In Person (Face-to-Face) Sampling
National samples for face-to-face surveys are typically drawn by an area probability method, which relies on US Census figures and maps of the country. For example, the 2012 ANES used a three-stage process to draw its national sample. This process is called a multistage area probability sample, and has a long history of use in face-to-face surveys. The three stages were:
Selecting census tracts from each of the nine US Census Divisions in the 48 contiguous states. (For the face-to-face survey, cost considerations precluded interviewing respondents in Alaska and Hawaii. Respondents in those states were, however, included in the Internet sample.)
Randomly selecting residential addresses within the census tracts identified in the first stage. The US Postal Service's Delivery Sequence File was used to identify residential addresses.
Selecting an eligible person to be interviewed in the households identified in the second stage. After an interviewer identified all household members, the interviewer's computer would randomly select one eligible person to be interviewed.
The 2012 ANES included an over-sample of African-Americans and Latinos. This was done in order to be able to make valid inferences to both of these groups. In a typical survey, the low number of African-Americans and Latinos (usually just over 10 percent of the sample each) generally results in too few respondents to represent the larger group. Over-sampling "beefs up" the validity of the observations by including more people. In order for the entire sample to be representative of the American public, the African-American and Latino sub-samples were weighted to bring them back to their true population parameters.