Pesticide Safety in Greek Plant Foods from the Consumer Perspective: The Importance of Reliable Information

1. Introduction

The agricultural sector has a significant role in achieving food security and improving food availability, especially in underdeveloped nations with high rates of malnutrition and poverty. Nevertheless, there is uncertainty regarding global agriculture's capacity to satisfy this need by augmenting food output [1]. Safe food is vital to achieve food security and healthy diets. Food safety can be described as the guarantee that food, when prepared and consumed in accordance with its intended purpose, will not pose any danger to the consumer. The presence of unsafe food poses a threat to the gradual achievement of the fundamental right to sufficient food within the framework of national food security, and undermines endeavors to ensure the preservation of health and overall well-being. The assurance of food safety is essential for the attainment of nutritious diets, expansion of market opportunities, economic progress, and overall food security. Food security is defined as the universal availability of adequate, safe, and nourishing food to sustain a healthy and productive life for all individuals [2].

Pesticides are frequently used across several agricultural sectors to enhance productivity and improve product quality [3]. Research indicates that agrochemicals have a positive effect on crop yields and food production overall. Ensuring enough food supplies for a growing population and addressing global food security challenges are critical priorities. Pesticides, if used efficiently, enhance our quality of life by protecting crops from pests, diseases and weeds, thereby minimising crop losses and ensuring a consistent food supply [4,5,6]. Therefore, as Cooper and Dobson (2007) noted [5], the prudent use of pesticides is of great importance to ensure the sustainability of agriculture and to protect public health. Pre- and post-harvest losses have a negative impact on poverty and malnutrition [4,7,8], making prevention of these losses one of the most important contributions of agrochemical use [9,10,11].

The use of pesticides has raised concerns regarding the potential impact on both the environment and human health [12]. However, there is an ongoing debate regarding the potential impact of pesticide residues on public health due to the challenges associated with assessing the significance of their occurrence in food [3,13]. Furthermore, the long-term effects of pesticide residues on human health are not fully understood and further research is needed to identify their potential hazards [14]. The main focus of existing research on the effects of pesticides on human health has been on cases of occupational exposure [12,15]. However, concerns have been raised about the potential adverse effects of pesticides on the health of the general population through food consumption. Several studies have shown that the presence of pesticide residues in food poses a potential threat to public health [15,16,17]. Numerous investigations have provided empirical support for the neurotoxic [18,19] and cytotoxic [20,21] properties of pesticides and their consequent influence on gene mutation, chromosomal impairment and DNA damage [16,22]. There is compelling evidence of a possible link between pesticide use and increased susceptibility to colorectal cancer [23]. Furthermore, prior research indicates a possible correlation between pesticide exposure and a higher vulnerability to Parkinson's disease [18,24]. Studies conducted earlier have demonstrated that the major method of pesticide exposure for young children residing in urban and suburban regions is via dietary intake [25,26]. This issue has caused great concern for the welfare of these children, given their susceptibility to potential neurological and neurodevelopmental effects, as well as their distinct behavioural characteristics [26,27,28,29]. However, Tago et al. (2014) [15] have contended that only a minimal number of health outcomes linked to pesticides can be classed as causal.

A number of recent studies have suggested that existing risk thresholds for pesticide exposure may not adequately account for potential synergistic effects resulting from exposure to a combination of pesticides [14,30,31,32,33,34]. However, there is still a lack of agreement on the consequences of interactions following exposure to combined chemicals, with concerns raised about the potential for synergistic interactions between many compounds at low doses, which could potentially increase toxicity [35,36]. The predictability of interactions is challenging due to various factors such as relative doses, routes, timing and duration of exposure. In addition, the biological persistence of mixture components and the specific biological targets involved also contribute to the complexity of predicting interactions [37]. In their study, Hernández and Lacasaña (2017) [38] concluded that synergistic effects in dietary exposure are rare and cannot be accurately predicted in terms of the toxicity of the individual components of the mixture.

