Are French wine consumers closer to Minerve or Bacchus?

Type de communication : Présentation oraleInternational audienceThe paper presents the results of a consumer survey carried-out in France in 2013, focusing on knowledge and involvement on wine. It intends to clarify whether involvement and knowledge with respect to wine are factors operating in wine drinking and purchasing habits. We assume that there is information asymmetry among consumers. The survey was carried out in two different cities, namely Dijon and Lyon. The first belongs to Burgundy, a famous wine producing region, which is not the case of the second. The 415 respondents were invited to answer a written questionnaire including twenty questions on key dimensions of product-oriented knowledge: processing, semantic and geography, and sixteen questions on involvement, all related to wine. One additional section of the questionnaire was devoted to the usual socio-demographic descriptors of the respondents. Each respondent obtained three scores on the basis of the level of knowledge shown through his/her responses among the three dimensions. The same was done with respect to involvement towards wine, including distinction between affective, cognitive and novelty involvement. The distribution of ratings issued from the above-mentioned coding of knowledge level according to wine indicates that semantic about wine is the dimension of knowledge most shared among the respondents, whereas the processing-related dimension for the studied products is more discriminating. Knowledge of wine's geography is in medium position. According to the literature, we may call experts those respondents with a high level of knowledge and novices those showing a low level of knowledge. As the level of knowledge is divided into three dimensions, namely processing-related, semantic or geographical, the classification of respondents into experts versus novices will not be fully reliable per se and needs to be refined by means of clustering analysis. K-means clustering was then used in order to better explain the diversity of knowledge for wine displayed by the respondents. This method segments respondents into clusters according to their level in various types of knowledge. K-means clustering is a non-hierarchical clustering procedure: objects are assigned into a user-specified number of clusters. Four significant segments of respondents were identified by means of this method of classification. Each cluster is described by its relative positioning according to the level of knowledge in each dimension