Testing geographical framing and substitution effects in spatial choice experiments Marije Schaafsma a,b,n , Roy Brouwer b,1 a Department of Environmental Economics, Institute for Environmental Studies, VU University Amsterdam, Amsterdam, Netherlands b Centre for Social and Economic Research on the Global Environment (CSERGE), University of East Anglia, Norwich, UK article info Article history: Received 8 April 2013 Accepted 22 April 2013 Keywords: Choice experiment Environmental valuation Water quality Choice set size abstract One of the main challenges in modelling spatial choices is the complexity resulting from the availability of multiple alternatives at different geographical scales. This study aims to test geographical framing and substitution effects in stated choice experiments by first increasing and subsequently reducing the geographical scale and associated set of choice alternatives in the experiment. Geographical framing effects are tested by comparing estimated choice models for differently sized choice sets. Testing these framing effects related to choice set size helps to inform decisions on choice set composition. The results indicate that changing the choice set size has little to no effect on preference parameters and estimated WTP values. However, the larger choice set is associated with higher error variance, suggesting higher choice task complexity. Crown Copyright & 2013 Published by Elsevier Ltd. All rights reserved. 1. Introduction Under the assumption of perfect information, decision-makers in neo-classical consumer choice theory are expected to evaluate all substitutes simultaneously when making choices. Stated preference (SP) studies for environmental valuation usually focus on single study sites. These studies have been criticised for drawing respondent attention to the single study site only, away from available substitutes and thereby inflating willingness-to-pay (WTP) values. Providing sufficient information about the good under valuation as well as its substitutes is a challenge given, for example, the time limitations of social surveys. Simple textual reminders of substitutes, suggested by the well-known NOAA panel (Arrow et al., 1993) to increase the reliability of SP studies, have been argued to be ineffective (Loomis et al., 1994; Whitehead and Blomquist, 1995; Kotchen and Reiling, 1999). The number of single-site SP studies applying contingent valuation (CV) and making an effort to account for substitution is limited. Brown and Duffield (1995) include the number of alternatives and Pate and Loomis (1997) account for the size (acreage) of possible substitutes. A small number of CV studies include multi-programme scenarios in which different goods are valued simultaneously to test for substitution and complementarity effects (Hoehn and Loomis, 1993; Cummings et al., 1994; Hailu et al., 2000; Payne et al., 2000). Cummings et al. (1994) and Neill (1995) argue that in order to elicit WTP estimates that account for substitution effects, respondents should be asked to value substitutes and study sites simultaneously. Alternatively, surveys should provide at least a good description of available substitutes, using pictures, maps or text. Contents lists available at SciVerse ScienceDirect journal homepage: www.elsevier.com/locate/jocm The Journal of Choice Modelling 1755-5345/$ - see front matter Crown Copyright & 2013 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jocm.2013.04.007 n Corresponding author at: Centre for Social and Economic Research on the Global Environment (CSERGE), University of East Anglia, Norwich, UK. Tel.: +44 1603 593747; fax: +44 1603 591327. E-mail address: m.schaafsma@uea.ac.uk (M. Schaafsma). 1 Tel.: +31 20 5985608; fax: +31 20 5989553. The Journal of Choice Modelling 8 (2013) 32–48