XXIV EuAWE 2017
2A
Bologna Conference
Consumers’ choice experiment for a wine bottle:
is sustainable certification important?
Giovanni SOGARI, Mario VENEZIANI, Davide MENOZZI, Cristina MORA
University of Parma
[email protected]
Abstract
Sustainability interest is increasing in the wine industry both from a
producer and consumer point of view. Although several studies have investigated
the importance of consumers’ preferences regarding wine choice, very little is
known about the role of sustainable claims.
The present study focuses more closely consumer preferences regarding
sustainable production via specific labelling for wine products. Especially, we
explore the relative importance among tradition attribute (Geographical
Indication), attachment to your region (Origin) and sustainability certification
claim (environmental and social importance).
Data were collected using a web-based stated choice experiment which
involved 245 individuals, segmented to be a representative sample of the Italian
wine drinking population. Participants were presented with eight simulated
choice set and asked to choose a preferred alternative between four profiles of
wine bottles and a no-choice option.
Our results show that the presence of a Geographical Indication on the
label is very much appreciated and it is considered the most important attribute.
However also the sustainable claim presents a positive utility supporting the
hypothesis that for a segment of consumers this indication might be relevant.
Interaction between attributes is also explained. Significant difference between
gender has been observed.
The results provide wine chain actors with valuable information and might
be useful to develop new strategies of market placement for wineries which
would like to implement sustainable programmes, as well as evidence for policy
makers about how new labelling system might be a key driver in the coming years.
Keywords: wine, conjoint analysis, sustainable label, certification
2a_sogari_veneziani_menozzi_mora.docx
Page 1/11
XXIV EuAWE 2017
1.
2A
Bologna Conference
INTRODUCTION
In the wine sector, sustainability has become a key issue in the global wine
business (Jones, 2012) and it embraces the aims to protect the environment and
landscape, biodiversity and ecosystems, to manage water resources and the
challenge of climate change. Moreover, to enhance the quality of the wine and its
competitiveness, the International Organisation of Vine and Wine considers also
as main priorities the management of economic resources and social aspects
inside and outside the winery (OIV, 2014).
Recently the Italian wine sector, which is important for the agri-food
business and widespread throughout all the country, has seen private associations
and consortia starting developing sustainable management of vineyards and their
ecosystems. In 2011 the Italian Ministry of the Environment, Land and Sea
developed a programme called V.I.V.A. Sustainable Wine, which established a
common methodology for environmental, social and economic sustainability
assessment using four indicators (Air, Water, Vineyard and Territory). In addition,
wineries which are enrolled in this programme have the opportunity to use the
logo with the indication of V.I.V.A. Sustainable Wine on the label of their
products.
Along with all this increasing interest for sustainable production in the wine
industry, there is no clear evidence whether such programmes lead to a growth in
the value perception of the product and the relative Willingness to Pay (hereafter
called WTP) for such characteristics. Moreover, as reported by Pomarici and
Vecchio (2013), obtaining reliable information for new attributes such as
environmental and social claims can be a difficult task.
Research background has shown how the most important factors that
influence consumers’ wine purchasing decisions are both attributes of the product
(taste, color, region of origin, brand), wine knowledge and experience, and
personal characteristics of the individuals. For instance, in the past many studies
have investigated the importance of Geographical Indication (hereafter called GI)
labelling regarding wine choice, but very little is known on consumers’
preferences for sustainable claims. In the last years few studies have been
investigated WTP for sustainable wine label in Italy (Pomarici and Vecchio, 2013;
Sogari et al., 2016) or investing consumer’s attitude towards sustainable wine
certification (Sogari et al., 2015). However, to our knowledge, little has been done
in the Italian contest to explore the trade-off between sustainable claim and other
wine quality attributes via choice experiment.
The specific objectives of this research is to determine and quantify the
relative importance placed by consumers on specific key attributes associated
with wine choices (GI, region of production and sustainability).
2a_sogari_veneziani_menozzi_mora.docx
Page 2/11
XXIV EuAWE 2017
2A
Bologna Conference
This study is a contribution to the current literature on wine preferences,
especially investing whether consumers might be willing to trade off the tradition
quality cues of a wine (i.e. denomination of origin) for a sustainable attribute,
which is offered to the respondent through the presence of “V.I.V.A. sustainable
wine” logo.
2.
