International Choice Modelling Conference, International Choice Modelling Conference 2015

Font Size: 
Decomposing the Utility of Alternatives from Mental Representations of Decisions: Bridging Choice Experiments and Means-End Chain Elicitation
Benedict G.C. Dellaert, Theo A. Arentze, Oliver Horeni, Harry J.P. Timmermans

Last modified: 11 May 2015


To improve their product offering and guide possible innovations marketing managers often need to develop a deep understanding of the choices that their customers make. When rich data exists on consumer decisions in the market, choice models can be used to derive the value of different product features for consumer choice. However, there are also many instances where market data provide only relatively little product and choice outcome variation and additional information is needed to develop further insights in consumer decisions. For example, the availability of shopping centers, recreational attractions, educational institutions, and hospitals, is typically highly stable over time and with only few alternatives that any given customer will consider. This lack of variation restricts the possibility to determine the relative importance to consumers of different aspects of the available alternatives by using choice models based on market data.

In response, techniques like conjoint analysis (Green and Srinivasan 1990) and choice experiments (Louviere, Hensher and Swait 2000) were developed to increase managers’ understanding of consumer choices by presenting more, and more varied, (hypothetical) alternatives to consumers than they routinely face in the market. This allows for exploration of the impact of a wide range of changes in product features on consumer choice. In addition, techniques such as means-end chain elicitation or laddering (Gutman 1982, Reynolds and Gutman 1988) were also developed to abstract away from the market choices that consumers make. However they do so in a different way, by probing for deeper insights into the underlying benefits that consumers look for in the choices that they make, and that drive their evaluations of attribute changes. This approach provides a richer understanding of the reasons why consumers buy products, something that also cannot be observed directly in the market. Recently, researchers have begun to combine these lines of thought in their research and have to included attribute-benefit relationships in models of consumer choice (Arentze, Dellaert and Chorus 2014; Ashok, Dillon and Yuan 2002; Chandukala, Edwards and Allenby 2011; Dellaert and Stremersch 2005; Luo, Kannan, Ratchford 2008).

The approach that we propose in this paper builds on this rich stream of research on choice modeling and attribute-benefit analyses, but in contrast to earlier work extends it in an opposite direction. In particular, rather than incorporating mental representations in models of consumer choice, we develop and empirically test a model of how consumers’ utility can be derived from their mental representations of decisions as observed in means-end chain elicitations or laddering. Theoretically, we start from the fact that consumers’ knowledge of complex decision problems enables them to mentally simulate the likely consequences of their decisions before implementing these decisions (Johnson-Laird 2001; Weber and Johnson 2006). By mentally connecting the possible choice of different alternatives to how these alternatives help achieve the consumer’s higher order needs, consumers can evaluate the attractiveness of different courses of action. To do so, consumers’ mental representations of decision problems include attributes of the choice alternatives that they face and link these alternatives to higher order benefits that reflect the consumer’s own personal needs (Gutman 1982; Liberman et al. 2007; Newell and Simon 1972; Reyna and Brainerd 1995).

However, inherent cognitive constraints restrict consumers’ mental representations of decision problems (Beach and Mitchell 1987; Johnson-Laird 2001; Loewenstein 2001). Therefore, we propose in line with information search theory (Hauser, Urban and Weinberg 1993, Moorthy, Ratchford and Talukdar 1997; Weitzman 1979) that additional decision components (i.e., attributes and benefits) are included in a mental representation only if the expected gain of evaluating alternatives on these decision components exceed the cognitive costs of doing so. We formalize this process in a utility theory-based model of the cognitive activation of different components of consumers’ mental representations of decisions as a reflection of their latent utility for the alternatives in these decisions. The consumer’s utility function drives the selection of attribute and benefit components (and their links) such that cognitive activation in the mental representation depends on whether variations in the level of a component have a strong enough impact (i.e., above the individual’s mental cost threshold) on the decision outcome to warrant their inclusion in the consumer’s mental representation of reality. This approach allows us to formally model the connection between selective mental representations and consumers’ preference in the context of choice. Thus, we can simultaneously model the consumer utility function, the attributes and benefits consumers consider in making their choices, and the mental connections between the alternatives of different choices and the different attributes and benefits.


The model and empirical analysis contributes to the literature by providing an alternative to earlier research that included attribute-benefit relationships in models of consumer choice (e.g., Ashok, Dillon and Yuan 2002; Luo, Kannan, Ratchford 2008), that lacked a structure to integrate mental representations and utility (Ter Hofstede, Steenkamp, and Wedel 1999), or was mainly qualitative in nature (Dellaert et al. 2008). A particular strength of the currently proposed approach is that the cognitive activation of attributes, benefits and their links in the model structure is directly integrated in the underlying utility function which provides for a structural explanation of which components are activated in the mental representation and which are not. This allows us to express the utility of the different components from the elicited mental representation only. We empirically illustrate the viability of the proposed approach we collected hard laddering data from 594 individuals from a nationally representative online panel. The data captured individuals’ mental representations of a hypothetical shopping decision problem. The results demonstrate that the approach represents a promising new alternative method to the use of choice experiments to decompose consumers’ utility of alternatives.

Conference registration is required in order to view papers.