Maybe, just maybe, those poll respondents don't actually think what they say they think, Langer suggests. Maybe they say all this for some other reason -- such as that they just don't like the president.
"I'd suggest that it's dicey at best to evaluate measurements on questions such as the president's birthplace as a 'belief,' as opposed to, for at least some respondents, merely an expression of antipathy," Langer wrote in an email. "We have called this 'expressed belief,' in contrast with 'affirmed belief.' The latter is an assertion of perceived factual reality; the former, message-sending."
As an example, Langer points to belief in global warming. While most Americans believe global warming exists, the fraction of global warming believers has shrunk over the past few years, according to polls, even as the scientific consensus has solidified in global warming's favor.I think Langer is correct, but I would go even further. Surely, some people---we don't know how many---answer polls about beliefs about factual matter symbolically: a person may tell a survey researcher that she believes President Obama is a Muslim or Kenyan or a Communist just to register her (intense) disapproval or antipathy toward the president.
I would also add my suspicion that some respondents do this sort of thing consciously and strategically. If a respondent who opposes President Obama's policy goals is aware that survey evidence of widespread doubt about the President's birthplace will undermine his political position, he benefits at the margin by anonymously falsely expressing doubt about the Obama's place of birth. If a respondent knows that evidence of widespread belief in anthropomorphic climate change will support the adoption of cap-and-trade legislation of which he disapproves, he may express doubts about the human origins of climate changes that he does not really hold. These responses are more than symbolic misrepresentations of reality; they are intentional and instrumental misrepresentations. In other words, some people may be lying to pollsters to manipulate an important input into the national policymaking process.