Italian Consumers’ Perceptions and Understanding of the Concepts of Food Sustainability, Authenticity and Food Fraud/Risk

Abstract

The present study investigates consumers’ perceptions and knowledge of food sustainability, food authenticity and food fraud/risk in Italy and whether their perception and knowledge differ according to demographic factors, consumption choices and preferences. To address these issues, a network analysis technique was applied to analyse the theoretical framework. Subsequently, the primary online survey data of 328 respondents in Italy were explored using principal component analysis, canonical correlation analysis and hierarchical cluster analysis. The results indicate that socio-economic determinants—above all, marital status, the presence of children, and annual net income—not only influence the propensity to seek information on the certification, sustainability labels and quality of products but also impact consumer perceptions of food fraud/risk. The findings provide a useful, informative tool for the protection of consumer health, which can be negatively impacted by the purchase of fraudulent food products. Despite the abundant literature on consumer perspectives and perceptions of food authenticity and the integrity of the food supply, to the best of this author’s knowledge, this study represents one of the first attempts to analyse consumer perceptions of certified food products and knowledge of food fraud/risk.

1. Introduction

Food authenticity, food sustainability and food fraud/risk are complex and multi-dimensional concepts. In general terms, food authenticity is perceived as an important aspect by consumers on an emotional level as it involves their trust in what they buy, and specific information provided by different product labels could signal sustainability attributes that drive consumers to make more sustainable food choices, food fraud/risk are not familiar to most consumers, and their understanding of the risks it poses is based on incomplete information [1]. Indeed, given that food fraud is a relatively new field of research, terminology such as food fraud, food forgery, food driven by economics or food crime is often used interchangeably in academia. One of the most accredited definitions in the literature is that of Spink and Moyer [2], who proposed seven types of food adulteration, including counterfeit products, diverting products from their intended markets, overruns, simulation, tampering, and theft.

Alternatively, the Food Standards Agency (FSA) defines food fraud as intentionally putting food on the market for financial gain with the purpose of misleading customers [3].

Moreover, food fraud motivated mainly by economic gain [4,5] is a significant problem in the food industry [6], with many cases of criminal organisations interfering, especially during the COVID-19 pandemic. However, the most recent report from Europol and the European Union Intellectual Property Office [7] suggests that criminal organisations could infiltrate more easily during the pandemic and affect the production of counterfeit products in the various stages of the food chain, starting with production upstream and ending with marketing downstream. This has happened, above all, because of a low or even a lack of coordination between the subjects that operate in various ways along the same chain. In addition, during the pandemic period, the same criminal organisations were facilitated by greater use of the digital world to contact consumers directly, find components and raw materials and use social media to their advantage. Food, in fact, particularly biscuits, pasta, chips and confectionery, in 2020, in the internal market of the European Union (EU), were the least counterfeit in the extra-EU and represented the second category of products, after clothing, most commonly seized at foreign borders.

Examples of food fraud include adding melamine to Chinese milk products to illegally increase the apparent nitrogen content, which implies an increase in protein content [8,9]; the use of cheaper and lower-quality oils to dilute extra virgin olive oils that can increase yields and decrease manufacturing costs [10,11]; and the ‘Horse Meat Scandal’, also known as ‘Horse Gate’ and other colloquial monikers, which was discovered in 2013 in various British and Irish markets [12]. These cases have contributed to increased consumer awareness and the introduction of legislative and global standard improvements, but there are still improvements that are required. This authenticity is essential for all consumers, but it can be especially problematic for those who have specific dietary needs or religious restrictions [3]. As a result, the prevention of counterfeit food requires consumer certainty over food authenticity and the integrity of the food supply [13]. However, the long chain that characterises the modern food production system has increased the distances between producers, processors and consumers. This has led to a reduction in knowledge and, consequently, to uncertainties among consumers about food risks [14].

A plethora of food studies [15,16,17] have shown how trust can become an important driver for consumer food choices as it helps to reduce levels of risk and uncertainty. However, there is a lack of empirical studies examining how consumers perceive the authenticity of food, food fraud/risk, and how their perceptions differ based on demographic factors, consumption choices, and preferences. Indeed, previous studies have concentrated primarily on consumer preferences concerning food safety attributes [18], local food, and the impact of food fraud incidents on how consumers feel and act [19,20,21]. Moreover, it is not clear whether the purchase of both authentic and fraudulent food is the same among different groups of consumers according to their socio-demographic characteristics. Consequently, this study used primary survey data to address this noticeable gap and provided novel and critical insights into how the socio-economic determinants can increase Italian consumers’ awareness, sustainable perception and knowledge about the purchase of authentic food, and food fraud/risk, as suggested by Cahyono [22].

The remainder of this article is organised as follows: firstly, in Section 2, network analysis leads to the presentation of the results of a literature review. Subsequently, the methodology applied is presented in Section 3. The results obtained are shown in Section 4, followed by Section 5, where there is a discussion of said results. Finally, Section 6 outlines the conclusions, the limitations and possibilities for prospective future research.

2. Theoretical Framework and Research Questions Formulation

To evaluate consumers’ beliefs about food authenticity and their knowledge of food fraud/risk, the VOS Viewer (version 1.6.20, Leiden University, Leiden, The Netherlands) tool for network analysis was used in text-mining research [23]. Scientific publications that are indexed in the scientific database Scopus were gathered to create a collection of metadata. The electronic search phase was conducted through the Scopus database, as it is the biggest and most complete bibliographic database and has been utilised in previous systematic reviews to identify and assess quality peer-reviewed publications [24]. The search was narrowed down by document type, thus including all articles. The strings used, with the Boolean operator AND, were (“consumers’ perception”) AND (“food authenticity”) for the first analysis and (“consumers’ knowledge”) AND (“food”) AND (“fraud/risk”) for the second. The search yielded 103 relevant articles published from 1997 to the present day for the first search but only 8 articles for the second stage of analysis, which were published from 2012 to the present day.

Figure 1 and Figure 2 display network visualisations based on the titles, abstracts, and keywords of selected articles. It shows 82 items grouped in eight clusters presented in different colours for consumer perceptions of food authenticity and only 8 items grouped in two clusters for consumer knowledge of food fraud/risk. This demonstrates how, in recent years, this second strand of research has been neglected by scholars both nationally and internationally.

