Predicting online purchase behavior : replications and tests of competing models

Online purchase behavior is definitely an interesting and relevant issue for marketeers today. In this paper, we report on a study into the antecedents of online purchase intention for B2C websites. In particular, this research juxtaposes two competing models that explain online purchase intention. The first model is trust-oriented and argues that online purchase intention is primarily predicted by trust in the company. The second model is website-oriented and argues that purchase intention is primarily predicted by usefulness and ease-of-use of the e-commerce website. In order to test to what extent each of these orientations has merit, a replication was carried out of the trust-oriented study by Jarvenpaa, Tractinsky, and Vitale (2000). The model was extended with the website-oriented constructs from by Chau, Au, and Tam (2000), first developed by Davis (1989). The replication study involved 227 undergraduate students. 64.4% of them had never bought online; 8.8% had bought on the intemet at least four times. Consequently, the results of the study are somewhat biased towards initiaZ purchase intention as opposed to repeat purchase intention. The research has a number of important findings. First, online purchase intention at the website is strongly determined by attitude towards online shopping at the website, providing support for the Theory of Reasoned Action in a website context. Second, perceived risk of shopping strongly influences attitude towards shopping. Trust in the company does not influence attitude directly, but indirectly through a significant impact on perceived risk. Third, perceived reputation influences trust, whereas perceived size does not, at least not in the case of low-value products. Fourth, website ease-of-use strongly and positively influences website usefulness. Finally, website usefulness does not significantly influence attitude towards shopping and online purchase intention. We conclude that trust-oriented models appear to be more appropriate to explain online purchase