Under current EU legislation, a sample remains compliant if multiple residues are present as long as each residue does not exceed the individual MRL set for each substance [39]. However, the European Food Safety Authority (EFSA) has published two cumulative risk assessments on dietary exposure to pesticides that have acute effects on the nervous system and pesticides that have chronic effects on the thyroid. They concluded, with varying degrees of certainty, that the combined dietary exposure did not exceed the regulatory assessment threshold in all European countries in 2014, 2015 and 2016 for pesticides with acute neurotoxic effects [40] and chronic thyroid effects [41]. According to EFSA and Dujardin (2021) [42], there was no substantial change in the exposure to pesticides associated with long-term thyroid effects and pesticides related to acute neurological effects during the period 2016 to 2018 compared to the previous reference period (i.e. 2014 to 2016).

According to the Special Eurobarometer Report published by the European Food Safety Authority [43], the highest proportion of people in the EU Member States who consider a high intake of fruit and vegetables to be an important part of a healthy diet is found among Greeks and Spaniards, with a prevalence of seventy per cent. Studies that examined the potential health benefits of fruit and vegetable consumption have shown that the positive effects of increased intake of these food groups outweigh any possible negative effects that may result from the presence of pesticide residues on them [44,45]. The research conducted by Sandoval-Insausti et al. (2020) [46], based on data collected from large cohort studies of health professionals, found no significant association between cancer risk and consumption of fruits and vegetables containing pesticide residues. Valcke et al. (2017) [45] suggest that it is advisable for public health messages to continue to promote regular and abundant consumption of a wide variety of fruits and vegetables among the general population.

The current body of research on the impact of pesticides on dietary health is inconclusive, making it difficult for the general public to access reliable information on the relationship between pesticides and health [47,48]. It is worth mentioning that there is inadequate discourse pertaining to the function of pesticides in sustainable food production [49]. Therefore, the evaluation of the risks and benefits associated with pesticides will remain a matter of public concern. It's essential to note that the public's perception of hazards may vary from those made by regulators, typically relying on empirical evidence [5,48,50,51,52] and specific risk assessment techniques [53]. The presence of pesticide residues was perceived as minimal by experts, but consumers rated these threats from neutral to substantial in a study by van der Vossen-Wijmenga et al. (2020) [52]. Therefore, their presence in food may increase risk perception [54,55,56].

According to Han et al. (2020) [57], there is evidence that the implementation of pesticide residue monitoring and control systems could effectively mitigate food safety concerns. According to Williams and Hammitt (2001) [58] and Han et al. (2020) [57], risk communication efforts to educate consumers about food safety issues must effectively address concerns about the reliability of regulatory processes and information sources. In addition, these efforts should focus on establishing appropriate information dissemination systems to bridge the gap between regulators and the public. Understanding the processes that ensure food safety is of paramount importance to consumers and includes both scientific knowledge and building trust through understanding systems and mechanisms [48]. A potential link between trust and risk perception could exist in situations where individuals have insufficient knowledge about an important issue. Siegrist (2021) [59] suggests that if consumers trust the official risk management organization, their perception of the technology is more likely to be positive and acceptable.

Our study specifically focuses on examining the influence of information dissemination on consumer perceptions regarding the safety of Greek food. The proliferation of information has increased significantly in recent times, primarily due to the advent of the Internet and other portable electronic devices that facilitate access to large amounts of data. The information explosion has brought about significant changes in the way individuals obtain and disseminate information. Moreover, it has played a crucial role in accelerating the widespread dissemination of news, knowledge and ideas on a global scale [60,61].

Harris et al. (2001) [62] suggest that consumers' impressions of the risks associated with pesticide residues are consistently influenced by emotional factors, perhaps leading to an overemphasis or exaggeration in their responses to novel information. According to Tiozzo et al. (2019) [63], media often present food hazard information when stories contain sensational aspects. This approach serves to maintain media attention and facilitate the progression of the story. The term 'post-truth era' now refers to a period in which subjective emotions and personal beliefs are more influential in shaping public opinion and decision-making processes, compared to objective facts and empirical data. This shift has been facilitated by the widespread availability of social media platforms, which enable the rapid dissemination of both accurate and misleading information, as well as the creation of echo chambers that reinforce pre-existing biases. As a result, trust in established institutions and authoritative individuals has declined [64,65], as the task of distinguishing between credible sources of information and misinformation has become increasingly difficult. According to Ueland et al. (2012) [66], cognitive processing of information from other sources and self-reflection on one's own circumstances are particularly important when considering dietary hazards. Lobb et al (2007) [67] found that the majority of consumers are exposed to a large amount of information about food hazards from a variety of sources, including the media, government, retailers and consumer organizations. Particularly in the digital age, the information landscape has changed, as a plethora of information is readily available and the amount of information on the Internet is challenging. As data is readily shared but rarely verified, its veracity is questionable [68,69]. Public trust, perceived expertise of the message sender and recipient, and demographic variables such as age, education level, ethnicity and socioeconomic status influence the reception of a food safety message [70].

