How Virtual Agents Influence Online User Behavior: An Experiment and Modelling Proposal Assessing Effects on Attitudes, Loyalty and Word-O-Mouth

Pablo Brice Lambert de Diesbach,Richard P. Bagozzi
DOI: https://doi.org/10.2139/ssrn.3397882
2019-01-01
SSRN Electronic Journal
Abstract:Virtual agents wear many different names according to their use contexts, functionalities, degrees of anthropomorphism. They can also be visible or incarnated, or not (incarnated agents are called Embodied virtual agents or EVAs). They have been widely used since the 2000’s and first generated a strong interest in the literature, which seemed to gradually decrease at the end of the last decade. Globally, their effects were assessed as positive on website use and interest towards the site visit. But most often they are not clearly modelized and we believe there is a lack of understanding the “why” virtual agents are accepted, and why they generate behaviors. Last, their effects not only on attitude but more precisely on detailled facets of attitudes, and on online loyalty and on the intention to recommend the online platform (word-of-mouth) has been very rarely conducted. This research is an intent to modelizer the effects of an embodied virtual agent on Attitudes towards a platform, and of Emotional reactions, to Attitude towards a brand, and to WOL and Loyalty intentions. We propose a scale measuring the attitude towards a website in which we pay a particular attention to avoid items redundancy, in order to precisely capture all facets of the attitude. Two attitudinal dimensions emerge: (Aws being the Attitude towards a website) Aws Functional, and Aws Hedonic. We conduct an experiment in a research laboratory of a major Telecom operator research center based in Europe, exposing a sample of a hundred web users to a navigation with a pre-purchase information session, oriented to a future purchase. Socio demographics are collected, as well as measures of Emotional reactions to the navigation, Attitude towards the platform (Functional and Hedonic), Attitude towards the brand, and Word-of-mouth intention. We proceeded to GLM analysis. The results show that the presence of an EVA significantly impacts both Loyalty towards the platform, and word-of-mouth concerning the platforms. Moreover, such effects being fully mediated by the two components of Attitude towards the website (Functional and Hedonic) and by Attitude towards the brand. The same mediation model is confirmed on two different websites tested in the laboratory.
What problem does this paper attempt to address?