In the original article, the analysis included output legitimacy as well as the variables we present in the article. The editor believed that this variable created concerns for common method bias, and thus we present these results only in Appendix A for interested readers. The authors did a number of tests that some have proposed to “test” for common method bias and concluded, at least based on these tests, that this was not the case.

For example, Holland, Spain, and Taiwan can be regarded as representative of Northern Europe, Southern Europe, and East Asia public administration traditions identified by the cited authors (see Meyer and Hammerschmidt 2010 Meyer, R. and G. Hammerschmidt. 2010. “The Degree of Decentralization and Individual Decision-Making in Central Government Human Resource Management: A European Comparative Perspective.” Public Administration 88(2): 455–478.[Crossref], [Web of Science ®] , [Google Scholar]; Berman, Moon, and Choi 2011 Berman, E., J. Moon, and H. Choi eds. 2011. Public Administration in East Asia: China, South Korea, Japan and Taiwan. New York: Taylor & Francis. [Google Scholar]). Some typical differences concern top-down leadership, stakeholder bargaining/consensus-building, techno-legalistic versus strategic-administrative decision making, and more. Hofstede notes further differences as well (individualism, uncertainty avoidance, long-term orientation, etc.) discussed also in the conclusion. However, all such differences are relative and highly aggregate that vary in specific settings and program areas.

In the case of the Netherlands and Spain, it was a self-administered web survey. In the case of Taiwan, the survey was administered through interviews conducted by the authors. As is the Taiwan practice, interviewers visited the interviewees in their offices, where interviewees completed survey questions.

In other research (see Meier and O Toole 2007 Meier, K. J. and L. J. O'Toole. 2007. “Modelling Public Management: Empirical Analysis of the Management-Performance Nexus.” Public Administration Review 9(4): 503–27.[Taylor & Francis Online], [Web of Science ®] , [Google Scholar]), more or less objective data are used, not collected by surveys. In this case, the test scores of students in national tests. Interesting though this is, it is not a measurement of networks but of individual organizations (that is, schools). And even this is under debate, since they are biased. They do not measure the quality of a school but the quality of pupils, which is something very different (schools that have less privileged students, for instance, may do very well and add much to students’ knowledge, but still score less than other schools, which received more privileged students from well-educated parents). So measurement of the network level in objective terms is hardly done and hardly possible since there are so many values and actors at stake with different judgements.

We used this way of constructing the variable because there are some indications that managers and others might overestimate the use of strategies. In this way, we made certain that the score was not inflated. We did check whether this led to different results compared to simply adding the scores, but no significant differences were found.