ArticlePDF Available

Abstract and Figures

Consumer behavior is key in shifts towards organic products. A diversity of factors influences consumer preferences, driving planned, impulsive, and unplanned purchasing decisions. We study choices among organic and conventional wine using an extensive survey among Australian consumers (N = 1003). We integrate five behavioral theories in the survey design, and use supervised and unsupervised machine learning algorithms for analysis. We quantify a gap between intention and behavior, and emphasize the importance of cognitive factors. Findings go beyond correlation to the causation of behavior when combining predictive prowess with explanatory power. Results reveal that affective factors and normative cues may prompt unplanned and spontaneous purchasing behavior, causing consumers to act against their beliefs.
Content may be subject to copyright.
1
Shifts in consumer behavior towards
organic products: theory-driven data
analytics
Firouzeh Taghikhah a*, Alexey Voinov a,b, Nagesh Shukla a, Tatiana Filatova b,a
a Center on Persuasive Systems for Wise Adaptive Living, Faculty of Engineering and Information Technology,
University of Technology Sydney, NSW 2007, Australia,
b Department of Governance and Technology for Sustainability, University of Twente, Netherlands
Email: Firouzeh.th@gmail.com, aavoinov@gmail.com, Nagesh.Shukla@uts.edu.au,
Tatiana.Filatova@uts.edu.au
* Corresponding Author
Abstract
Consumer behavior is key in shifts towards organic products. A diversity of factors
influences consumer preferences, driving planned, impulsive, and unplanned
purchasing decisions. We study choices among organic and conventional wine using
an extensive survey among Australian consumers (N=1003). We integrate five
behavioral theories in the survey design, and use supervised and unsupervised
machine learning algorithms for analysis. We quantify a gap between intention and
behavior, and emphasize the importance of cognitive factors. Findings go beyond
correlation to the causation of behavior when combining predictive prowess with
explanatory power. Results reveal that affective factors and normative cues may
prompt unplanned and spontaneous purchasing behavior, causing consumers to act
against their beliefs.
Keywords:
Organic food; emotion; habit; impulsive purchasing; data mining; explainable artificial
intelligence.
2
1. Introduction
Demand-side policies can significantly contribute to tackling climate change issues
and managing environmental resources. The practical implementation of these
policies is vital for conserving the ecological-service for the future and controling the
exploitation of natural capital assets. The challenge for policymakers is how to change
the consumption patterns and increase the demand for environmentally-friendly
products, which triggers the market forces to make the adoption of sustainable
practices economically attractive to suppliers and producers.
For example, since the introduction of chemicals in the 19th century, viticulture has
significantly contributed to a wide range of environmental issues, particularly those
related to land and water pollution. By excluding agrochemicals from vineyards,
organic agriculture helps preserve biodiversity and the overall quality of
agroecosystems (Rugani et al. 2013). Wines produced with organically grown grapes
have a higher content of antioxidants (30%) (Vrček et al. 2011) and lower content of
orchatoxins (Gentile et al. 2016). Consumer choices and their willingness to pay
(WTP) more for organic wines can support farmers in expanding organic vineyards
(Taghikhah et al. 2020b). In fact, it can be a game-changing strategy contributing to
the economy, ecology, and society.
The demand-side argument is prominent in the ongoing debate about how to
increase the organic wine market share. It highlights the need to investigate the
characteristics of consumer segments willing to purchase the organic food, identify the
factors influencing their decisions, target influential factors in each consumer segment,
and develop segment-specific marketing strategies if we wish to nudge behavior
3
towards organic consumption. Prior studies report various factors drivers of
consumers’ decisions in purchasing organic wine. The key factors include price
(Panzone 2014), perceived that health and environmental benefits (Loose & Lockshin
2013), region of origin (Trinh et al. 2019; Yang & Paladino 2015), brand (Ryan &
Casidy 2018), superior taste and quality (Kim & Bonn 2015), as well as socio-
demographics including age, gender, and income (D’Amico et al. 2016). More recent
studies have highlighted the relative importance of occasions like hosting friends and
gift-giving (Boncinelli et al. 2019), wine consumption and shopping frequency
(Pomarici & Vecchio 2014), and drinking frequency (Pomarici et al. 2016) as predictors
of consumers shift from conventional to organic wine.
In the context of pro-environmental behavior, the literature highlights a discrepancy
between consumers’ stated intentions and their actions, known as the intention-
behavior gap. Even though consumers demonstrate WTP for products with
sustainability cues, and their intentions are high, these do not necessarily translate
into the actual purchasing behavior. With regard to organic wine, the literature focuses
on identifying determinants of WTP; yet, this is rarely differentiated from real
purchasing behavior. An exception is a study by Schäufele and Hamm (2017), who
confirm the inconsistencies between intentions to purchase organic wine and the
actual behavior among low-income consumers, identifying prices as the primary
purchasing barrier. Poor quality and inferior taste are other reported reasons for
avoiding organic purchases (Mann et al. 2012; Stolz & Schmid 2008).
Impulsive and unplanned purchasing behaviors appear to interrupt the intention-
behavior relationship. According to the literature on consumer behavior, affective
factors as well as cognitive and normative factors, can trigger behavior change
4
(Russell et al. 2017). The non-cognitive factors, such as emotions, impulse
tendencies, and personal goals, may underlie the failure to translate consumers’
intentions into actions. Yet, to the best of our knowledge, there have been no
quantitative studies to date that have investigated the relative importance of these
factors as they relate to organic wine purchasing.
Moreover, quantitative research predicting consumers’ intentions and behavior for
purchasing organic wine has, to date, been dominated by statistical models. While
these models can successfully reveal the relationship between variables, their
predictive power and accuracy, as compared to machine learning (ML) algorithms, are
low, especially when dealing with a high number of observations and attributes.
Indeed, they are powerful tools in identifying unexpected patterns and emergent
proprieties of underlying phenomena of interest. Pattern verification is a useful
application of ML for confirming whether suggested behavioral theories exist.
Our study aims to explore the determinants of heterogeneity in organic food
purchasing intentions and behaviors. To identify the behavioral factors driving
purchasing decisions, we consider behavior change theories from psychology and
developed a conceptual framework that integrates five relevant theories. We focus on
organic wine as a case study and surveyed 1,003 Australian consumers living in the
City of Sydney. The collected data enable to quantitatively assess the impact of socio-
demographics, shopping and wine consumption patterns, and behavioral factors on
consumers’ stated intentions and behavior for purchasing organic wine. Our findings
reveal factors that cause the intention-behavior gap in pro-environmental food
consumption.
5
This article makes a number of innovative contributions to the literature on consumer
behavior.
(i) It examines the influence of affective factors, including emotions, impulse
tendencies, and personal goals, as well as cognitive factors, especially social
norms, in the context of wine purchasing. To the best of our knowledge, this
is the first study that has fully explored how this set of attributes affects
preferences for organic food by integrating the strength of multiple behavioral
theories. While many papers in this field focus on studying willingness to pay
and intentions for buying organic wine, we focus on the gap between
intentions and behavior.
(ii) It goes beyond the traditional analysis in empirical consumer behavior studies
by applying both supervised and unsupervised ML methods. Besides
increasing the accuracy of predictions, we explain why AI arrived at a specific
decision by identifying the most influential factors. These methodological
advances provide new insights into different consumer segments, identify the
causality and mechanisms of decisions related to organic products, and, most
importantly, verify whether the behavioral patterns will continue to function as
expected over time. We apply explainable AI techniques to open the "black
box" of ML in consumer behavior area so that the results can be understood
by humans.
(iii) It provides empirical insights for industry and policymakers when promoting
organic food and can contribute to the facilitation of demand-side solutions in
the transition to sustainable agriculture. The demand-side policy is attractive
to producers (farmers) and policymakers because the diffusion of organic food
consumption is expected to reduce the cost of developing organic farms and
6
help overcome the trade-offs between economic, health, and environmental
goals.
The remainder of this paper is organized as follows. We begin with explaining the
proposed theoretical framework used to develop the survey (Section 2) and describe
the methodological aspects, data collection, and the analysis process (Section 3).
Section 4 presents the results, and Section 5 discusses them in the context of existing
literature. We conclude with discussing the implications for practice and outline
potential avenues for future research.
2. Theoretical framework
Behavior change theories are widely applied to understand the internal, external, and
interpersonal factors driving individual actions. To provide a more holistic perspective
on pro-environmental purchasing behavior, we refer to the principles of Stern’s buying
theory (Stern 1962) that classify decisions as planned, impulsive, and unplanned.
Planned purchasing behavior is time-consuming, information-searching, norm-
dependent, semi-bounded rational decision making. In contrast, unplanned
purchasing behavior refers to decisions driven by atmospheric store-related stimuli
(e.g., promotions, posters) or habits (context-dependent stimuli) without any
preliminary planning or even actual need. Impulsive purchasing refers to rapid,
spontaneous decisions driven by an individual’s impulse tendency (i.e., a sudden,
irresistible urge). Internal stimuli cause impulsiveness in response to mood swings,
excitement, or unpleasant situations. Research shows that the use of sensory cues,
such as the addition of scent or music, can influence consumers’ emotions and
impulse purchasing behavior (Helmefalk & Hultén 2017).
7
To encompass the complexity of consumer behavior and various stages that lead
one to a purchasing decision, we develop a theory-grounded framework (Figure 1).
Namely, we combine the strength of relevant theories to understand the influence of
cognitive and affective factors behind a variety of purchasing behavior:
1. Theory of Planned Behavior (TPB) (Ajzen 1991) to account for factors driving
planned decisions,
2. Theory of Interpersonal Behavior (TIB) (Triandis 1977) to integrate the
influence of emotions,
3. Impulsive Buying Theory (IBT) (Stern 1962) to capture factors driving
impulsive purchasing,
4. Alphabet Theory (AT) (Zepeda & Deal 2009) to integrate the role of habits,
5. Goal Framing Theory (GFT) (Lindenberg & Steg 2007), to account for a variety
of goals.
This framework allows to comprehensively explore purchasing decisions in different
situations (e.g., shopping environment), understand the influence of context on the
action (e.g., occasions), identify potentials to influence preferences (e.g., social
media), and bridge the gap between intention and behavior. Appendix A1 describes
these theories and the literature review on consumer behavior for organic food in
detail; Taghikhah et al. (2020a) discuss the further rationale for integration.
8
Figure 1. Conceptual model of the determinants of organic wine purchasing behavior.