To begin, the first strand of the literature, according to Figure 1, has various clusters designated with different colours, including red, green, blue, yellow, purple, light blue, orange, and brown.

The red cluster, which comprises 19 items, focuses attention on the traceability of products. Indeed, recently conducted studies [25,26] have provided valuable insight into the origin and authenticity of food products and have also provided valuable feedback on consumer perceptions and interests. For example, the study by Castellini et al. [27] was intended to comprehend the consumer’s attitude towards ‘omic’ technologies applied to organic vegetables and to discover the primary psychological and socio-demographic factors that influence their appeal.

The green cluster covers 14 items and is mostly focused on how consumers perceive the authenticity of ethnic food and restaurants. For example, by using semi-structured in-depth interviews, Arviv et al. [28] provide novel and critical insights into how Israeli-Jewish consumers judge the authenticity of ethnic food and restaurants. The work of Reiher [29] is also similar, as it introduces the analytical category ‘feelings of authenticity’ and discusses the way Japanese food entrepreneurs, chefs, and workers make and sell Japanese food in Berlin. Chang et al. [30] conducted a study on food souvenirs using the means-end chain approach to explore consumer values purchased by tourists and subsequently identify attributes of food souvenirs that match these values.

The blue cluster consists of 13 items. In this case, recent work focuses the analysis on how different sustainability labels can influence consumers’ perception and their evaluation of alternative food networks [31].

The yellow cluster with 10 items is more aligned with the aims of this paper because it deals with consumer perceptions of the authenticity of certain categories of food products. For instance, Fernández-Sánchez et al. [32] point out that the authenticity of cheeses, perceived by consumers of different sociological generations in Mexico, is a combination of the pleasure of consuming the product, the link to rural life and new consumer values. Knowledge of consumer perspectives on the authenticity of cheeses is influenced by age, sociocultural, ethical, political and consumer factors. Furthermore, Marozzo et al. [33], by testing traceability through a series of questions centred around topics related to product authenticity and sustainability, explore the sensitivities and perceptions of Asian consumers concerning organic olive oil.

The purple cluster, with nine items, provides important insights into consumer behaviour and perceptions concerning healthy food. For example, a recent study conducted in Slovakia by Predanócyová et al. [34] provides valuable information for producers in terms of marketing and communication strategies that can be used by policy makers to improve food policy to promote public health in society.

The light blue cluster with eight items deals with cause-related marketing (CRM). Perceptions of CRM in fast food restaurants were reflected in consumer brand evaluations in an African context [35].

The orange cluster has eight items and includes a recent study by Gaiato et al. [36] that analyses how consumer perceptions of the animal welfare (AW) label affect brand equity (BE). The sample consisted of 334 Brazilian consumers of chicken breast. However, controlling for different levels of consumer knowledge about AW, higher knowledge was associated with higher BE. The results suggest that there may be plausible explanations for consumers’ lack of awareness, understanding, confusion and scepticism regarding the authenticity of the label, suggesting areas for future research.

Two items are included in the final brown cluster, which explores the relationship between place of origin, natural scarcity, food authenticity and the impact of COVID-19 on consumers’ perceptions and emotional attitudes towards space. For example, Thompson and Kumar [37] pointed out that products in traditional supermarkets are relatively disconnected from their place of origin and their producers. The results of the study show a clear link between natural scarcity and authenticity. The perception of natural scarcity products is associated with authenticity, and a sense of authenticity is highly desirable to consumers and is considered a ’rare and coveted commodity’ in contemporary marketing.

In the same vein, Garner and Hollenbeck [38] delineate the link between natural scarcity and authenticity. Finally, research by Gan et al. [39] shows how COVID-19 has affected consumers’ perceptions and emotional attitudes toward food markets, highlighting how the post-epidemic era is more multi-dimensional and optimistic.

The second strand of the literature that relates to the present work deals with consumer knowledge of food fraud/risk: a strand of research, as outlined in the introduction, that has received scant attention nationally and internationally.

In general, consumers do not know what they are eating and where it comes from. In addition, food fraud is not easy to detect and does not necessarily cause food safety incidents [4]. However, food fraud is related to food safety [40], mainly because of its potential to cause health problems.

The network analysis (Figure 2) shows eight items grouped in only two clusters. The red cluster includes five items and mainly deals with consumer perceptions of food fraud. Indeed, the fulcrum of this literature strand cluster is represented by the study by Théolier et al. [1]. In this work, the authors stress that the concept of food fraud is unfamiliar to many consumers, and their understanding of the risks involved is based on incomplete information. Furthermore, the authors suggest that consumers’ knowledge of food fraud is instinctive and limited, and their understanding of the risks involved is also based on incomplete information. From a consumer perspective, there is a need for communication and education on the management and detection of food fraud. For this reason, action needs to be taken at a local level, as it appears that ‘consumers’ are not a homogeneous group.

Moreira et al. [41] point out that the information provided on food labels is useful, but the way it is presented may limit consumer interest and understanding. Food fraud is recognised by more than half of the respondents, with a greater understanding of those practices that pose a risk to public health than those related to economic motivations. The age and education of consumers influenced their perception of the information provided on food labels and their confidence and knowledge of food fraud. The findings on respondents’ perceptions could be used by the food industry to improve the design of food labels and increase consumer understanding.

Another example of work in this area is an online survey conducted by Djekic and Smigic [42], which investigates how food fraud is perceived by consumers in Serbia and Montenegro. In the Serbian population, older or highly educated respondents are aware of different types of fraudulent activities such as substitution, mislabelling, concealment and counterfeiting. Women, the younger population and students were more aware of dilution. Consumers felt that trust was the most important factor when buying food. The highest level of agreement regarding food fraud concerned the risk to consumer health and the view that food control services are the most responsible actors in the food chain continuum.

A further study that made use of an online questionnaire was carried out by Marozzo et al. [33]. It highlighted the relevant role of food sustainability and authenticity concepts as ‘risk mitigants’ in relation to food fraud and negative issues related to COVID-19.

