Table of Contents  |  Search Technical Documentation  |  References

NAEP Sample Design → NAEP 2002 Sample Design → 2002 State Assessment Sample Design → School Sample Selection for the 2002 State Assessment → Computation of Measures of Size → Establishment of Measures of Size for Each Jurisdiction

NAEP Technical DocumentationEstablishment of Measures of Size for Each Jurisdiction

The measures of size for most of the State NAEP 2002 jurisdictions were set according to the algorithm established for school sample selection. There were a number of jurisdictions for which the target student value of 6,300 could not be achieved using the algorithm, because of small student populations¹. These became take-all-school states, so that the expected number of hits exceeds 1 for every school in these jurisdictions (for both grades). All schools in the jurisdiction are in the sample.

In many of the take-all-schools jurisdictions, the targets could not be met easily. Some of these became take-all-students jurisdictions: every eligible student was assessed. For these jurisdictions, one can view the measure of size for each school as equal to the expected number of hits in the school (school enrollment divided by 60), and the upper bound u_j for the burden as infinity.

In two jurisdictions that were take-all-schools, but did not need to be take-all-students, we set the upper bound for burden u_j as higher (4 and 5 respectively) in order to achieve higher targets (closer to 6,300).

Note that ultimately the targets needed to be reduced in many of the smaller states, even after increasing maximum burdens. This reduction was because the burden of a full sample was deemed too great, under the National Assessment Governing Board policy. The option to reduce the sample is described below.

There were a number of jurisdictions that were offered the sparse state option. In these cases, the student populations tended to be spread over a large number of small schools. Any jurisdiction whose expected school sample size (after calculation of b_j) exceeded 120 was offered the sparse state option, which changed targets and parameters to reduce the expected school sample size below 120. This reduced the burden for the schools in these states.

¹The names of the jurisdictions are omitted to protect confidentiality.

Printer-friendly Version