3. Methodology
3.1. Data collection
Relying on the theoretical framework (Figure 1), we design a questionnaire to elicit
data on the corresponding variables (discussed in detail in Appendix A2). The
questionnaire includes 7 sections consisting of 35 questions about (i) socio-
demographic characteristics (10 questions), (ii) shopping and drinking-related style (7
questions), (iv) habits (1 question), (v) attitudes (3 questions), PBC (2 questions), (vi)
social networks (3 questions), (vii) personal goals (4 questions), (viii) emotions, and
(ix) impulse tendency (1 question). We align the questions for assessing habits,
shopping patterns and emotions and impulsiveness with previous studies (e.g.,
Verplanken and Orbell (2003), Ogbeide (2013), Watson et al. (1988)). We use a
multiple-question approach in assessing each variable to improve the quality of
9
results. They were measured on a Likert scale ranging from 1–5. An expert panel
consisting of two academic researchers and one practitioner reviewed and validated
the questions.
In September 2019, the online survey was conducted in 32 suburbs of the City of
Sydney through Qualtrics online customer panel (https://www.qualtrics.com). The
respondents (18+ years old) were chosen randomly. We ran a one-stage pilot study
(Npilot=50) to test the consistency of questions and responses. We check the internal
consistency of the survey using Cronbach’s alpha. The test result shows a good
consistency at 0.766, which confirms the validity of the designed questions for
assessing factors.
We acknowledge that self-reported items do not always reflect the actual behavior in
stated-preferences studies like surveys. However, the choice of what wine to buy is a
regular decision, which stays in the memories of consumers. In this case, consumers
were not thinking of a hypothetical decision when filling in our questionnaire; they were
explicitly asked about a decision that is learnt and is practiced on an almost weekly
basis. Our questionnaire explicitly asked respondents to remember whether they had
purchased organic wine and what share of their actual past purchases was organic.
3.2. Methods of analysis
3.2.1. Data pre-processing and correlation analysis
For standardization, the variables containing discrete sequences of values, such as
age, shopping frequency, shopping size, family size, etc. are normalized with the min-
max normalization method to values between 0 and 1, to scale the differences in the
10
ranges of the continuous variables. We apply a binary encoding procedure to all
categorical variables in our dataset to convert them into binary variables. Our final data
set includes 1003 responses and 89 variables. The descriptive statistics reveal the
characteristics of respondents and the distribution of key variables (Section 4.1). We
use Spearman’s rank correlation to assess the strength and direction of the
relationships between the nine latent variables representing behavioral factors
(Section 4.2). This allows us to validate the proposed conceptual framework (Figure
1). We consider coefficients greater than +0.4 (and smaller than -0.4) as indicators of
a strong relationship, while those between 0.2 and 0.4 (-0.4 and -0.2) as of a moderate
correlation. The strength of a correlation depends on the context and sample size. As
common in social sciences (Cohen (1992, 2013), coefficients around 0.3 and 0.5
represent moderate and strong correlations, respectively. However, for large sample
sizes, a moderate correlation coefficient can be considered as significant as a strong
correlation in a small sample, meaning that this relationship is unlikely to occur by
chance.
3.2.2. Supervised learning: Classification
To reach beyond correlations towards implying causation of the behavior, we use
classification, the most commonly applied supervised learning approach, to predict the
probability that a consumer prefers organic to conventional wine. We consider 6
classes of intentions and 5 classes of behavior for purchasing organic wine. The
consumers with no willingness to pay for organic wine are labelled as class (1) and
those with willingness to pay for organic wine up to 10%, 20%, 30%, 40%, and 50%
are labeled class (2), (3), (4), (5), and (6), respectively. Similarly, for predicting
behavior, labels are assigned to consumers who purchase only conventional wine
11
(class (1)), organic wine up to 25% (class (2)), organic wine between 25% and 50%
(class (3)), organic wine between 50% and 75% (class (4)), and organic wine 75% or
more (class (5)). We test both parametric (logistic regression, LR) and nonparametric
(support vector machine, SVM) classification algorithms (Cortes & Vapnik 1995), as
well as the Decision Tree (DT) (Quinlan 1990) and Random Forest (RF) (Ho 1998)
algorithms to identify the best performing method for classification of our data.
Appendix A3 provides the details of these classification algorithms.
Parametric algorithms assume that a linear combination of variables and coefficients
can be fitted to a line, whereas nonparametric algorithms construct the model based
on the similarities between patterns in data, without making any assumptions. While
the selection of methods depends mainly on the characteristics of the data, higher
flexibility and predictive power are generally expected for nonparametric algorithms.
However, data requirements and overfitting issues should be carefully controlled when
using these algorithms. SVM finds the best prediction model using an optimization
process to minimize the error function. DT uses conditional control statements in a
flowchart-like structure to predict outcomes. Previous studies have reported better
performance of ensemble methods like RF for classification, where multiple predictive
models (in this case, trees) vote for the class assigned to a given sample so as to
decrease biases and variances in predictions. The partitioning ratio for training and
testing for each of these methods is set to 70% vs. 30%, respectively.
3.2.3. Unsupervised learning: Clustering
To identify hidden patterns or distinct groups based on their similarities in our dataset,
we use clustering, the most common unsupervised learning approach for exploratory
data analysis. Density-based clustering algorithms such as DBSCAN automatically
12
detect the number of clusters and are suitable for cases where the clusters are not
compact and well-separated (Ester et al. 1996). In contrast to ad-hoc methods that
divide records based on one attribute, this method includes all attributes when
computing the cohort outliers. Hence, one may reveal hidden patterns and groups in
survey data and study their characteristics along known dimensions to potentially
attribute various behavioral factors to consumption patterns. Partitioning methods
(e.g., K-means) and hierarchical clustering work by finding spherical-shaped clusters
or convex clusters, while DBSCAN identifies arbitrary-shaped clusters under fewer
restrictions. However, since our database is highly dimensional and scattered, this
algorithm fails to detect clusters of consumers with similar properties. Hence, we utilize
its extension HDBSCAN - designed to deal with high-dimensionality. HDBSCAN
uses a technique to hierarchically represent every possible cluster generated by
DBSCAN and extract a set of flat clusters (Campello et al. 2013). We applied
HDBSCAN on the pre-processed dataset with 89 dimensions (Section 3.2.1). As the
algorithm fails to extract meaningful clusters when using all 89 dimensions (the noise
is 70%), we further use the principal component analysis (PCA) method to gradually
reduce the dataset dimensions, minimize the clustering noise, and increase the
density of resulting clusters. PCA identifies six dimensions where the clustering noise
is the lowest, while its density is the highest. Further, relying on the HDBSCAN
recommendations for selecting parameters, we use the approach proposed by
Rahmah and Sitanggang (2016) to tune its hyper-parameters. Appendix A4 provides
details of HDBSCAN and the settings for its hyper-parameters.
13
4. Results
4.1. Descriptive analysis
Table 1 compares the socio-demographic characteristics of the City of Sydney
population (collected from ABS - 2016 census) with the collected sample (own survey
- 2019). The results indicated that, except for the educational level, the sample is
representative of the population. Nevertheless, as discussed in Section 4.2 below,
education is only moderately correlated with intention and behavior, and indicating that
the possible education gap between our sample and the local population should not
affect the main conclusions of the study.
Table 1. Socioeconomic distribution in the City of Sydney (LGA) and the survey sample.
Factors City of Sydney LG
A
Survey sample
Total number of households 85,423 1,003
Gende
r
Female (%)
Male (%)
47%
53%
41%
59%
Median age group 30-40 years old 36-45 years old
Median total income AU$75,001 to AU$150,000 AU$75,001 to AU$150,001
Average household size 2 2
Education level
Postgraduate Degree levels (%)
Graduate Degree level (%)
School education level (%)
17.9%
39.9%
42.2%
51.1%
36.8%
12.1%
Table 2 provides summary statistics for the socio-demographic characteristics of the
survey respondents. It shows that (1) gender statistics are balanced and can
adequately reflect differences, (2) the majority of consumers are highly educated and
work full time in the management and engineering occupations, (3) the income level
of more than two-thirds of the consumers is higher than the average income, between
14
AU$ 75 and AU$ 250 thousand, and (4) about half of the respondents are singles and
half are couples.
Regarding consumers’ patterns of wine purchasing and consumption, the results
indicate that the majority of respondents surveyed visit wine shops more than once a
week and purchase more than five wine bottles per month. More than 70% of
consumers purchase the same brand of wine quite often and report drinking wine 2 to
5 times a week.
Table 2. Socio-demographic characteristics of surveyed consumers.
Socio-demographic items
Gender Household annual income Household size
Male 59% less than 45 thousand AU$ 9% One 15%
Female 41% 45-75 thousand AU$ 14% Two 29%
Age 75-150 thousand AU$ 38% Three 26%
18-25 years 10% 150-250 thousand AU$ 26% Four 0%
26-35 years 25% More than 250 thousand AU$ 13% Five 13%
36-45 39% Occupation Six 47%
46-55 16% Engineering 19% Seven and more 18%
56-65 6% Education 12% Employment status
66 and more 4% Sales and service 15% Full-time employed 78%
Education Management 29% Part-time employed 10%
Primary 2% Other 26% Retired 4%
Secondary 10% Student 5%
Graduate 39% Unemployed 3%
Post-Graduate 51%
Table 3 presents the summary statistics for the behavioral factors related to
purchasing behaviors. The results showed that, on average, consumers have positive
attitudes towards organic wine and positive emotions during shopping. Most
15
consumers report high habitual (0.69) and low impulsive purchasing (0.29).
Consumers distinguish between organic and conventional wine and like the taste of
organic (more than 0.72), whereas the advice of staff, choice of other people at the
shop, and social media are not significant predictors of wine choice (less than 0.33).
While wine availability is important to our respondents, they indicate no concern for
price (comparing 0.56 to 0.37).
Table 3. Importance of behavioral factors among survey respondents.
Behavioral
factors
Sub factors (related
theory) Measures Average Standard
deviation
Cognitive
Attitude (TPB)
Trust on organic wine 0.74 (0.19)
Environmental knowledge of
organic wine 0.73 (0.17)
Health knowledge of organic wine 0.72 (0.16)
Perceived Behavioral
Control (TPB)
Importance of wine price 0.37 (0.35)
Importance of wine availability 0.56 (0.28)
Habit (AT) Automaticity of purchasing 0.69 (0.21)
Hedonic goals (GFT)
Taste 0.85 (0.29)
Difference and distinction 0.78 (0.36)
Likeness 0.72 (0.39)
Gain goals (GFT)
Change of price at the shop
(switch preference) 0.42 (0.35)
Change of availability at the shop
(switch preference) 0.32 (0.37)
Normative
Social norms (TPB)
Frequency of socializing about
wine 0.57 (0.3)
Purchasing wine for occasions 0.65 (0.47)
Advice of family and friends 0.75 (0.15)
Normative goals (GFT) Staff and others at shop 0.33 (0.23)
Social media 0.17 (0.26)
Affective Emotions (TIB & IBT) Positive emotions 0.75 (0.25)
Spontaneous urge (IBT) Impulse tendency 0.29 (0.3)
Our results indicate a significant gap between intention (Figure 2.a) and behavior in
organic wine purchasing (Figure 2.b). We consider WTP more for organic wine as an
indicator of individual intention and the proportion of purchased organic wine in the
shopping basket as an indicator of actual behavior. The respondents are asked to
assume that the average price of a wine is $10 per bottle. More than 80% of
consumers have a positive intention for purchasing organic wine (Figure 2.a).
