Objectives and Aims

This research group proposes the development of a next generation joint process-outcome referencing modelling framework in which the choices made by agents are conditioned on rules adopted by each agent in assessing the attributes packaged in the definition of existing and potential alternative products and services. The BCG research plan will heavily focus on the development of new ways of introducing heuristics and biases into choice analysis that can account for decision strategies such as those listed in Table 1.

The attribute-based approaches include the elimination-by-aspects (EBA), lexicographic choice (LEX), and majority of confirming dimensions (MCD) strategies. The alternative-based approaches include the weighted additive (WADD), satisficing (SAT), and equal-weight (EQW) strategies. Individuals construct strategies depending on the task demands and the information they are faced with. Given the position, that rationally-adaptive behavioural models are more likely to be behaviourally valid descriptions of choice behaviour, we can clearly discount the WADD strategy, since it assumes that all information is processed (this remains a testable assumption, however).

Elimination by Aspects (EBA) involves a determination of the most important attribute (usually defined by the attribute with the highest weight/probability) and the cut-off value for that attribute (i.e., a threshold). An alternative is eliminated if the value of its most important attribute falls below this cut-off value. This process of elimination continues for the second most important attribute, and so on, until a final alternative remains. Thus, the EBA strategy is best characterized as a 'threshold' attribute processing strategy. The LEX strategy, in its strictest sense, involves a direct comparison between alternatives on the most important attribute. In the event of a tie, the second most important attribute is used as a comparison, and so on until an alternative is chosen. The LEX strategy is thus best characterized as a 'relative comparison' strategy. Thus, we can clearly differentiate two classes of attribute processing strategies: threshold and relative comparison.

A major deficit in these strategies is that although they assume selectivity in attribute processing across different decision task contexts, they assume consistency in attribute strategy within the same decision context. In other words, once a strategy is selected for a given task (or choice), it does not change within the task. To date no research has looked at process heterogeneity and modelled it jointly (i.e., endogenously) with choice outcomes. The proposed BCG will focus on this.