ABSTRACT
The micro level behavioral foundation for how quality reviews of experts and word of mouth (WOM) affect demand for experience goods has received scant attention in previous research. Taking an experimental approach, the present study examines how quality reviews and peer recommendations influence consumers’ decisions to buy red wine. Four main findings are presented for a sample of Norwegian wine consumers. First, consumers prefer wines that have obtained very good quality reviews to wines getting OK quality reviews. Second, consumers prefer peer-recommended wines to non-recommended wines. Third, the effect of a very good quality review is greater for non-recommended wines than for peer-recommended wines. Fourth, some of these effects on wine buying decisions are contingent on price level, and there is also some heterogeneity to their magnitude. The results are mainly in concert with the proposed hypotheses.
Acknowledgements
Thanks to Andrew Musau, two anonymous reviewers and the coordinating editor for comments on a previous draft of this paper.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1 In addition to the direct effect on demand, experts’ reviews may influence consumer demand indirectly through WOM; see Jacobsen (Citation2015).
2 For similar works on fiction novels, see Ashworth, Heyndels, and Werck (Citation2010), Chevalier and Mayzlin (Citation2006) or Ponzo and Scoppa (Citation2015).
3 See Aqueveque (Citation2015) for studies on the role of WOM as an antecedent of wine buying behavior.
4 Other sources of information are search engines like Google and wine blogs; for the latter, see Cosenza, Solomon, and Kwon (Citation2015). Other more heuristic quality cues that might reduce consumers’ uncertainty are wines’ region of origin and producer reputation; see Lockshin and Corsi (Citation2012) for a general review of consumer (search) behavior in the wine market.
5 More precisely, the Heckman method needs exclusion restrictions to work properly (Wooldridge, Citation2010). In practice, however, it is difficult to come up with independent variables that affect the decision to drink wine but not the frequency of drinking or vice versa. For more on the identification problems of sample-selection models, see Vance and Ritter (Citation2014).
6 Most of the subjects were members of wine clubs affiliated with the Association of Norwegian Wine Clubs (www.nvkf.no). Thanks to NVKF for informing about our study on their web site.
7 Unfortunately, it is impossible to know how many of the contact persons who received the email and, thus, how many wine club members who actually faced the decision to answer the electronic questionnaire. For this reason, it is difficult to arrive at a well-informed opinion about the actual response rate. Yet for strict statistical power considerations the sample size is more than sufficient.
8 Equivalent ordered logistic regression models were also estimated to examine whether OLS regression was justified. These analyses gave the same main conclusions as those reported for the OLS models (results available on request).
9 Both wine drinking and wine purchase frequency were tested as controls in preliminary analyses. Since these variables had practically no explanatory power in regard to the two dependent variables, they were discarded from the final set of analyses.
10 A one-sided test of significance is defensible since all the hypotheses are directional.