Research | Funded Projects

SUPPORT BUYERS' DECISIONS THROUGH STRUCTURAL EQUATION MODELING

1. BackgroundD’Avanzo & Kuflik, (2013) presented a study combining an E-Commerce literature survey, an E-Commerce websites' analysis, and a survey of online consumers opinions, pointing to a gap that exists between sellers' services and consumers’ expectations/choices. Empirical evidence suggests that this gap can be bridged turning to non-compensatory strategies (i.e. heuristics), implemented as web services on the E-Commerce websites. 2.ObjectivesThis project aims at introducing a further analysis, based on Factor Analysis and Path Modelling, that is able to identify significant non-compensatory choice mechanisms among those emerged from previous experiments.The previous analysis of 594 E-Commerce websites allowed for the identification of 68 different variants of services, reported in Figure 1 ($IMM_PER_0001). For instance, a typical set of non-compensatory strategies is represented by “product selection”, that groups together service variants such as product selection by categories, types, prices, brands, sales percentage and alphabetical order. Most of the services identified are heuristic shopping aids, since they enable consumers employing a very limited number of searches criteria, without in depth analysis or comparison among attributes.Starting from the taxonomy reported in Figure 1, through the project a combination of Factor Analysis and Path Modelling will be employed to look for a model of consumer choice mechanisms that could explain the choice strategies identified during a previous analysis.The model identified has a threefold aim. First, it is supposed to select only a subset out of 68 of relevant choice mechanisms that seem to be preferred and employed by the consumer when buying online. Second, whereas a grouping of the choice mechanisms was proposed, as depicted in Figure 1, the model employed may confirm the similarities that were proposed according to the literature. Third, the model formulated may provide a finer-grained division in the same category (i.e., browsing by category) as the previous model. As regards the path modelling part, instead, the model should be able to show significant correlation about constructs and possible causal relations among them.3. Methodology adopted to accomplish the objectivesThe most important contribution of this work is the employing of the Factor Analysis and Path Modelling (SEM) to identify significant non-compensatory choice mechanisms (i.e., heuristics) among those emerged from previous experiments, as depicted in Figure 1. In this sense the taxonomy of mechanisms and their relations with the different levels of the taxonomy represents a conceptual model, just the starting point for every good approach via SEM. Groenland, & Stalpers, (2012) claim that a conceptual model “describes a process comprised of concepts and causal relationships between these concepts”, where concepts are intended characteristic, cognition, an affect, a mental process, or an act, such as a specific form of behaviour. In the sense the model employed is expected to show both measurement and structural parts that make up the final model. Whereas the former part will translate consumers’ choice mechanisms from the questionnaire into psychological constructs, represented as latent variables, the latter will show how the identified constructs relate to each other.Figure 1 – The figure reports 68 services identified during the website analysis. All services identified are grouped using a taxonomy made by different levels of categorization, according to (Gabbott, & Hogg, 1998). In the following of this analysis it is only considered the choosing phase.References1. Gabbott, M. & Hogg, G. (1998). Consumers and Services John Wiley & Sons, Chicester, UK. 2. Groenland, E. A., & Stalpers, J. (2012). Structural equation modeling: A verbal approach.Nyenrode Research Paper, (12-02)