study, for example, 55% of 1,526 likely voters said they believed an opposition party could win more votes than the PRI. In addition, 56% said they believed people were afraid to speak freely about politics and government.

And when asked to name the principal problem in Mexico today, the overwhelming majority listed unemployment, the economy, poverty, personal security and corruption--traditional weaknesses for an incumbent party.

In an effort to square such apparent contradictions with the overwhelming preference for the PRI recorded by pollsters, Belden said the ruling party’s lead “relies on a lack of enthusiasm for the alternative candidates” and indicates that, despite their dissatisfaction, voters “are not yet willing to commit to change at the hands of another party.”

The Harris poll also reflected negative attitudes toward the government. In discussing that and the possibility that respondents lied to protect themselves, Taylor said: “Is it possible this poll is very wrong? Of course that is possible. There are many reasons why polls are wrong.”

Taylor said the poll tried to account for lying and fear.

But in addressing the question of whether opinions will change in the final week--indeed, whether the polls will affect voter sentiment--Taylor was not so certain. “That always makes me very nervous and anxious,” he said.

Worcester, in particular, cited what is known as “the boomerang effect” in suggesting that the flood of polls giving Zedillo a huge lead could work against the ruling party by making its voters complacent and by pushing marginal opposition supporters into voting.

Carlos Castillo is counting on it. Castillo is the president of Fernandez’s party, which he said has been conducting polls of its own but keeping the results confidential. “With all the polls saying the PRI is very strong, we think people will vote against the PRI,” he said.

But, Castillo added, there is another reason his party is not publishing its poll results.

“The polls show, for example, that there are a lot of people who know for which candidate they will vote but will not say publicly. If these people were going to vote PRI, surely they would say PRI.

“But at this point, it is all just guessing. I believe the only good poll is the poll on Election Day.”