An overview of consumer concerns about food fraud in selected countries in Southeast Asia is provided by Soon-Sinclair et al. [5]. A cross-sectional online survey with 1393 valid responses was conducted in Indonesia, Malaysia, the Philippines, Thailand and Vietnam. To reduce the large data set of nominal variables, multiple correspondence analysis (MCA) was first performed. Increased demand for food fraud control influenced concerns about food fraud, perceived risks of different types of food fraud, information sources from media and personal networks, information sources from credible organisations and personal experience of food fraud.

The blue cluster includes three items and deals with the analysis of differences among producers, processors and distributors regarding perceptions and knowledge of food fraud [43]. It also examines the criminality of food fraud incidents, providing an overview of the most common cases of food fraud worldwide from 2010 to 2020. It makes suggestions about possible solutions to minimise food fraud incidents, such as an increase in the level of risk-based inspections. The establishment of more productive monitoring, the implementation of food protection systems in the supply chain and the implementation of better ingredient control and certification, as suggested by Jurica et al. [44], can reduce food risks. Similarly, Vågsholm et al. [45] list further methods to reduce food fraud, such as the promotion of food safety culture; greater information about the food chains; the use of health epidemiological indicators, sensors and block chains; improved industry/private standards; and the application of a system approach from farm to fork aimed at reducing information asymmetry and building trust and social capital for all stakeholders within the meat value chain.

Earlier studies tended to analyse different perceptions of food fraud and associate risk levels with different consumer groups. For example, Charlebois et al. [15] found a positive relationship between educational level and consumer perceptions of food fraud in their study on Canadian consumers. Another study [46] conducted using Austrian consumers highlighted how educational level and sex influence perceptions of food fraud. Consumers who are more educated tend to distrust food label information and food regulators and take action against food fraud. More females than males pay attention to food labels before purchasing food [47]. Another factor that alters the perceptions of consumers is family income. Gupta and Panchal [48] argued that consumers with higher family incomes tended to have a greater awareness of food fraud.

A recent study carried out using online questionnaires by Moreira et al. [41] found that among a sample of 308 Portuguese consumers, the main socio-demographic characteristics that influenced the respondents’ perceptions of food labelling and food fraud were age and education, whereas sex, marital status, salary and lifestyle were not significant.

Despite existing research, there is still a limited number of studies focusing on the level of perception of Italian consumers towards the purchase of authentic food products and their level of knowledge regarding food fraud/risk. Gaining a deeper understanding of these needs is crucial for developing effective communication tools that could assist policymakers, responsible producers, and manufacturers in making decisions about technological investments (such as blockchain) to address food fraud.

Hence the following three research questions (RQs) are proposed:

RQ1: Do the socio-demographic characteristics of consumers increase their level of perception about the purchase of authentic food products?

RQ2: Do the socio-demographic characteristics of consumers increase their level of knowledge of food fraud/risk?

RQ3: Are the socio-demographic characteristics, consumer preferences and choices concerning where to purchase, of the well-established “homogeneous” consumer groups, critical to ensuring that they do not engage in the purchase of inauthentic, fraudulent food?

3. Materials and Methods

3.1. Questionnaire Design

Italian consumers who were over eighteen participated in a web-based survey between June 2023 and April 2024. Two stages were used to enrol study participants. In the first stage, individuals were identified within the networks of university students. Individuals were notified that their involvement in the survey was voluntary and confidential. They were requested to complete a questionnaire using the Google Forms platform.

In the second step, these individuals were invited to share the link with friends, family, and colleagues using a snowball method via social media networks [5].

The questionnaire was tested to determine if there were any ambiguities in understanding the questions; it was preceded by a short description of the study’s objectives and a request for consent. Specifically, 27 questions were inserted in the questionnaire, which was divided into two sections, in order to facilitate understanding for the respondents and obtain targeted answers.

The first part of the questionnaire, in line with previous studies [1,41,46,48,49,50,51], aimed to collect the respondents’ main socio-demographic characteristics. Nine determinants (gender, age group, marital status, presence of children, education, household members, occupational status, annual net income and residence) were chosen that can influence perceptions of food authenticity, food fraud and food risk (

Table 1. Variables used in the analysis.

Variable	Type	Coding
Gender	Registry	(Male = 0; Female = 1)
Age group	Registry	(18–24 = 1; 25–45 = 2; 46–60 = 3; >60 = 4)
Marital status	Registry	Single = 1; Married = 2
Presence of children	Registry	(No = 0; Yes = 1)
Education	Registry	(Primary school = 1; Secondary school = 2; University Degree = 3; Professional qualification = 4)
Household members	Register	1 person = 1; 2 persons = 3; 3 persons = 3; 4 persons = 4; More than 4 = 5
Occupational status	Registry	(Student = 1; Housemaker = 2; Employee = 3; Self-employed = 4; Temporary worker = 5; Freelancer = 6; Pensioner = 7; Inactive = 8; Unemployed = 9)
Annual net income (€)	Registry	(No income = 0; 0–5000 = 1 5001–10,000 = 2; 10,001–15,000 = 3 15,000–25,000 = 4; >25,000 = 5)
Residence		(Urban = 0; Rural = 1)
(Q1) When you buy food, how do you usually behave?	Consumer choice and preference	(I tend to buy the same products = 1; I usually change = 2; I try to vary the consumption of food, also based on offers = 3)
(Q2) When you buy a food product, what do you focus more on?	Consumer choice and preference	(Date of packaging = 1; Expiry date = 2; Ingredient = 3; Place of origin = 4; Weight = 5; Price = 6; Nutritional value = 7)
(Q3) Where do you usually buy food?	Consumer choice and preference	(Local store = 1; Supermarket = 2; Discount = 3; Certified organic stores = 3; E-Commerce = 4; Points of sale chosen by parents = 5)
(Q4) Did the lockdown (due to the spread of COVID-19) affect the way you purchased food?	Consumer choice and preference	(No = 0; Yes = 1)
(Q5) Which of the following certifications do you know?	Food certification and labelling	(PGI = 1; PDO = 2; Biological = 3; DOC = 4; DOCG = 5; UTZ = 6; Fair-trade = 7; I am aware of all the certifications mentioned = 8; I am not aware of the certification acronyms mentioned = 9)
(Q6) Do you generally buy certified food?	Food certification and labelling	(No = 1; Yes = 2; Only for some products = 3)
(Q7) When buying a food product, do you usually view the labels on the packaging?	Food certification and labelling	(No = 0; Yes = 1)
(Q8) Which of the following alternatives gives the exact name “food fraud of a product”?	Knowledge and awareness of agri-food fraud.	(Diversity of origin = 1; Deformity of quality and quantity of products = 3; None of the previous = 3)
(Q9) Are you afraid to buy and/or consume, without your knowledge, a fraudulent food product?	Knowledge and awareness of agri-food fraud.	(No = 0; Yes = 1)

).

