International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
An incentive-compatible experiment to model wine preferences considering extrinsic and intrinsic attributes
David Esteban Palma, Juan de Dios Ortúzar, Luis Ignacio Rizzi, Stephane Hess

Last modified:  1 March 2019


Despite many technical advances in food health, safety and sustainability in the last few decades, adoption of new product and services ultimately depends on consumers preferring them in the marketplace. This makes it paramount to understand how and why consumers choose to purchase given products. In the case of food and beverages, choice is an extremely complex process: (i) it is influenced by both extrinsic and intrinsic attributes of the product; (ii) it is dynamic, as choice may change depending on a consumer’s previous experience with the product; (iii) it is highly dependent on the purchase and consumption contexts; (iv) preferences vary widely among consumers, and (v) demand may not be monotonically decreasing with price, as price is often used as a cue for quality by consumers.

All these difficulties have lead researchers on food and beverage choice to focus on specific stages or aspects of the choice process, instead of on the whole. And even though there are behavioural models encompassing the whole purchase, consumption and re-purchase process (notably Grunert 2005, with his Total Food Quality Model), few studies have attempted to operationalize it (e.g. Mueller et al., 2010).

In this paper, we present a framework to study the choice of food and beverages based on a simplified version of the Total Food Quality Model. This model describes a three stage process: (i) purchase, (ii) tasting or consumption and (iii) re-purchase. During the first stage, consumers can only perceive extrinsic attributes, i.e. attributes that can be perceived before tasting the product such as the packaging and health claims. Based on these attributes, consumers build an expected quality of the product and decide their purchase based on it. Note than in the first stage, price can act as both a cue for quality and as a strain on consumers’ budget restrictions, therefore having both a positive and negative effect on the probability of purchasing the product. During the second stage, consumers taste the product, therefore perceiving the product’s intrinsic attributes (mainly taste and aroma) and building a sensory perception based on those, i.e. how much they ‘liked’ the product. Finally, an interaction between the expected quality and the sensory perception determines the experienced quality, i.e. how good consumers think the product is after looking at it and tasting it. Consumers’ experienced quality will determine if they will re-purchase the product in the future.

The described behavioural model can be applied not only to food, but to any experience product, i.e. those whose quality cannot be fully determined before consumption. For example, medicine and its effectivity, travel modes and their safety and comfort, holiday destinations and their touristic interest, etc. In all these cases price and the unobserved attribute can be associated, and the continuous consumption of these products or services will not depend only on their observable attributes, but also on consumers’ subjective experience.

Using wine as a case study, we developed an incentive-compatible experiment measuring the effect of both extrinsic and intrinsic attributes on choice, while considering the dynamic nature of the purchase, taste and re-purchase process, and controlling for the effect of price as a cue for quality. In addition, we allowed participants to choose more than one product, just as happens in the actual marketplace. We reproduced the whole process in a single experiment. First participants had to choose wine on a shelf without tasting the product (therefore simulating the first purchase). Then participants blindly tasted wines, and finally - having full information about the product - they were able to actually buy one or more wines (re-purchase stage). We jointly analysed data from all three stages using a discrete-continuous model (Bhat 2008), allowing us to compare the relative importance of expected quality and sensory perception on the actual purchase decision.

On a sample of 130 Chilean consumers, we found strong preference toward Carménère and Shiraz grape varieties. We also found a positive preference toward higher alcohol content and a strong use of price as a cue for quality. We found sensory perception to be roughly 33% more important than expected quality, implying that the re-purchase decision is mostly determined by the tasting experience, and extrinsic attributes being key mostly during the first purchase. We also found that higher income individuals preferred to buy bottles of different wines over more of the same wine.

Our results highlight the relevance of considering the problem of choosing food and beverages (and potentially other products and services) in its full complexity. It is only by reproducing the whole purchase, taste (experiencing) and re-purchase process that we can measure the relative importance of extrinsic and intrinsic attributes. Allowing purchases of more than one bottle also allows us to recognize variety-seeking behaviour that would otherwise go undetected. Finally, building on our earlier work (Palma et al. 2016) we once more confirmed the use of price as a cue for quality and that not correcting adequately for this effect leads to bias and erroneous conclusions.

While the proposed experimental setting is promising, there is still much room for improvement, as important factors such as consumption context and the passing of time between purchase and repurchase were not considered in our experiment. Despite these limitations, the procedure seems to work as expected, and applications to bigger samples could provide deeper insights into the phenomena, such as perception changes before and after tasting.


Bhat, C. (2008) The multiple discrete-continuous extreme value (MDCEV) model: Role of utility function parameters, identification considerations, and model extensions. Transportation Research 42B, 274 – 303.

Grunert, K. (2005) Food quality and safety: consumer perception and demand. European Review of Agricultural Economics 32, 369 – 391.

Mueller, S.; Osidacz, P.; Leigh Francis, I.; Lockshin, L. (2010) Combining discrete choice and informed sensory testing in a two-stage process: Can it predict wine market share? Food Quality and Preference 21, 741 – 754.

Palma, D.; Ortúzar, J. de D.; Rizzi, L.I.; Guevara, C.A.; Casaubon, G. and Ma, H. Modelling choice when Price is a cue for quality: a case study with Chinese consumers. The Journal of Choice Modelling 19, 24 – 39.

Conference registration is required in order to view papers.