WITH only 103 days until Mexicans vote in what could be the closest presidential election in decades, a controversy has erupted over the validity of political polls.
And in Mexico, this isn't just salsa for political pundits.
The country has a 65-year history of single-party rule and elections marred by fraud allegations. Political analysts say the polls are setting up public expectations. If the electoral results do not jibe with the polls, opposition parties are likely to organize protest marches that could lead to violent unrest - as has happened after local elections in the recent past.
The polls are critical, says political commentator Sergio Sarmiento. ``We need some proof because we Mexicans don't give credibility to electoral results.''
But the credibility problem now extends to the polls.
No one questions that the ruling Institutional Revolutionary Party (PRI) candidate is the odds-on favorite - it is the size of the PRI lead that is hotly disputed.
In mid-April, the pro-government newspaper published a poll showing the PRI candidate with 63 percent of support, versus 16 for the center-left Democratic Revolutionary Party (PRD), and 12 percent for the National Action Party (PAN).
The same week, the magazine Este Pais, published a poll by Market Research and Opinion International (MORI) that gave the PRI only 46 percent, the PRD 32 percent, and the PAN 22 percent. A May 11 MORI poll shows the PRI with 33 percent, and the PAN and PRD in a dead heat with 15 percent. The ``undecideds'' lead the tally with 36 percent. But snap polls taken since the May 12 debate between presidential candidates show the PAN has surged into second place.
Why the big difference between polls done by MORI and El Nacional?
Miguel Basanez, the director of MORI, cites the ``Nicaragua syndrome.'' Before the 1990 presidential elections in Nicaragua, all but one pollster predicted a sweeping Sandinista victory. The Sandinistas lost by a wide margin.
``In a traditional Catholic society, lacking in democracy, the people are afraid to express their true opinion,'' Mr. Basanez says.
Unlike the other pollsters, Basanez does his polling in the street where people apparently express their opinions more freely -
in anonymity - than surveys done door-to-door. Some Mexican political analysts cite a poll done by the daily newspaper Reforma in March that supports Basanez's methodology. Reforma interviewed two groups. One group was surveyed in their homes, and 35 percent expressed support for the PRI candidate. In another group, interviewed on the street, only 25 percent preferred the PRI.
Mr. Sarmiento says Mexico's politics and culture make reliable polling difficult. ``Polls are like asking for directions. In Mexico, it's not the same as asking for directions in Madison, Wisconsin. A Mexican will tell you how to get someplace even if he doesn't know, out of embarrasment.''
President Carlos Salinas de Gortari's pollster, Ulises Beltran, questions the MORI methodology and the assumption that Mexico is a special case.
``Polling in the street is problematic,'' Mr. Beltran says. ``It's impossible to determine the margin of error because you have no control over the population sample. Every sample is completely different. And, the people in the street are not a true cross-section of Mexican society.''
He also dismisses the Nicaragua case as a fluke. ``Nicaragua is taken as a paradigm for polls in an authoritarian climate. But Nicaragua is one case among hundreds. El Salvador has similar characteristics, and the polls were very successful in predicting the outcome of the latest elections there.''
Beltran notes that polling and market research in Mexico is a $100-million-a-year industry with ample experience in producing accurate surveys on issues people are reluctant to talk about, such as illegal drug use and abortion.
The problem, Beltran says, is that the message is tarnished by the messenger. ``The lack of media credibility in Mexico hurts the credibility of the poll results.''
Of course, Mexicans have cause to question media bias. El Nacional ran a large, front-page headline in April proclaiming the PRI as the choice of 63 percent of those polled. But on page three, the PRI lead dropped to 45 percent. Why? The percentages on page one were calculated without including the 28 percent of the voters who polled ``undecided.''
PRD campaign adviser Ricardo Pascoe does not have an argument with MORI poll results but claims many pollsters ``lend themselves to government manipulation for monetary reasons.'' MORI director Basanez alleges government officials have pressured him to change his results.
But presidential pollster Beltran and other pollsters interviewed call such accusations unfounded and unwarranted. ``If Miguel can explain to me how he got those results [referring to a recent poll],'' Beltran said at a recent breakfast with foreign correspondents, ``I will retire from the business and dedicate myself to writing history books.''
In a recognition of the importance of reliable polls, President Salinas has waded into the controversy. His administration will finance a national, independent poll to be conducted by Belden Russonello, a Washington-based firm. A number of Mexican pollsters, including Beltran and Basanez, will be on a committee to oversee the methodology.