>The Washington Post article provided by Jan Werner looking more closely
>at poll results in Iraq is useful. There is also one methodological
>issue that needs careful consideration when surveys are done in places
>like Iraq and indeed in other countries where perceptions of sponsorship
>may well affect answers. In a 1990 survey-based experiment in
>Nicaragua, whether interviewers were perceived to come from the
>government or from the opposition party had an important impact on the
>results. (A brief summary of the experiment appeared in the NYTimes,
>op-ed page, March 7, 1990; a full report was in the Amer. J. Political
>Science, 1992, 36: 331-50.) Whether a poll in Iraq is thought to be
>sponsored by Americans or not, and whether the interviewers are seen to
>be of a particular ethnicity or religion (or Sunni as against Shia), may
>be as relevant as race of interviewer has turned out to be in the United
>States.