The NAAL sample size is not large enough to provide literacy estimates (direct estimates) of adequate precision for most states and counties. The NAAL direct literacy estimates (i.e. standard survey estimates) can satisfy the demand for the participating states, but not for other states, including larger states. However, there remains a need for adult literacy estimates, particularly at the lowest literacy level, for all states and smaller jurisdictions within the states.
This need—to provide small-area projections—has led to the development of model-dependent techniques for small-area projection. The small-area projections are predictions of how the state would have done if NAAL data collection had conducted a survey within its border. These estimates are not entirely based on data collected in NAAL. They are predictions based on auxiliary data available for the small-area together with any NAAL data collected within the small-area. The resulting projections are often known as “indirect” projections to distinguish them from standard or “direct” estimates.
Unlike the direct estimates that are derived directly from survey data, small-area projections borrow information from other, related data, such as Census data, and predict the estimate of interest using a statistical model based on the NAAL assessment data. The auxiliary variables (1) will be drawn primarily from the 2000 Census data for the specific small-area, and (2) will include variables that are selected based on their relationship to the percentage of adults at the lowest literacy levels. These variables include education, English-speaking capability, immigration, racial and ethnic minority status, age, employment status, type of employment (managerial, professional, technical, service, labor), urban/rural status, and poverty status.
Accessing Small-Area Projections
NCES has developed a user-friendly webpage to allow easy access to the small-area projections. The webpage includes the percentage of adults at the lowest level of literacy (i.e., adults who were placed at Below Basic in prose literacy as well as those who did not take the assessment because of the language barrier). The website also includes the estimates and confidence intervals, with explanations provided of their meaning and their limitations—to indicate the level of accuracy a user can expect with these estimates.