Interestingly, only 4% of consumers are exclusively organic wine buyers (i.e.,
purchase organic in 75-100% of cases), with an additional 17% of consumers
16
indicating they are frequent organic wine buyers (i.e., between 50-75% of their wine
shopping basket is organic). Still, 60% of respondents indicate that less than 50% of
their wine shopping basket is organic, while 20% had never purchased organic wine
before. Our further analysis aims to explore what stands behind various consumption
decisions.
(a) Intention to purchase organic wine (b) Wine purchasing behavior
Figure 2. Distribution of intention and behavior for purchasing organic wine (in percentage; N=1003).
4.2. Correlation analysis
Table 4 presents the correlation matrix for behavioral factors, perceived behavioral
control (PBC), social norms, emotions, habits, impulse tendencies, hedonic, gain, and
normative goals. Overall, attitudes and emotions are the most strongly correlated with
the other variables, while the weakest correlations are between the gain goals and
other variables. We find that habits are strongly positively correlated with hedonic and
normative goals. As expected, habits correlate negatively with the impulse tendency
(-0.46), meaning those who stick to certain products are less prone to spontaneous
shopping. In general, customers with negative attitudes and feelings, and who are
against norms and habits, tend to purchase wine more impulsively.
17
Table 4. Triangular matrix of correlations among latent constructs of behavior (bold, underlined values
represent strong correlations, and italic values show moderate correlations).
V
ariables
Cognitive
Attitude
_
PBC
0.31 _
Hedonic
goal 0.48 0.23 _
Gain
goal -0.24 -0.1 -0.22 _
Habits
0.59 0.23 0.42 -0.18 _
Normative
Social
norms 0.47 0.24 0.46 -0.16 0.49 _
Normative
goal 0.48 0.18 0.36 -0.13 0.52 0.5 _
Affective
Emotions
0.56 0.24 0.45 -0.2 0.61 0.49 0.54 _
Impulse
tendency -0.42 -0.18 -0.24 0.13 -0.46 -0.35 -0.45 -0.51 _
Variables
Attitude
PBC
Hedonic
goal
Gain
goal
Habits
Social
norms
Normativ
e goal
Emotions
Impulse
tendency
Cognitive Normative Affective
Furthermore, we calculate the correlation matrix for the relationships between wine
purchasing intentions and behavior and all the database variables. Table 5 shows that
both intention and behavior are strongly and positively correlated with hedonic goals
(likeness, taste, distinction), attitudes (health belief, environmental belief, and trust),
habits, emotions, social norms (special occasion and socializing), and shopping and
drinking-related patterns (wine drinking frequency, purchasing frequency, shopping
size, time spent at the wine shop, and the average price paid for wine). At the same
time, demographics, including gender, family size, education, and income, are
moderately correlated with intention and behavior. Moreover, the relationships
between impulse tendency, wine substitution (if the products are unavailable), and
organic wine purchasing intention and behavior are negative. Appendix B presents the
details of the correlation analysis for all database variables.
18
Table 5. Correlations between intention and behavior for purchasing organic wine and other variables,
where strong correlations are bold and underlined, and moderate correlations are in italics.
Organic
purchasing
intention
Organic
purchasing
behavior
Socio-
demographic
factors
Gender -0.28 -0.33
Retired 0.31 0.38
Household size 0.2 0.32
Average household education 0.37 0.34
Average household income level 0.28 0.31
Shopping and
drinking-related
patterns
Average wine shopping size per month 0.43 0.51
Wine drinking frequency 0.45 0.53
Wine purchasing frequency 0.5 0.64
Time spent in wine shops 0.42 0.45
Loyal to certain brand of wine 0.26 0.28
Average price paid for wine 0.54 0.6
Maximum price willing to pay for wine 0.26 0.22
Behavioral factors
Cognitive
Like organic wine 0.49 0.61
Distinction between organic and conventional wine 0.47 0.51
Perceive organic wine tastier 0.48 0.56
Habitual wine purchasing 0.45 0.53
Environment belief for organic wine 0.57 0.5
Health belief for organic wine 0.53 0.51
Trust in organic wine 0.59 0.56
Price importance for purchasing wine 0.32 0.29
If price increases, cheaper substitution -0.27 -0.2
If price increases, no substitution -0.32 -0.19
If price increase, loyalty 0.51 0.33
If unavailable, no substitution 0.28 0.34
If unavailable, cheaper substitution -0.2 -0.22
If unavailable, expensive substitution -0.15 -0.2
Normative
Influence of family 0.37 0.4
Influence of friends 0.33 0.35
Influence of other shopper 0.39 0.46
Influence of social media 0.46 0.53
Frequency of talking about wine when socializing 0.41 0.45
Organic wine for special occasion 0.51 0.66
Affective Positive emotions during shopping 0.46 0.63
Impulsive/spontaneous shopping -0.31 -0.29
19
4.3. Supervised machine learning: Classification analysis
While correlation analysis for the entire dataset elicits only linear, rough associations,
its results cannot be used for describing nonlinear relationships and making
predictions. Moreover, correlations analysis fails to describe the causalities. Hence, to
imply the causation of organic wine purchasing and select the classification algorithm
with the highest accuracy, efficiency, and prediction power, we compare the
performance of SVM, LR, DT, and RF in predicting consumers’ intentions (4.3.1) and
behavior (4.3.2). The comparison helped us to select the best performing algorithm in
our survey data, understand the causal factors of organic wine buying behavior, and
derive predictive models for consumers’ preferences.
4.3.1. Predicting consumers’ intentions to purchase organic wine
We test the considered supervised algorithms on the 6 classes of intentions (Section
3.2.2) and also combine classes 2 and 3 as well as 4 and 5 to decrease granularity
(Figure 3). The highest accuracy in predicting the likelihood that a consumer will have
an intention to buy organic products is achieved if we consider 4 combined classes of
intention: “not willing to pay” (a premium), “willing to pay 10% and 20% more”, “willing
to pay 30% and 40% more”, and “willing to pay 50% and higher more.” In all cases,
RF outperform the other algorithms (DT, SVM, and LR), while LR had the lowest
accuracy (Figure 3). Nonparametric algorithms are better able to handle homogeneity
amongst classes, resulting in higher accuracy and higher efficiency in processing
complex and highly dimensional datasets. Appendix C1 provides the details of the
analyses and the decision tree resulting from RF model for predicting 4 classes.
20
Figure 3. Comparing the performance of the algorithms (i.e., support vector machine (SVM), logit
regression (LR), decision tree (DT), random forest (RF)) in predicting consumers’ intentions across
three models. The original 6 classes of intention range from not willing to pay a premium for organic
products (class 1) to willing to pay more than 50% for organic (class 6).
Apart from delivering predictive models, RF provides a deeper understanding and
useful information about the relative importance of different variables affecting overall
accuracy (Table 6). We find that for organic wine intention, consumers’ trust in organic
wine has the highest predictive power, followed by environmental belief in organic wine
and the average price paid for a bottle of wine (importance weights varied between
0.04 and 0.06 in the three models). On the contrary, factors such as age, loyalty, wine
availability, and special occasions are least important (importance weight of 0.02, only
in one model). Besides trust in organic farming, environmental belief about organic
wine, positive emotions, higher payment for wines, more hedonic motivations, habitual
purchasing, and high-frequency wine drinking and purchasing are associated with
greater intention to purchase organic wine.
21
Table 6. The importance of factors in predicting intention according to the Random Forest analysis
(variables repeated in the three models are indicated with *; the most important factor and numbers
are underlined and bolded). The numbers indicate the weights, where 0.06 has the highest and 0.02
has the lowest influence on the predictions.
Factors Variables used in RF model Importanc
e in 6 class
model
Importanc
e in 5 class
model
Importanc
e in 4 class
model
Behavioral factors
Cognitive
Like organic wine* 0.02 0.03 0.02
Perceive organic wine tastier - 0.03 0.03
Trust in organic wine* 0.05 0.05 0.06
Environmental belief about organic wine* 0.05 0.05 0.05
Health belief about organic wine - 0.03 0.03
Habitual wine purchasing* 0.03 0.03 0.03
Distinction between organic and conventional wine 0.02 - 0.02
Wine price importance 0.02 - 0.02
Wine availability importance - - 0.02
Normative
Talking about wine when socializing* 0.02 0.03 0.03
Organic wine for special occasion - - 0.02
Family and friend influence 0.02 - 0.02
Other shoppers influence 0.02 - 0.02
Wine shop staf
f
influence 0.02 0.03 -
Social media influence on wine choice - - 0.02
Affective Positive emotions* 0.03 0.04 0.05
Impulsive shopping tendencies* 0.03 0.03 0.02
Shopping and
drinking-related
patterns
Average price paid for wine* 0.05 0.04 0.06
Time spent in wine shop * 0.02 0.03 0.02
Wine purchasing frequency* 0.03 0.04 0.04
Average wine purchasing size* 0.02 0.03 0.04
Wine drinking frequency* 0.03 0.03 0.03
Frequency of comparing different wine prices 0.02 - 0.02
Loyalty to a certain brand - - 0.02
Socio-
demograph
ic factors
Household average income 0.02 - 0.02
Household highest education - - 0.03
Age 0.02 - 0.02
Household size 0.02 - 0.02
Gende
r
- - 0.02
4.3.2. Predicting consumers’ likelihood of purchasing organic wine
We assess the accuracy of the predictive models of the different algorithms for
estimating the probability of purchasing organic wine. Similar to intention prediction,
RF outperformed the other algorithms, but SVM had the worst performance. Moreover,
DT and LR demonstrated comparable performance, except in predicting 3 classes,
where DT outperformed (Figure 4). Appendix C2 provides the details of the analyses,
and the decision tree resulted from RF model for predicting 3 classes.
22
Furthermore, we measure the importance of all predictor variables and keep the
significant variables in the model. However, there is no full agreement among models
about the importance of the variables. For example, the 5-class model indicates that
positive emotions and the average price paid for wine had the strongest influence,
while the 4-class model indicates that special occasion is the most important factor
(for more details, please refer to Appendix C3). Thus, we test the performance of the
models when the intention variable is included in our analysis as another predictive
factor.
Figure 4. Comparing the performance of the different algorithms (i.e., support vector machine (SVM),
logit regression (LR), decision tree (DT), random forest (RF)) in predicting wine purchasing behavior.