The second part (consisting of 9 questions), following the recent food consumer studies approach with network analysis [1,27,28,29,32,33,42,52], focused on consumer choices, preferences for different types of food authenticity and knowledge and consumption habits of certified food products. The following table (Table 1) provides an overview of the variables collected in the survey and used in the analysis.

3.2. Survey

The study was conducted with a convenient sample of 328 voluntary responses received after a ten-month data collection period (June 2023–April 2024). As this was an exploratory study, the sample was considered appropriate. Other previous studies have used restricted samples [11,35,36,41,53,54,55,56,57].

3.3. Methods

This analysis used a quantitative approach. Indeed, the information collected through an online questionnaire was processed using “Stata” version 16, a free software environment for statistical computing and graphics. In the first step of the analysis, principal component analysis (PCA) and canonical correlation analysis (CCA) techniques were used to determine the links between the two sets of variables, in this case, the consumers’ socio-demographic characteristics and the respondents’ perceptions of food authenticity and knowledge of food fraud/risk. In a second step, hierarchical cluster analysis (HCA) was applied to identify the well-established “homogeneous” consumer groups of respondents based on their socio-demographic characteristics and on their responses to the nine questions (Table 1).

First, PCA was performed to reduce the observed variables to a set of uncorrelated latent variables, where the latent variables are computed directly from a correlation matrix. To define the number of dimensions to retain, the following criteria were used: eigenvalues of >1.00 [58]. Through the PCA, factors are generated by grouping defined variables that are correlated according to their variance. Considering a series of variables defined in the sample as x₁, x₂, … x_n, a new set of variables will be generated z₁, z₂, … z_p, uncorrelated with each other with a decreasing explanatory capacity. In any case, p < n, and every z_j (where j = 1, …, p) is extracted as a linear combination between original variables (x_i), with an explanatory capacity similar to the original set of variables. A factor is defined as follows:

z_j = α_j1 x₁ + α_j2+ x₂ +…+ α_jn x_n j = 1, …n

Through a factor analysis using categorical variables (PCA), an initial set of items is reduced to a more manageable group. The reduction in variables into factors can be performed because a great deal of variability in data can often be explained by a small number of k principal components, which is lower than the initial number of variables x_n [59].

Second, a CCA was applied, which is an exploratory statistical method, to assess correlations between two sets of variables by maximising correlation among them. Similar to PCA, it is utilised as a way to reduce data.

Indeed, CCA essentially examines the relationship between multiple dependent and independent variables. To investigate the relationship between variables, CCA uses independent canonical functions that optimise the correlation between linear composites as its underlying principle, known as canonical variates. The canonical correlation coefficient is used to measure the strength of association between the variable sets under consideration. The correlation between independent variables and their respective canonical variates is determined by canonical loadings, whereas canonical cross-loadings measure the correlation between each observed variable and the opposite canonical variable.

The methodology was described in detail by Noori [60].

After performing the PCA, the HCA based on Euclidean distances was conducted on the first PCs, explaining a relevant part of the original cumulative variance. Ward’s classification algorithm was used. This method is distinct from all others since it uses an analysis of variance approach to evaluate the distances between clusters. We can refer to Ward [61] for details concerning this method, which is regarded as very efficient; it tries to find the partitions P_n, P_n−1,…, P₁ in a manner that minimises the loss associated with each grouping and to quantify that loss in a form that is readily interpretable.

The Ward method relies upon the well-known decomposition: T = W + B, where T is the total sum of squares (SS) of the observations, W is the within-clusters SS, and B is the between-clusters SS. In general, passing from k + 1 to k clusters, W tends to increase (less homogeneity in the new cluster with the addition of new units), while, of course, B decreases: at each step of the Ward procedure, the clusters joined together are the two with the minimum increase in W.

Regarding the subsequent choice of an optimal number of clusters, it is common practice to repeat the analysis for different numbers of clusters and then calculate the objective function.

$$\mathrm{R}{2}_{(k)}={\displaystyle \frac{{\mathrm{B}}_{\left(k\right)}}{\mathrm{T}}}$$

This procedure is sometimes called the elbow method since, when reporting on a graph the number of clusters k (on the horizontal axis) and the R2_(k) values (on the vertical axis), a good choice for the number of clusters is the k where the graph presents an “elbow”, i.e., a sudden change in slope.

4. Results

4.1. Summary Demographics

Table 2. Demographic information (N = 328).

Characteristics	Absolute Value	% of Respondent
Gender
Male	127	38.7
Female	201	61.3
Age group
18–24	133	40.5
25–45	95	29.0
46–60	80	24.4
>60	20	6.1
Marital status
Single	121	36.9
Married	196	59.8
Presence of children
No	141	43.0
Yes	187	57.0
Education
Primary school	92	28.0
Secondary school	110	33.5
University Degree	104	31.7
Professional qualification	22	6.7
Household members
1 person	132	40.2
2 person	83	25.3
3 person	43	13.1
4 person	55	16.8
More than 4	15	4.6
Occupational status
Student	114	34.8
Housemaker	49	14.9
Employee	109	33.2
Self-employed	15	4.6
Temporary worker	8	2.4
Freelancer	2	0.6
Pensioner	19	5.8
Inactive	8	2.4
Unemployed	2	0.6
Annual net income (€)
No income	146	44.5
0–5000	37	11.3
5001–10,000	20	6.1
10,001–15,000	29	8.8
15,001–25,000	49	14.9
>25,000	47	14.3
Residence
Urban	195	59.5
Rural	133	40.5
Total	328	100

