ABSTRACT
Collaborative governance processes have become a popular mechanism for addressing complex environmental problems. Their success is premised, in part, on the assumption that they promote learning among diverse participants, who are then better equipped to develop creative, consensus-oriented environmental management actions. Significant gaps remain, however, in our understanding of how collaborative governance processes foster learning and what impact increased learning has on policymaking outputs. To investigate these relationships, this study provides one of the first empirical applications of Heikkila and Gerlak's collective learning framework. Key framework concepts are operationalized via interview data and existing literature and then measured via survey data collected from participants in a collaborative environmental governance process in Colorado, U.S. Findings indicate that both internal and exogenous contextual factors affect how much an individual learns within a collective context. Additionally, participants who report more learning also more strongly agree that the process produced favorable outputs and outcomes. These findings advance theories of learning in collaborative contexts and inform process design to maximize learning.
Acknowledgments
The author would like to thank interviewees and survey respondents for their participation in this study, as well as Dr. Deserai Crow, Dr. Thomas Koontz, Dr. Tanya Heikkila, Dr. Amanda Carrico, Dr. Elizabeth Albright, and anonymous reviewers for their thoughtful comments and methodological guidance on earlier versions of this manuscript.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes on contributors
Elizabeth Koebele, Ph.D., is an Assistant Professor of Political Science at the University of Nevada, Reno. She earned her Ph.D. in Environmental Studies from the University of Colorado Boulder. Her research focuses on the role of collaboration in environmental governance processes, with an emphasis on water and natural hazards management.
ORCID
Elizabeth A. Koebele http://orcid.org/0000-0001-9133-2710
Notes
1 Following Bennett and Howlett (Citation1992) the term ‘policy learning’ is used to describe various types of learning within policy processes.
2 In Round 1, learning was coded as an outcome of the collaborative process. In Round 2, learning was coded for separately, with subcodes related to specific groups of actors who learned and types of information learned.
3 Collinearity testing confirmed the independence of the internal contextual factor variables, despite their strong correlations. All variables achieved a Tolerance greater than .1 (values = .512–.785), and a VIF less than 10 (values = 1.273–1.953), allowing for the rejection of multicollinearity.
4 Due to the clustered nature of the data (i.e. respondents belong to one of 9 Roundtables or the IBCC), it was necessary determine if respondents’ scores on the Learning Process or Learning Products variables are more similar within groups than they are across groups. An intraclass correlation (ICC) analysis produced a correlation coefficient near 0 (ρ = −.09) for both variables, with design effects of .15 and .25 respectively. Because these scores fall below the threshold of a design effect of 2 that is commonly used to determine when multi-level modeling is needed to cope with cluster effects, I proceeded with regular multiple regression techniques.
5 Model 1: F(2, 88) = 9.651, p ≤ .01; Model 2: F(4,86) = 6.732, p ≤ .01.
6 Model 3: F(2, 88) = 11.809, p ≤ .01; Model 4: F(4, 86) = 7.255, p ≤ .01; Model 5: F (5,85) = 15.300, p ≤ .01.