Explaining positive deviance in public sector reforms in development

Matt Andrews


Public sector reforms are commonplace in developing countries. Much of the literature about these reforms reflects on their failures. This paper asks about the successes and investigates which of two competing theories best explain why some reforms exhibit such positive deviance.

These theories are called ‘solution and leader driven change’ (SLDC) and ‘problem driven iterative adaptation’ (PDIA). They are used to analyze data emerging from a case survey involving thirty cases from Princeton University’s Innovations for Successful Society (ISS) program. The bulk of evidence from this study supports a PDIA explanation, but there is reason to believe that SLDC hypotheses also have value. It seems that PDIA and SLDC are two viable paths through which positive deviance can emerge; although PDIA seems to provide the wider path for more positive deviance.

The 30 case studies chosen for this research are examples of reforms that have yielded functional improvement in some of the toughest contexts in development, where reform failure is far more normal. As such, they are considered abnormal successes—or positive deviants—and are the kinds of reform experiences where one would hope to learn a great deal.

The sample used emerged from a purposive sampling process focused on selecting cases of public sector reform in developing nations that: (i) are considered abnormally successful (because they fostered functional improvements in the way governments work), (ii) have enough information to allow analysis, and (iii) cover a variety of public sector reform types. Thirty studies were chosen because they come from countries once considered fragile states and have generated sustained improvement in institutional performance and economic growth.

Key findings:

  • The evidence emerging from the case survey seems more supportive of problem driven iterative adaptation (PDIA) as a theory explaining positive deviance. In support of major PDIA hypotheses, the majority of reforms in the sample seem to have been motivated by a problem, implemented through flexible processes involving experimentation, and led by groups.
  • Also supporting PDIA is the fact that over 90 percent of these reforms ultimately produced hybrid products that did not resemble pure-form best practices. This strong evidence is particularly impressive given potential bias in the ISS sample towards an SLDC narrative (as discussed).
  • However, there is also evidence that solution and leader driven change (SLDC) has some value in the discussion about positive deviance. There were some reforms that seem to have been motivated by clearly defined solutions, implemented according to plan, led by a single leader, and ultimately proved successful in introducing a functional ‘best practice’ product. This evidence suggests that SLDC and PDIA offer two different paths for producing positive deviance in public sector reform in developing countries. The PDIA path appears much wider than the SLDC path, seemingly explaining more of the cases than SLDC. Future work should explore why this is so.
  • It is also interesting that some cases of positive deviance seem to be explained through a softened version of PDIA (PDIA-lite); where positive deviants are motivated by both problems and a proposed (partially identified) solution but where flexible implementation processes still ensure solutions are shaped to the context. Future research needs to examine whether this kind of PDIA-lite intervention is more common than an extreme version.
  • The evidence from this study needs to be qualified, however, and applied to practice with care. The study is limited because of a lack of counterfactual cases, for instance. It is unclear whether elements of PDIA or SLDC would be more or less present in less successful reform cases. Future research could build on the work in this paper by expanding the sample to include both positive deviants and more-normal reform failures. Adopting a larger and more varied case sample in future research could also help in managing potential bias one must expect when drawing cases from one source (as in this study).


Andrews, M. (2013). Explaining positive deviance in public sector reforms in development, CID Working Paper No. 267, October 2013. Center for International Development, Harvard University.