Regarding the model accuracy, the inclusion of intention leads to no improvements.
However, we find that the average price paid for wine is consistently the most
important factor in predicting organic wine behavior, as shown in Table 7 (importance
weights between 0.07 and 0.1). Shopping and drinking-related patterns play a similar
role in predictor behavior, as observed in relation to intention. Consumers who more
frequently purchased more bottles of wine, reported drinking more often, and spend
more time at the shops were more likely to purchase organic wine. Behavioral factors,
including cognitive (i.e., intention, attitude, habits), normative (i.e., purchase
occasions, social media), and affective (only emotions) are other emergent proxies for
23
organic wine purchasing behavior. Finally, socio-demographic factors appeared to be
unimportant in predicting purchasing decisions.
Table 7. The importance of factors in organic wine purchasing behavior according to random forest
analysis (variables repeated in three models are indicated with * and the most important factor is
underlined).
Factors Variables used in RF model Importance
in 5 class
model
Importance
in 4 class
model
Importance
in 3 class
model
Behavioral factors
Cognitive
Intention for purchasing wine* 0.05 0.05 0.06
Trust organic wine* 0.03 0.02 0.02
Health belief about organic wine* 0.03 0.03 0.02
Environmental belief about organic wine* 0.03 0.02 0.02
Habitual wine purchasing* 0.03 0.04 0.03
Like organic wine* 0.03 0.03 0.07
Distinction between organic and conventional wine - - 0.05
Normative
Influence of social media* 0.03 0.04 0.04
Organic wine for special occasion* 0.04 0.03 0.08
Influence of other shoppers 0.02 - -
Affective Positive emotions* 0.04 0.04 0.03
Shopping
and
drinking-
related
patterns
Average price paid for wine* 0.07 0.10 0.09
Wine purchasing frequency* 0.04 0.04 0.06
Time spent in wine shop* 0.03 0.06 0.04
Wine drinking frequency* 0.03 0.03 0.04
Average wine purchasing size* 0.03 0.04 0.03
Socio-
demograp
hic
factors
Income 0.02 - -
4.4. Unsupervised machine learning: Cluster analysis
Here, we determine how the data is distributed in the space, explore what groups of
similar examples exist within the data, and examine whether their characteristics can
be described by behavioral theories. The HDBSCAN method identified three hidden
heterogeneous clusters of consumers (Figure 5). The size of each cluster varied from
a minimum of 63 (7%) for cluster 1 to a maximum of 326 (33%) and 327 (33%) for
clusters 2 and 3, with 29% of data labelled as noise. Although this percentage of noise
24
may seem high, the literature (e.g., Chen et al. (2018) and Maurus and Plant (2016))
indicated that such a level of noise in the data is common in density-based algorithm
studies. We compared the characteristics of clusters in terms of the different variables.
Clusters exhibit significant differences in terms of demographics (e.g., income,
education), behavioral factors (e.g., attitudes, habits, emotions), and shopping and
drinking-related patterns (e.g., wine drinking, purchasing frequency), see Figure 6. We
label these clusters as non-organic (Section 4.4.1), occasional organic (Section 4.4.2),
and organic segments (Section 4.4.3).
25
Figure 5. HDBSCAN results with three clusters (1, 2, and 3) in six dimensions. Clusters 1, 2, and 3 are represented by circles, diamonds, and
triangles, respectively. Cluster 0 is noise. The distributions show a clear clustering, where data falls into three groups or types. The clouds
present the density of clusters on each dimension.
26
Figure 6. Variables according to which the three clusters (1, 2, and 3) are segregated. Special occasion (no=0, yes=1) and Gender are binary variables
(male=0, female=1). The clusters are clearly different according to most of the variables, while there are some overlaps in others.
27
4.4.1. Non-organic segment: Impulsive behavior
Cluster 1, the non-organic segment, mainly represents conventional wine
consumers. They report the lowest wine consumption and usually purchased items
spontaneously. The gap between their higher intention (WTP 20% more) and lower
organic purchasing behavior (organic wine purchasing less than 25%) is well
explained by high impulsiveness, in line with IBT and affective events theory. The wine
drinking and shopping frequency of this cluster were the lowest. Conventional wine
consumers expressed negative feelings during shopping. They did not like the taste
of organic over conventional wine or reported no distinction between the two, implying
that hedonic goals were not activated. Yet, according to GFT, hedonic goal is one of
the main drivers of behavior. Although they reported that health and environmental
impacts are relatively important decision factors, they were less convinced that organic
products have health and environmental benefits and did not trust them. Social norms
influenced the wine purchasing decisions of these consumers very slightly. They
stated that, in case of an increase in the price of their favorite wine, they would look
for a cheaper substitute. In fact, they reported less loyalty to a certain brand of wine
compared to other clusters. Regarding socio-demographic factors, consumers in this
cluster were mainly poorly educated, lower-income women who had small-size
households.
4.4.2. Occasional organic segment: Planned behavior
Cluster 2, the occasional organic segment, represents the bulk of the consumers with
the highest potential for organic wine adoption. The intentions and behavior of these
consumers were well aligned (WTP 10-20% more for organic and purchasing 25-50%
28
of wines organic), indicating planned wine purchasing behavior, which is aligned with
TPB. For this cluster, the price was by far the main driver preventing organic wine
purchasing decisions in this cluster: when the price of organic wine increases, they
are unlikely to purchase it anymore (no substitution). Although the average price paid
for wine in this cluster was similar to cluster 1, organic wine was mostly purchased for
special occasions. In general, these consumers believed in the environmental and
health benefits of organic wine consumption. Still, due to its high price, they only
purchased it for celebrations or as a gift. Compared to conventional consumers,
occasional organic consumers had relatively higher education, income, family size,
brand loyalty, and interest in drinking organic wine and were less prone to impulsive
wine shopping.
4.4.3. Organic segment: Unplanned behavior
Consumers in cluster 3, the organic segment, were mainly men with the highest
education and highest income levels, living in big families. The average share of
organic wine in their basket was more than 50%, higher than their reported intention
(WTP varied between 20-50%). They based their choice primarily on normative goals
and habits. On the one hand, the influence of family, friends, and other shoppers
choices on their wine purchasing decisions was the highest. They looked for more
information about different wines from social media and sought the advice of others
when selecting wine (in line with GFT). On the other hand, they were generally happy
during shopping (in line with IBT) and tended to buy items habitually (in line with AT).
Thus, the characteristics of this class are representative of unplanned wine purchasing
behavior. Consumers in this cluster are strongly concerned with the health and
environmental impacts of their food choices. Changes in the price of wine have a low
29
impact on their demand, and their average price acceptance is at a maximum. In other
words, the price elasticity of this cluster is low, and if the prices of products increase,
consumers will continue to purchase at higher prices.
5. Discussion
Our findings confirm the presence of planned, unplanned, and impulsive behaviors
when shopping for wine. The following discussion of the results highlights several
factors that can explain consumers’ wine preferences.
Regarding the cognitive factors, RF models showed that trust adds substantially to
the prediction of intentions (similar observation was made by Kim and Bonn (2015) as
well). In line with D’Amico et al. (2016), the present study found that environmental
consciousness and curiosity were associated with consumer WTP a premium for
organic wines. When it comes to purchasing behavior, health attributes were found to
be an important motivator for purchasing organic wine. This finding is consistent with
the studies of Rana and Paul (2017) and Yadav (2016). Having said that, we found
that consumers in cluster 2 mainly purchase conventional wine, despite their positive
attitudes towards the health and environmental beliefs associated with organic
products. Hence, we could not confirm that attitudes strongly predict behavior, as
hypothesized by IBT and AT. Prior studies have reported contradictory results
regarding the importance of taste on organic wine purchasing behavior (Mann et al.
2012). Nevertheless, our classification and clustering analyses were consistent with
the study by Kim and Bonn (2015), in which American consumers reported taste as
an important factor in their wine choice, favoring GFT when explaining consumers’
behavior.
30
The influence of habits on more WTP for organic wine has not been sufficiently
explained by the results of other studies in this context. Most wine-related studies
define habits as the repetition of behavior as assessed by frequencies of shopping and
drinking (e.g., Pomarici et al. (2016) and Vecchio (2013)), whereas, here, we
considered habit as cognitively effortless and automatically initiated behavior, as
assessed by the automaticity-specific index (Gardner et al. 2012). Our findings
highlighted that while habitual purchasing as suggested by AT is important for
promoting both organic wine purchasing intention and behavior, it could shed light on
establishing stable shift towards organic consumption and its causality is not
confirmed. Contrary to our expectations, habits did not override intention in directing
behavior, and intentions remained significantly and equally predictive of behavior in all
models: consumers choose wine mindfully rather than habitually. Gardner et al. (2015)
referred to the temporal self-regulation theory to explain similar observations in terms
of unhealthy snacking behavior, where strong self-control inhibits the habit. Similar to
other organic wine studies (e.g., Pagliarini et al. (2013)), we found that WTP more for
organic wine (intention) strongly influenced organic wine behavior. The classification
results showed that, on average, consumers with higher WTP for organic wine also
had a higher probability of buying it. However, our cluster analysis detected clusters
of consumers with relatively higher intentions and lower behavior for organic wine
(clusters 1 and 3). A similar gap between intention and behavior for organic wine has
been described by Schäufele and Hamm (2018), who found that attitude and price are
the barriers to organic wine adoption.
Regarding the normative factors, we found that normative support, as provided by
social media and purchasing occasions, was relevant in determining consumers’
organic wine purchasing behavior. This result highlights the influence of wine reviews
31
and recommendation systems on consumers’ choices as suggested by GFT. It also
highlights new potentials and opportunities for social media to assist businesses and
industries to influence consumers’ preferences. Szolnoki et al. (2018) and Dolan and
Goodman (2017) both recently investigated the application of social media for
promoting wine. Moreover, in line with the study of Boncinelli et al. (2019), in the
current study, consumers valued organic wine more for special occasions rather than
personal consumption. Concerning the clustering results, this statement stands true
for 33% of consumers (occasional segment), while for the rest, it might not be the
case, as occasions only partially influenced their wine purchasing decisions.
Regarding the affective factors, our findings demonstrate that happier, positive, and
optimistic consumers are more likely to pay more for organic wine. Consistent with the
study by Danner et al. (2016), positive and negative emotions were predictive of WTP
more for organic wine. The influence of impulsive tendencies on organic wine
purchasing decisions was more prevalent in the cluster analysis. On the one hand,
consistent with IBT, impulsiveness caused by negative emotions may prompt
spontaneous behavior that may, in turn, drive the consumer towards purchasing more
conventional wine. On the other hand, unplanned decisions triggered by habits and
normative cues may lead to higher organic purchasing if a consumer experiences
positive emotions. Therefore, we can relate the effects of emotions to either impulsive
or habitual behavior. Despite the importance of impulsiveness in predicting wine
purchasing decisions, we only found one study, by Feldmann and Hamm (2015), that
has highlighted the influence of spontaneous purchase situations.
