Abstraction and product categories as explanatory variables for food consumption

Abstract

This study sets out to explore variation in the consumer choice structure in relation to three types of products with different levels of search, experience and credence attributes. The decision structures of rice, wine and functional food (experience, search and credence products) consumers were explored by means of laddering. The results suggest the presence of an emotional component in foods that increases in complexity (becomes more abstract) with the number of credence attributes. The findings recommend adjusting the complexity of advertising campaigns and product positioning to the type of product being promoted.

Keywords:

• consumer behaviour
• laddering interview
• means-end chain
• economic information theory
• abstraction level

JEL Classification::

• Q13
• M31
• M37

Notes

¹ Note that although most of the literature assumes the hierarchical structure of the MEC theory, this has been challenged by some authors in recent years (Van Rekom and Wierenga, 2007, among others). This study assumes the hierarchical structure because it is the most widely accepted.

² Concrete attributes are those properties or characteristics of the product, service or behaviour that may be preferred or sought by consumers; abstract attributes are those that cannot be verified without consuming the product and must therefore be inferred from internal or external information sources. Functional consequences are the benefits the customer derives directly from the attributes of the product or service consumed.

³ Psychological consequences are the more personal, or social and less tangible consequences. Instrumental values are intangible goals related to the behavioral channels through which terminal values can be satisfied. Finally, terminal values are preferred end states.

⁴ Hierarchical Value Map 1 (HVM1) represents the ‘most important’ or ‘best’ linkages, in the sense that any other choice of cut-off value or values must produce a HVM based on smaller and thus less strongly-associated linkages between levels of abstraction. HVM1 is also the least complicated and most easily interpreted of all possible HVMs, since it displays the least number of elements. Similarly, HVM2 is constructed by repeating the above process but defining the relevant cut-off levels at the second largest cell entry at each level of abstraction. Thus additional linkages are created by using a less stringent requirement (second highest versus first highest), which results in a more complex HVM. Continuing in this way, choosing successively smaller cell entries, in the manner described above, a sequence of HVMs can be created.

⁵ With respect to sample size, note that a large number of studies using this methodology are based on samples of a similar size to ours: Nielsen et al. (1998) used a sample of 90 subjects in Denmark, 50 and the UK and 50 in France to assess consumers’ attitudes towards vegetable oil; Fotopoulos et al. (2003) surveyed 49 subjects to examine their attitude towards organic products; Vannoppen et al. (2001) surveyed a sample of 40 subjects to identify their motive to purchase quality labelled apples; etc.

⁶ Although ATP was designed for use with large samples, it is approved for samples of 50+. Some examples of published works in which this methodology has been used with relatively small samples include Russell et al. (2004), where the APT technique is applied to a sample of 55; Ter Hofstede et al. (1998), where a sample of 50 subjects are surveyed for their attitudes towards olive oil and vegetable oil using this hard laddering technique.

⁷ A functional dairy product with an immunological effect (similar to Actimel).

⁸ The importance of the attributes was measured on a 5-point Likert scale, where 5 indicated the highest level of importance.

⁹ Missing scores correspond to attributes that were specific to one or two products.

¹⁰ The main differences between products according to the test statistics are in search attributes, followed by experience attributes.