Legend: n represents the number of interviewees; (%) represents their share in the sample.

describes the demographic characteristics of the 328 respondents. The sample set consists of 61.3% women and 38.7% men. The majority of the sample is of the 18–24 age group (40.5%), followed by 29% of the 25–45 age group, with an average age of 25.5 years old. A total of 196 respondents (59.8%) were married, and 121 (36.9%) were single. According to family status, 187 (57%) respondents had children, and 141 (43%) had none; 132 (40.2%) respondents lived alone, followed by 25.3% with two household members. Most of the sample (33.5%) had a high school diploma. Regarding respondents’ occupational status, 34.8% were students, 33.2% were employed, and 14.9% were homemakers, followed by pensioner (5.8%), self-employed (4.6%), temporary-worker (2.4%), inactive (2.4%), freelancer (0.6%) and unemployed (0.6%). The majority of the respondents (44.5%) did not perceive income, followed by 14.9%, who fell in the class of EUR 15,001–25,000. Finally, 59.5% of the respondents lived in urban areas and 40.5% in rural areas.

4.2. Identification of the Important Socio-Demographic Determinants

PCA and CCA gave a comprehensive analysis of the relationship between respondents’ socio-demographic determinants, overall perception, and knowledge about food authenticity and food fraud/risk to answer the first and second RQs.

With the PCA, the eigenvalues, the variance proportion and the cumulative variance proportion were calculated and are shown in

Table 3. Eigenvalues analysis.

PCs	Eigenvalue	Difference	Variance Proportion	Cumulative Variance Proportion
PC1	3.94	2.65	0.22	0.22
PC2	1.28	0.03	0.07	0.29
PC3	1.25	0.04	0.07	0.36
PC4	1.21	0.07	0.07	0.43
PC5	1.14	0.04	0.06	0.49
PC6	1.10	0.03	0.06	0.55
PC7	1.07	0.06	0.06	0.61
PC8	1.01	0.06	0.06	0.67

. The eight extracted PCs accounted for 67% of the total variance in the data, respecting the rule of at least 50% [62].

Table 4. Correlation among starting variables and PCs.

Variables	PC1	PC2	PC3	PC4	PC5	PC6	PC7	PC8
Gender	−0.04	−0.18	0.44	−0.07	0.19	0.42	0.18	0.03
Age	0.43	−0.04	−0.10	0.02	0.03	0.11	0.04	−0.02
Marital status	0.49	−0.04	−0.05	0.05	0.01	0.01	0.02	−0.03
Presence of children	0.47	−0.01	−0.07	0.07	−0.01	0.01	−0.01	−0.01
Education	0.36	0.01	0.14	0.00	0.07	−0.24	−0.05	−0.04
Household members	0.42	−0.08	−0.05	0.06	0.00	0.06	0.02	−0.08
Occupation	0.09	0.39	0.08	−0.44	0.04	−0.11	−0.14	0.40
Annual net income	0.14	0.31	0.47	−0.19	0.21	0.02	−0.08	0.16
Residence	−0.02	0.01	−0.35	0.12	0.29	0.39	0.44	0.29
q1	−0.01	0.03	0.42	0.31	−0.30	−0.26	0.31	0.05
q2	0.00	0.48	−0.30	−0.10	−0.03	−0.05	0.17	0.16
q3	−0.02	0.01	−0.04	0.19	0.31	−0.58	0.44	−0.07
q4	−0.07	0.00	−0.26	0.07	0.45	−0.13	−0.44	−0.18
q5	0.01	0.32	0.07	0.51	0.16	0.26	0.02	0.21
q6	0.00	0.29	0.14	−0.13	0.15	0.27	0.14	−0.72
q7	0.02	−0.32	0.04	−0.46	0.40	−0.13	0.33	0.06
q8	−0.02	−0.28	0.21	0.25	0.36	−0.01	−0.32	0.26
q9	0.06	−0.33	−0.11	−0.20	−0.33	0.13	0.04	0.18

shows the eigenvectors, which assess the coefficients for the formation of PCs. The determinants that can be considered significant are those with coefficients for formation that are greater than or equal to 0.40 [63]. The meaning of each PC is determined by the original variables correlated with it. The correlations indicate high loadings for four variables. These, in decreasing order, are marital status, presence of children, annual net income and age for component 1 (PC1). All these variables concern the socio-demographic characteristics of the respondents. As a result, it is possible to define PC1 as a synthesis of the “Family status” of the participants. For its part, PC2 captured the “Participants attentive to product attributes”, with a high loading (0.48) on q2. Component 3, “Economic status and food choices”, has a high correlation (0.47) with annual net income, gender (0.44) and q1 (0.42). These results indicate that the participants’ food choices are driven mainly by their annual net income and by their gender. Component 4 has a positive correlation (0.51) with q5 and a negative correlation with q7 (−0.46) and occupation (−0.44). Therefore, it can be defined as “Participants who know certified products”. Indeed, the negative correlations between the vision of food labels on packaging and occupation indicate that participants who know certified products well do not read the food labels.

Component 5 led to the creation of “Participants attentive to food labelling”, with positive loadings on q4 and q7. Component 6 led to “Places of purchase of food as gender question”, with a positive correlation with gender and a negative correlation with q3.

Component 7 has positive correlations with residence and q3 and a negative correlation with q4, so it can be defined as “Residence as a condition for the purchase of food”. The negative correlations with q4 indicate that during the lockdown, the purchase of food products was independent of the area of residence. Finally, component 8 indicates that occupational status reduces the purchase of certified products, leading to it being named “Occupational status and purchase of certified food”.

The nine questions were taken as the response data, and the eight PCs were treated as the predictor set. The CCA results are shown in

Table 5. CCA of the data sets.