Regarding the shopping and drinking-related patterns, the classification method
indicated that the average price paid for wine was the strongest predictor and the
32
source of heterogeneity in the average behaviors of consumers. A higher price
acceptance increases the likelihood that a consumer is more willing to pay a premium
for organic wine. In the literature, the findings are mixed regarding the importance of
price for buying wine (Huang et al. 2017); however, our results are in line with the
studies of Schäufele and Hamm (2018) and Di Vita et al. (2019) who reported that, for
the majority of consumers, price is the pivotal driver of wine choices. Another
interesting result of the current study is that while consumers state they generally pay
little attention to wine prices (about 70% of respondents), they actually base their
organic wine purchasing decisions primarily on ‘price’. While wine prices were
considered to be the best predictor of organic wine purchasing behavior according to
the RF model, the HDBSCAN model identified clusters that have equal average price
acceptance, but the proportion of organic wine in their shopping baskets differed (refer
to Figure 6, where organic wine in the shopping basket was less than 25% for cluster
1 and between 25-50% for cluster 2). The type of consumer behavior can explain this
inconsistency in results; the wine purchasing decisions of cluster 2 consumers are
more planned, whereas the decisions of cluster 1 consumers are more impulsive. The
conventional segment consumers may change their preference for organic
consumption if they experience positive emotions (like joy and contentment) during
shopping and practice more planned buying as hypothesized by TIB. Interestingly, for
the organic food segment, cluster 3, food price was the most important wine attribute,
and that is why their high WTP more for organic wine (between 20% and 50%) cannot
lead to full adoption of organic wine. The present findings seem to confirm GFT and
support the findings of Janssen et al. (2020), where both conventional and organic
food consumers reported that price was the most important attribute when making
purchasing decisions.
33
Apart from the average price paid, variables such as the duration of shopping,
average purchasing size, and the frequency of purchasing and drinking wine were
found to be strong predictors of both intentions and behavior. It seems that consumers
who spend a long time in the shop searching for products are likely to be willing to pay
more for organic wine. Further, the more wines purchased per month, the higher the
likelihood of intentions and behavior for purchasing organic wine. In line with previous
studies, such as those by Pomarici and Vecchio (2014), higher frequencies of
consuming and purchasing wine are related to a higher WTP more for organic wine.
Regarding socio-demographics, in agreement with the study by Zepeda and Deal
(2009), the classification results indicated that socio-demographic factors have the
lowest predictive power and are poor proxies for intention and behavior models.
However, our clustering results revealed significant differences in income, education,
household size, and gender between organic and conventional wine consumers.
6. Conclusions
Our findings have important implications for both theory and practice. From a
theoretical perspective, they underscore the importance of considering impulsive and
unplanned, as well as planned behavior, in understanding food purchasing. We argue
that organic purchasing decisions result from an interplay between these factors, as
explained by different social theories. Relying only on TPB and disregarding the
presence of interruptive factors between intention and behavior means that we are
unlikely to adequately capture the decision-making processes for organic food
purchasing. TIB, AT, GFT, and IBT can explain the intention-behavior gap for different
consumer segments when purchasing organic products.
34
6.1. Managerial implications
From a practical perspective, the classification results suggest that, for the average
person, price is still an obstacle to purchasing organic food. The clustering results
provide strong evidence of the influence of impulsive, habitual, and normative cues as
well as the dual role of emotions in choosing organic products in three distinct
consumer segments. In fact, we would have highlighted these two factors (trust and
price) as the most important attributes in wine purchasing if we had only used
classification algorithms.
Sales promotions and government subsidies for organic products can support
organic purchasing and, at the same time, change consumer consumption habits to
help the environment. Retailers can have an organic section in their stores specifically
designed to facilitate this behavior. Encouraging a greater sense of joy and happiness
in the store, and using social media to advertise a range of organic products, may be
other effective mechanisms to change wine purchasing behavior. We may be ignoring
the influence of affective factors if we rely only on the results of the classification
analysis. Future research would benefit from examining the efficacy of these
interventions in shifting behavior towards organic consumption.
6.2. Limitations and future directions
This study has a number of limitations that suggest several potential directions for
future research. Firstly, the reliance on self-reported behavior rather than conducting
observational experiments is a limitation. Survey respondents are prone to social
desirability bias in reporting their intention for organic products, and their behavior can
only be interpreted as a reported preference; it is not their real behavior. Thus, the
35
findings of this study are based on stated preferences and are experimental in nature.
One possible future direction is to compliment survey data with real market
transactions that provide realistic, robust results. Another limitation of this study is
relevant to the geographical constraints of the sample and the generalizability of the
results. Our data were collected from one region of a major city in Australia and there
is a possibility that the results are more closely aligned to the perspectives of these
particular residents. Thus, the results cannot be generalized to the entire national
population. Future research may choose to broaden the participant recruitment
process or conduct a comparative study on the differences and similarities between
organic wine preferences of Australians across different states and other populations.
Finally, the impact of packaging, region of origin, grape variety, and other extrinsic
characteristics on organic wine purchasing can be explored in future research.
This said the article provides important contributions to the literature. Using
explainable AI techniques, we advance the methodological principles of empirical
research on retailing and consumer behavior. Besides, our study provides new
insights into the role of emotions and norms in decisions making process for food with
sustainability characteristics. Further, the presented ML algorithms can be used to
inform the extended supply chain framework (Taghikhah et al. 2019) to predict
consumer motivation and behavior for green products across a variety of segments.
The results also allow us to further calibrate and test the agent-based model, ORVin
(Taghikhah et al. 2020a), developed to quantify the cumulative impacts of organic wine
preference changes among heterogeneous consumers prone to behavioral biases
and social interactions.
36
Acknowledgments
Authors wish to thank the editor and three anonymous reviewers for their valuable comments
and suggestions on this manuscript.
We would like to thank PERSWADE center, Australian Mathematical Sciences Institute
(AMSI) and Australian Postgraduate Research Intern (APR.Intern) for funding the survey.
We also thank Dr Muhammad Saqib, from Centre for artificial intelligence at University of
Technology Sydney and Ivan Bakhshayeshi, from University of New South Wales, for their
technical help with data analysis.
References
Ajzen, I. (1991). The theory of planned behavior. Organizational behavior and human decision
processes, 50(2), 179-211. doi:https://doi.org/10.1016/0749-5978(91)90020-T
Boncinelli, F., Dominici, A., Gerini, F., & Marone, E. (2019). Consumers wine preferences
according to purchase occasion: Personal consumption and gift-giving. Food Quality
and Preference, 71, 270-278. doi:https://doi.org/10.1016/j.foodqual.2018.07.013
Campello, R. J., Moulavi, D., & Sander, J. (2013). Density-based clustering based on
hierarchical density estimates. Paper presented at the Pacific-Asia conference on
knowledge discovery and data mining.
Chen, Y., Tang, S., Bouguila, N., Wang, C., Du, J., & Li, H. (2018). A fast clustering algorithm
based on pruning unnecessary distance computations in DBSCAN for high-
dimensional data. Pattern Recognition, 83, 375-387.
doi:https://doi.org/10.1016/j.patcog.2018.05.030
37
Cohen, J. (1992). Statistical power analysis. Current directions in psychological science, 1(3),
98-101.
Cohen, J. (2013). Statistical power analysis for the behavioral sciences: Academic press.
Cortes, C., & Vapnik, V. (1995). Support-vector networks. Machine learning, 20(3), 273-297.
D’Amico, M., Di Vita, G., & Monaco, L. (2016). Exploring environmental consciousness and
consumer preferences for organic wines without sulfites. Journal of cleaner production,
120, 64-71. doi:https://doi.org/10.1016/j.jclepro.2016.02.014
Danner, L., Ristic, R., Johnson, T. E., Meiselman, H. L., Hoek, A. C., Jeffery, D. W., & Bastian,
S. E. (2016). Context and wine quality effects on consumers’ mood, emotions, liking
and willingness to pay for Australian Shiraz wines. Food Research International, 89,
254-265. doi:https://doi.org/10.1016/j.foodres.2016.08.006
Di Vita, G., Pappalardo, G., Chinnici, G., La Via, G., & D’Amico, M. (2019). Not everything has
been still explored: Further thoughts on additional price for the organic wine. Journal
of cleaner production, 231, 520-528. doi:https://doi.org/10.1016/j.jclepro.2019.05.268
Dolan, R., & Goodman, S. (2017). Succeeding on social media: Exploring communication
strategies for wine marketing. Journal of Hospitality and Tourism Management, 33, 23-
30.
Ester, M., Kriegel, H.-P., Sander, J., & Xu, X. (1996). A density-based algorithm for discovering
clusters in large spatial databases with noise. Paper presented at the Kdd.
Feldmann, C., & Hamm, U. (2015). Consumers’ perceptions and preferences for local food: A
review. Food quality and preference, 40, 152-164.
doi:https://doi.org/10.1016/j.foodqual.2014.09.014
Gardner, B., Abraham, C., Lally, P., & de Bruijn, G.-J. (2012). Towards parsimony in habit
measurement: Testing the convergent and predictive validity of an automaticity
subscale of the Self-Report Habit Index. International Journal of Behavioral Nutrition
and Physical Activity, 9(1), 102.
38
Gardner, B., Corbridge, S., & McGowan, L. (2015). Do habits always override intentions?
Pitting unhealthy snacking habits against snack-avoidance intentions. BMC
psychology, 3(1), 8. doi:10.1186/s40359-015-0065-4
Gentile, F., La Torre, G. L., Potortì, A. G., Saitta, M., Alfa, M., & Dugo, G. (2016). Organic wine
safety: UPLC-FLD determination of Ochratoxin A in Southern Italy wines from organic
farming and winemaking. Food control, 59, 20-26.
Helmefalk, M., & Hultén, B. (2017). Multi-sensory congruent cues in designing retail store
atmosphere: Effects on shoppers’ emotions and purchase behavior. Journal of
Retailing and Consumer Services, 38, 1-11.
doi:https://doi.org/10.1016/j.jretconser.2017.04.007
Ho, T. K. (1998). The random subspace method for constructing decision forests. IEEE
transactions on pattern analysis and machine intelligence, 20(8), 832-844.
Huang, A., Dawes, J., Lockshin, L., & Greenacre, L. (2017). Consumer response to price
changes in higher-priced brands. Journal of Retailing and Consumer Services, 39, 1-
10.
