Margin of Error
In den letzten Tagen geisterte eine Meinungsumfrage durch die Medien, in der sich Zweidrittel der Befragten für Ahmadinejad als Präsidenten ausgesprochen haben. Hierzu nun Joshua Muravchik:
–––Consider: You’re a citizen of Iran, rated by Freedom House as one of the two dozen least free countries in the world. Secret police are everywhere, as are storm troopers called “white shirts.” Many of your follow citizens are in prison, exiled or dead for dissent.
Your phone rings, and the caller, whom you don’t know, explains that he or she wants to ask your opinion on a few dozen controversial political issues and also how you intend to vote.
The call comes from abroad. Why? Because the pollsters are afraid to ask their questions from inside Iran. Nonetheless, they assume that you won’t be afraid to answer frankly. What do you do?
Only a reader who went to Ballen and Dougherty’s Web site and ploughed through their original 75-page report would know the answer.
It turns out that their callers reached 1,731 people. Fully 730 refused to speak to them. Of the remaining 1,001, half (501) either refused to say how they’d vote or said they didn’t know or had no preference.
That left just 500 who answered the presidential poll — 338 who said they favored Ahmadinejad, 162 who expressed support for another candidate. Voila, the two-to-one margin that Ballen and Dougherty trumpeted: 19.5 percent for Ahmadinejad to only 9.5 percent for the others.
To reach their conclusions, Ballen and Dougherty assumed that the opinions of the 71 percent who either hung up or refused to divulge a presidential preference also must have been two-to-one for Ahmadinejad.