Demand for sanitation in Salvador, Brazil: A hybrid choice approach Andreia C. Santos a, * , Jennifer A. Roberts a , Mauricio L. Barreto b , Sandy Cairncross a a London School of Hygiene and Tropical Medicine, London, United Kingdom b Instituto de Saúde Coletiva, Universidade Federal da Bahia, Brazil article info Article history: Available online 3 March 2011 Keywords: Sanitation Sewerage system Hybrid choice model Mixed logit Brazil abstract Funds to promote access to water and sanitation in developing countries are scarce and most of the investments come from the national governments and households sources, not international resources. In many of these countries, mainly in middle income countries, households are paying direct taxes to access these services, and understanding what determines their choice and motivation is fundamental to promote access to them. It has been argued that is not enough to supply a wide range of alternatives unless the individuals can recognise their benefits and sustainability. The objective of this paper is to understand the choice of sanitation technology by residents in the city of Salvador, Brazil. We propose a unique hybrid choice model that incorporates a set of latent attitudinal variables and explains how the demographic factors within a household influence choice. The substantial difference of our hybrid choice model from descriptive frameworks is that it integrates choice and latent variables (such as attitudes and preferences) allowing us to model explicitly the cognitive process that influences sanitation adoption, draw conclusions from cognitive variables associated with individuals’ socio-economic and demographic characteristics, and establishes a causal pathway among these variables. The results show that the attributes of health protection, accessibility, privacy, and house modern- isation were what households cared about when opting for flush toilet and sewerage connection, rather than the high cost and consequent household socio-economic status associated with them. The hybrid model is statistically consistent with these findings, and seems to fill the gap between behavioural theory and discrete choice models applied to sanitation. Ó 2011 Elsevier Ltd. All rights reserved. Introduction Funds to promote access to water and sanitation in developing countries are scarce and most of the investments come from the national governments and households sources, not international resources (Shordt, van Wijk, & Brikke, 2004). In many of these countries, mainly in middle income countries, households are paying direct taxes to access these services, and understanding what determines their choice and motivation is fundamental to promote access to them. It has been argued that is not enough to supply a wide range of alternatives unless the individuals can recognise their benefits and sustainability (Persson, 2002). Jenkins (1999) demonstrated that the few latrines installed in rural Benin were paid for by the households, and that the deter- minants of household choice for sanitation were associated with factors other than the health messages disseminated by governmental and non-governmental agencies. Her findings indi- cated the necessity for deeper investigation of the demand-side aspects, understanding the cultural and historical context, needs, perception and preferences of households, as a means to improving coverage and sustainability. Few studies have explored the demand for sanitation. Most have applied the conventional contingent valuation approach that continually fails in giving an accurate prediction of consumer behaviour. The method assumes that the maximum amounts people will pay to avoid a loss (willingness to pay) and the minimum compensation necessary for them to accept it (willing- ness to accept) is equivalent in probabilistic contexts (Diamond & Hausman, 1994; Knetsch & Sinden, 1994). However, experiments have demonstrated that individuals weigh gains and losses differ- ently, and they have greater sensitivity towards losses than towards gains, contradicting the main assumption of this approach and making it poorly suited to providing answers to policy questions facing planners (Kahneman, 2003; Knetsch & Sinden, 1994). On the other hand, the traditional economic demand model is not able to * Corresponding author. E-mail address: andreia.santos@lshtm.ac.uk (A.C. Santos). Contents lists available at ScienceDirect Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed 0277-9536/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.socscimed.2011.02.018 Social Science & Medicine 72 (2011) 1325e1332