IDENTIFYING RELEVANT ATTRIBUTES
In the design of a conjoint experiment, one of the first and most crucial
steps is to define the attributes and the levels that influence preferences and
choices (Hair et al., 2010). In a wine decision-making process, sensory
characteristics and price factors are not enough to guide the choice; consumers
assess a broader range of information before a purchase. The number of
attributes to consider for getting a realistic description of a wine product concept
is very large. However, CA requires choosing a limited number of attributes to
avoid information overload and at the same time to consider all the relevant
characteristics to elicit consumer preference.
Based on literature review and preliminary focus groups (Sogari et al.,
2013), four independent attributes qualifying the wine were decided to be used
for our experiment: price, Geographical Indication (GI), region of production
(Origin) and sustainable logo.
3.
METHOD
3.1. Experimental design
Conjoint analysis, also called trade-off measurement, is a technique used
when the aim is to measure the trade-off among attributes to predict hypothetical
scenarios and the performance of new products or services.
Lancaster (1966) consumer theory and McFadden (1974) Random Utility
Theory (RUT) are considered the basis for discrete choice experiment. The
assumption of RUT is that consumers are considered to maximize utility selecting
the product with the most desired set of attributes from a set of choices (Barjolle
et al., 2013). However, in a choice-making process, consumers have to make
trade-offs as their ideal combination of levels is not necessarily available among
the alternatives displayed (James and Burton, 2003).
We chose to use Choice-Based Conjoint (CBC) analysis, considered the most
widely used conjoint approach, in which consumers are asked to select a product
from a given set of alternatives (choice set), simulating as much as possible the
actual purchasing process. In addition, the interactions among attributes are also
better measured rather than the traditional CA (individually ranking or rating
profiles) or the Adaptive Conjoint Analysis (Orme, 2006). In fact, respondent data
can be estimated at the disaggregate level (individual) or gathered in
2a_sogari_veneziani_menozzi_mora.docx
Page 3/11
XXIV EuAWE 2017
2A
Bologna Conference
homogeneous groups to measure the part-worth utilities for each level of the
attributes simultaneously and also their interactions (Hair et al., 2010). Each
profile is a function of different attributes of product (including price) and each
attribute (explicative variable) varies at different levels (Table 1). The collected
preference information is used as a basis to estimate the part-worth utilities for
each level. Each part-worth coefficient expresses the contribution of a particular
level of an attribute to the total utility of a product when that level is present. The
relative importance of an attribute can be defined as the weight that the
consumer places on each attribute when selecting a product during the buying
process.
Table 1. Experimental design attributes and levels
No.
Attribute
Levels
1
Price
2
Geographical Indication
3
Region of production
4
Sustainable label
• till 3.00€;
• 3.10-5.00€;
• 5.10-7.00€
• >7.00€
• DOC - the Controlled Origin Denomination
• None
• Produced in your region
• Not produced in your region
• VIVA - Sustainable wine logo
• None
Source: own elaboration
A “no-choice option” was added to every choice set to make the conjoint
choice experiment more realistic. In this way the respondent is not forced to
make a choice which might lead to less accurate predictions when analyzing the
data. The “no-choice option” creates the same circumstances of a real purchase
situation when customers do not like any product in the market and therefore
prefer to continue looking for better alternatives which meet their requirements
(Vermeulen et al., 2008; Lusk and Hudson, 2004).
Sawtooth Software Incorporated (SSI Web version 8.3.8), which is a
software system for creating Web-based questionnaires, was used to generate an
efficient full factorial design which includes all the possible combinations of the
attributes for each level. Four attributes, namely, price, Geographical Indication,
region of production and sustainable label were tested.
Once the number of attributes and their levels are chosen, the design of the
experiment is obtained. Orthogonality must be maintained which suggests no
correlation between the attributes and independency from each other. A total of
32 possible profiles (4 × 2 × 2 × 2) were identified. Based on the literature (Hair et
al., 2010) it was considered that 8 choice set is a reasonable number of profiles to
2a_sogari_veneziani_menozzi_mora.docx
Page 4/11
XXIV EuAWE 2017
2A
Bologna Conference
manage for the respondents without negative effects on the quality of the data
due to respondent fatigue.
Each simulated choice set consisted of four wine bottles with each label
characterized by attribute bundles with various attribute levels and a “no-choice
option”.
3.2. Data collection
A web-based questionnaire was created using Sawtooth Software. The
choice to use an on-line tool to collect data instead than a more tradition face-toface choice experiment interview was supported from our need to have a sample
of respondents from all Italian regions and not just from a limited geographical
area. The wide web use spread among many Italian citizens of different age,
regions and gender allows to trust in the online questionnaire tool.