Janssen, M., Schäufele, I., & Zander, K. (2020). Target groups for organic wine: The
importance of segmentation analysis. Food quality and preference, 79, 103785.
Kim, H., & Bonn, M. A. (2015). The moderating effects of overall and organic wine knowledge
on consumer behavioral intention. Scandinavian Journal of Hospitality and Tourism,
15(3), 295-310. doi:https://doi.org/10.1080/15022250.2015.1007083
Lindenberg, S., & Steg, L. (2007). Normative, gain and hedonic goal frames guiding
environmental behavior. Journal of social issues, 63(1), 117-137.
doi:http://10.1111/j.1540-4560.2007.00499.x
Loose, S. M., & Lockshin, L. (2013). Testing the robustness of best worst scaling for cross-
national segmentation with different numbers of choice sets. Food quality and
preference, 27(2), 230-242. doi:https://doi.org/10.1016/j.foodqual.2012.02.002
39
Mann, S., Ferjani, A., & Reissig, L. (2012). What matters to consumers of organic wine? British
Food Journal, 114(2), 272-284. doi:https://doi.org/10.1108/00070701211202430
Maurus, S., & Plant, C. (2016). Skinny-dip: Clustering in a Sea of Noise. Paper presented at
the Proceedings of the 22nd ACM SIGKDD international conference on Knowledge
discovery and data mining.
Ogbeide, O. A. (2013). Consumer willingness to pay premiums for the benefits of organic wine
and the expert service of wine retailers.
Pagliarini, E., Laureati, M., & Gaeta, D. (2013). Sensory descriptors, hedonic perception and
consumer’s attitudes to Sangiovese red wine deriving from organically and
conventionally grown grapes. Frontiers in psychology, 4, 896. doi:
https://doi.org/10.3389/fpsyg.2013.00896
Panzone, L. A. (2014). Why are discounted prices presented with full prices? The role of
external price information on consumers’ likelihood to purchase. Food Quality and
Preference, 31, 69-80. doi:https://doi.org/10.1016/j.foodqual.2013.08.003
Pomarici, E., Amato, M., & Vecchio, R. (2016). Environmental friendly wines: a consumer
segmentation study. Agriculture and agricultural science procedia, 8(2016), 534-541.
Pomarici, E., & Vecchio, R. (2014). Millennial generation attitudes to sustainable wine: an
exploratory study on Italian consumers. Journal of cleaner production, 66, 537-545.
doi:https://doi.org/10.1016/j.jclepro.2013.10.058
Quinlan, J. R. (1990). Probabilistic decision trees. In Machine learning (pp. 140-152): Elsevier.
Rahmah, N., & Sitanggang, I. S. (2016). Determination of optimal epsilon (eps) value on
dbscan algorithm to clustering data on peatland hotspots in sumatra. Paper presented
at the IOP Conference Series: Earth and Environmental Science.
Rana, J., & Paul, J. (2017). Consumer behavior and purchase intention for organic food: A
review and research agenda. Journal of Retailing and Consumer Services, 38, 157-
165.
40
Rugani, B., Vázquez-Rowe, I., Benedetto, G., & Benetto, E. (2013). A comprehensive review
of carbon footprint analysis as an extended environmental indicator in the wine sector.
Journal of cleaner production, 54, 61-77.
doi:https://doi.org/10.1016/j.jclepro.2013.04.036
Russell, S. V., Young, C. W., Unsworth, K. L., & Robinson, C. (2017). Bringing habits and
emotions into food waste behaviour. Resources, Conservation and Recycling, 125,
107-114. doi:https://doi.org/10.1016/j.resconrec.2017.06.007
Ryan, J., & Casidy, R. (2018). The role of brand reputation in organic food consumption: A
behavioral reasoning perspective. Journal of Retailing and Consumer Services, 41,
239-247.
Schäufele, I., & Hamm, U. (2017). Consumers’ perceptions, preferences and willingness-to-
pay for wine with sustainability characteristics: A review. Journal of cleaner production,
147, 379-394. doi:https://doi.org/10.1016/j.jclepro.2017.01.118
Schäufele, I., & Hamm, U. (2018). Organic wine purchase behaviour in Germany: Exploring
the attitude-behaviour-gap with data from a household panel. Food Quality and
Preference, 63, 1-11. doi:https://doi.org/10.1016/j.foodqual.2017.07.010
Stern, H. (1962). The significance of impulse buying today. Journal of marketing, 26(2), 59-
62. doi:https://doi.org/10.1177/002224296202600212
Stolz, H., & Schmid, O. (2008). Consumer attitudes and expectations of organic wine.
http://www.isofar.org/modena2008/prceedings.html
Szolnoki, G., Dolan, R., Forbes, S., Thach, L., & Goodman, S. (2018). Using social media for
consumer interaction: An international comparison of winery adoption and activity.
Wine Economics and Policy, 7(2), 109-119.
Taghikhah, F., Voinov, A., & Shukla, N. (2019). Extending the supply chain to address
sustainability. Journal of cleaner production(229), 652-666.
41
Taghikhah, F., Voinov, A., Shukla, N., & Filatova, T. (2020a). Exploring consumer behavior
and policy options in organic food adoption: Insights from the Australian wine sector.
Environmental science & policy, 109, 116-124.
Taghikhah, F., Voinov, A., Shukla, N., Filatova, T., & Anufriev, M. (2020b). Integrated modeling
of extended agro-food supply chains: A systems approach. European Journal of
Operation Research(Accepted).
Triandis, H. C. (1977). Interpersonal behavior: Brooks/Cole Pub. Co.
Trinh, G., Corsi, A., & Lockshin, L. (2019). How country of origins of food products compete
and grow. Journal of Retailing and Consumer Services, 49, 231-241.
Vecchio, R. (2013). Determinants of willingness-to-pay for sustainable wine: Evidence from
experimental auctions. Wine Economics and Policy, 2(2), 85-92.
doi:https://doi.org/10.1016/j.wep.2013.11.002
Verplanken, B., & Orbell, S. (2003). Reflections on past behavior: a self-report index of habit
strength 1. Journal of applied social psychology, 33(6), 1313-1330.
doi:https://doi.org/10.1111/j.1559-1816.2003.tb01951.x
Vrček, I. V., Bojić, M., Žuntar, I., Mendaš, G., & Medić-Šarić, M. (2011). Phenol content,
antioxidant activity and metal composition of Croatian wines deriving from organically
and conventionally grown grapes. Food Chemistry, 124(1), 354-361.
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures
of positive and negative affect: the PANAS scales. Journal of personality and social
psychology, 54(6), 1063.
Yadav, R. (2016). Altruistic or egoistic: Which value promotes organic food consumption
among young consumers? A study in the context of a developing nation. Journal of
Retailing and Consumer Services, 33, 92-97.
Yang, Y., & Paladino, A. (2015). The case of wine: understanding Chinese gift-giving behavior.
Marketing Letters, 26(3), 335-361. doi:https://doi.org/10.1007/s11002-015-9355-0
42
Zepeda, L., & Deal, D. (2009). Organic and local food consumer behaviour: Alphabet theory.
International Journal of Consumer Studies, 33(6), 697-705.
doi:https://doi.org/10.1111/j.1470-6431.2009.00814.x
... Sun et al., 2022). Similarly, the cognitive association is also playing a key role to develop value perception in the customer for any brand because the markets are mature and the customer is well educated and well informed (Jang & Hsieh, 2021;Nassani et al., 2013).Therefore, they never go for purchasing any product without any prior knowledge (Taghikhah et al., 2021). Moreover, marketing innovation is also playing a key role to develop customer satisfaction (Yalley, 2021), because the purpose of marketing innovation is to provide the customer product or service with benefits that are crucial and satisfactory for the needs of the consumers (Di Crosta et al., 2021;Gutter et al., 2010). ...
... Fourthly, the findings of hypothesis 4 explain that there is a significant relationship between consumer lifestyle and consumer satisfaction. Similarly, the consumers are divided into innovators, early adopters, early majority, middle majority, late majority, and laggards, so according to their purchasing power and lifestyle (Gutter et al., 2010;Kutaula et al., 2022;Schiessl et al., 2022).The brands should modify their products and marketing strategies to provide enough information and satisfaction to become differentiate brand in the market that is ultimately the target of the consumer to purchase when they get involved in their life and social circle (Di Crosta et al., 2021;Goraya et al., 2020;Taghikhah et al., 2021;Xiao et al., 2011). Additionally, brands are working on the strategy of providing emotional importance to the consumers because the consumer who are emotionally attached to the brands, just got satisfaction from the same brand (Hoe & Mansori, 2018;Lee et al., 2022;Shankar, 2021).As result, they become potential and loyal consumers of that particular brand in the large or target market (Eichinger et al., 2022;Emekci, 2019;Khraim, 2011;Ottosson & Kindström, 2016). ...
Article
Full-text available
Brand equity development has become a challenge for modern brands. Marketing innovation and changing consumer value perceptions are causing consumers to leapfrog, posing a challenge for brand equity development. The data for this study was collected with random sampling techniques. 450 respondents were considered to collect data on the Likert scale questionnaire. The Partial Least Square (PLS) method of data analysis was used in this study for data analysis. According to the findings of the study, brand personality, customer satisfaction, consumer lifestyle, marketing innovation, and consumer changing value perception all have a significant impact on brand equity. The significant framework of this study is a contribution to the body of knowledge as it describes the significant relationship between different variables critical for brand equity development. Furthermore, this study provides business implications for brands working in Pakistan to improve brand equity with realistic results to influence the consumers for brand equity.
... Second, we break down these effects by product categories (vice and virtue), which differ with respect to short-and long-term consumption goals and can clarify whether and to what extent the partner's presence during shopping encourages or reduces the pursuit of these goals. In addition to the basic distinction between vice and virtue, we add organic labeling, for two reasons: on the one hand, this labeling has become enormously important in practice (Taghikhah et al., 2021), and on the other hand, it is applicable to the logic of the vice-virtue distinction in the sense that the term "organic" is often associated with "healthy" and thus not with "vice/unhealthy/tasty" (Raghunathan et al., 2006), which in principle ensures that an "organic vice" product is perceived as less "vice" than a vice product without this label. This argument provides nuance to the meaningfulness of the distinction between vice and virtue, which in this sense is a novelty in the literature. ...
... In the wake of the sustainability megatrend and the accompanying increase in ecological awareness among consumers and companies (McDonagh and Prothero, 2014;Mittelstaedt et al., 2014), organic products have experienced above-average growth in the recent past (e. g., Taghikhah et al., 2021). Therefore, we distinguish products into organic and regular (nonorganic) products. ...
