In European wine-producing countries, previous research has shown that ROO is one of the most important attributes in consumers’ purchase processes; in some cases, even more so than price. Perrouty et al. (2006) used a discrete choice method to survey European wine purchasers (from four countries), and found that ROO equity is significantly moderated by the other attributes of the wine. In Italy, Casini et al. (2009) showed that, when consumers are choosing a wine for a dinner at home with their friends, they firstly tended to select a wine they had tried previously, and secondly matched the wine to the food. However, these Italian consumers seemed to give less importance to the origin of the wine, although the vast majority of them usually bought wines produced in their own region. Some studies of the Spanish market have shown the relevance of the ROO and the PDO (Martínez-Carrasco et al., 2006; Bernabéu et al., 2008, 2013; Kallas et al., 2013). Except for some regions (Martínez-Carrasco et al., 2006; Bernabéu et al., 2008), consumers have a strong preference for locally produced wines as against national and imported wines.

Regarding “New World” wine countries such as Australia or USA, where grape varietal marketing has played a significant role in consumers’ purchase processes, ROO is also gaining importance in wine marketing since the end of 1990´s (Bruwer & House, 2003; Johnson & Bruwer, 2007). The importance of the ROO effect in these markets has been analysed by various authors. Bruwer & House (2003) found that ROO ranked third behind price and grape variety/wine style in Australian consumers’ wine purchase intention in a retail outlet. According to Johnson & Bruwer (2007), the wine’s region is the most important quality-predicting information on wine labels in the California market, with over 85% of consumers taking the region into account when evaluating a wine label. Indeed the region was the most often reported source of information, above brand, vintage, appellation, state, or country. Similarly, Atkin et al. (2007) found that the geographic origin of the wine was the most important attribute for both Australian and New Zealand consumers when choosing wines; in addition, local wines were preferred over all the other choices presented. A study by Atkin & Johnson (2010) found brand to be the commonest attribute used by consumers, followed by two geographical elements (country and region). Less important were other attributes such as vintage, state, label, medals won, alcohol content, appellation, and organic production attributes. Lacey et al. (2009) found that, for Australian consumers in a fine restaurant setting, the region followed only grape variety in being the strongest attribute they used in their risk reduction strategies when selecting the wine to go with their meal. However, other studies in Australia report different results. For instance, the relative importance of ROO among the main wine product attributes was found to be just fifth in a study by McCutcheon et al. (2009), after quality, price, style and variety.

Besides the study of the importance given to the place-of-origin, an important question that still remains to be important and should be analysed in detail is the existence of certain antecedent and moderating variables of the place-of-origin effect (Pharr, 2005). The literature review carried out for this study allows highlighting some of the main conclusions obtained by the scholars about these variables:

As we noted above, most of the studies on the place-of-origin effect have focused on the nation level and further research is needed on the ROO effect (González & Villanueva, 2001; Pharr, 2005; De Francesco et al., 2012). Based on the literature review, we designed the structural model shown in Fig. 1a with which to study the ROO effect in the Spanish wine market. The specific objective was to determine the variables that explain and moderate the intention to purchase regional wines. The model proposes direct and indirect effects of the following antecedent variables on the purchase intention: the region’s image, the region’s image as a producer in this product category, the perceived quality of the region’s products, and consumer ethnocentrism.

The model poses a number of ideas and hypotheses that had been suggested by a prior qualitative study carried out by the research team. This was an exploratory, qualitative, longitudinal (2002-2008) study, the objective of which was to assess the views and beliefs of the managers of wineries in a wine-producing region of Spain about the importance of regional origin and the PDO for consumers (García-Galán et al., 2012, 2014).

Ahmed & d’Astous (2007) showed that perceptions of a product’s quality are influenced by the image consumers have of its country of origin. Similar findings have been reported in other studies, such as those by González & Villanueva (2001), Srinivasan et al. (2004), and Verlegh et al. (2005). Van Ittersum et al. (2003) showed how consumers’ attitudes towards an ROO influence their perceptions of quality, although only for one of the two product categories they analysed in their study. In the case of the product under study in this present work, Lee & Lockshin (2011) found that the image of Australia has a strong positive influence on the perceptions of its wines. Given the above, we posit the following hypothesis:

H₁: The better the overall image consumers have of a region, the better their perceptions of the quality of its products.

