To select eligible participants, the main NAAL used four stages of selection:
The NAAL sample design did not call for selecting representative samples within every state. To do so would increase the cost enormously. Instead, an optimal sample design was employed that first required the selection of counties from across the nation without a constraint for having a sample in every state. There are 3,140 counties in the U.S.
From the 3,140 counties, about 2,000 geographic areas were formed by grouping contiguous counties together with the objective of reaching a minimum size needed under the NAAL sample design.
After the 2000 geographic areas were formed, they were grouped into 100 strata (groups), based on census division, race/ethnicity distribution, per capita income, and MSA status. The purpose was to create homogeneous groups. From each stratum, one geographic area was selected, resulting in 100 counties (or groups of counties) in the main NAAL sample.
Because the NAAL sample is based on counties, and every county has a chance of being included in the sample, some states may not have any NAAL sample. However, the sample is still statistically representative of the nation.
Within each of the 100 strata, the larger the geographic area (comprising of single or multiple counties), the higher the probability of inclusion in the NAAL sample. The smallest geographic area has a population of about 16,000, with the median being about 35,000.
Within each selected PSU, area segments (census blocks or groups of blocks) were selected with a probability proportional to the number of housing units they contained. However, area segments that were classified as high-minority (more than 25 percent Black or Hispanic) were oversampled. Oversampling of minorities was necessary to ensure that the minority samples would be large enough to conduct meaningful analyses.
Field staff visited all selected area segments and prepared a list of housing units located within those segments using block maps from the 2000 Census. From this list, households were systematically selected within each segment. Minority households in high-minority segments were given a higher chance of being selected.
For each selected household, field staff determined the number of age-eligible household members (16 years and older). Field staff were trained to record the names, relationship, sex, age, and race/ethnicity of all household members. When needed, screening questions were given orally in Spanish by those field staff who were bilingual.
In households with fewer than four eligible members, one person was randomly selected. In households with four or more eligible members, two persons were randomly selected. Also, in some households, no one was eligible.