Canonical Variates	1	2	3	4	5	6	7	8
Canonical correlations	10.000	1.000	0.996	0.979	0.786	0.594	0.276	0.038
Sig. of F	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Variable	1	2	3	4	5	6	7	8
q1	−0.415	−0.520	0.180	0.312	−0.390	−0.206	−0.432	−0.354
q2	−0.105	0.016	−0.063	0.035	0.328	0.090	−0.186	0.141
q3	−0.138	0.067	0.292	0.439	0.096	−0.160	0.333	0.211
q4	0.989	0.755	−0.220	0.457	0.228	−0.837	0.006	−0.895
q5	0.051	−0.135	−0.171	0.149	0.031	0.293	0.191	−0.149
q6	−0.399	0.616	−0.432	0.087	−0.587	0.139	0.110	0.296
q7	−0.053	0.707	11.240	−0.020	−0.035	13.120	−0.358	−0.728
q8	0.672	−0.189	0.120	0.207	−0.334	0.276	−0.296	0.757
q9	−0.050	−0.358	0.425	−10.553	0.050	−0.023	12.925	0.312
Covariate	1	2	3	4	5	6	7	8
PC1	−0.040	−0.028	0.008	−0.059	−0.002	0.088	0.113	0.477
PC2	−0.328	0.058	−0.556	0.285	0.354	0.096	−0.369	0.092
PC3	−0.116	−0.240	0.060	0.124	−0.653	0.202	−0.487	0.066
PC4	0.168	−0.490	−0.275	0.518	−0.134	−0.124	0.423	−0.023
PC5	0.433	0.494	0.123	0.495	−0.015	0.427	−0.002	0.045
PC6	0.083	0.014	−0.507	−0.428	−0.196	0.558	0.268	−0.204
PC7	−0.683	−0.016	0.356	0.217	0.050	0.380	0.335	−0.186
PC8	0.290	−0.589	0.266	−0.081	0.487	0.425	−0.247	−0.042

. Correlation coefficients for canonical variates 1, 2, 3, 4, 5, 6, 7 and 8 were 10.000, 0.9998, 0.9955, 0.9793, 0.7861, 0.5941, 0.2756, 0.0375 respectively, with the significance of F equal to 0.00, indicating that they are all significant. To determine the dominant determinants, their outstanding coefficients with the highest values in each group were analysed and are highlighted in Table 4.

Considering the above-mentioned results, marital status, the presence of children and annual net income were the dominant variables in the demographic characteristics.

Regarding the set of questions, q5, q4 and q2 had a positive sign, while q6, q3 and q7 had a negative sign.

4.3. Identification of Well-Established “Homogeneous” Consumer Groups

In order to answer the third RQ, the HCA performed on the eight factors identified by the PCA led to defining the following three customer groups, named according to their socio-economic characteristics (

Table 6. Socio-demographic characteristics of the well-established cluster of consumers.

Variable	Young and Single Consumers	Married Consumers	Educated Consumers
Gender
Male	35.43	17.32	47.24
Female	43.28	28.36	28.36
Age group
18–24 years old	99.25	0.00	0.75
25–45 years old	0.00	40.00	60.00
46–60 years old	0.00	37.50	62.50
>60 years old	0.00	55.00	45.00
Marital status
Single	100.00	0.00	0.00
Married	0.00	40.31	59.69
Presence of children
Yes	0.00	40.31	59.69
No	100.00	0.00	0.00
Education
Primary school	79.35	11.96	8.70
Secondary school	53.64	20.91	25.45
University Degree	0.00	37.50	62.50
Professional qualification	0.00	27.27	72.73
Household members
1 person	100.00	0.00	0.00
2 person	0.00	34.94	65.06
3 person	0.00	51.16	48.84
4 person	0.00	38.18	61.82
More than 4	0.00	46.67	53.33
Occupation status
Student	58.77	29.82	11.40
Housemaker	24.49	38.78	36.73
Employee	28.44	18.35	53.21
Self-employed	33.33	13.33	53.33
Temporary worker	50.00	12.50	37.50
Freelancer	0.00	0.00	100.00
Pensioner	42.11	10.53	47.37
Inactive	37.50	12.50	50.00
Unemployed	50.00	0.00	50.00
Annual net income (EUR)
No income	50.68	31.51	17.81
0–5000 euro	40.54	18.92	40.54
5.001–10,000 euro	40.00	15.00	45.00
10,001–15,000 euro	31.03	17.24	51.72
15,000–25,000 euro	22.45	22.45	55.10
>25,000 euro	31.91	14.89	53.19
Residence
Urban area	65.81	27.35	6.84
Rural	41.35	35.34	23.31

) and the awareness, perception and knowledge of food authenticity and food fraud/risk (

Table 7. Participants’ responses to the following questions:.

	Uninformed Consumers	Conscious Consumers	Informed Consumers
(Q1) When you buy food, how do you usually behave?
I tend to buy the same products	54.55	68.35	47.86
I usually change	18.94	16.46	22.22
I try to vary the consumption of food, also based on offers	26.52	15.19	29.91
(Q2) When you buy a food product, what do you focus more on?
Date of packaging	16.67	20.25	14.53
Expiry date	14.39	12.66	17.09
Ingredient	17.42	17.72	11.11
Place of origin	7.58	11.39	6.84
Weight	9.85	13.92	13.68
Price	16.67	11.39	13.68
Nutritional value	17.42	12.66	23.08
(Q3) Where do you usually buy food?
Local store	20.45	21.52	17.09
Supermarket	19.70	26.58	23.93
Discount	34.85	43.04	23.08
Certified organic stores	10.61	7.59	14.53
E-Commerce	4.55	0.00	11.11
Points of sale chosen by parents	9.85	1.27	10.26
(Q4) Did the lockdown (due to the spread of COVID-19) affect the way you purchased food?
No	30.30	48.10	37.61
Yes	69.70	51.90	62.39
(Q5) Which of the following certifications do you know?
PGI	38.64	31.65	23.93
PDO	21.21	26.58	19.66
Organic	10.61	20.25	19.66
DOC	10.61	13.92	16.24
DOCG	6.82	2.53	6.84
UTZ	0.00	0.00	0.00
Fair-trade	3.03	2.53	0.00
I am aware of all the certifications mentioned	5.30	0.00	9.40
I am not aware of the certification acronyms mentioned	3.79	2.53	4.27
(Q6) Do you generally buy certified food?
No	42.42	41.77	37.61
Yes	18.18	20.25	27.35
Only for some products	39.39	37.97	35.04
(Q7) When buying a food product, do you usually view the labels on the packaging?
No	58.33	36.71	67.52
Yes	41.67	63.29	32.48
(Q8) Which of the following alternatives gives the exact name “food fraud of a product”?
Diversity of origin	25.00	31.65	35.04
Deformity of quality and quantity of products	37.88	22.78	24.79
None of the previous	37.12	45.57	40.17
(Q9) Are you afraid to buy and/or consume, without your knowledge, a fraudulent food product?
No	33.33	8.86	34.19
Yes	66.67	91.14	65.81

).