In addition to the aforementioned effect that the overall image of a place has on the perception of its products’ quality, various studies have shown that this image also affects purchase intentions. In particular, it has a direct and positive influence on the final variable of our model (Van Ittersum et al., 2003; Srinivasan et al., 2004; Verlegh et al., 2005; Ahmed & d’Astous, 2007; Lee et al., 2013). On the basis of these results, we therefore posit the following hypothesis:

H₂: The better the overall image consumers have of a region, the greater their purchase intention towards its products.

Another of the conclusions drawn from the literature review was that the perception of quality of the products of a given origin directly and positively influences the purchase intention or preference towards products of that origin (González & Villanueva, 2001; Van Ittersum et al., 2003). More recently, Lee & Lockshin (2011) have shown that consumers’ beliefs about the wines of some given place are positively related with the purchase intentions. For this reason, we posit the following hypothesis:

H₃: The better the perception of quality consumers have of a region’s products, the greater their purchase intention towards them.

As mentioned above, there have been several studies that have examined the role the overall image of a place plays in the purchasing behaviour of consumers towards its products. However, according to Van Ittersum et al. (2003), when the study is focused on a specific product category, the predictive value of this variable is usually not high. They state that it is also necessary to measure the region’s image as a producer in that product category. In this regard, there have as yet been few studies that have included this variable. Nevertheless, we posit the following hypothesis:

H₄: The better the image consumers have of a region as producer of a specific product category, the better their perceptions of quality of that region’s products.

Van Ittersum et al. (2003) attempted to show that this variable has no direct influence on consumer preferences for the region’s products. They were unable to accept this hypothesis, however, since it was not confirmed in one of the two categories they studied. In the present study therefore, we posit the following hypothesis:

H₅: The better the image consumers have of a region as producer of a specific product category, the better their purchase intention towards the region’s products.

Furthermore, the aforecited authors consider that consumers’ overall attitude or image about a region can directly influence their image of it as a producer of a particular category of product. We therefore posit the following hypothesis:

H₆: The better the overall image consumers have of a region, the better their image of it as producer of a specific product category.

Finally, we included the consumer ethnocentrism variable in the model as numerous studies have found it to be positively related to both domestic purchasing behaviour and the perceived quality of domestic products (e.g., Luque et al., 2004; Dmitrovic et al., 2009). The hypotheses to be tested are:

H₇: Consumers’ ethnocentric tendencies are directly and positively related to the purchase intention towards their region’s products (domestic purchasing behaviour).

H₈: Consumers’ ethnocentric tendencies are directly and positively related to their perceptions of quality of their region’s products.

In addition to analysing the relationships between the variables of the structural model, a further objective was to determine the moderating role that a consumer’s familiarity with the product category can play in some of the relationships of the model. Studies such as those by Perrouty et al. (2006), Atkin & Johnson (2010), and Atkin & Newton (2012) have shown that consumers who are more expert and involved in the purchase of wine give greater importance to the ROO information in their purchasing process. Other studies that have included this variable in the general place-of-origin context are those of Insch & McBride (2004) and Srinivasan et al. (2004). In these cases, however, the results have been contradictory as to whether such a moderating effect exists, and if so, in which sense. Our working hypotheses were therefore (Fig. 1b):

H₉: Familiarity with the product category moderates the relationship between a region’s the image as producer of the product category and the perceived quality of the products from that region.

H₁₀: Familiarity with the product category moderates the relationship between the perceived quality of a region’s products and the purchase intention towards those products.

The same as with familiarity with the product category, consumers’ level of knowledge of an origin or of its products may play a moderating role in the relationships between the variables of the structural model (González & Villanueva, 2001; Lee & Lockshin, 2011; Gázquez et al., 2012). Therefore, our last two hypotheses were (Fig. 1b):

H₁₁: Familiarity with an origin’s products moderates the relationship between a region’s image as a producer of a specific product category and the perceived quality of that region’s products.

H₁₂: Familiarity with an origin’s products moderates the relationship between the perceived quality of that region’s products and the purchase intention towards those products.

