value of the variable SEX assigned on the school roster was used. If SEX was still missing, it was imputed from the respondent's name. On any records for which this could not be done unambiguously, this variable had a value of 1 or 2 randomly assigned.