Abstract
The relative influence of objective and sensory attributes' effects on wine prices have intrigued scholars recently. In this regard, a study by Lecocq and Visser (Citation2006) concluded that the objective attributes were of primary importance. Against this background the purpose of this article is two-fold: (i) to establish that Lecocq and Visser's (Citation2006) conclusion might be premature and (ii) to demonstrate how any study aiming at comparing the relative strengths of different sets of explanatory variables might be carried out more effectively.
Acknowledgements
I thank Iver Andreas Heggen and Bredo Molthe for the permission to use their data.
Notes
1 There are a number of severe problems associated with the use of stepwise regression or automated-variable selection techniques in general; see e.g. Fox (Citation1997, pp. 356–358). However, I do not follow up this empirically here.
2 The regression assumption of noncorrelated errors seems difficult to satisfy in these data. Hence, Huber/White robust SE are reported.
3 I do not address the quality-price relationship in this article. In general, the introduction of (subjective) quality measures within the price hedonic framework raises new econometric problems (Oczkowski, Citation2001; Thrane, Citation2004).
4 The adjusted R 2 measures are 0.593 (Model 1) and 0.615 (Model 2).
5 These reductions in percent might be interpreted as the spurious part of the associations between the sensory attributes and the wine prices.
6 It must be noted that the standard errors of this regression are incorrect. As such it should only be thought of as a heuristic device to quantify possible indirect effects in the present context.