The media play an important role in risk communication [71]. To ensure the effectiveness of risk communication, it is imperative to provide comprehensive details about the characteristics of the risk and its associated benefits, as well as to acknowledge the uncertainties inherent in risk management and to present available alternatives for risk reduction [72]. The influence of the media on consumers' attitudes and perceptions of risks related to food safety has been explored in previous studies [73,74]. The phenomenon of risk amplification by the media has been extensively studied in the scientific literature [71]. According to Hohl and Gaskell (2008) [55], extensive media coverage has a greater tendency to amplify risk perceptions and create a greater desire for measures to minimize such perceptions. The media often prioritize dietary hazards based on criteria that are more consistent with newsworthiness than rankings provided by experts [63,75,76]. Journalists may deem a story worth reporting despite the presence of ambiguous claims, as exemplified in the context of risk communication [77]. A study conducted by Tiozzo et al. (2019) [63] revealed disparities in the reporting of food hazards between government monitoring systems and mass media. The study conducted by Lobb et al. (2007) [67] showed that the level of trust individuals have in the information presented by the media has a direct impact on their propensity to purchase a food product. The researchers also found that trust in public authorities plays a moderating role in the relationship between fear and purchase likelihood, particularly among consumers with higher levels of education. Education acts by reinforcing the positive effects of trust in information provided by public institutions and by modifying the influence of subjective norms. According to Kehagia and Chrysochou (2007) [78], the Greek media is very sensitive when it comes to providing the public with comprehensive information on food hazards. Furthermore, there is an observed prevalence of exaggerated concerns in media portrayals of food dangers, namely with the presence of pesticide residues in food. The journalistic representation of risk frequently fails to acknowledge the intrinsic uncertainty linked to probability, so fostering a feeling that the risk will inevitably result in adverse consequences. Additionally, because these hazards are the ones that receive a lot of media attention but have low probability but great potential repercussions, they are often overestimated in this coverage [77]. This phenomenon has the potential to magnify risks, making them more likely to occur or more likely to cause harm [76].

In contrast, the study conducted by Koch et al. (2017) [79] found that media exposure was associated with a higher level of understanding of the regulatory elements of pesticides, which in turn led to a reduction in reported levels of perceived risk. Access to reliable and specialized sources that provide risk assessments or comprehensive descriptions of the strict regulations governing the trade and use of pesticides is of paramount importance. Non-adherents were significantly less likely to receive information about pesticides from knowledgeable sources than supporters of the benefits of pesticide use over threats [56]. In addition, individuals may use media coverage as a cognitive shortcut for understanding complicated issues, especially in situations where they have no direct expertise in risk management [63,80].

In a recent Eurobarometer survey conducted on behalf of EFSA, Greeks ranked first in terms of personal interest in food safety [43]. Previous Eurobarometer surveys [81,82] show this trend stable in recent years. This is supported by the findings of Simoglou and Roditakis (2022) [56], who found that Greek consumers were especially concerned about the impact of pesticide residues on their own health and that of their families. The present study aimed to improve our understanding of how Greek consumers perceive the safety of plant foods, their sources of information about plant protection products and food safety, and the predictive variables related to their attitudinal beliefs. In this respect, the purpose of the study is to answer the following research questions:

RQ1: How do Greek consumers perceive the safety of Greek plant-foods regarding pesticide residues compared to other EU Member States?

RQ2: Which socio-demographic and attitudinal variables predict Greek consumers' personal perspectives on Greek plant food safety?

RQ3: What is the role of information sources in the formation of consumer perceptions of the safety of Greek plant foods?

