NAEP Technical DocumentationDevelopment of Final School Nonresponse Cells for the 2004 Assessment

Limits were placed on cell sizes and adjustment factors to prevent unstable nonresponse adjustments and too large nonresponse factors. All initial weighting cells with less than 6 cooperating schools or adjustment factors greater than 3.0 were collapsed with suitable adjacent cells. If necessary, initial weighting cells were generally collapsed in reverse order of the cell structure.

Nonresponse adjustments were performed separately for public and private schools within the assessment type (operational, mathematics bridge, and reading bridge), age category (9, 13, and 17), and grade classification for the given age (upper and lower grade). This split created 36 subsamples total.

For each of the 18 public schools subsamples, cells with the most similar race/ethnicity strata within a given location stratum were collapsed first. If further collapsing was required after all levels of race/ethnicity strata were collapsed, cells with the most similar location stratum were collapsed within the NAEP region. Cells were never collapsed across NAEP region. Also, schools within certainty and non-certainty PSUs were kept separate.

For each of the 18 private schools subsamples, cells with the most similar race/ethnicity strata within a given private school type stratum were collapsed first. If further collapsing was required after all levels of race/ethnicity strata were collapsed, nonresponse cells were collapsed across different private school types. Cells were never collapsed across NAEP region. Also, schools within certainty and non-certainty PSUs were kept separate.

Last updated 17 April 2009 (JL)

Printer-friendly Version