Motivation and aims

The choice modeling literature has experienced a blossom of research aimed at unveiling and modeling different attribute processing strategies. It is widely acknowledged that respondents might adopt different processing strategies when making their choice so as to simplify the decision process (Heiner et al. 1983; Payne et al. 1993). However, as observed by Hensher (2014, p.2), ‘What we do not yet have enough accumulated wisdom on is the identification of a small set of processing rules that might be the best descriptors of the way in which individuals process information in hypothetical (via choice experiments) and real markets.’.

Several extensions of standard discrete choice models have been proposed to take into account different attribute processing strategies. A great deal of studies have been focusing on modeling attribute non-attendance, either using information drawn from respondents, or inferring it from the model estimates (Hensher et al. 2005; Hensher 2006; Hensher et al. 2007; Puckett and Hensher 2008; Hensher and Rose 2009; Puckett and Hensher 2009; Scarpa et al. 2009; Carlsson et al. 2010; Hess and Hensher 2010; Scarpa et al. 2010; Campbell et al. 2011; Hole 2011; Alemu et al. 2013; Colombo et al. 2013; Hess et al. 2013; Hole et al. 2013; Kehlbacher et al. 2013; Kragt et al. 2013; Lagarde 2013; Scarpa et al. 2013).

Another stream of work has investigated the role of lexicographic preferences, assuming respondents choose only on the basis of a specific attribute, level or alternative (Sælensminde 2001; Scott 2002; Rosenberger et al. 2003; Gelso and Peterson 2005; Campbell et al. 2006; Lancsar and Louviere 2006; Hess et al. 2010). Furthermore, Cantillo and Ortúzar (2005) and Swait (2009) have put forward a two-steps elimination by aspect decision strategy, according to which respondents initially eliminate alternatives where attributes do not reach a certain level and, successively, a choice is made among the remaining options.

Another approach is based on the stated importance attached to the attributes. For instance, Balcombe et al. (2014) have proposed to incorporate the stated ranking of the attributes in the utility function using a mixed logit framework, where specific weights are given to the attribute parameters depending on their stated ranking.

This paper aims at contributing to the stream of research devoted to improving and widening the knowledge of heuristics in choice modeling, introducing an hypothesis on the anchoring effect of salient attributes in the choice exercise. In a nutshell, we test the following behavioral hypothesis: when respondents have some difficulty-irrespective of the reason-in making the choice between alternatives, they use a simplifying strategy that consists in conditioning the choice on the attribute that they perceive as the most important. In our model, the individual choice is characterized by a mixture of a fully compensatory behavior and a lexicographic behavior. The latter is the result of a simplifying strategy, or in other words a sort of response bias, rather than of a truly non compensatory preference structure.

If some elicitation effect is present in the choice data it may lead to biased results in the coefficients’ estimates and monetary valuations. This paper provides a framework to detect and properly take into account a potential source of bias, hence leading to more reliable estimates.

Model

We model our behavioral hypothesis employing a constrained latent class framework, with i individuals, j alternatives, t choice tasks, k attributes and s segments (or classes). A two-classes mixture model is necessary to allow for the anchoring to salient attributes and the fully compensatory behavior. We define segment S=1 as capturing the former, whereas in S=2 the trade-off between all of the attributes is modeled. An alternative specific constant (ASC_Most) characterizes the deterministic component of the utility function conditioned on segment 1. This constant indicates which of the alternatives contains the salient attribute. Remarkably, after the series of choice tasks, respondents were asked to provide the ranking of the attributes, since in this application we employ attribute stated importance to capture saliency. Instead, the utility function conditioned on segment 2 has the effect of this constant constrained to zero, whereas the coefficients attached to the attributes are estimated. Error terms are assumed to be IID Gumbel.

In a nutshell, each respondent is associated to a segment S=1 with probability α, where only the most important attribute drives the choice; simultaneously he or she is associated to the remaining classes with probability 1-α, where a fully compensatory behavior is allowed. Remarkably, practitioners may be interested in modeling more than two latent segments, jointly modeling the role of salient attributes and preference heterogeneity in a discrete fashion.

Preliminary results

The proposed processing strategy is applied to two datasets concerning preferences towards energy (respectively with 216 and 860 respondents) and comparing its results with standard discrete choice models, including the Multinomial Logit, Latent class and Random Parameters Logit, considering the following:

• The sign and significance of the coefficient attached to ASC_Most;
• Goodness of fit measures: Log-likelihood, AIC, AIC3, BIC, CAIC;
• In-sample predictions;
• Cross validation predictions;
• Out of sample predictions.

Overall, preliminary results provide evidence supporting our behavioral hypothesis: respondents might be focusing on attributes which are salient for them. Most importantly, failure to take this into account leads to poorer fit and substantially different welfare valuations. 

