Assessing Instrument Mixes through Program‐ and Agency‐Level Data: Methodological Issues in Contemporary Implementation Research

Abstract

Abstract Theories of policy instrument choice have gone through several “generations” as theorists have moved from the analysis of individual instruments to comparative studies of instrument selection and the development of theories of instrument choice within implementation “mixes” or “governance strategies.” Current “next generation” theory on policy instruments centers on the question of the optimality of instrument choices. However, empirically assessing the nature of instrument mixes is quite a complex affair, involving considerable methodological difficulties and conceptual ambiguities related to the definition and measurement of policy sector and instruments and their interrelationships. Using materials generated by Canadian governments, this article examines the practical utility and drawbacks of three techniques used in the literature to inventory instruments and identify instrument ecologies and mixes: the conventional “policy domain” approach suggested by Burstein (1991 ); the “program” approach developed by Rose (1988a ); and the “legislative” approach used by Hosseus and Pal (1997 ). This article suggests that all three approaches must be used in order to develop even a modest inventory of policy instruments, but that additional problems exist with availability and accessibility of data, both in general and in terms of reconciling materials developed using these different approaches, which makes the analysis of instrument mixes a time‐consuming and expensive affair.