This recent survey conducted in Iran claims that most Iranians believe that their authoritarian government is actually largely democratic, support the government's efforts to develop nuclear weapons, and are hostile to the United States (hat tip: Jeffrey Friedman).It is perfectly possible that these poll results really do reflect the majority opinion of the Iranian public. However, the researchers failed to consider an important factor that might bias their results: Iran is a repressive dictatorship where the government often punishes those who express dissident views that question the official line.
In such an environment, Iranians who disagree with the government's positions might hesitate to express those disagreements to foreign pollsters. They might instead toe the official line in order to protect themselves from the secret police. Obviously, good pollsters will promise respondents anonymity. But Iranians - who have lived in a police state for decades - might well distrust the sincerity of such promises. For all they know, the "pollster" who has contacted them is actually a government agent trying to ferret out dissent. Even if they believe the pollster to be sincere, there is still no guarantee that the conversation between pollster and respondent won't be overheard by the secret police.
Fear of punishment gives inhabitants of repressive societies strong incentives to lie to pollsters if they disagree with the government line. Public opinion scholars have long been aware of this problem. For example, this classic study of Nicaraguan public opinion under the Sandinista communist dictatorship showed that many Nicaraguans were only willing to express their true opinions about the government if they believed the pollsters to be affiliated with anticommunist opposition parties. In his brilliant 1995 book Private Truths, Public Lies, economist Timur Kuran documents numerous similar examples of people in repressive societies hiding their true views in order to avoid punishment by the state.
None of this proves that citizens of repressive societies never genuinely support their rulers. Adolf Hitler was genuinely popular in 1930s Germany, for example. It does, however, show that researchers should not take pro-government poll responses in such countries at face value. Scholars have understood the problem for a long time. There is no longer any legitimate excuse for polling organizations to ignore it.
I don't blame Iranians who may have lied to Western pollsters out of fear. I do blame pollsters who fail to consider the possibility that such lying may affect their results.