2. Materials and Methods

Data collection was facilitated by a questionnaire hosted on Google Forms and distributed via email, Viber and Facebook Messenger. The survey was also disseminated through online news forums and magazines. The questionnaire was accompanied by a short document describing the objectives of the study, a consent statement, an explicit privacy statement with reference to the GDPR regulation, and the names and contact details of the researchers. After 10 days, follow-up emails were sent or social media reposts were made to mitigate the risk of non-response. The large sample size (1846 respondents) provides an in-depth insight into Greek attitudes towards food safety.

The survey was conducted between 6 and 31 March 2021. The questionnaire was divided into two sections: socio-demographic information and attitudes. It included closed 5-point Likert scale questions about their perceptions or attitudes. The Likert scale response levels were as follows: 1 = strongly disagree, 2 = somewhat disagree, 3 = neither disagree nor agree, 4 = somewhat agree, and 5 = strongly agree, or 1 = never, 2 = rarely, 3 = occasionally, 4 = frequently, and 5 = routinely, depending on the context.

The data collected through the questionnaire were first subjected to descriptive statistical analysis. Following Simoglou and Roditakis (2022) [56], the median was used as an appropriate measure of central tendency to present and interpret the results. The non-parametric Kruskal-Wallis and Mann-Whitney U tests were used to compare ordinal variables.

Principal component analysis (PCA) was used to determine the basic information structure contained in the original interrelated variables and to summarize it using a smaller set of composite variables. An eigenvalue threshold greater than 1 was used as a criterion for the number of principal components (PCs) retained. The oblique (maximum) rotation resulted in more simplified PC loads, with each variable load on an individual PC. In the final analysis, only variables with loadings greater than or equal to 0.60 were retained. The adequacy of the PCA was assessed using the Kaiser-Meyer-Olkin (KMO) test, which uses values between 0 and 1 as a measure of sampling adequacy, and Bartlett's test of sphericity, where a significant result indicates that at least some pairwise correlations among variables are non-zero [83].

McDonald's ω reliability coefficient of internal consistency was calculated and reported for scale variables loading on a single PC. To obtain a single measure for each PC, the variables loading on a single PC were combined using composite scores for further analysis [83].

Participants' perceptions of the safety of Greek food products in comparison with products from other EU member states were used as the dependent variable in a binary logistic regression to identify potential predictors. Sociodemographic variables and factors retained from the PCA were included in the model as potential predictors. The calculated odds ratios (OR) and 95% confidence intervals (CI) are presented. Each independent variable in the model was subjected to the Wald test for statistical significance. Finally, performance metrics such as specificity and sensitivity, which represent the proportion of true negative and true positive observations predicted by the model, and AUC (area under the ROC curve, which represents the trade-off between true positive and false positive rates), which is an overall test of predictive accuracy and indicates the degree of discrimination between true positive and false positive values of the estimated model, are considered. Large AUC values (greater than 0.5-1) indicate an excellent model fit [83].

For the purpose of logistic regression analysis, participants' opinion variables were split into two levels with a binary outcome: ''favour'' = 1, after grouping the Likert response levels somewhat agree and strongly agree, and ''disfavour'' = 0, after grouping the Likert response levels strongly disagree, somewhat disagree'' and neither disagree nor agree, following Simoglou and Roditakis (2022) [56].

Latent class analysis was used to identify potential underlying consumer categories that could explain the observed patterns across cases [84] in an effort to further clarify the significance of information sources on participants' perceptions of the safety of Greek plant foods. As class-defining variables, the original variables pertaining to the information sources used by participants for pesticide information were chosen. To characterize the differences between classes in greater detail, a chi-squared correlation test and a Mann-Whitney U test were conducted on the nominal background variables and the ordinal focus variables, respectively.

The statistical analyses were carried out using Jamovi 2.4.2 [85] and Jasp 0.17.3 [86], both of which use the R programming language.

