Pew Interview: Mitofsky on the Mexican Exit Polls

Exit Polls Legacy blog posts

This week’s belated focus on the U.S. exit poll controversy was well timed, as exit polls once again made news, this time in the Mexican elections.  An exit poll conducted by Warren Mitofsky for Mexican television network Televisa showed the election too close to call, and the vote count reported Felipe Calderón ahead of his main opponent, Andrés Manuel López Obrador, by razor thin margin (35.88% to 35.31%).   So how could the exit polls in Mexico have been right in Mexico, but off in the 2004 U.S. elections?  The Pew Research Center brings us the explanation from Warrent Mitofsky in an interview conducted by pollster Andrew Kohut.

The interview is definitely worth reading in full, but here is the most pertinent excerpt:

What is the biggest difference between conducting an exit poll in Mexico compared to doing one in the United States? And why do you think this poll was so much more accurate than the exit polls for the 2004 presidential election?

There are a few important differences: One is that the response rate is a lot better in Mexico. Also, we train the interviewers in person rather than by telephone. Third is that when we do the research we cannot get maps, which causes a problem. But we get good records, good cooperation.

And we did one thing that hopefully I learned from the 2004 election: We insisted in the personal training of the interviewers that they absolutely stick to the sampling rate. And if there is a refusal they continue the count to the next sample person; they don’t substitute anyone in between.

So if I understand correctly, if I, the interviewer, am to talk to every fifth person coming out the door and that person refuses, I don’t go to the next person. Instead I go to the next fifth person who comes out?

Right, there was no bias in this exit poll at all. In a typical Mexican exit poll we overstate the PRI, which was the government party that ran the country for the past 70 years, but which finished third this time around. We usually overstate PRI and understate the other two parties. This time the other two were understated by a fraction of a percent and offset each other so there was essentially no bias in the exit poll. I did the same thing in another exit poll elsewhere and there was no bias there either.

Despite Mitofsky’s report of "absolutely no leaks in the results," Greg Palast has argued, oddly, that exit poll showed "the voters voted for the progressive candidate."    See this blog post for a contrary opinion.  Mexico’s PRI party also seems to disagree

Mark Blumenthal

Mark Blumenthal is the principal at MysteryPollster, LLC. With decades of experience in polling using traditional and innovative online methods, he is uniquely positioned to advise survey researchers, progressive organizations and candidates and the public at-large on how to adapt to polling’s ongoing reinvention. He was previously head of election polling at SurveyMonkey, senior polling editor for The Huffington Post, co-founder of Pollster.com and a long-time campaign consultant who conducted and analyzed political polls and focus groups for Democratic party candidates.