Pilot studies have been carried out to improve the clarity of the questions
and to evaluate the total duration of the questionnaire and adjustments have
been done to allow a range of 8-10 minutes to fill in the survey.
From September to December 2014 the questionnaire was mailed out to a
random sample of over 2000 households resident in all Italian regions and, after
eliminating uncompleted answers and respondents who answer too fast, 245
completed and valid responses were collected. Economic and technical
constraints make it difficult to gather a large number of respondents. However,
the sample is representative of our ideal target population which is composed by
wine drinkers in Italy, segmented by region of residence, age and gender.
Responses from participants who had consumed wine within the last month were
considered valid.
the questionnaire starts with (1) a set of wine consumption habits patterns,
(2) a set of importance of wine attributes, (3) the choice experiment, (4) a set of 8
candidate statements on sustainable wine labelling, and (5) a set of sociodemographic questions.
Wine consumption habits
How many times have you been drinking wine glasses in the last month?
The first question was the so called “filter question” which excluded all the
respondents who have not drunken any glass of wine in the last month (“no
one”). In this way we assured to have a sample of people who can be considered
wine consumers based on their drinking habits in a recent period of time. At the
same time we have a frequency of consumption in the last month which allowed
us to segment our sample in daily (“one per day” or “more than one every day”)
or weekly (“one per week” or “3-4 times per weeks”) wine drinkers.
2a_sogari_veneziani_menozzi_mora.docx
Page 5/11
XXIV EuAWE 2017
2A
Bologna Conference
Where do you usually buy wine?
The second question emphasizes on the place of purchasing. Max five
possible answers were showed – as suggested from the literature – in order to
facilitate the respondent; however an open question (“other”) was allowed to
specify any places not already indicated.
The location of purchasing is important because it might indicate the
degree of experience with the product and leads to the identification and
differentiation between experienced and deep involved consumers (buy “at the
wineries” or “in a wine shop”) and not much interested users (“at the
supermarket” or “at the wholesale”). More difficult to categorize consumers who
generally buy wine online. Such target might be purchasing wine directly from the
producer’ website or from a general online shop.
At what range of price do you usually buy a bottle of wine for your home consumption?
The last question on consumption habits wants to highlight the budget that a
consumer spends on average for wine. Based on the price per bottle we can
differentiate among consumers who spend a small amount of money (“up to 3€”),
average (“3.10-5.00€” and “5.10-7.00”) and higher budget (“more than 7.00€”).
However, this data should be analyzed taking in consideration several sociodemographics parameters such as the revenue of the respondents and household size.
What attributes do you consider the most important for choosing a wine?
14 attributes which mostly influence the choice of a wine have been taken in
consideration. Apart price and presence of discounts, the others can be classified
among intrinsic (sensory characteristics, color) and extrinsic wine features (brand
name, grape variety, geographical indication, eco and social certifications, previous
experience, etc.). At the end it was proposed an open question (“other”) in which the
respondents were free to specify any other personal factors. Some interesting
answers have been given, such as the “year of production”, “alcohol per volume” or
“recommendation from wine guide”.
This question has twofold objectives: (1) confirming (or not) the results of the
choice experiment and (2) obtaining a measure of the importance of all other
attributes which are not included in our CBC analysis.
In our CA experiment respondents are placed in a purchase context (Barjolle et
al., 2013) asked to choose their preferred wine. To make the exercise more realistic
and motivate the respondent’s participation, in the online questionnaire it was
created a layout which reminded a shelf in the shop with the image of several wine
bottles on the background. The four alternatives were shown as images of bottle with
their respective attributes on the label and the “no-choice” option (Figure 1). An
image of the sustainable logo was used to illustrate the presence of this attribute on
the label, as in the real market.
2a_sogari_veneziani_menozzi_mora.docx
Page 6/11
XXIV EuAWE 2017
2A
Bologna Conference
Figure 1 Example of screen shot of bottle’s images
4.
RESULTS
Table 2 summarizes the demographics statistics of our sample (n=245) and the
sample from Italian wine consumption habits drawn from the National Statistics
Institute census data (ISTAT, 2013), classified by population over 18 years by region of
residence and gender. Our data are in line with the 2013 census for all the three
parameters: we have some percentage points more of younger people (18-35 years)
and less in the range over 56 years. The percentages also fit well regarding the regions
of residence for North and Centre area and lack some points in the regions of South of
Italy.