4.3.1. Cluster 1: Young, Single and Uninformed Consumers

Of the 328 participants, this cluster has the largest number of participants (132, 40.24% of the total). The consumers in this first group are young and single and predominantly from urban areas (65.8%). Half were unemployed (50%) and had a low income or no income at all and a lower educational qualification. Moreover, in this cluster, the level of knowledge of food certifications is restricted only to some typologies (PGI, Fair-trade) and is accompanied by poor knowledge of what types of products can be certified and which are of good quality. In addition, more than 55% of the respondents usually buy the same products. Regarding the age variable, nearly all of the consumers (99.25%) in this first group fall into the 18–24 age group and do not have a level of knowledge that would allow them to avoid unwanted purchases. However, participants of this first group, when buying food products, generally only focus their attention on the food safety attributes, mainly the ingredients and nutritional values. The price is important to about 17% of the respondents in this group. This is corroborated by the fact that about 35% choose discount stores as a place to purchase food products because price, as argued by De Irala-Estévez et al. [64], is an important factor in food choice, especially for low-income consumers like those belonging to this first group of consumers.

Regarding knowledge about food fraud/risk, about 38% of the respondents perceived it as a problem relating to the quality and quantity of products, and about 67% said they were afraid to buy and/or consume a fraudulent food product. For these reasons, information about the importance of making informed purchases and viewing product labeling so as not to risk purchasing fraudulent products may be provided. Given the young age of most of the respondents in this group, this could be performed through social media.

4.3.2. Cluster 2: Married and Conscious Consumers

This second cluster contains 79 participants (24% of the total). Respondents of this second cluster have an advanced qualification (upper secondary school diploma, Bachelor’s degree and/or Master’s degree) and a good knowledge (or thorough understanding) of certified food products. However, given the marital status in the cluster analysis, it was found that consumers who are married and have children have a good knowledge of the high-quality types of food certification (PDO, ORGANIC). Nevertheless, more than 43% chose discount stores as a place to purchase food products. Roughly 47% of the respondents of this group live with families of more than four members, and when buying food products, they generally focused their attention on the date of packaging and the weight, which is more than the participants of the other two groups. Furthermore, the participants of this group are more conscientious about their food choices when compared to the other two groups; indeed, when buying food products, 63% usually view the labels on the packaging.

Regarding the level of knowledge of food fraud/risk, about 46% perceived food fraud as more than an incorrectly stated origin, low-quality or an incorrect quantity on a product. Moreover, almost all of the consumers in this group (91%) were afraid to buy and/or consume fraudulent food products without their knowledge. This may be due to the fact that most of the participants (55%) belong to the oldest age group (>60 years old).

4.3.3. Cluster 3: Educated and Informed Consumers

This last group has 117 participants (36% of the total) and is characterised by those living in rural areas (65.8%); 52.1% of members are female, with more than 62.5% aged between 45 and 60 years of age. This group had the highest number of participants with professional qualifications (72.7%). It was also the most diverse group in terms of occupation (mainly company and state employees and the self-employed). Participants in this cluster have a higher degree of awareness (or overall understanding), and they have a higher level of income and education (62.5% have a university degree). This may help to explain their moderate knowledge (or thorough understanding) of food certification. In fact, these consumers have a higher income than the other groups (53% earn more than 25,000 Euro net annually) and have more knowledge and experience in buying good quality food.

Finally, as can be observed in Table 7, more than 40% of consumers do not have an exact conceptualisation of food fraud, and about 66% of them are afraid to buy and/or consume fraudulent food products.

5. Discussion

This study investigated how consumers perceive food authenticity and food fraud/risk in Italy. It also sought to ascertain whether consumer perceptions and knowledge differed according to demographic factors, consumption choices and preferences.

First, the results derived from the application of PCA, CCA and HCA showed that all socio-demographic determinants considered in the analysis were important. In order of importance, these determinants were marital status, the presence of children, annual net income, gender, residence, household members, age and employment status. These results are in contrast to the findings of Moreira et al. [41]. In the latter study, carried out using a sample of 308 Portuguese consumers, scholars revealed that the main socio-demographic characteristics that influenced the respondents’ perceptions of food labelling and food fraud were age and education. In contrast, the study found that gender, marital status, salary and lifestyle were not significant.

Second, regarding the responses to the nine questions, the factors with a positive sign were the most significant: consumer awareness of certified products (q5), the level of attention given to food product attributes (q2) and the method of purchase during the lockdown period (q4). Those factors with a negative sign were consumer propensity to buy certified products (q6), the choice of places to purchase food (q3) and the attention given to food labels on packaging (q7).

Indeed, the CCA results showed that PC1 (Family status) impacted positively on the participants’ response concerning a more exact definition of “food fraud of a product” and negatively on consumers’ methods of purchasing food during the lockdown period and on the consumers’ attention to food labels. Considering that the majority of respondents were in the second group labelled “Married and conscious consumers” when buying food, they typically read the labels on the packaging.

The participants’ attention to product attributes (PC2) was correlated negatively with only the question about consumers’ viewing of labels on packaging. The annual net income had a low impact on the consumers’ knowledge about food fraud/risk. The first group, “Young, single and uninformed consumers”, stands out for its members’ low income and educational level. It is possible to see how socio-economic characteristics not only influence the propensity to have information about what types of products can be certified as of good quality but also affect the way consumers perceive food fraud. It may well be explained, in line with Moreira et al. [41], by the possibility that a consumer with a high income has the opportunity to buy more expensive products and consequently has access to products with higher safety control.

In line with Charlebois et al. [46], educated consumers are more likely to distrust the information on food labels. In fact, the group labelled “Educated and informed consumers”, whose members had a more advanced level of education and a higher income than average, seems to have a greater knowledge of the various types of food certification. This group stands out by interpreting labels as a sign of a food’s quality.

