The Use of Matched-Pairs Research Design in Industrial Surveys

Abstract

Surveys of manufacturing plants frequently require a representative sample of plants in a particular industry or geographical area. In this paper the usefulness of an alternative research design based on matched pairs of plants is evaluated. The design is illustrated with the use of a case study of the subcontract linkages of manufacturing plants. A new statistic is introduced which treats, in a more rigorous manner than conventional tests, cases where no differences were detected between pairs.