Surveys provide widely cited measures of political knowledge. Do unusual aspects of survey interviews affect these measures? An experiment on a nationally representative sample of over 1200 Americans provides an answer. Respondents are randomly assigned to one of four groups. A control group...