Table 2. Descriptive statistics of the respondents (n=245)
Characteristics
Our sample
N
(%)
ISTAT 2013
(%)
Gender
Male
Female
Missing values
Total
139
91
15
245
57
37
6
100
61
39
100
Age
18-35 years
36-55 years
Over 56 years
Total
59
92
94
245
24
38
38
100
21
38
42
100
Regions of residence (Italy)
North West
North East
Centre
South and Islands
Total
68
62
48
67
245
28
25
20
27
100
27
21
21
31
100
2a_sogari_veneziani_menozzi_mora.docx
Page 7/11
XXIV EuAWE 2017
2A
Bologna Conference
Results of part-worth utilities were obtained using SPSS (21.0) statistical
software packages. Positive and negative values will suggest whether a level has been
more or less appreciated.
Logistic Regression (LR) was used to analyses the data. LR can be used to
estimate the values of these part-worths which measure the value or attractiveness
that the respondent puts on each level of the attribute in question. The four levels of
the price interval have been translated in four singular values. As the three attributes
are categorical, dummy variables have been appropriately defined (Sánchez and Gil,
1998).
The parameter estimates of the LR for main effect variables are listed in Table 3.
Table 3. Parameter estimates
Variables
Intercept
Price
Geographical Indication
Region of production
Sustainable label
Note: *** significant at the 0.01 level
Parameter estimates
Total sample
Male
-1.381***
-1.658***
-.222***
-.209***
1.430***
1.685***
.612***
.767***
.725***
.657***
Source: own elaboration
Female
-1.009***
-.279***
1.190***
.472***
.934***
First of all, as expected, the coefficient for the price is negative (-.222). Second,
as reported in the literature examined so far, the presence of a GI mark on the label is
very much appreciated (1.430), suggesting that this attribute plays a relevant role in
the making decision process.
For our aims, it is interesting to mention that the sustainable label attribute
shows a positive utility (0.725) slightly higher than the region of production (0.612).
This latter result could arise from the fact that for the majority of consumers wine is
associated to a specific area (Geographical Indication) but not deliberate giving
priority to the wine from the region of residence as it happens for other food
products. Probably due to the fact that some Italian regions are more famous and
well-known from producing good quality wine compared to others. Moreover, this
result supports the hypothesis that sustainable claim might be relevant for a segment
of consumers.
Willingness to pay is calculated as –(βx / βprice), where x represents the
attribute of interest, and price is the estimated price coefficient. Therefore, starting
with the first attribute the GI is value as –(β1/βprice)= -(1,430/-,222)= 6.44€, followed
by sustainable logo (3.26€) and region of origin (2.75€).
On the other hand, the attribute’ importance which tell us in percentage terms
the impact of an attribute on the overall product’s utility, is calculated taking in
consideration the difference between the maximum and minimum level values of
2a_sogari_veneziani_menozzi_mora.docx
Page 8/11
XXIV EuAWE 2017
2A
Bologna Conference
each attribute and divided for the sum of all differences obtained for all the attributes
(Green et al., 2001; Orme, 2010). The relative importance of an attribute can be
defined as the weight that the consumer places on each attribute when selecting a
product during the buying process. Excluding the price attribute, table 4 shows how
Geographical Indication has the highest relative importance (51.7%), followed by the
sustainable label (26.2%) and region of origin (22.1%). When we segment our sample
by the gender we notice how females prefer much more the sustainable attribute
(36%) over the fact that wine has been produced in region of residence (18.2%) as
well as giving lower importance to the GI mark (45.8%) compared to the mean of the
sample (51.7%).
Table 4. Relative importance of each attribute
No.
Attribute
1
Geographical Indication
2
Region of production
3
Sustainable label
Source: own elaboration
Sample
51.7
22.1
26.2
Male
54.2
24.7
21.1
Female
45.8
18.2
36.0
Considering that consumers do not assess attributes individually, first and
second order interactions (interactions between two and three attributes) have also
been included in this analysis. Therefore, we conducted a market share simulation,
using the estimated parameters from LR model, to investigate the probability that
consumers will prefer the product of interest.
A wine produced in the region of residence with the sustainable label and the
GI mark has an odds-ratio of 4.181. Thus, the odds of a sustainable wine of the region
of residence of the consumer being chosen will more than quadruple with the
presence of a GI mark. On the other hand, the odds to be chosen for a wine produced
outside your region but with a GI will double with the presence of a sustainable label.
As already mentioned, the GI mark listed on the label is found to extremely increase
the market share for all types of wine.
5.
CONCLUSIONS AND LIMITATIONS
In this paper we presented the outcomes of a choice experiment method,
where respondents have been presented with a set of wine labels alternatives,
differing in terms of attributes and levels, and asked to choose the most preferred.