Many purchase decisions take place in social relationships, and yet few studies have specifically investigated couples’ purchase decisions made during shopping about products for later joint consumption. We hypothesize that romantic partners purchase more when they shop together than individually and that this effect is strong for vice products, particularly those without an organic label. For our empirical study, we asked romantic partners shopping together in a real-life context to make purchase decisions together or individually (our main experimental condition) in a self-programmed web store that offered 88 product variants (differing in category [vice/virtue] and labeling [with/without organic label]). Participants then filled out an online questionnaire on site. Results of a sequence of nested generalized linear models show that making purchase decisions together increases purchase amount (number of items selected) and purchase value (quantities selected multiplied by the corresponding willingness to pay), especially for vice products without organic labeling. In a second study, we benchmark these effects by comparing them with the effects of individual decision making and varying consumption mode (joint vs. individual consumption), using data from an online survey that followed the same structure as the main study. These effects, again estimated through generalized linear models, are negligible. Our findings strongly support the “accomplice” (rather than the “minder”) role of romantic partners in shopping. Therefore, retailers should target couples, encourage them to shop together, and emphasize joint consumption as a shopping goal.
... Al respecto, Taghikhah et al. (2021) revelan que los factores afectivos también pueden tener un importante impacto en la conducta de compra de productos orgánicos. Asimismo, destacan que encontraron diferencias significativas en los grupos que consumen productos orgánicos respecto a los factores sociodemográficos ingreso, educación, tamaño del hogar y género. ...
Article
Full-text available
Objetivo: analizar los motivadores, las barreras y los factores sociodemográficos que inciden en la compra de productos agroecológicos y la relación que existe entre estas variables. El conocimiento de los motivadores y barreras que inciden en la compra de estos productos repercute en los planes y estrategias de mercadotecnia, principalmente de tipo social. Diseño metodológico: el enfoque de este estudio es cuantitativo, correlacional y transversal. La recopilación de la información se realizó a través de la aplicación de 719 encuestas a clientes potenciales de productos agroecológicos. Los métodos de análisis de la información empleados fueron la estadística descriptiva y la estadística no paramétrica, efectuando pruebas de independencia Chi a partir de las tablas de contingencia cuadrado correspondientes y la obtención de los coeficientes de correlación de Spearman, Tau b de Kendall, y un modelo de regresión logística binaria. Limitaciones: el trabajo de campo se hizo en el mes de noviembre de 2021, tiempo en el que había restricciones de movilidad impuestas por la pandemia COVID-19, lo que resultó en una alta tasa de rechazo para efectuar las entrevistas personales que se requirieron, por lo que, a pesar de haber logrado la muestra, esta situación puedo haber afectado la representatividad. Hallazgos: los resultados demuestran que, si el comprador tiene la motivación para contribuir con el medio ambiente o para convertirse en un consumidor responsable es más probable que tenga una alta frecuencia de compra de productos agroecológicos. Asimismo, se observaron los siguientes resultados: a mayor edad de los clientes potenciales, mayor frecuencia de compra; a menor edad de los clientes potenciales, más motivación tienen para elegir los productos agroecológicos si tienen un buen sabor, entre otros.
... Shukla et al. [10] utilized supervised and unsupervised machine learning algorithms for investigation for correlation to the causality of consuming behavior prediction when integrating predictive ability with explanatory capacity and offer a more comprehensive understanding of proenvironmental business consuming behavior. It is unable to obtain 2 ...
Article
Full-text available
Consumption behavior prediction reveals customer attributes, personal preferences, and intrinsic laws. Organizations would benefit from knowing further about customer needs and business desires by monitoring client behavior to provide more precise recommendations and boost acquisition rates. The economics of the customer, buyer groupings, and product quality are only a few of the numerous variables that influence customer behavior. The key issue that has to be resolved at this time is how to filter out useful information from these vast amounts of data to forecast customer behavior. For customer consumption behavior prediction and analysis with an advanced quantitative research process, we proposed the multiobjective evolutionary algorithm, which significantly boosts the accuracy of consumption behavior predictions. The dataset is initially gathered based on consumer preferences and behaviors as the essential information for the entire prediction model. Min-max normalization is used as a component of the preprocessing of the data to get the elimination of redundant and superfluous data. The Word2vec model is utilized for feature extraction, and boosted ant colony optimization (BACO) is employed to choose the best features. Utilizing the suggested multiobjective evolutionary algorithm (MOEA), the predictions are made. The suggested system’s performance is assessed, and the metrics are contrasted with more established methods. The findings demonstrate that the suggested MOEA technique performs well than the traditional ML, XGB, AI, and HNB algorithm methods in terms of accuracy (95 percent), quality of prediction (97 percent), precision (99 percent), recall (93 percent), F 1 -score (98 percent), and prediction time (50 seconds). Hence, the outcomes show that the regression model is sustainable. The suggested consumption behavior prediction system has demonstrated its efficiency in boosting profitability.
... The effective payment method has a role in the satisfaction of the customer when they are purchasing any product or service from any brand (Aliasghar, Sadeghi, & Rose, 2020). Indeed, the modern world demands the modern way of payment and in online businesses (Abdelaziz, 2021), the payment method is critical for the customer, because they want a reliable and effective payment method to make the transaction with the business entity (Taghikhah, Voinov, Shukla, & Filatova, 2021). Similarly, it is also noted that the brands that are failed to provide an effective payment method to the customer for the transaction of services in result, these brands are failed to achieve customer satisfaction and the appropriate level of brand loyalty in the target market (Rahmadani, Schaufeli, Stouten, Zhang, & Zulkarnain, 2020). ...
Article
Full-text available
The online brands in Pakistan are facing brand loyalty issues as the market is new and emerging for Pakistani consumers. The less satisfaction of consumers with different operations of online businesses is the fundamental reason for this loyalty challenge. The purpose of this study is to identify different factors that are contributing to the brand loyalty of online businesses in Pakistan. The Likert scale questionnaire was used to collect data from the Pakistani people with a random sampling technique. The Partial Least Square (PLS) method is used to test the relationship between different variables of the study. The study highlights that effective payment method, low distribution charges and low transit time has key responsibility for brand loyalty of the Pakistani customers with moderating role of effective management. The study is a contribution to literature and knowledge as the framework of the study contains significant variables. The practical implications of this study would provide a way for online businesses in Pakistan to develop brand loyalty.
... In fact, a gap between intentions and behaviors is typically reported in the literature. In this gap, cognitive and affective factors, together with normative cues, seem to play a crucial role and may prompt unplanned and spontaneous purchasing behavior, causing consumers to act against their beliefs (Taghikhah et al., 2021). To address the attitude-behavior-gap, Sch€ aufele and Hamm (2018) and Sch€ aufele et al. (2018) showed that consumers' preferences for organic products and sustainability concerns strongly determine purchases of organic wine (as in Olsen et al. (2012)) and that, generally, consumers' attitudes were in agreement with purchase behavior. ...
Article
Purpose Among the growing interest towards market segmentation and targeted marketing, the current study adopted a scientometric approach to examine the literature on wine selection and preferences. The current review specifically attempts to shed light on the research that explores the determinants of wine preferences at multiple levels of analysis. Design/methodology/approach CiteSpace was used to compute a Document Co-Citation Analysis (DCA) on a sample of 114,048 eligible references obtained from 2,846 publications downloaded from Scopus on 24 May 2021. Findings An optimized network of 1,505 nodes and 4,616 links was generated. Within the network, impactful publications on the topic and thematic domains of research were identified. Specifically, two thematic macro-areas were identified through a qualitative analysis of papers included in the 7 major clusters. The first one - “Methods of Wine Making” - included clusters #0, #3, #5, #6 and #18. The second one - “Consumers' Attitudes and Preferences Towards Wine” - included clusters #1 and #2. The first thematic macro-area included more technical aspects referring to the process of wine making, while the second thematic macro-area focused more on the factors influencing individuals' preferences and attitudes towards wine. To reflect the aims of the current paper, publications giving light to the “Consumers' Attitudes and Preferences Towards Wine” macro-area were analyzed in detail. Originality/value The resulting insights may help wine makers and wine sellers optimize their work in relation to market segments and to the factors influencing individuals' purchasing behaviors.
... Health, sustainability, material quality, and environmental preservation are the main drivers of organic food consumption today. Taghikhah et al., 2021Australia, 1003 Organic and conventional wine ...
Article
Full-text available
Global organic food consumption has increased significantly in the last decade. Organic product agriculture is a one-of-a-kind technique that strikes a balance between environmental sustainability and consumer safety by developing a good client relationship with the end customer or consumer group. According to environmental studies, organic farming is less damaging to the environment than traditional agricultural methods. Recent studies show that customers who eat organic foods have lower pesticide exposure, which results in fewer human diseases. Organic food has more nutrients. However, the evidence to support this is lacking, and no well-designed human study has shown any direct health benefits or disease prevention benefits from consuming organically produced food. Furthermore, some researchers mention that for some types of plants, the nutrients in the case of conventional products were better. This study aims to identify the benefits of organic foods over conventional foods and consumers' beliefs about health and environmental benefits associated with organic foods. Secondary data for this research have been gathered from different international journals and the Internet, utilizing information from a variety of scholarly publications. The author addresses the present state of organic farming, as well as the benefits and disadvantages of consuming organic and conventional food,. The article has also investigated the effects of organic and conventional food production on both health and the environment.
... There has been a biased tendency towards the use of traditional statistical models in the subject area, whereas little to no efforts have been devoted to implementing innovative techniques ( [44] is a rare case in point). Despite the capability of traditional statistical methods to reveal the relationships between parameters, their accuracy and predictive powers compared to machine learning (ML) algorithms are relatively low, particularly when handling a high volume of attributes and observations [45]. If used properly, ML methods can unearth the emergent properties and unexpected patterns of the underlying phenomena of interest. ...
Article
When residential rooftop solar photovoltaic (PV) systems are widely accepted across society, the uptake of home battery energy storage systems is closely tied to the PV-status quo and the behaviour previously taken by households. This study proposes that a decision of acceptance or rejection of PV systems is the past behaviour of the battery adoption decision. This antecedent role of PV behaviour may spark two attitudinal changes: (a) feelings of regret, which may occur among PV adopters, stemming from a positive experience of using the system, or among non-adopters due to their rejection of the system, and (b) feelings of despair, which may arise among current PV users due to dissatisfaction and discontent with the system. While regret positively changes consumers’ attitude towards battery adoption expediting an earlier purchase, despair could preclude battery installation. Through a survey of 557 households in South East Queensland, Australia, this study investigated factors driving attitude change. Instead of traditional statistical methods, machine learning algorithms were adopted to derive data-driven models of attitude change, allowing for higher prediction accuracy and a determination of the latent causalities. The main findings indicate that perceived attitudes towards financial and non-financial benefits, followed by informal peers, best estimate the attitude change, whereas traditional sociodemographic factors, knowledge and affordability may not engender a shift. Leveraging from this new paradigm can encourage current PV consumers to take another step and become earlier battery adopters. Failure to recognise these dynamics may breed despair and turn current innovators and early adopters of PVs into late adopters or rejectors of battery systems.
