Table of Contents  |  Search Technical Documentation  |  References

NAEP Weighting Procedures → Weighting Procedures and Variance Estimation for the 2000 National Main Assessment → Reporting Weights Main → Initial Student Weight for the 2000 National Main Assessment → School Nonresponse Adjustment for the 2000 National Main Assessment

NAEP Technical DocumentationSchool Nonresponse Adjustment for the 2000 National Main Assessment

The school nonresponse adjustment procedure inflates the weights of schools that were assigned a given subject, and participated, to compensate for those that did not participate in that subject. School nonresponse leads to the loss of sample data that must be compensated for in the weights. Similar to the school trimming procedure, the purpose of the nonresponse adjustment procedure is to reduce the mean square error of survey estimates. While the nonresponse adjustment may reduce bias from the loss of sample, it increases variability among the survey weights leading to increased variances. However, it is assumed that the reduction in bias outweighs the increase in the variance, thereby reducing the mean square error of survey estimates.

The school nonresponse procedure is conducted separately by grade and subject. The adjustment factors are computed separately within nonresponse adjustment cells. Initial nonresponse adjustment cells vary by grade and school type. In public schools, grades 4 and 12, nonresponse cells are defined by the first three digits of the primary sampling unit (PSU) stratum. The PSU stratum reflects NAEP region, metropolitan statistical area (MSA) status, sample type, and various socioeconomic characteristics such as percent minority and percent of persons 25 years or older with a college degree. In grade 8 public schools, the cells are formed by NAEP region, type of location, and sample type. In private schools for all three grades, the cells are defined by religious affiliation, sample type, and National Association of Independent Schools (NAIS) membership status.

At times, initial cells are collapsed to improve the stability of the adjustment factors, especially for the smaller assessment subjects. Most cells that need collapsing contain small numbers of cooperating schools. Occasionally, cells with low response rates are collapsed.

In each nonresponse cell h, the nonresponse adjustment factor SCH_NRAF_h is calculated as follows:

where

• SCHBWT_i represents the school base weight for school i in nonresponse cell h for subject s,

• SCH_TRIM_i represents the school trimming factor for school i in nonresponse cell h for subject s,

• B_h represents the set of all eligible schools assigned subject s in nonresponse cell h, and

• C_h represents the set of all eligible schools assigned subject s in nonresponse cell h that ultimately participated in subject s.

Nonresponse adjustments assume that nonresponse occurs at random within the categories within which adjustments are made. Some degree of bias could result to the extent that this assumption is false.

In past assessments, school nonresponse adjustment accounted for the difference in estimated grade enrollment between the substitute and the corresponding original school. In 2000, a more traditional nonresponse adjustment scheme was used which did not include this substitution adjustment. The substitution adjustment factor was part of the within-school student sampling weight component in the student base weight calculation.

Printer-friendly Version