- Surveys & Programs
- Data & Tools
- Fast Facts
- News & Events
- Publications & Products
- About Us
Table of Contents | Search Technical Documentation | References
NAEP Technical DocumentationStratification for Private Schools
The five explicit strata for the private school samples in fourth, eighth, and twelfth grades were Catholic schools (RC), Lutheran schools (LU), Conservative Christian schools (CC), other private schools with known affiliation status (OP), and private schools with incomplete affiliation status (DK). The implicit stratification within these strata was based on Census division, urbanization status (type of location), and percent minorities (Black, Hispanic, and Native American). Extensive collapsing of the implicit strata was carried out to assure that the expected number of schools within each implicit stratum was reasonably large.
The starting point was Census division, which was combined with another division within the same Census region if the expected number of schools to be sampled within the division was less than six (within the explicit stratum). The relationship between Census division and Census region is shown in the following table.
| CENSDIV | Census division | CENSREGN | Census region |
|---|---|---|---|
| 1 | New England | 1 | Northeast |
| 2 | Middle Atlantic | 1 | Northeast |
| 3 | East North Central | 2 | Central |
| 4 | West North Central | 2 | Central |
| 5 | South Atlantic | 3 | South |
| 6 | East South Central | 3 | South |
| 7 | West South Central | 3 | South |
| 8 | Mountain | 4 | West |
| 9 | Pacific | 4 | West |
| SOURCE: U.S. Department of Education, Institute of Education Sciences, National Center for Education Statistics, National Assessment of Educational Progress (NAEP), 2002. | |||
For example, in the final strata defined for private school sampling, for grade 4 Lutheran schools, the Northeast and South Census regions were collapsed into a single stratum (1). The two Central Census divisions were not collapsed (3 and 4). The two Western Census divisions were collapsed into their Census region (8).
The second stratification variable, nested within the first, is type of location. A level of type of location is collapsed with one or more adjacent levels if the expected number of schools to be sampled within that level is less than six. The following combinations of type of location are considered permissible: 1, 2, 3, 4, 5, 6, 7, 8; 1-2; 3-4; 5-6; 7-8; 1-2-3-4; 5-6-7-8, and 3-4-5-6-7-8. (1 and 2 are combined first as they are both central city; 3 and 4 are combined first as they are both urban fringe; 5 and 6 are combined first as they are both town designations; 7 and 8 are combined as they are both rural. 5-6 and 7-8 are put together as they are both nonurban. 3-4 is combined with 1-2 [urban fringe with central city], or with 5-6-7-8 [urban fringe with nonurban]).
The third stratification variable, nested within the geographic and location type strata, is based on percentage of Black students, Hispanic students, and Native American students enrolled in the grade1. As above, cutpoints on this stratifier were established so that each cell had an expected number of schools sampled greater than or equal to six.
The method used to sort the data prior to selection was the serpentine sort, which executes multiple sorts within a stratum such that bordering sample units are the most similar with respect to the sort variables.
1These three minorities tend to have lower achievement levels than White students and Asian/Pacific Islander students, hence stratification using this dichotomization should improve the precision of achievement level estimates.
Last updated 08 July 2008 (PE)
