Random regret minimization for consumer choice modeling: Assessment of empirical evidence

Abstract

This paper introduces to the field of marketing a regret-based discrete choice model for the analysis of multi-attribute consumer choices from multinomial choice sets. This random regret minimization model (RRM), which has recently been introduced in the field of transport, forms a regret-based counterpart of the canonical random utility maximization paradigm (RUM). This paper assesses empirical results based on 43 comparisons reported in peer-reviewed journal articles and book chapters, with the aim of finding out to what extent, when, and how RRM can form a viable addition to the consumer choice toolkit. The paper shows that RRM and hybrid RRM-RUM models outperform RUM counterparts in a majority of cases, in terms of model fit and predictive ability. Although differences in performance are quite small, the two paradigms often result in markedly different managerial implications due to considerable differences in, for example, market share forecasts.

Heterogeneous Valuation of Quality Dimensions of Railway Freight Service by Chinese Shippers: Choice-Based Conjoint Analysis

“…More recently, several papers offered the random regret minimization (RRM) model as an alternative way of modeling choice behavior; examples include Chorus (38) and Chorus et al (39). The RRM approach is based on the notion that, when choosing, people anticipate and aim to minimize regret rather than maximize utility.…”

“…Choice probabilities for the RUM and the RRM models, respectively, are as follows (both in MNL form): Furthermore, on many occasions, a hybrid model (in which some attributes are processed by utility maximization and others by regret minimization) fits the choice data best (39). Behaviorally, such a hybrid model implies that some attributes are assumed by the researcher to be processed by the decision maker in a linear-additive RUM way while others are assumed to be processed in an RRM fashion.…”

Random Regret Minimization and Random Utility Maximization in the Presence of Preference Heterogeneity: An Empirical Contrast

“…This paper set out to explore the difference between RUM and RRM when preference heterogeneity is accounted for through random parameters, in contrast to the predominant assessment in the literature under the non-random parameter MNL form or an error components model form with non-random parameters (as reviewed by Chorus et al 2014).…”

“…Finally, ongoing research should investigate the contrasts between various other semi-compensatory RUM forms (i.e., process heuristics including attribute non-attendance and RRM), recognising that in the current paper we have chosen to stay with a model form that is fully compensatory (in respect of attributes). Chorus et al (2014) provide an audit on many studies, the majority being MNL with standard RUM and RRM forms. It is worth noting, however, that Leong and Hensher (2015) compared the Relative Advantage Model (RAM) (based on the smoothed regret function of the RRM model), and the fully compensatory RUM with RRM and found that although model fit differences were small, a comparison shows that the RAM model empirically outperforms the standard RUM and RRM models, and a hybrid RUM-RRM model in all eight data sets analysed.…”

“…This recognition of heterogeneity in preferences is something that should matter and hence be investigated whether the underlying behavioural assumption of a model is RRM or random utility (RUM). Extensions to build in preference heterogeneity under RRM through random parameters are rare although there are few studies using error components models to allow for correlated alternatives (see Chorus et al 2014). In addition, some studies have estimated models in which a subset of attributes is subject to random utility maximisation (RUM), and another subset subject to RRM Stathopoulos 2013).…”