To test the proposed relationship model, we applied the partial least squares (PLS) regression technique (Chin, 1998). All computations were done using the SmartPLS software package. The data came from a survey of regular or occasional purchasers of wine in the region of Extremadura (Spain). The study’s technical data sheet is given in Table 1. The sampling procedure was non-probabilistic, by gender and province of residence quota, with the aim of achieving a sample representative of the region. The final sample size was 427 individuals.

The questionnaire included various questions to measure each of the model’s variables. To measure the overall image that the respondents have of the region, we adapted the scale proposed by González & Villanueva (2001). This has five items with which to measure the image of a country: quality of life, wealth, technological level, education level, and political stability. We excluded this last item because in general it is not an element that differentiates regions of the same country. We added three more items: one relating to the attractiveness of the region as a tourist destination, another to the effectiveness of the region’s political management, and one to the region’s overall image. Thus, the final scale comprised seven items. The respondents scored their responses on a 7-point Likert scale.

To measure the image that the respondents had of the region as a wine producer, we took as referent the work of Van Ittersum et al. (2003). According to those authors, the image of a country or region as a producer of a product category has three dimensions: the human factor, the climate factor, and the natural environment factor. They therefore introduced items referring to the qualifications of the region’s workforce, to its people’s culture and experience, to its climate, and to its soil types, among others. For the purposes of the present study, we adapted some of those items to the specific case of wine. The final scale used comprised seven items.

The scale to measure the perceived quality of the region’s wines was based on the proposals of Klein et al. (1998) and González & Villanueva (2001). The former used four items, three of which were retained for the present study – those relating to overall quality, to the quality-price relationship, and to the respondents’ confidence regarding the quality of the region’s wines. The excluded item referred to the good workmanship of the product, an aspect associated with other industrial products, but not with wine. With respect to the scale proposed by Klein et al. (1998), this included items on the workforce, technological level, quality, reliability, design, and value in relation to the product’s price. Most of them do not apply to agro-food products, but the item regarding the quality of the region’s product in comparison with those produced in other regions was retained. Thus, the final scale used comprised four items.

To measure the respondents’ purchase intentions towards the region’s products, we took items from the scales proposed by Olsen et al. (1993), and Klein et al. (1998), eliminating those that did not apply to agro-food products. The final scale used comprised four items that the respondents scored on a 7-point Likert scale.

To measure the level of the respondents’ ethnocentrism as consumers, we chose to use the CETSCALE scale created by Shimp & Sharma (1987). This has been validated for different countries and contexts (e.g., Olsen et al., 1993; Klein et al., 1998; González & Villanueva, 2001; Luque et al., 2004; Srinivasan et al., 2004; Jiménez & San Martín, 2010; Dmitrovic et al., 2009). It consists of 17 items scored on a 7-point Likert scale. The minimum score for a respondent would thus be 17 points, and the maximum 119 points. For the present study, we adapted the items to refer to a regional level since the original scale referred to countries.

The respondents’ familiarity with the product category was measured by four questions about the wine-making process and types of wine. The respondents could choose one of the three options available: “true”, “false”, or “do not know”. With this system, we divided the respondents into two groups: high (expert) and low levels of knowledge. For a respondent to be considered an expert in the product category they had to answer three of the four questions correctly. It was felt that this system was more objective than evaluating the respondent’s knowledge about the products by means of a self-assessment question (Ahmed & d’Astous, 1999).

Finally, to measure the respondents’ familiarity with the products from their region, we included an open question in the questionnaire in which the respondents were asked to enumerate the region’s wine PDOs that they knew. Those respondents who recognized the Ribera del Guadiana PDO were considered to have a sufficient level of familiarity with wine from Extremadura.

To analyse the moderating effect of both types of familiarity, comparisons were made using the multi-group analysis (Chin, 2000). In this procedure, a Student’s t-test is calculated using Eq. [1] below. This equation derives from a Student’s t-distribution with m+n–2 degrees of freedom, where S_p (Eq. [2]) is an estimator of the pooled variance of the standard errors, m and n are the sample sizes of the two subgroups, and the SEs are the standard errors of the structural model’s standardized coefficient for each group.