WHAT PORTUGUESE CONSUMERS CARE ABOUT WHEN BUYING WINE. HOW IMPORTANT IS THE DESIGN OF THE BOTTLE?

the


INTRODUCTION
The extensive and heterogeneous competition existing in the wine market makes for a very complex consumer purchasing process. As a result, there are diverse ways of differentiating wine brands. Wine has many different attributes, both extrinsic and intrinsic, which can be considered by the consumer as criteria for choosing a brand. Some examples of intrinsic attributes are the grape variety, the age of the wine, the production process (organic or conventional) and the alcohol content. In terms of extrinsic attributes, the country or region of origin, brand, quality certifications and awards, price, and attributes related to the design of the bottle, such as the type of bottle, type of seal, design of the front and back labels and the information contained on them can be highlight.
In recent years, there has been a boom in new designs for wine bottles in traditional markets such as Portugal: different images on the labels, new colours, types of brand name, among others. This raises the question of the importance of design elements on wine packaging when purchasing wine. Wine packaging can be considered to consist of the following elements: a) the type of bottle; b) the seal and its system; c) the brand name appearing on the label; d) additional information on the label; e) the information appearing on the back label in text and graphic form.
As indicated by Ruivo (1996), the design of the packaging is the consumer's first point of contact when standing in front of shelves where hundreds of bottles with identical characteristics are on display. In this regard, the packaging and the label have, in terms of wine marketing, an undeniable influence and are a fundamental part of any wine consumption promotion or process (Thomas, 2000;Rocchi and Stefani, 2006).
Understanding wine purchasing criteria is essential if producers and distributors are to develop effective marketing strategies. For that reason, research on these purchasing criteria has gained prominence in recent years. In Portugal, despite being the 11 th ranked country in global wine production (OIV, 2017), studies about purchasing behaviour and the evaluation of wine purchasing criteria are relatively scarce (Panzone and Simões, 2009;Duarte et al., 2010;Freitas Santos and Cadima Ribeiro, 2012;Caldas and Rebelo, 2013;Madureira and Nunes, 2013;Silva et al., 2014;Drennan et al., 2015;Nunes et al., 2016). Moreover, if attention is focused on the importance of the product's visual design, there are still gaps in the knowledge due to the limited number of papers that analyse this and the lack of consistency in the results obtained. This paper aims to provide additional knowledge about the importance to purchasers of the design of wine packaging.

Background
Research on wine purchasing criteria has been extensive since the 1990s. Most studies allow to understand the importance for consumers of certain extrinsic attributes, such as price or origin, compared to some intrinsic attributes such as the grape variety or the age of the wine. However, as has been indicated, there are not many papers in this line of research that have included one or more attributes related to the design of the wine packaging in their analysis. Below, some of these studies are briefly discussed. Perrouty et al. (2006) conducted a study in France, Austria, Germany and the United Kingdom, with a sample of 1,162 consumers. The results showed that expert consumers valued, in order, the region of origin, bottle design and, finally, the price. For non-expert consumers the order of importance was different, with the region of origin again being the first attribute, followed by price and the bottle design. These results are an example of how the importance of each purchasing criterion varies depending on the level of knowledge of the consumer. Lockshin et al. (2009) conducted a study in Australia on 1,233 wine consumers with the aim of assessing the impact of a set of attributes on their purchasing decision. Among the attributes analysed, various were related to the design of the wine bottle (style of the label, colour of the label, material, sealing system of the bottle and reference to awards and medals on the label), as well as the brand, country and region of origin, price, price discounts and alcohol content. Their results showed that attributes related to the design of the bottle had little weight in the purchasing decision of consumers, with brand and price being the most important attributes.
Along a similar line, Mueller and Szolnoky (2010a) concluded, in a study conducted in the U.S., that the origin of the wine is more important in consumer purchasing decisions than attributes related to the design of the bottle, such as its style, information on or colour of the label, type of bottle or type of seal.
In contrast to previous studies, Jarvis et al. (2010) conducted a study in Australia on a sample of 200 university students that led them to conclude that the symbols or images included on the label were the attributes that most influenced the purchasing decision and the attribute of the least importance was the origin of the wine. These results, which contradict those found in previous studies, could be due to the different purchasing and consumption behaviour of young consumers. Krystallis and Chrysochou (2010) carried out a study in Greece with 567 individuals. Their results allowed them to rank certain attributes in order of importance, with the first of these being price, followed by quality, packaging, previous experience, grape variety, region of origin and, finally, brand. Similar results were found in the study by Chrysochou et al. (2012).
In another study conducted on 521 consumers in Germany, Mueller and Szolnoky (2010b) concluded that the most important criterion was prior information about the wine, followed by price, packaging and region of origin, among other less important attributes. The study also demonstrated the existence of segments of consumers who place a different level of importance on the attributes depending on their frequency of consumption, level of knowledge about the product and place they purchased the wine.  analysed the importance of certain wine purchasing criteria in the Australian market using two statistical techniques: the best-worst scaling (BWS) method and a discreet choice experiment. In the first, the respondents directly valued the attributes without visual support, and they found that brand was the most valued attribute, followed by price and region. The visual elements and design of the bottle and label were clearly the least valued. In contrast, in the second study, consumers placed greater importance on the style of the label, followed by the brand, in their purchasing decision. The colour of the label had a greater influence than the region and type of bottle. The contrast between the two studies reveals the effect that images have on the opinions expressed by the respondents. These authors also found that there were no significant differences according to the frequency of consumption or the interest in and level of knowledge about wine. In contrast, they did find differences in the valuation of purchasing criteria according to age and gender.  analysed 331 regular wine consumers in Australia in order to identify the importance to them of 10 different types of information that may appear on the label of a bottle, as well as the price. The results revealed that these elements on the label have very little influence on the buyer's decision when compared to price. The study identified various segments of consumers, but the socio-demographic variables of gender, age and income level were not statistically significant. Bernabéu et al. (2012) conducted a study in Spain with the participation of 421 wine consumers. Their results showed that the order of importance of the different attributes is as follows: previous experience, region of origin, price and recommendations from family and friends. The study demonstrated that the other seven attributes analysed (grape variety, age, brand, alcohol content, design of the bottle and label, gastronomic combination and organic production) were less important to consumers. The analysis found significant differences between segments based on certain variables such as gender, age, income level and education level.
Elliot and Barth (2012) conducted a study on 404 young Canadians and, in contrast, concluded that the most highly valued attributes when selecting a wine were those related to the design of the bottle. Other attributes such as the type of wine, alcohol content, country of origin, producer or year of harvest were classified as less important. It was again found that the age of the consumer may significantly influence the purchasing criteria used.
In the Portuguese market, Madureira and Nunes (2013) interviewed 171 consumers to understand the importance they placed on the one hand on a set of intrinsic attributes, and on the other, on a set of extrinsic attributes. Among these, the most important by far was the region of origin. In terms of the elements of the packaging design, the most important was the design of the label, this being more important than the information about the producer shown on the label or the awards received. However, the results of this study also showed that the importance of the purchasing criteria varies according to the consumer's frequency of consumption and level of knowledge.
In Australia, MacDonald et al. (2013) interviewed 1,229 consumers in order to analyse ten purchasing criteria that included extrinsic and intrinsic attributes. Their results showed that the most important attribute was taste, followed by price, brand and aroma. The information on the back label and the appearance of the bottle were in seventh and eighth position in the rankings, respectively.
Lastly, Nunes et al. (2016) analysed the wine purchasing preferences of 250 Portuguese consumers, using the BWS method. The results showed that having tried the wine previously was the criterion that most influenced the purchasing decision. Region of origin and recommendations from friends were also relevant. Alcohol content and the characteristics of the bottle were the attributes with the least importance. The study identified significant differences depending on the age and gender of the respondents.
The conclusion obtained from this literature review is evident: the results demonstrated by the different studies are heterogeneous, varying according to the country where the study was conducted and the type of wine consumer (frequency of consumption, level of knowledge, socio-demographic characteristics, among others). Within this heterogeneity, it can be argued that in many cases origin has been identified as the main purchasing criterion, even ahead of price. Intrinsic attributes such as the grape variety, age of the wine, alcohol content and production system (organic or not) are less important. The attributes related to the design of the bottle show contradictory results. Some authors such as Elliot and Barth (2012) and Jarvis et al. (2010) defend the relative importance of design compared to other attributes. However, in general terms it can be argued that most of the studies have concluded that attributes related to the design of the wine bottle are less important in the purchasing decision than other extrinsic attributes (Lockshin et al., 2009;Mueller and Szolnoky 2010b;Chrysochou et al., 2012;Madureira and Nunes, 2013;Nunes et al., 2016).

Objectives and research design
The main objective of the research conducted is to determine the most important wine purchasing criteria for Portuguese consumers, with a special focus on the design of the bottle. Specifically, the objective is to answer the following three research questions: To achieve the proposed objective, a survey was conducted on regular and occasional wine consumers over the age of 18 and resident in Portugal (Table I).

29
The questionnaire included a block of questions about the respondent's purchasing and consumption of wine. Specifically, they were asked about the usual place of consumption (at home, outside the home or both), the frequency of consumption, the frequency of purchase, the usual establishment of purchase (specialised store, supermarket, among others) and the level of knowledge about wine. This last question was designed in a self-assessment format, so individuals had to decide whether they considered themselves to have wine knowledge that is superior to, similar to or inferior to that of the average consumer.
Moreover, the respondents had to decide the importance they place on 21 attributes in their decision to purchase a bottle of wine. The valuation was carried out using a scale from 1 (not at all important) to 7 (extremely important).
The questionnaire was subject to a pre-test among 15 academics and professionals linked to the wine sector, in order to identify any possible research errors in its design and implementation. Table IV contains the average scores awarded by the respondents to the 21 wine purchasing criteria. This allows to respond to the first two research questions raised in the objectives. As can be seen, there are very big differences in the importance placed on the different criteria, with taste being the most important for the sample as a whole, with an average score of 6 on a scale from 1 to 7. The least valued attribute is the colour of the bottle, with an average of 3.4. Although two intrinsic attributes are the two most important (taste and aroma), it can be seen that the intrinsic and extrinsic attributes alternate in the order of importance for wine purchasing criteria. The attributes related to price and the origin of the wine are among the most important, but other intrinsic attributes are considered little in the purchasing decision, such as the grape variety, colour or alcohol content. In any case, the attributes related to the design of the wine bottle are of little importance in the purchasing decision. Only the type of cap or seal of the bottle and the information contained on the label are of medium importance among the set of criteria used in the study.  (Table V) which explain 66.76% of the variance. The KMO value is 0.846, which is higher than the 0.7 reference value that guarantees the validity of the analysis. Moreover, the reliability of the scale is also guaranteed due to obtaining a Cronbach's Alpha value of 0.873. The first factor contains most of the intrinsic attributes: colour of the wine, age or aging of the wine, grape variety, organic production, alcohol volume/content and information on the label about the characteristics of the wine. For this reason, it is called Organoleptic characteristics. Factor 2 is formed by the following attributes: country of production, region of production, reference to the designation of origin and brand or winery. This factor brings together extrinsic attributes and is called Origin.

RESULTS AND DISCUSSION
Factor 3 contains the attributes related to design: shape of the bottle, colour of the bottle and attractiveness of the bottle's label, so this is called Design. Factor 4 includes the taste and aroma of the wine, although there is also an element of the bottle's design: the type of seal. We have chosen to call this factor Taste and Aroma.
Factor 5 covers the following attributes: winning of an award at a national or international fair, recommendations from friends and acquaintances and recommendations from sellers. This factor brings together extrinsic factors and is called Recommendation. Finally, Factor 6 is formed by only two attributes: price and promotions and offers, so it has been called Price.
After the completion of the factorial analysis, we carried out segmentation through a Cluster K-Media analysis, in which the researcher determines, a priori, how many segments they want to have, or in other words, the K value to be defined. This study analysed the dendrogram to detect the most suitable K value. After a test with K=4, various segments were obtained that were too small and had few differences in terms of their purchasing behaviour. For this reason, we finally opted for a value of K=3. The results are shown in Table VI.
These three segments of purchasers have different purchasing behaviours for all the factors analysed, except for Factor 1. According to the ANOVA analysis (Table VII), the difference in means between the three segments is statistically significant at the 1% level, except for organoleptic characteristics.
Segment 1, which represents 37.5% of the sample, stands out for not placing much weight on price as a factor and for attributing greater importance to the taste and aroma factor. Origin is fairly important in their purchasing decision. This segment can be classified as price insensitive purchasers.

