Dealing with Heterogeneous Preferences Using Multilevel Mixed Models Begon ˜a A. Farizo, John Joyce, and Mario Solin ˜o ABSTRACT. One of the main issues on the research agenda regarding stated preference methods concerns the heterogeneity of preferences either within or be- tween individuals. We present a multilevel mixed model (MMM) to capture heterogeneity in determin- istic utility components, instead of simply leaving them to random components. MMM captures hetero- geneity at different levels: individuals, locations, and groups of individuals sharing other characteristics. The results show that individuals’ surroundings help to capture heterogeneity, and that can be controlled by specifying these aspects as predictors for this be- havioral model. Therefore, MMM may contribute to the identification of the underlying structure affecting environmental decisions. (JEL D62, Q51) I. INTRODUCTION In recent years, most environmental valu- ation studies have managed the heterogeneity of preferences among individuals. In the past, most of this heterogeneity was incorporated through socioeconomic variables, but few studies show the relevance of the context of where the individual performs the valuation (Seong-Hoon, Poudyal, and Roberts 2007; Brereton, Clinch, and Ferreira 2007; Solin ˜o, Farizo, and Campos 2009; A ´ lvarez-Farizo, Gil, and Howard 2009; Solin ˜o, Joyce, and Farizo 2013). It is logical to think that an in- dividual living in a suburb with very low en- vironmental quality may appreciate improve- ments in her or his surroundings on a different scale than another individual living in high- quality environmental conditions. Therefore, individual decisions derive from multiple log- ics and reflect a variety of past and present experiences at different levels. Environmental valuation uses techniques, for example, choice modeling, that are well Land Economics • February 2014 • 90 (1): 181–198 ISSN 0023-7639; E-ISSN 1543-8325  2014 by the Board of Regents of the University of Wisconsin System developed for marketing, where the consumer is the focus of the research. In contrast, en- vironmental goods have no market in which consumer sovereignty can be defended. Indi- viduals do not share the same feelings, beliefs, or attitudes, and even if choices are the same, these individuals have different motivations (Fishbein and Azjen 1975). Individuals’ iden- tities, experiences, and principles are key to understanding their decisions. This heterogeneity can be included in econometric analysis through the use of ap- proaches such as random parameter and latent class models (Train 2003; Morey, Thacher, and Breffle 2006; Johnston 2007; Breffle, Morey, and Thacher 2011; among many oth- ers). However, preference heterogeneity is more complex and nuanced than previous re- search would suggest. Preferences for “good things” can always be anticipated, and it is generally assumed that environmental goods are appreciated by some people. On the other hand, environmental problems have multiple facets, as some have consequences only at the local level, whereas others have global con- sequences. The perception of these problems is varied. There are similarities and differ- ences in what people want, and some of these similarities and differences are geographically distributed. To understand this phenomenon, we must focus on people’s geophysical sur- roundings and find patterns of behavior across regions, districts, neighborhoods, or river ba- sins and the determinants of these patterns. The authors are, respectively, research fellow, Insti- tute for Public Goods and Policies (IPP), Spanish Na- tional Research Council (CSIC), Madrid, Spain; Se- nior Water Economist and Theme Leader for Water Economics, Stockholm International Water Institute, Stockholm, Sweden; and research fellow, Forest Re- search Center (CIFOR), National Institute for Agri- culture and Food Research and Technology (INIA), Madrid, Spain.