The objective in forming the nonresponse adjustment classes is to create as many classes as possible that are internally as homogeneous as possible, but such that the resulting nonresponse adjustment factors are not subject to large random variation. Consequently, all initial nonresponse adjustment classes deemed unstable were collapsed with suitable neighboring classes so that:
the combined class contained at least six sessions and
the resulting nonresponse adjustment factor did not exceed 1.35 (in a few cases a factor in excess of 1.35 was permitted).
When 100 percent of the schools in a jurisdiction responded, no action was taken for a school adjustment class that contained fewer than six sessions. Although there is no need adjustment for school nonresponse in these cases, this procedure could have an effect on the final definition of the student nonresponse adjustment classes.
Some nonresponse adjustment classes had too few cases to support analysis. The number of cases was increased by collapsing adjacent levels of minority status (or median household income/achievement level if minority information was missing). Metropolitan and nonmetropolitan schools were combined together only if there were less than six cooperating schools after collapsing across all levels of minority status (or median household income levels, if minority/achievement status information was missing). No collapsing was done across sample type (S2 /S3).