Table of Contents  |  Search Technical Documentation  |  References

NAEP Sample Design → Sample Design for the 2001 Assessment → National Main Assessment Sample Design in 2001 → Student Sample Selection for the 2001 National Main Assessment → Oversampling Black and Hispanic Students for the 2001 National Main Assessment

NAEP Technical DocumentationOversampling Black and Hispanic Students for the 2001 National Main Assessment

In public schools with high minority (Black and Hispanic) enrollments, schools were assigned a measure of size twice the size of other low minority schools, therefore increasing their probability of selection, and indirectly increasing the number of Black and Hispanic students in the sample.

In public schools with low minority enrollment, an oversample of Black and Hispanic students was selected. The procedure was as follows. After the initial sample was selected, the non-selected Black and Hispanic students were identified and listed. All such extra Black and Hispanic students were sampled to a total that, in expectation, was the same number of Black and Hispanic students as were already selected. In practice, if the number of non-selected students was less than the number of selected students, then all non-selected Black and Hispanic students were to be assessed also. Otherwise, Black and Hispanic students were sampled so that their overall within-school probability of selection was twice the rate of other students.

The sampling of additional Black and Hispanic students was carried out using designated line numbers, as indicated on the session assignment form used to generate the samples of students in each school. In this way, the necessary information as to the selection probability of each student was retained for use in weighting. No reliance was placed on information generated in the field. Field supervisors had only to follow the prespecified sampling instructions.

Since the aim was to oversample by a factor of two where possible, but never more than two, the overall rate of oversampling of Black and Hispanic students was instead less than two. That is because in smaller low minority schools there were no students remaining who had not already been assigned to a session. The weighting procedures ensured that the results were not biased as a result of the relative under-representation of Black and Hispanic students from smaller low minority public schools.

Printer-friendly Version