References

Alemu M H, Mørkbak M R, Olsen S B, Jensen C L (2013) Attending to the reasons for attribute non attendance in choice experiments. Environmental and Resource Economics 54:333-359

Balcombe K, Bitzios M, Fraser I, Haddock-Fraser J (2014) Using attribute rankings within discrete choice experiments: An application to valuing bread attributes. Journal of Agricultural Economics 0(2):446-462

Campbell D, Hensher D A, Scarpa R (2011) Non-attendance to attributes in environmental choice analysis: a latent class specification. Journal of Environmental Planning and Management 54 (8):1061-1076

Campbell D, Hutchinson W G, Scarpa R (2006) Lexicographic preferences in discrete choice experiments: consequences on individual-specific willingness to pay estimates. Fondazione Eni Enrico Mattei, Nota di lavoro 128.2006

Cantillo V, Ortúzar J (2005) A semi-compensatory discrete choice model with explicit attribute thresholds of perception. Transportation Research Part B 39 (7):641-657

Carlsson F, Kataria M, Lampi E (2010) Dealing with ignored attributes in choice experiments on valuation of Sweden’s environmental quality objectives. Environmental and Resource Economics 47:65-89

Colombo S, Christie M, Hanley N (2013) What are the consequences of ignoring attributes in choice experiments? Implications for ecosystem service valuation. Ecological Economics 96:25-35

Gelso B R, Peterson J M (2005) The influence of ethical attitudes on the demand for environmental recreation: incorporating lexicographic preferences. Ecological Economics 53(1):35-45

Gibrilde T J, Allenby G M, Brazell J D (2006) Models for heterogeneous variable selection. Journal of Marketing Research 43(3):420-430

Heiner R A (1983) The origin of predictable behavior. The American Economic Review 73(4):560-595

Hensher D A, Rose J, Greene W H (2005) The implication on willingness to pay of respondents ignoring specific attributes. Transportation 32:203-222

Hensher D A (2006) How do respondents process stated choice experiments? Attribute consideration under varying information load. Journal of Applied Economics 21:861-878

Hensher D A, Rose J, Bertoia T (2007) The implication on willingness to pay of a stochastic treatment of attribute processing in stated choice studies. Transportation Research E 43(1):73-89

Hensher D A, Rose J (2009) Simplifying choice through attribute preservation or non-attendance: implications for willingness to pay. Transportation Research E 45:583-590

Hensher D A (2014) Process heuristics in choice analysis: An editorial. The Journal of Choice Modelling 11:1-3

Hess S, Hensher D A (2010) Using conditioning on observed choices to retrieve individual-specific attribute processing strategies. Transport Research Part B 44:781-790

Hess S, Rose J M, Polak J (2010) Non-trading, lexicographic and inconsistent behaviour in stated choice data. Transportation Research Part D 15(7):405-417

Hess S, Stathopoulos A, Campbell D, O’Neill V, Caussade S (2013) It’s not that I don’t care, I just don’t care very much: confounding between attribute non-attendance and taste heterogeneity. Transportation 40(3):583-607

Hole A R (2011) A discrete choice model with endogenous attribute attendance. Economics Letters 110:203-205

Hole A R, Kolstad J R, Gyrd-Hansen D (2013) Inferred vs. stated attribute non-attendance in choice experiments: A study of doctors’ prescription behaviour. Journal of Economic Behavior & Organization 96:21-31

Kehlbacher A, Balcombe K, Bennet R (2013) Stated attribute non attendance in successive choice experiments. Journal of Agricultural Economics 64:693-706

Kragt M E (2013) Stated and inferred attribute attendance models: A comparison with environmental choice experiments. Journal of Agricultural Economics 64:719-736

Lagarde M (2013) Investigating attribute non-attendance and its consequences in choice experiments with latent class models. Health Economics 22:554-567

Lancsar E, Louviere J (2006) Deleting ‘irrational’ responses from discrete choice experiments: a case of investigating or imposing preferences? Health Economics 15(8):797-811

Payne J W, Bettman J R, Johnson E J (1993) The adaptive decision maker. Cambridge University Press, Cambridge UK

Puckett S M, Hensher D A (2008) The role of attribute processing strategies in estimating the preferences of road freight stakeholders under variable user charges. Transportation Research E 44:379-395

Puckett S M, Hensher D A (2009) Revealing the extent of process heterogeneity in choice analysis: an empirical assessment. Transportation Research A 43 (2):117-126

Rosenberger R S, Peterson G L, Clarke A, Brown T C (2003) Measuring dispositions for lexicographic preferences of environmental goods: integrating economics, psychology and ethics. Ecological Economics 44: 63-76

Scarpa R, Gilbride T, Campbell D, Hensher D A (2009) Modelling attribute non-attendance in choice experiments for rural landscape valuation. European Review of Agricultural Economics 36(2):151-174

Scarpa R, Thiene M, Hensher D A (2010) Monitoring choice task attribute attendance in non-market valuation of multiple park management services: does it matter? Land Economics 86 (4):817-839

Scarpa R, Zanoli R, Bruschi V, Naspetti S (2013) Inferred and stated attribute attendance in food choice experiments. American Journal of Agricultural Economics 95:165-180

Scott A (2002) Identifying and analysing dominant preferences in discrete choice experiments: An application in health care. Journal of Economic Psychology 23(3):383-398

Swait J (2009) Choice models based on mixed discrete/continuous PDFs. Transportation Research Part B 43 (7):766-783

Sælensminde K (2001) Inconsistent choices in stated choice data: use of the logit scaling approach to handle resulting variance increases. Transportation 28:269-296