Consumers’ knowledge of certified products reduces their level of fear of buying or consuming fraudulent food products. Age did not influence the way food was purchased during the lockdown period, nor did it have an impact on the viewing of food labels on packaging. However, it was a determinant for a more exact name, “food fraud of a product”. In accordance with some authors [32,41], this may be due to the fact that age influences a person’s knowledge of food fraud and reveals the preferences of older consumers, as suggested by Grunert and Aachmann [65], for PDO-labelled products.

Among the other socio-demographic characteristics considered, in contrast to the findings of Moreira et al. [41], age influenced the perception of the information displayed on food labels, consumer confidence and knowledge about food fraud.

5.1. Theoretical Contribution

In light of the above, the theoretical contribution of this preliminary study was to extend the empirical studies into how consumers perceive the authenticity of food and food fraud/risk and to highlight how, in accordance with Cahyono [22], their perceptions differed depending on demographic factors, consumption choices and preferences in a context of uncertainty about food authenticity and food fraud/risk. The study’s findings suggest that family status has more influence not only on consumer’s knowledge of what types of products can be certified and which are of good quality but also on the way in which they perceive food fraud/risk.

5.2. Practical and Policy Implications

Practical and policy recommendations are provided in this study. It is important to acknowledge food fraud as a long-standing but emerging threat to the safety and quality of foods in national food safety systems, particularly in Italy, where the level of consumer concern about food fraud is high. To reduce consumer concerns and increase their trust in food, it is necessary to improve control of food frauds perceived as the most dangerous by consumers.

For this reason, the practical implications of this research, in line with recent studies [1,27,28,29,32,33,42,52], are useful for the protection of consumer health. Indeed, the perceptions of respondents regarding food authenticity and food fraud/risk observed in the present study could serve as guidelines for the food industry to improve food label design to enhance consumer understanding and usefulness of labels. Furthermore, in accordance with Moreira et al. [41], adding clear and legible information can increase consumer confidence in a period of low confidence in the food industry due to the recent food fraud scandals.

Implementation of educational programs to increase consumer knowledge about food labelling and fraud is essential. For this reason, in line with some authors [1,5,21], acquiring information on how consumers value authentic food when there is risk and uncertainty related to food fraud and which factors affect consumers’ evaluations would assist policymakers, responsible producers and manufacturers in making decisions about technological investments (such as blockchain) to address food fraud. However, consumers have different perceptions of food authenticity, varying degrees of knowledge and diverse experiences of fraud. This is due to the fact that their awareness, as reported in several studies [66,67,68,69,70,71], is influenced by their age, their geographical location, their country of origin, and stakeholder risk communications. With regard to the latter, scientific data show that, while in developed countries, it is possible to test and monitor fraudulent products because it is quite easy to find a list of the most common fraudulent food products [72], in developing countries, as suggested by Boatemaa et al. [73], basic foodstuffs are not subject to the same regulatory, economic and social frameworks. For this reason, consumers can not be considered as one homogenous group but can be divided, as has been performed for this study, into different groups based on their socio-economic characteristics, with the aim of developing personalised food risk communications.

6. Conclusions

By performing a network analysis technique and a primary survey, this study intends to investigate Italian consumers’ awareness, perception, and knowledge regarding buying authentic food, as well as food fraud and risk. A sample of 328 voluntary responses was analysed to answer three RQs. In response to the first RQ (i.e., do the socio-demographic characteristics of consumers increase their level of perception about the purchase of authentic food products?), this study revealed that all socio-demographic determinants considered in the analysis were important. In order of importance, these determinants were marital status, the presence of children, annual net income, gender, residence, household members, age and employment status.

In response to the second RQ (i.e., do the socio-demographic characteristics of consumers increase their level of knowledge of food fraud/risk?), the findings show how socio-economic characteristics not only influence the propensity to have information about what types of products can be certified and of good quality but also affect the way consumers perceive food fraud.

Finally, in response to the third RQ (i.e., are the socio-demographic characteristics, consumer preferences and choices concerning where to purchase, of the well-established “homogeneous” consumer groups, critical to ensuring that they do not engage in the purchase of inauthentic, fraudulent food?), the findings of the HCA reveal that the three well-established “homogeneous” consumer groups identified have different consumer preferences, choice, and level of awareness of food fraud/risk. Indeed, for the first group labeled “Young, single, and uninformed consumers”, while the price is an important factor in food choice, food fraud/risk is perceived as a problem related to the quality and quantity of products.

For the second group, it was found that Married and conscious consumers have a good knowledge of high-quality types of food certification, mostly because when buying food products, they usually look at the labels on the packaging and perceive food fraud as more than just an incorrectly stated origin, low quality, or incorrect quantity on a product.

Finally, the third group of “Educated and informed consumers”, thanks to their more advanced level of education and a higher income than average, have more knowledge and more experience in buying good quality, but they do not have an exact conceptualisation of food fraud, and the majority of consumers belonging to this group are afraid to buy and/or consume fraudulent food products.

The main value of this analysis is its comprehensiveness in terms of the degree of consumer awareness and consumer perceptions of authentic food, their knowledge about food fraud/risk, and the analysis carried out in terms of consumer categories. The implications of this research are useful for the protection of consumer health, which can be negatively impacted by the purchase of fraudulent food products.

This paper does not come without limitations, which provides opportunities for future research.

First, we must stress that, as an exploratory study, the purpose of the present work has been to gain preliminary insights rather than draw firm conclusions concerning the perceptions of respondents regarding food authenticity and food fraud/risk. In addition, the sample is based on convenience and snowball sampling and is restricted to a single country, Italy. Thus, one direction for further research could be to collect data randomly with a cross-national sample in order to enhance the generalisation. Furthermore, future research should be conducted on consumers’ purchasing practices, including where they purchase food from, how they prepare food, and how they store it, as these factors could have an impact on their risk perception.

The second limitation of the study is the general measurements of perceptions. This study did not identify any particular food item, even though food fraud may affect some food categories more than others. It would be preferable in future research to use a range of food products for the measurement so that the constructed validity would be less ambiguous. However, the number of respondents was deemed sufficient for an exploratory study of this magnitude.