NAEP Technical DocumentationStratification Variables for the 2009 State Assessment

			Stratification by Urbanization Classification Stratification by Race/ethnicity Classification Stratification by Achievement Data and Median Income Missing Stratification Variables

The implicit stratification of public schools for the NAEP 2009 state assessment involved three dimensions within a jurisdiction. Public schools were stratified hierarchically by the following variables:

• urbanization classification, combined NCES urban-centric locale with whether or not the school is in a Trial Urban District Assessment (TUDA) district;
• race/ethnicity classification; and
• achievement score on state tests or median income of the ZIP Code area where the school is located.

The urbanization classification classifies schools by as many as 12 types of NCES urban-centric locale (e.g., city, suburb, town, rural), which includes classification as to the size of the city or town and the distance a rural area is from an urbanized area. This variable is nested within an indicator of whether the school is located in a TUDA district.

Student race/ethnicity status variables describe schools by the relative magnitude of Black, Hispanic, Asian and Pacific Islander, and American Indian and Alaska Native enrollments in particular jurisdictions and represented in schools. Race/ethnicity classification is based on the Common Core of Data (CCD) race/ethnicity variables. The race/ethnicity cells are nested within the urbanization classification (within each jurisdiction). 

The last stratification dimension is achievement score or median household income. Achievement data were used if they were available for a particular jurisdiction and grade. If achievement data were not available, median household income from the 2000 Census of the ZIP Code area where the school was located was used.

Missing values for stratification variables were imputed.

The implicit stratification in this three-fold hierarchical procedure was achieved via a "serpentine sort." This sort was accomplished by alternating between ascending and descending sort order on each variable successively through the sort hierarchy. Within this sorted list the schools were arranged in serpentine order by achievement data or median household income within each cell determined by the two higher classifications (TUDA/urbanicity classification, race/ethnicity classification), with ascending order for achievement data/median household income used in every other cell, and descending order for achievement data/median household income used in the remaining cells, giving an ascending-descending-ascending-descending pattern. These TUDA/urbanicity classification and race/ethnicity classification cells were also sorted in serpentine order. Within each TUDA/urbanicity classification, race/ethnicity classification cells were sorted by ascending order for one TUDA/urbanicity classification, followed by descending order for the next TUDA/urbanicity classification, and so on.The following table shows an oversimplified example to illustrate the ascending-descending-ascending-descending pattern of the serpentine sort.

The third dimension of stratification differed for schools in the National Indian Education Study (NIES) oversample. These schools were implicitly stratified by the percentage American Indian/Alaskan Native students within the school instead of achievement scores or median household income. 

Printer-friendly Version