The 2000 national main samples were selected using multistage probability based sample designs. Samples represented fourth-, eighth-, and twelfth-grade students in public and private schools in the 50 states and the District of Columbia. At grade 4, mathematics, reading, and science were assessed. At grades 8 and 12, only mathematics and science were assessed.
The 2000 sample design oversampled specific student groups to enhance the reliability of NAEP estimates for the following three groups:
Black and Hispanic students;
The oversampling of these groups was achieved at either the school or student sampling stages and, in the case of Black and Hispanic students, at both stages.
For all 2000 assessment samples other than the grade 8 public school sample:
First stage sampling units were individual counties or groups of counties called primary sampling units (PSUs).
Second stage units were elementary or secondary schools.
The third stage of sampling was the assignment of sample types and session types to sampled schools. In NAEP assessments, the "session type" identifies assessment subject, while the "sample type" identifies whether accommodations are provided to SD/LEP students and the type of inclusion criteria applied for a given assessment.
The fourth and final stage was selection of students within schools and their assignment to session types (assessment subject area).
The eighth-grade public school sample was part of a pseudo-integrated design with the eighth-grade state sample, which allowed state and national samples to be combined after data collection to produce a larger sample that would yield somewhat more precise state estimates and substantially more precise national estimates. This design provided data necessary for NCES to explore the feasibility of conducting an integrated national and state sampling operation in assessments beginning in 2002. Should this design prove successful, future assessments would no longer require a separate national sample in years during which state assessments were being conducted.
The basic grade 8 public school sample involved a stratified sample with state as the primary stratification. Except in six sparsely populated states, the national main eighth-grade public school sample used a three-stage probability sample design, modeled after the state sample:
The first stage of selection was the sampling of schools.
The second stage of sampling was the assignment of session type and sample type to schools.
The third stage was student sampling.
For efficiency, the sample design in the six sparse states—Alaska, Idaho, Montana, North Dakota, South Dakota, and Wyoming—used geographic clustering (PSUs) at the first stage, similar to the grade 4 and 12 public school samples.