Saturday, November 25, 2006

When Is a Poll Not a Poll? by Maritza Ramirez Agena

When Is a Poll Not a Poll? Sometimes Numbers Do Lie
Maritza Ramirez de Agena (November 24, 2006)November 25, 2006 Ipsos, an international polling company, conducted a poll related to the upcoming Venezuelan presidential election for the Associate Press that was released (officially) on November 24th. In the AP-Ipsos poll, when voters were asked: “For whom would you vote?” The results show: Chavez 58%; Manuel Rosales 25%; a 33 point lead.Ipsos is a company with over 30 years of experience, based in Paris, as Mr. Jaime Seijas, one of Ipsos’ directors in Venezuela explained to me today during a phone conversation. Mr. Seijas told me that Ipsos opened its office in Caracas in 1999. I called Ipsos-Venezuela to ask a few questions about the methodology used in the study. I was also curious about whether or not Ipsos-Venezuela was affiliated with other market research firms in the country. I was told that Juan Vicente Zerpa, the Country Manager, was in Mexico attending an annual meeting. Mr. Jaime Seijas was kind enough to take my call.Mr. Seijas said the study was conducted in 21 of the 23 Venezuelan states. He confirmed that 75 municipalities were selected and stratified by region, city size, and household income. The sample included individuals from the following groups: A,B,C 20%, D 40% and E 40%. All this information so far, does not raise suspicion as to the introduction of bias into the results.The problem is that the Venezuelan arm of Ipsos may have inadvertently ignored a factor that could render their survey, invalid; the “fear factor. “Ipsos conducted the study door to door! Yes, in a “random manner” (every fifth household in a clockwise fashion from a designated starting point). Nonetheless, please note that the Venezuelans who opened their homes to strangers to answer questions related to Chavez and their voting intension were thinking only these things:• How do I know this person is a pollster?• How do I know he/she is not associated with the government?• Why are they coming to my home?• Who will have access to this information?• What are they going to do with this information?• I cannot answer freely; they have my home address and know who I am and where my family lives!• If I say I have no intension to vote for Chavez, I may be fired!• If I say I have no intension to vote for Chavez, I may get hurt by hard-core Chavistas; my family can get hurt too!It does not matter that Ipsos has 30+ years experience conducting survey-based studies. They have no way to successfully conduct opinion polls in countries where the government publicly threatens its citizens with massive layoffs and physical violence if they do not support the incumbent regimes.It really does not matter that Ipsos interviewed people in 21 out of 23 states in Venezuela or how representative the sample was. It does not matter how accurately they stratified the sample in the 75 municipalities selected to conduct the study. None of this matters, because the individuals they interviewed are not free to answer questions about their government. Venezuelans are not free to vote for whom they want to vote, they are not even sure their vote will be private. Venezuelans are not certain that the government will not suffer reprisals against them if they vote for Manuel Rosales. I would like to quote Minister Rafael Ramirez during a speech given to employees of the Venezuelan Oil Company a few weeks ago:“Those who seem to have forgotten that we are in the middle of a revolution will be beaten until they get it!”I consider it imperative for the international media, not to ignore the fragile condition of democratic freedoms in Venezuela. I would like to take this opportunity to ask for discretion and professionalism; to dig under the covers to find the real story. Regarding the poll published by AP-Ipsos, I think their statement: “There are other sources of potential error in polls, including the wording and order of questions and the political and social context in which interviews are conducted” was greatly underestimated. It is clear the “fear factor” was not filtered by the methodology implemented. The results cannot be considered representative, even less so, accurate.
For more information, please contact:
Maritza Ramirez de Agena

Maru Angarita
My blog is: