Choosing a lifestyle? Reflection of consumer extrinsic product preferences and views on important wine characteristics in Germany

Risius, Antje; Klann, Björn-Ole; Meyerding, Stephan G. H. Article Choosing a lifestyle? Reflection of consumer extrinsic product preferences and views on important wine characteristics in Germany Wine Economics and Policy Provided in Cooperation with: UniCeSV - Centro Universitario di Ricerca per lo Sviluppo Competitivo del Settore Vitivinicolo, University of Florence Suggested Citation: Risius, Antje; Klann, Björn-Ole; Meyerding, Stephan G. H. (2019) : Choosing a lifestyle? Reflection of consumer extrinsic product preferences and views on important wine characteristics in Germany, Wine Economics and Policy, ISSN 2212-9774, Elsevier, Amsterdam, Vol. 8, Iss. 2, pp. 141-154, https://doi.org/10.1016/j.wep.2019.09.001 This Version is available at: https://hdl.handle.net/10419/284481 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/4.0/ Choosing a lifestyle? Reflection of consumer extrinsic product preferences and views on important wine characteristics in Germany Antje Risius, Bj€ orn-Ole Klann, Stephan G.H. Meyerding * Received 7 September 2018; revised 18 August 2019; accepted 4 September 2019 Available online 10 October 2019 Abstract Wine is a product, in which heterogeneous facets are marketed abundantly. Quality selections like origin, ratings, sustainable productions are (increasingly) important to the sector. Little is known about how and when the common quality scale interferes with other quality attributes and how lifestyle factors determine changing preferences. Hence, the study investigates which attributes are preferred by consumers and how they attach to lifestyle and consumption habits. The consumer survey (N¼962) itself used a direct (questionnaire) and an indirect rating technique (choice experiment) to elicit preferences for selective attributes for wine and the attachment to a wine-related lifestyle. Most important factors in the wine-related lifestyle approach were used to describe different consumer segments derived from latent class analysis based on choice preference. Latent class unveiled five different consumer segments. With regard to the importance of other extrinsic attributes (e.g. organic production) growing societal demand for ethical consumption, ‘organic’, ‘medals and awards’seem not to (yet) be of high relevance for the product wine. The geographical site of production is a quality attribution of high importance to consumers. Nonetheless the largest consumer segment found, was considered to specifically look for further information. It seems of relevance to understand information-seekers behavior better and the nature of information, they would like to get. ©2019 UniCeSV University of Florence. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Keywords: Wine-related lifestyle; Wine quality; Conjoint analysis; Consumer marketing; Wine preference 1. Introduction In nowadays, wine is one of the most heterogeneous products on the food market (Lockshin and Hall, 2003). This not only holds true to the marketing of the established quality scheme in Germany and Austria (‘Pr€ adikatsweine’), but also with regard to brands, stories, trust and renown-producers (Bernetti et al., 2006; Lockshin and Corsi, 2012; Oczkowski, 2001; Perrouty, d’Hauteville and Lockshin, 2006). The grape orientated quality scale that has been worked on since 1971 (Deutsches Weininstitut, 2015) and is backed up by law and enforced by state agencies (Diaz-Bone, 2005). This system focuses on the grape quality. Special requirements apply for the premium wines (‘Pr€ adikatsweine’)(Deutsches Weininstitut, 2015). This classification system, started to be established in markets in the 80ies, succeeded to place value and high acceptance to a differentiated quality regarding grape varieties and taste eso to say ‘chemical properties’of the wine. This approach reversed a development of the sector towards low, but abundant quality and gave reason to base a marketing differentiated in selective quality attributions. The selection of a wine with simultaneous consideration of a multitude of attributes requires knowledge on the consumer side. However, only a few consumers have this knowledge of wine. In particular, food retailers do not create a remedy for this knowledge deficit, since wine is purchased completely anonymously (Petzoldt et al., 2007). Regarding the selection of groceries, consumers with comparatively low product knowledge or little experience with the product in particular, pay more attention to extrinsic characteristics (e.g., price, origin). These characteristics are used as an indication of the *Corresponding author. E-mail address: [email protected] (S.G.H. Meyerding). Peer review under responsibility of UniCeSV, University of Florence. HOSTED BY Available online at www.sciencedirect.com ScienceDirect Wine Economics and Policy 8 (2019) 141e154 www.elsevier.com/locate/wep https://doi.org/10.1016/j.wep.2019.09.001 2212-9774/©2019 UniCeSV University of Florence. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). quality of intrinsic attributes such as taste. In terms of wine, empirical research has revealed that the most important key information consumers determine during the decision-making process for the purchase is the origin and price of a wine. Country-specific differences apply to this knowledge (Petzoldt et al., 2007). Considering market numbers, wine is bought predominantly in food retail stores without professional advice. Moreso, consumers tend to reinforce extrinsic features rather than assessing the intrinsic quality of the product (Lockshin and Corsi, 2012). This is especially true nowadays, since sustainable product attributes like ethical production or organic production are getting more and more relevant to society (Valor et al., 2014; Lockshin and Corsi, 2012; Zander and Hamm, 2010). As such, different extrinsic quality regimes have been established. One alternative quality classification system is more related to the ‘terroir’concept, which has its origin in France and uses the AOC (appellation d’origine contr^ ol ee: protected designation of origin) classification system. Some winemakers united to found the VDP Verband deutscher Pr€ adikatsweingu ¨ter: Association of German Premium Wineries) to establish their own quality system. This system focuses on local production including soil and the microclimate of the vineyard, as well as on the craftsmanship of the winemaker (R€ ossel and Beckert, 2013). The wine’s region of origin has a high impact on the purchase decision when the consumer has a considerable wine knowledge (Bernetti et al., 2006; Perrouty, d’Hauteville and Lockshin, 2006). Nonetheless, climate change challenges the wine production as it changes production regions according to the production site. Moreso lifestyle changes and societal demand for ethical question pose the question, of whether extrinsic sustainable quality attributions are of higher importance than the ‘terroir’or/and the quality oriented classification system. This paper aims to analyze the importance of other extrinsic attributes (e.g. organic production) against the wellestablished quality classification system based on grape variety and sensory attributes of taste as well as the local place of production (‘terroir’) in reflection of changing societal demands and consumption habits for wine. 2. Methods To segment German wine market consumers, the study applied a choice-based conjoint analysis and clustered consumers by similar preferences using a latent class analysis. To describe the segments, the wine-related-lifestyle approach (Bruwer et al., 2002) was used. The following describes the methodological design and describes how the survey has been conducted. Design of the empirical experimental approach ‘The winerelated lifestyle approach’is an instrument introduced by Bruwer et al. (2002). The questions and attributes are based on the food-related lifestyle approach from Grunert et al. (1997). It links consumer’s lifestyle features, based on values for life, to product selections and consumption of wine (Grunert et al., 1997). The 80-item questionnaire measures five lifestyle dimensions to probe information on desired quality or attributes, ways of shopping, wine consumption situation, wine drinking rituals, and desired consequences of wine consumption. All items are rated on a seven-point Likert-type scale, ranging from ‘totally disagree’to ‘totally agree’(Bruwer and Li, 2017). The 80 items (statements) were adjusted and translated to fit the German wine market. The original scale was reduced to 60 items: statements concerning place of consumption and other questions concerning membership of a wine club were removed. Preferences for wine attributes were assessed both directly and indirectly. Directly, both, consumers, were asked to rate 14 different attributes of a wine on a seven-point Likert-type scale (1 ¼not important at all; 7 ¼very important). The attributes were based on previous studies (Barber and Almanza, 2006; Mueller et al., 2010; Rocchi, 2006; S aenz-Navajas et al., 2013). The reason for using this approach was to investigate if the results might differ from those seen in the choice experiment: choice experiments might be less biased by socially desirable behavior in answering (Lusk and Schroeder, 2004; Meyerding, 2016; Meyerding and Merz, 2018). Also, the participants were directly asked about their willingness to pay (WTP) for a bottle of German wine for four different occasions: WTP for good quality; WTP for an unknown occasion; WTP for a business occasion; and WTP for a celebrity occasion. Special occasions are considered because both groups of participants, consumers and wine producers, might tend to spend more on wine for social events like birthdays, weddings, family celebrations, and other important occasions (Thiene et al., 2013). The final part of the survey was a choice-based conjoint analysis (‘choice experiment’). In this indirect survey technique, participants were asked to choose from a set of product alternatives as if they were choosing in a supermarket (ref. to Fig. 1). Every product is a combination of product features from a defined set of attributes (Desarbo et al., 1995). Hence, this indirect survey techniques allows the indirect, nonetheless systematic, measurement of preferences as a reflection of the direct ratings described above. In this study this included the measurement of preferences for the attributes Brand, Origin, Sweetness, Price, Awards, Quality Level, Grape variety and Production quality (ref to Table 1). As such consumers could select one out of a given set of wine with selected product attributes or a No-Buy-Option. We have chosen five different brands which represent a variety of prices and quality levels in the German wine market. By using conditional pricing for each brand, we aimed to ensure realistic price levels. The brand Rotk€ appchen is one of the most famous in Germany when it comes to sparkling wine; the company also produces different kinds of wine. We selected this particular brand, which has a reputation as a relatively low priced robust wine with a medium quality, because consumers directly recognize the brand when looking at its logo. They might have the perception of a relatively good price-performance ratio, even if the performance is rather low. At the other end of the spectrum, Robert Weil is a premium wine producer with a vinery with the highest seal in terms of quality in Germany (VDP). 142 A. Risius et al. / Wine Economics and Policy 8 (2019) 141e154 His wine has a high price but the products are known for their excellent quality among connoisseurs (Wiegelmann, 2017). The next wine producer, Lukas Kesselring, is an ambitious winemaker from Palatinate, which is one of the most important wine regions in Germany. He values not only quality but also organic products. His wines can be purchased in wine retail stores or other grocery shops with a good selection. The products’ prices vary between five and ten euros per bottle, which is a moderate price for a high-quality wine (Wiegelmann, 2017). The Zimmermann Graeff &Mu ¨ller Weinkellerei: Genossenschaft buys grapes from many different regions and winemakers with the goal of producing an extremely cheap wine for discount stores and their pricesensitive customers. In return, the quality is relatively low compared to the other products in the sample (Sautter et al., 2017). Finally, the Lemberger Land Genossenschaft is an association that highly values quality. Thus, this producer offers wine with a better quality in the price range of four to eight euros per bottle (Sautter et al., 2017). Overall, we set a base price for each brand with a variation of ±25 percent. By choosing different levels for each attribute, we attempted to offer a wide range of the products available on the market. The combination of these attributes and their different levels resulted in 45,000 different choice sets. Of these, ten choice sets were arbitral chosen based on a d-efficient design structure established by the software Sawtooth 9.5.2. In each of the choice sets, each respondent was asked to evaluate which wine they would buy based on the five product profiles (example in Fig. 1). By analyzing the respondents’ choice for this selective set of choice sets, it is possible to draw inferences concerning preferences for attributes and use the estimated preference structure to evaluate scenarios of interest (Hensher et al., 1998). Within the analysis, respondents’ choices are converted into ‘utilities’for each level of attributes by using multinomial logit regression (Louviere et al., 2010). The main goal of a CBCA is to identify a product’s key attributes. We only used German wines in our choice experiment because one focus of this survey was on the different Fig. 1. Choice set of the online survey. Table 1 Attributes and their Levels used in the CBCA. Attribute Attribute levels Attribute Attribute levels Brand Rotk€ appchen Awards AWC bronze Robert Weil AWC gold Lukas Kesselring DLG gold ZimmermannGraeff &Mu ¨ller Weinkellerei None Lembergerland Kellerei Quality Level VDP.Gutwein Origin Palatinate VDP.Erste Lage Rheinhessen Auslese Rheingau Qualit€ atswein Wu ¨rttemberg Kabinett Nahe Grape variety Silvaner (white) Sweetness Dry Sp€ atburgunder (red) Medium dry Riesling (white) Sweet Dornfelder (red) Price Lower price Protugieser Weibherbst (ros e) Average price Production quality Organic Higher price Non-organic 143A. Risius et al. / Wine Economics and Policy 8 (2019) 141e154 (extrinsic) quality levels that only apply to wines produced and sold in Germany. We intended to select wines with different features that have a high or low market share and from various subcategories to examine the attributes used by the consumer in the selection process. 2.1. Data analysis The following section describes the data processing with regard to the wine-related lifestyle measurement, at first, and with regard to choice experiment and latent class analysis in the later. In order to reduce the WRL measurement to most relevant factors, we conducted an exploratory factor analysis. First, we checked the data by performing the general acceptance test. For that, we have chosen the Kaiser-Meyer-Olkin measure of sampling adequacy and the Bartlett’s test of sphericity. Table 2 shows the results of the test, which indicate that the data are suitable for factor analysis. The goal of factor analysis is to reduce the number of items by finding correlations between them and abstracting a smaller number of factors. Because there are no predefined dimensions, explorative factor analysis was chosen (Wedel and Kamakura, 2000). Determining number of factors. We used Kaiser’s criterion (Kaiser, 1958) to determine factors with a eigenvalue greater than 1.0. In total, eleven factors were identified. Ledesma and Valero-Mora (2007) suggest the use of Horn’s parallel analysis (PA) instead of Kaiser’s criterion because the latter tends to over-extract the number of factors (Ledesma and Valero-Mora, 2007). Horn’s PA compares the shown eigenvalues with those from randomly generated data from a Monte Carlo-based simulation. Applying this technique resulted insix factors identified and shown in Table 3. Determining descriptors of the factors. For determining the factors, a varimax normalized factor loading method was used. Every item was given a value for each factor. A high value predicts membership of the item to a factor. Table 7 shows the six factors. The six factors explained 43.04 percent of the variance of the data. The mean values, standard deviations, and factor loadings of all items are presented in Appendix 1. Latent class (LC) models are techniques for investigating unobservable classes that differ in the utility attached to particular product attributes between the classes but are more similar within (Kamakura and Russell, 1989). The software Lighthouse Studio 9.5.2, which was used in the present study, predicts part-worth utilities for every attribute level and assigns each respondent a probability of membership for each of the latent classes (Sawtooth, 2004). The LC model is based on the random utility framework with the following function: Uni=C¼bcXin þεni=C As such, the utility of the n th respondent is estimated for a particular class cfrom selecting an alternative ifrom the available five choice options. S is a linear combination of attributes of the part-worth’s b c , and an error term. X ni is depicting the choicespecific product attributes. Heterogeneity is captured by different latent classes, so each class cgets its own utility parameter vector b c . It is assumed that the errors εni are IID and follow a Type I distribution. That provides them with the probabilistic response function (Mueller et al., 2010): pni=CðiÞ¼elcðbcXinÞ,X jεS elcðbcXjnÞ Other authors found that results of a latent class analysis are preferable to segmenting solutions in the case of a choice experiment (Desarbo et al., 1995; Moore et al., 1998). Table 2 Measure of sample accuracy according to Kaiser-Meyer-Olkin. Test Acceptable result Actual result Status KMO measure of sampling adequacy >0.6 0,9 Acceptable Bartlett-test of sphericity p <0.05 0,0 Acceptable Degrees of freedom 1770,0 Approximate x 2 20338,3 Table 3 Description factors of the wine-related-lifestyle. Factor Description Eigenvalue Individual % Cumulative % Factor 1 Savor and looking for new experience 12,62 21,03 21,03 Factor 2 Wine knowledge and connoisseur tendencies 4,10 6,83 27,86 Factor 3 Information procurement for the purchase 3,40 5,67 33,54 Factor 4 Purchase decisions 2,60 4,33 37,87 Factor 5 Reason for wine consumption 1,66 2,77 40,64 Factor 6 Price consciousness 1,44 2,40 43,04 Table 4 Model selection for latent class segmentation. Classes Log-likelihood AIC CAIC BIC Average max. membership 220060,65 40227,29 40674,94 40621,94 0,96 319314,10 38788,21 39463,90 39383,90 0,95 418832,72 37879,44 38783,18 38676,18 0,94 5¡18360,70 36989,39 38121,18 37987,18 0,94 618016,43 36354,86 37714,70 37553,70 0,93 717812,33 36000,66 37588,54 37400,54 0,92 817670,59 35771,19 37587,12 37372,12 0,92 144 A. Risius et al. / Wine Economics and Policy 8 (2019) 141e154 To date, there are no distinct statistical information criteria by which to decide over the number of latent segments. In the present study, a multiple criteria approach was chosen, using the Akaike information criterion (AIC, Akaike, 1987), the consistent Akaike criterion (CAIC, Bozdogan, 1987), and the Bayesain information criterion (BIC, Schwarz, 1978), in order to achieve the optimal number of segments (Nylund et al., 2007). Nylund et al. (2007) suggest that BIC and CAIC are superior to the AIC when it comes to choosing the most appropriate number of segments. Nevertheless, it is not only statistical information that should be taken in account. Other decision factors are interpretability and group size are also of considerable importance (Sawtooth, 2004). We used BIC and CAIC with the known limitations to make our decision. As can be seen in Table 4, the CAIC and the BIC decrease until class 5 and then stay fairly consistent at higher class numbers. Hence the relative gain in information of more consumer segments over increased predictive power gets smaller. Thus, we committed to the five-group solution. 2.2. Survey assessment An online survey was conducted in November 2017 with 962 participating wine consumers. The survey was created using Lighthouse Studio 9.5.2 software from Sawtooth. At the beginning, the participants were asked to provide some socioeconomic information and were screened for the wine consumption (only wine consumers were included). Then they were asked about their consumption habits. The participants were asked if they were a winemaker by profession. Consumers answered the wine-related lifestyle (WRL) items firstly. Following, the WRL statements were direct questions about wine preferences concerning color, sweetness, consumption occasions, and favorite country of origin, as well as the place of purchase and the frequency of visit. These data helped to underline the segments with a foundation of consumer information. After that, the participants were asked to rate (on a seven-point Likert-scale) their wine knowledge and their involvement level when buying wine. Following, their willingness-to-pay (WTP) for a bottle of German wine for different occasions was ascertained. The last part of the survey was the choice experiment (see experimental design above). The recruitment took place via market research institute, which holds a respondent’panel. Quotes for sampling were set for gender, age, and gross household income. Age as an indicator delivers insights about purchasing and consumption habits that change at certain life stages (Hall, Binney, &Barry O’Mahony, 2004; Thach and Olsen, 2006). The quotes for the sampling were based on the latest numbers of the Federal Statistical Office in Germany (Statistisches Bundesamt, 2017). Table 5 Sociodemographic characteristics of the consumer sample (N¼962). Characteristics Description Frequency Share (%) sample Share (%) Germany* Age (years) 16e19 37 3,7 4,8 20e29 118 11,8 14,1 30e39 147 14,6 14,5 40e49 176 17,5 15,7 50e59 176 17,5 18,7 60e69 123 12,3 13,9 70 þ227 22,6 18,9 Gender Female 477 47,5 50,7 Male 527 52,5 49,3 Income <1,300V158 15,7 18,2 1,300V2,600V305 30,4 31,6 2,600V3,600V209 20,8 19,4 3,600V5,000V174 17,3 16,8 >5,000V158 15,7 14,0 Marital status Single 316 31,5 24,4 Married 492 49,0 61,2 Divorced 121 12,1 8,2 Widow/er 64 6,4 6,3 Prefer not to say 11 1,1 n.A. Persons in household 1 243 24,2 23,8 2 459 45,7 38,1 3 160 15,9 17,2 4 106 10,6 15,6 5 36 2,7 5,3 Education Apprenticeship 339 33,8 47,2 Technical college degree 296 29,5 7,8 Technical college former GDR 34 3,4 1,0 Bachelor’s degree 69 6,9 1,9 Master’s degree 51 5,1 1,2 Diploma 125 12,5 12,8 Doctorate 29 2,9 1,2 No finished education level 61 6,1 26,9 Note. * Reference: German Federal Statistical Office. 145A. Risius et al. / Wine Economics and Policy 8 (2019) 141e154 Gross household income is important for analyzing a rough budget consumers can spend on wine. The minimum age to participate in the survey was 16, which is in line with the minimum age to purchase wine in Germany by law. The questionnaire was designed and distributed in German language. 3. Results In this results section, the sample is first described. After that, the results of the direct questions and the part-worth utilities of the product characteristic attributes resulting from the choice-based conjoint analysis, using a hierarchical bias mechanism, of consumers are presented. Thereafter, the latent segments are identified using latent class analysis. In a last subsection, the results of the wine-related lifestyle approach are presented to describe the different consumer segments and the willingness-to-pay for a bottle of wine for different occasions is shown. 3.1. Sample description The demographics of the consumer sample are summarized in Table 5. The consumer sample included 962 participants aged over 16. A little more than half of the sample is male, and the average age of the respondents is 51.7 years. The education level of the participants is higher than of the general population, which is a common issue in online surveys (Granello and Fig. 2. Characteristics’ importance for wine choice. Note. Scale from 1 (not important at all) to 7 (very important). Consumers (N¼962). Fig. 3. Part-worth utilities as a result of the CBCA for consumers (N¼962) Note. Attribute importance from left to right: producer 24.37%, origin 7.26%, grape variety 20.64%, medals and awards 5.91%, quality level 6.64%, sweetness 26.22%, price 5.03%, organic 3.92%. 146 A. Risius et al. / Wine Economics and Policy 8 (2019) 141e154 Wheaton, 2004). The consumer sample is representative in terms of age, gender, and income. 3.2. Importance of characteristics for wine (direct rating) Fig. 2 presents the importance ascribed to characteristics when choosing a wine, using the mean value for each item. Sweetness is the most important criteria for consumers, when it comes to selecting wine, followed by the official quality level, grape variety and region of production. Medals and awards, along with VDP quality level, are only of medium interest, while the least important factors for the consumer are bottle shape and vineyard. 3.3. Preference of Characteristics for Wine (Indirect Choice Rating) Fig. 3 shows consumer preference for the selective wine attributes measured by part-worth utilities. Thus, the columns represent the respective part-worth of each attribute level in relation to a chosen reference. The higher the resulting partworth value, the greater the benefit for the consumer and the greater the likelihood that he will purchase a product with this characteristic (attribute level) when shopping for wine. The relative attributes’importances are given in the notes under Fig. 3. Table 6 Part-worth utilities (rescaled) of the different attribute levels for consumer segments (convergence limit for log-likelihood: 0.01). Attribute Level Segment 1 (15.9%) Segment 2 (18.5%) Segment 3 (35.2%) Segment 4 (14.5%) Segment 5 (15.8) Brand Rotk€ appchen 64 32 27 95 89 Robert Weil 48 51 70 105 146 Lukas Kesselring 7 30 55 61 102 Zimmermann-Graeff &Mu ¨ller Weinkellerei 51 0 22 72 121 Lembergerland Kellerei 55 54 76 039 Region Palatinate 7 2 11 11 23 Rhinehessen 0 3 28 11 33 Rhinegau 14 10 122 24 Wu ¨rttenberg 716 13 20 34 Nahe 15 11 24 19 20 Grape variety Silvaner (white) 11 193 49 514 Sp€ atburgunder (red) 14 219 78 11 73 Riesling (white) 30 190 98 20 49 Dornfelder (red) 28 227 116 15 4 Protugieser Weibherbst (ros e) 27 64 46 11 42 Medals and awards AWC Bronze 8 8 611 46 AWC Gold 2 10 1 22 DLG Gold 2 9 25 117 None 00000 Quality level VDP.Gutwein 7 12 8670 VDP.Erste Lage 17 421 16 17 Auslese 314 14 65 Qualit€ atswein 17 14 5 3 13 Kabinett 413 17 17 35 Sweetness Dry 284 43 21 181 31 Half dry 42 43 121 44 42 Sweet 242 87 99 225 10 Price Lower than average 7 7 24 18 2 Average 10 11 18 6 14 Higher than average 3 4 624 16 Organic Organic 123 19 925 Non-organic 1 23 19 925 Relative importance (%) Brand 14,90 13,17 18,30 25,06 33,38 Region 3,58 3,36 6,52 5,27 7,24 Grape variety 7,24 52,49 26,72 4,29 15,29 Medals and awards 2,14 3,24 6,23 2,97 8,13 Quality level 4,14 3,32 4,65 4,12 16,95 Sweetness 65,80 16,23 27,53 50,72 9,07 Price 2,03 2,34 5,19 5,28 3,81 Organic 0,17 5,85 4,86 2,29 6,13 147A. Risius et al. / Wine Economics and Policy 8 (2019) 141e154 Overall, consumers primarily look for the attribute ‘sweetness’(measured from dry, medium dry to sweet) followed by the producer, the grape variety and the region of production. Within, the attribute levels ‘dry’(attribute category ‘sweetness’), ‘Sp€ atburgunder (red)’(attribute category ‘grape variety’), and producers ‘Zimmermann-Graeff&Mu ¨ller’ as well as ‘Lembergerland Kellerei’(attribute category ‘producer’) are preferred most. Even though the production quality ‘organic’had a positive part-worth coefficient, this attribute was lower in attribute importance than the overall producer impacts, or other attribute levels from the categories ‘sweetness’or ‘grape variety’: ‘Medium dry’and ‘Dornfelder (red)’, respectively. Similar to the attribute value for ‘organic’, ‘medals and awards’seem to be not very decisive, when it comes to consumer preference schemes. 3.4. Consumer target groups for extrinsic wine qualities In order to understand consumer trends in a changing society for the heterogeneous product wine better, consumer segmentation was conducted using a latent class analysis. Table 6 shows the estimation results (rescaled part-worth utilities) of the five-group segmentation for German wine consumers, as well as the relative importance of each attribute. The impact of each attribute on the product utility is shown for each segment (Orme, 2010). The segmentation revealed five latent classes, four of which with equal sizes (rough estimates for latent class membership probability 15%) and one larger segment (35.2%). That the attribute ‘dry’of the category ‘sweetness’is most important for Segment 1, whereas the attribute ‘Sp€ atburgunder (red) is most relevant to segment 2. Segment 3 seems to be considering more different attributes at the same time and does not have a distinct profile with regard to selevtive attributions. Overall, ‘grape variety’and ‘sweetness’are the most important categories for Segment 3. For Segment 4, the attribute ‘sweet’is the most important characteristic, ‘Rotk€ appchen’is it’s most decisive brand. Members of this group have the least interest in organic origin, awards, and quality. Segment 5, of all the consumer segments, has the most ambition to buy wine from certain wine producers. For this segment, quality, organic origin, origin in general, and grape variety are not unimportant when it comes to the purchase of wine. The least important factor is the price of the wine. 3.5. Wine-related lifestyle and the WTP measurement After considering different consumer segments with regard to understanding preferences for wine consumption, the following aims to relate the consumer heterogeneity captured through the latent classes to lifestyle trends and WTP measures as well as consumption habits. Table 7 presents the mean values and standard deviations of the factors of the WRL for each segment. In Table 8, the WTP for wine is shown for the consumer segments. In Table 9, the characteristics of the segments concerning wine consumption habits are shown. In Appendix 2, the demographic characteristics of the consumers are provided. Segment 1: Wealthy, educated and passionate dry wine drinker (15.9 percent). This cluster is slightly dominated by women, mostly aged 50þ, with the highest income compared to the other clusters. The members are usually married and have a high level of education. Over 48.8 percent of the members drink wine more than once a week. They have a tendency to dislike ros e wines and have a strong preference for dry wine (97 percent). Members of this group have the highest WTP for wine, which is consumed on special occasions with an average spend of 17.08 euros. These segment members have the highest value for ‘wine knowledge and connoisseur tendencies’(WRL factor 2). The value for ‘purchase decisions’(WRL factor 4) is the lowest. Segment 2: Hedonistic red wine drinker (18.5 percent). Most of this group is male and is slightly older than the average participant. The members of this segment have a slightly lower household income than the average and are mostly married; 10.1 percent have a Bachelor’s degree as their highest educational level. The other members of this segment have an average education. With 2.36 members per household, the participants who belong in this segment have the largest households in terms of number of persons living in a household. Over half of the segment members drink wine weekly or even more often. There is a strong tendency towards red wine (94.1 percent), which is mostly dry or medium dry. The WTP for wine on a business occasion is at 12.66 euros. This value is Table 7 Differences in Mean Values (SD) of the WRL factors of each Segment (N¼962). Factors Segment 1 Segment 2 Segment 3 Segment 4 Segment 5 Savor and looking for new experience 0,08 (0.840)ab 0,06 (0.916)ab 0,17 (1.088)b 0,34 (0.858)a 0,08 (1.086)b Wine knowledge and connoisseur tendencies 0,52 (0.861)c 0,32 (0.862)c 0,07 (0.980)b 0,38 (0.897)a 0,48 (1.061)a Information procurement for the purchase 0,04 (1.013)b 0,01 (1.038)b 0,15 (0.947)b 0,04 (0.980)b 0,35 (1.005)a Purchase decisions 0,13 (0.972)a 0,06 (1.093)ab 0,04 (0.937)ab 0,04 (0.994)ab 0,21 (1.036)b Reason for wine consumption 0,04 (1.047)a 0,07 (1.003)a 0,11 (0.959)b 0,08 (0.964)a 0,06 (1.062)a Price consciousness 0,08 (0.933)ab 0,06 (0.958)abc 0,07 (0.981)bc 0,22 (1.058)c 0,23 (1.062)a Note. Items were assessed by means of Likert scales (1 ¼totally disagree; 7 ¼totally agree). Superscripts stand for significant mean differences at the 0.05 level based on Tukey’s post hoc testing. 148 A. Risius et al. / Wine Economics and Policy 8 (2019) 141e154