Article
While agent-based modeling (ABM) has become one of the most powerful tools in quantitative social sciences, it remains difficult to explain their structure and performance. We propose to use artificial intelligence both to build the models from data, and to improve the way we communicate models to stakeholders. We use machine learning to facilitate model development directly from data. As a result, beside the development of a behavioral ABM, we open the ‘blackbox’ of purely empirical models. With our approach, artificial intelligence in the simulation field can open a new stream in modeling practices and provide insights for future applications.
Article
Organic food consumption is seen as a key strategy to alleviate both environmental and health problems. Although consumer purchasing of organic food has regularly been studied, major gaps exist in the literature. Knowledge is missing on how contextual factors, such as pandemics (e.g., COVID‐19 pandemic), affect individuals' purchasing of organic food. Therefore, the aim of this research is to examine the effect of a pandemic on organic food purchasing. To provide evidence on this effect, data collected at two points in time (before the COVID‐19 pandemic and during the first wave of the COVID‐19 pandemic) from 429 German consumers was analyzed with structural equation modeling. The results showed that pandemics positively influence both consumer quality consciousness (β = .116) and health consciousness (β = .106) and thereby enhance organic food purchasing. However, pandemics were not found to shape a consumers' environmental consciousness (β = −.005). Additional analyses showed that the effects of a pandemic are not equal for all consumer segments and that consumers' income occupies—different than consumers' age, gender, and education—a decisive role. For instance, pandemics promote consumers' health consciousness only for consumers of lower than of higher income. These findings yield the diverse implications for practitioners and public policy.
Article
Full-text available
The current intense food production-consumption is one of the main sources of environmental pollution and contributes to anthropogenic greenhouse gas emissions. Organic farming is a potential way to reduce environmental impacts by excluding synthetic pesticides and fertilizers from the process. Despite ecological benefits, it is unlikely that conversion to organic can be financially viable for farmers, without additional support and incentives from consumers. This study models the interplay between consumer preferences and socio-environmental issues related to agriculture and food production. We operationalize the novel concept of extended agro-food supply chain and simulate adaptive behavior of farmers, food processors, retailers, and customers. Not only the operational factors (e.g., price, quantity, and lead time), but also the behavioral factors (e.g., attitude, perceived control, social norms, habits, and personal goals) of the food suppliers and consumers are considered in order to foster organic farming. We propose an integrated approach combining agent-based, discrete-event, and system dynamics modeling for a case of wine supply chain. Findings demonstrate the feasibility and superiority of the proposed model over the traditional sustainable supply chain models in incorporating the feedback between consumers and producers and analyzing management scenarios that can urge farmers to expand organic agriculture. Results further indicate that demand-side participation in transition pathways towards sustainable agriculture can become a time-consuming effort if not accompanied by the middle actors between consumers and farmers. In practice, our proposed model may serve as a decision-support tool to guide evidence-based policymaking in the food and agriculture sector.
Article
Full-text available
Organic food has important environmental and health benefits, decreasing the toxicity of agricultural production, improving soil quality, and overall resilience of farming. Increasing consumers’ demand for organic food reinforces the rate of organic farming adoption and the level of farmers' risk acceptance. Despite the recorded 20% growth in organically managed farmland, its global land area is still far less than expected, only 1.4%. Increasing demand for organic food is an important pathway towards sustainable food systems. We explore this consumer-centric approach by developing a theoretically- and empirically-grounded agent-based model. Three behavioral theories – theory of planned behavior, alphabet theory, and goal-framing – describe individual food purchasing decisions in response to policies. We take wine sector as an example to calibrate and validate the model for the case study of Sydney, Australia. The discrepancy between consumer intention and purchasing behavior for organic wine can be explained by a locked-in vicious cycle. We assess the effectiveness of different policies such as wine taxation, and informational-education campaigns to influence consumer choices. The model shows that these interventions are non-additive: raising consumer awareness and increasing tax on less environmentally friendly wines simultaneously is more successful in promoting organic wine than the sum of the two policies introduced separately. The phenomenon of undercover altruism amplifies the preference for organic wine, and the tipping point occurs at around 35% diffusion rate in the population. This research suggests policy implications to help decision-makers in the food sector make informed decisions about organic markets.
Article
Full-text available
In today's growing economy, overconsumption and overproduction have accelerated environmental deterioration worldwide. Consumers, through unsustainable consumption patterns, and producers, through production based on traditional resource depleting practices, have contributed significantly to the socio-environmental problems. Consumers and producers are linked by supply chains, and as sustainability became seen as a way to reverse socio-environmental degradation, it has also started to be introduced in research on supply chains. We look at the evolution of research on sustainable supply chains and show that it is still largely focused on the processes and networks that take place between the producer and the consumer, hardly taking into account consumer behavior and its influence on the performance of the producer and the supply chain itself. We conclude that we cannot be talking about sustainability, without extending the supply chains to account for consumers' behavior and their influence on the overall system performance. A conceptual framework is proposed to explain how supply chains can become sustainable and improve their economic and socio-environmental performance by motivating consumer behavior toward green consumption patterns, which, in turn, motivate producers and suppliers to change their operations.
Article
Full-text available
With the rapid adoption of social media by consumers, it is increasingly important for retailers to investigate and consider their use and adoption of social media, and to know which activities are most effective. This can differ by product line and geographical location. Further, for complex products such as wine, with specific consideration of higher price segments, consumers frequently search for more information before purchase. This study investigates the social media adoption and activities of 1173 wineries located in Germany, the USA, New Zealand and Australia. The results show that Facebook is the main platform that wineries use to engage with consumers, but that the actual reasons social media is used vary. Winery size and the number of hours spent working on social media also varies across respondents and countries. The findings suggest that wineries need to develop a clear purpose for using social media and then adapt to the needs of the consumers in their respective markets.
Article
This article reviews the scientific literature on local food from the consumer's perspective and analyses findings through the application of the Alphabet Theory-a newly developed theoretical framework for consumer behavior towards alternative food choices. As consumers' interest in local food has steadily increased in the past fifteen years, so has the number of research studies on consumers' attitudes and purchase behavior with regard to local food. A literature search was carried out on three online catalogues using the search terms taken into account. In all, the literature search returned 550 scientific articles. This paper provides an overview of 73 relevant publications, summarizes the main results, and identifies research gaps in the context of the Alphabet Theory. One major result was that, unlike organic food, local food is not perceived as expensive. Nevertheless, consumers are willing to pay a premium for local food. In mostly quantitative studies, consumer characteristics , attitudes, and purchase behaviors with regard to local food were assessed. Research gaps were identified in various areas: cross-national (cultural) comparisons, influence of different types of products (fresh vs. non-perishable, processed vs. non-processed, or plant vs. animal products), origin of foodstuffs used to produce local food as well as the influence of personal and social norms on the formation of attitudes towards local food. This contribution appears to be the first review of scientific articles from the field of local food consumption to present an overview on international research and to identify research gaps.
Article
In light of low and stagnating market shares of organic wine, there is an ongoing debate about growth potential for organic wine. A recent study revealed that even consumers of organic food did not necessarily purchase organic wine regularly. The aim of this contribution was to analyse the wine preferences of organic food consumers and identify promising new target groups for organic wine. We conducted choice experiments in Germany (N = 600) and analysed the data with mixed logit models and latent class models, revealing interesting differences between the results of the two approaches. While the mixed logit models suggested ‘organic’ was the most important wine attribute, the latent class models challenged this proposition. While three of four consumer segments had a strong preference for organic, only one segment in the red wine model (and no segment in the white wine model) gave organic highest priority. Just like non-organic consumers, many organic food consumers seem to use price or country of origin as the most important quality cue for wine. The comparison between the results of the choice experiments and the participants’ stated normal purchase behaviour suggested there is growth potential for organic wine. Apparently, consumers of organic food would buy more organic wine if their preferred type and variety of conventional wine would be available in organic quality at similar price levels.
Article
Several studies have been conducted on organic wine consumption, but no specific study has yet fully explored how the set of attributes explored by existing research affects the additional price for organic wine. To fill this gap, the objective of this paper was to examine whether and to what extent consumers are inclined to pay for buying organic wine and what are the attributes that significantly influence the additional price of organic wine compared to conventional one. With this aim, a quantitative study over a representative sample of wine consumers in Sicily (Italy) was carried out. Results allowed to observe that consumers attached greater importance to personal motivations such as environmental protection, distinctness and curiosity as well as to explicit label information such as brand renown and local production. In addition, male gender and income are positively correlated to the willingness to pay an additional price for organic wine. Our results have important implications for the actors involved in the wine sector as the adoption of marketing practices explicitly related to the label and motivations attributes can lead to value augmentation of organic wine that could increase consumers’ valuation for it.
This study proposes an alternative approach to the usual cognitive investigation of the purchasing behaviour for food products by country-of-origin (COO). The new approach is based on two well-known behavioural models: the Negative Binomial Distribution and the Dirichlet model. Using two large datasets of actual purchasing of COOs of wine and butter in the UK, the study finds that both models are capable of describing and predicting COO purchasing behaviour in a probabilistic and polygamous manner (not loyal to one origin). The study also identifies several important patterns of COO purchases that are associated with these models, including the double jeopardy pattern and the duplication of purchase pattern. These patterns can be used to develop strategy to grow product sales based on COO. They can also be used to evaluate the effects of marketing interventions, such as origin-based promotions and communications based on COO.
Article
Aware of the significance of impulse buying and wishing to anticipate possible changes in the market for its products, the Wm. Wrigley Jr. Company commissioned Stanford Research Institute to study the market for impulse items as it has developed in the past decade and as it is likely to develop during the 1960s. This article is drawn from the study findings, on the nature and significance of consumer impulse buying.
Article
We tested the hypothesis that individuals may act differently when buying a bottle of wine for themselves than they do when buying wine as a gift. Using a between-subject design, we estimated the differences in consumer preference for these two occasions. We conducted a choice experiment on 618 Italian wine consumers and included the attributes of price, geographical indication (i.e. IGT, DOC, or DOCG), organic claim, and brand (i.e. famous producer or a non-famous producer). By applying an error component random parameters logit model, we detected relevant differences between the two scenarios in terms of the relative importance of the studied attributes. The gift-giving scenario was further investigated using a latent class model, which identified three segments of consumers; we profiled these according to personal attitudes and wine knowledge. Our results show a relevant heterogeneity among consumers’ preferences for the gift-giving scenario, with geographical indication having a low impact and brand and organic claim playing a pivotal role. This study provides relevant insights for winemakers and retailers regarding diversifying marketing strategies.