First, results provide valuable information on consumers’ wine decision making
when sustainable attribute is involved and might contribute to the discussion of public
authority policies for new sustainable claims labelling. Second, these findings suggest
insights for private companies to develop strategies given the opportunity to evaluate
the trade-off between Geographical Indication and sustainable claims on a wine label.
Results suggest that Geographical Indication has a strong impact to influence the
choice of a wine, however the presence of a sustainable label might add value to the
2a_sogari_veneziani_menozzi_mora.docx
Page 9/11
XXIV EuAWE 2017
2A
Bologna Conference
product. Significant differences have been found between males and females, with
the latter valuing more sustainable label and less the GI mark.
Companies which would like to implement sustainability programmes should
be aware that consumers value positively the presence of a sustainable claim on the
label of a bottle. This might also provide consumers with more information about the
mission of the winery and the characteristics of the product in order to enable them
to make well-considered choices based on sufficient information.
As a final remark, these findings provide some inputs for the actors involved in
the promotion of sustainable behavior and enables researchers to further investigate
aspects of the ethics of consumer choice.
References
Barjolle, D., Gorton, M., Ðorđević, J.M. and Stojanović, Ž. (2013). Food Consumer
Science. Theories, Methods and Application to the Western Balkans. Dordrecht:
Springer Science+Business Media.
Green, P.E., Krieger, A.M. and Wind., Y. (2001). Thirty Years of Conjoint Analysis:
Reflections and Prospects. Interfaces, 31, pp.56-73.
Hair, J.F., W.C. Black, B.J., Babin and Anderson, R.E. (2010). Multivariate Data
Analysis. 7th ed. Upper Saddle River, NJ: Pearson Education, Inc.
James, S. and Burton, M. (2003). Consumer preferences for GM food and other
attributes of the food system. The Australian Journal of Agricultural and Resource
Economics, 47, pp. 501-18.
Jones, G.V. (2012). Sustainable vineyard developments worldwide. OIV bull.,
85(49), pp.971-73.
Lancaster, K.J. (1966). A New Approach to Consumer Theory. The Journal of
Political Economy, 74(2), pp.132-57.
Lusk, J.L. and Hudson, D. (2004). Willingness-to-Pay Estimates and Their Relevance
to Agribusiness Decision Making. Review of Agricultural Economics, 26, pp.152-69.
McFadden, D. (1974). Conditional logit analysis of qualitative choice behavior. In P.
Zarembka, ed. Frontiers in econometrics Academic Press. New York. pp.105-42.
OIV (2004). RESOLUTION CST 1/2004.
Orme, B.K. (2010). Getting Started with Conjoint Analysis: Strategies for Product
and Pricing Research. Research Publishers LLC.
Pomarici, E. and Vecchio, R. (2013). Millennial generation attitudes to sustainable
wine: an exploratory study on Italian consumers. Journal of Cleaner Production,
(66), pp.537-45.
2a_sogari_veneziani_menozzi_mora.docx
Page 10/11
XXIV EuAWE 2017
2A
Bologna Conference
Sánchez, M. and Gil J.M. (1998). Consumer Preferences for Wine Attributes in
Different Retail Stores: A Conjoint Approach, International Journal of Wine
Marketing, Vol. 10 Iss: 1, pp.25 – 38.
Sogari, G., Mora, C., and Menozzi, D. (2013). Consumers’ perception of organic
wine. A case study of German and Italian young consumers. In: Lun, L.M., Dreyer,
A., Pechlaner, H. and Schamel, G. (eds.) 2013, Wine and tourism. A value-added
partnership for promoting regional economic cycles. Proceedings of the 3 rd
Symposium of the Workgroup Wine and Tourism of the German Society of
Tourism Research (DGT). EURAC book, Vol. 62. p.101-112.
Sogari G., Mora C., Menozzi D. (2016). Factors driving sustainable choice: the case
of wine. British Food Journal Vol 118/3. http://dx.doi.org/10.1108/BFJ-04-20150131
Sogari G., Corbo C., Macconi M., Menozzi D., Mora C. (2015). Consumer attitude
towards sustainable-labelled wine: an exploratory approach. International Journal
of Wine Business Research11/2015; 27(4):312-328. DOI:10.1108/IJWBR-12-20140053
Vermeulen, B., Goos, P. and Vandebroek, M. (2008). Models and optimal designs
for conjoint choice experiments including a no-choice option. International Journal
of Research in Marketing, 25(2), pp. 94–103.
2a_sogari_veneziani_menozzi_mora.docx
Page 11/11