Studies of national identity distinguish between ethnic and civic nations and have sought to identify these alternative conceptions of national identity in public opinion. The standard measurement technique is to assess the normative content of American national identity by asking survey respondents to rate the importance of particular traits for making someone a “true” American. We argue that such measures are problematic, chiefly because of the impact of nonrandom measurement error. We explore the influence of using ranking measures instead of ratings, using a survey experiment conducted on a nationally representative sample of Americans in 2008. The ranking method is superior for distinguishing between ethnic and civic conceptions of nationhood and, therefore, for predicting preferences on issues such as immigration policy. We develop a new statistical method that effectively “converts” ratings into scores that approximate rankings, resulting in the creation of more valid measures of both ethnic and civic national identities.