NAEP Technical DocumentationAdjustments for Nonresponse at the Student Level

			Student Nonresponse Adjustment Factors

An additional factor that affects the base weight for a selected student is absent students who are invited to the assessment but do not attend the scheduled sessions (either original or makeup).

For the 2000 state assessment, the initial student nonresponse classes for assessed students were formed using several variables. These variables were based on information from the sample design, age of the student, final collapsed school nonresponse cells, and the actual monitor status (or assigned monitor status, if the actual monitor status was not available) at the session level. The first of these variables was an indicator of whether or not a student was excluded from the assessment. A variable was then created from combining SD/LEP status and the sample type for the student.

Within these categories, the initial student nonresponse adjustment classifications were further defined depending on the SD/LEP status of a student. For all schools except Department of Defense Domestic Dependents international schools, if a student was SD or LEP, then the class was formed by urbanicity cross-classified by student age. Age was used to classify students into two groups (for grade 4, those born in September 1989 or earlier and those born in October 1989 or later; for grade 8, those born in September 1985 or earlier and those born in October 1985 or later). If a student was neither SD nor LEP, then the initial nonresponse adjustment class was formed by urbanicity cross-classified by student age (as defined above), by the quality control monitoring status, then finally by minority status as collapsed for the school nonresponse. For the Department of Defense Domestic Dependents international schools, the nonresponse adjustment cells for SD and LEP students was student age cross-classified by the minority status variable as defined for the school nonresponse adjustment classes.

Following creation of these student nonresponse adjustment classes, all unstable classes were identified for possible collapsing with other classes. A class was considered to be unstable when either of the following conditions was true for the given class:

number of responding eligible students was fewer than 20 or
nonresponse adjustment factor exceeded 1.5.

All classes deemed unstable in the previous step were collapsed with other classes using the following rules:

Do not collapse across SD/LEP and non SD/non LEP.
If within cells defined by the cross-classification SD-LEP/non SD/LEP status and sample type within the SD/LEP categories, all of the adjustments are one, no adjustments are made.
For non SD/LEP classes, collapse across minority status within monitor status. If that does not work, collapse across monitor status within minority status.

More collapsing was necessary only if the resulting classes had fewer than 15 responding eligible students. Collapsing then continued within the successive variables until the class size was no longer deficient or until a "set" boundary that could not be crossed was reached. In the case of SD or LEP students, more collapsing was necessary to eliminate the rare situation in which all students in a class were nonrespondents.

Last updated 13 August 2008 (KL)

Printer-friendly Version