This paper describes procedures used to stratify and refine an original sample of 951 families with preschool children living in the Appalachia area, for a study to provide information on the target audience for Appalachia Educational Laboratory's Home-Oriented Preschool Education Program (HOPE). The procedure was to force-match the sample distribution to the population distribution of the U.S. Census Bureau on an important background variable, the educational level of the mothers; and to use a secondary background variable, the occupational category of the heads of households, as a further guide in removing original sample cases so as to force the match. The original sample of 951 parents was reduced to 699 cases through use of these stratification variables in the matrix sampling procedures. (MS)