TABLE VII
ANOVA results

Resultados da ANOVA
Segment 2 does not place importance on the recommendations factor, nor on origin or design. It most values the price factor. They represent 14.4% of the sample and can be classified as price sensitive purchasers, as they seem to base their purchasing decision mainly on this variable.
Finally, segment 3 is the largest (48.1% of the sample) and is characterised by placing less importance on origin as a factor, but a lot of weight on the price factor. It is formed by individuals who are price sensitive but positively value design, taste and aroma, and recommendations. This segment can be classified as value for money purchasers. This is the segment in which design elements have the greatest importance in the purchasing decision.
Once these segments had been identified, the sociodemographic profile and the purchasing and consumption behaviour of the individuals forming part of each segment are described. The variables for which significant differences were found were the level of knowledge about wine, the frequency of consumption, the place of consumption and the level of monthly household income. No significant differences were found between the segments for the usual shopping establishment, gender and age variables.
With regard to the level of knowledge, Pearson's Chi-Squared established statistically significant differences at 1% ( In terms of the frequency of consumption, Pearson's Chi-Squared establishes statistically significant differences at 1% (Table IX). The price insensitive purchasers segment is especially formed by individuals with a high frequency of wine consumption (40.9%). The price sensitive purchasers segment is formed by individuals with a low frequency of consumption (57.1%). And, again, the value for money purchasers segment contains individuals with heterogeneous characteristics.  In terms of the place of consumption (Table X), the price insensitive purchasers' segment and the value for money purchasers segment are characterised by consuming wine equally at home and outside of the home (50.6% and 45.7% respectively). In contrast, the price sensitive purchasers segment has the habit of mostly consuming wine outside of the home.
Pearson's Chi-Squared establishes significant differences between the segments at 5%. Finally, with regard to the level of household income, Pearson's Chi-Squared showed significant differences between the three segments at 1% (Table XI). The price insensitive purchasers segment is mainly formed (64%) by consumers with high incomes.
The price sensitive purchasers segment is mainly formed (58.7%) by consumers with low incomes.
Finally, the value for money purchasers segment once again manifests itself as the most heterogeneous.  is purchased. In this regard, it was found that the attributes related to the design of the packaging are in the group of less important criteria, with the shape and colour of the bottle being of very little importance. These results are in line with those obtained by some previous studies, such as those of , Bernabéu et al. (2012), Madureira and Nunes (2013), MacDonald et al. (2013) and Nunes et al. (2016).
It was also identified three different segments based on their purchasing behaviour: a price insensitive segment, another made up of price sensitive purchasers and a final segment that contains purchasers who seek value for money. This final segment places more importance on the attributes related to the design of the packaging. The profile of these three segments is very different, showing significant differences related to the level of knowledge about wine, the frequency of consumption, the place of consumption and income level. These results are similar to those obtained by Madureira and Nunes (2013) in terms of the influence of the frequency of consumption, but are contrary to those obtained by Nunes et al. (2016), who indicated that this variable is insignificant. With regard to the study by Nunes et al. (2016), the results presented here are also contrary to those found for the influence of the gender, age and education level variables, which have been found to be insignificant in this study.
Taking into account that Portugal produces almost 7 million hectolitres per year (OIV, 2017) and is the 5 th ranked country in terms of wine consumption, consuming around 4.8 million hectolitres in 2016 (IVV, 2018), we believe that the information deriving from the results of this research can guide the design of the marketing strategies of the Portuguese wine industry and of the managers of the 24 Protected Designations of Origin existing in the country. According to the research results, the target group when launching wine brands with new and modern designs seems to be the middle segment of consumers. These are the consumers who are looking for value for money, who have an average level of knowledge about wine, an average frequency of consumption (neither regular nor occasional) and who usually purchase it for consumption at home (either because they prefer to consume wine at home or because they do so equally at home and outside of it).
The literature review revealed that wine purchasing behaviour is not universal and differs significantly according to the country, especially when it comes to old wine world (for example, Portugal or Spain) and new wine world (for example, Australia or the United States). For this reason, these results cannot be extrapolated and we believe that new research analysing the importance of the design of wine bottles must be conducted in other geographic markets. Moreover, as a complementary line of research, studies should be carried out to identify which types of wine bottle design (type of bottle, colours, images) are most attractive to different types of consumers.