Combining Sources of Data to Estimate Preferences for an Environmental Resource Edward R. Morey and William S. Breffle Department of Economics Campus Box 256 University of Colorado Boulder CO 80309-0256 contact: edward.morey@colorado.edu November 7, 2003 Abstract The objective of this research is to develop a random-utility model of preferences that is estimated with three types of data: stated preference choice data, stated preference frequency data, and revealed preference use data. A probit model is specified, and the application is angler preferences for Green Bay. The model takes advantage of the relative strengths of the different types of data; the combination of data not only allows for the estimation of how individuals trade off different characteristics of the site, but also how the angler’s proportion of trips to Green Bay would change if its characteristics changed. This methodology for multiple data sets is especially useful when the commodity of interest is unique, as is Green Bay. The estimated parameters are used to estimate the compensating variation per fishing day (and per Green Bay fishing day) for the elimination of Green Bay fish consumption advisories. * This work has benefitted greatly from comments and suggestions from Vic Adamowicz, David Allen, Bob Baumgartner, Rich Bishop, Don Dillman, David Layton, Pam Rathbun, Bob Rowe, Paul Ruud, V. Kerry Smith, Roger Tourangeau, Don Waldman, and Michael Welsh.