3. Results

3.1. Characteristics of Survey Participants

All Greek regions were represented by a total of 1,846 respondents to the survey. The population under study is comprised of plant-based food consumers between the ages of 18 and 65 who reside in both urban and rural areas of Greece (the mainland and the Islands). The survey respondents' sociodemographic details are shown in Appendix A, Table A1. Both genders (females 48.5%) and all age groups were adequately represented. The majority of respondents (45.1%) resided in the south of Greece (including Athens), while 26.6% resided in the centre and 29.1% in the north. In accordance with Simoglou and Roditakis (2022) [56], the age groups were reduced to three for the purposes of the analyses. Of the participants, 22.5% were between the ages of 18 and 34, 58.1% were between the ages of 35 and 54, and 19.4% were older than 55. The vast majority of participants possessed at least a high school diploma and were primarily civil servants (44.1%), private employees (18.6%), self-employed individuals (12.1%), university students (11.7%), and farmers (5.3%). In addition, a number of personal habits relating to free time, smoking, sports participation, and vegetarianism were recorded (Appendix A, Table A1).

3.2. Participants' perspectives on the safety of Greek-produced plant foods in terms of pesticide residues in comparison to those of other EU member states.

The frequency distribution of participants' responses to the research question on whether or not they perceive Greek plant foods to be as safe as those from other EU member states (RQ1) was determined as follows. Strongly disagree (median = 1): n = 125 (6.77%); somewhat disagree (median = 2): n = 290 (15.71%); neither disagree nor agree (median = 3): n = 499 (27.03%); somewhat agree (median = 4): n = 600 (32.50%); Strongly agree (median = 5): n = 332 (17.99%). The median of the responses is 4 (IQR: 1), which in terms of central tendency indicates partial agreement with the statement. Following Simoglou and Roditakis (2022) [56], a binomial proportion test was applied after dividing the response rates into two levels with a binary outcome, i.e. 'agree' and 'disagree'. In accordance with the null hypothesis that the two levels are equally likely (p = 0.50), a non-significantly higher proportion of positive responses was found. The proportion of unfavourable responses was 0.495% (95% CI: 0.472 - 0.518%), n = 914, while the proportion of favourable ones was 0.505% (95% CI: 0.482 - 0.528%), n = 932 (p = 0.692).

3.3. The variables predicting the participants' attitudes towards the safety of Greek-produced plant foods research question

Principal components and logistic regression analyses identified socio-demographic and attitude variables as significant predictors of Greek consumers' perceptions of the safety of Greek-produced plant foods relative to those from other EU Member States with respect to pesticide residues (RQ2).

3.3.1. Principal components underlying the participants' attitudes

The analysis included 22 original variables with loadings greater than 0.60. Six principal components (PCs) were retained on the basis of the criterion of eigenvalues, as they explained a significant amount of variance (64.6%) and were considered to reasonably represent the underlying structure of the data (Appendix A, Table A2). The range of reliability coefficients (McDonald's) was 0.698% to 0.865%. The sum of squared loadings (eigenvalues) was greater than 1.0 for each PC. Bartlett's test of sphericity (p 0.001) and the KMO measure of sampling adequacy (0.829) indicated that the correlation matrix was appropriate for principal components analysis.

Four variables account for 13,502% of the variance in the first principal component, which summarizes variables representing "Official information sources" used by participants to learn about pesticides. As information sources, official websites, newsletters from public institutions, scientific periodicals, and agronomists had the strongest correlation with the first PC. Variables loading onto the second PC included participant opinions regarding pesticides' contribution to the national income and to increased food production, the inevitability of pesticide use, and assertions that pesticides' correct application safeguards the user or consumer. Therefore, this is related to "Perceived pesticide benefits" and accounts for 12.706% of the variance. The television and radio, the electronic and printed press, and social media were variables loading to the third principal component, "General information sources on pesticides", which accounts for 10.863% of the variance. Variables loading onto the fourth PC included food labelling and traceability information, and the safety of certified and integrated crop management food products. This is known as "Confidence in traceability" and it accounts for 10.205% of variation. The fifth PC consisted of variables representing "Perceived pesticides' risk", which accounts for 9,008% of the variance. Here contained are variables regarding the participants’ health being at risk due to pesticide residues, their uncertainty regarding the health of their own people, and their concerns regarding pesticide residues in food. The sixth PC was correlated with the participants' perceptions of the safety of plant food consumption, the low risk of consuming fruits and vegetables, and the existence of a framework for the control of pesticide residues in food. This variable is labelled "Perceived plant-food safety" and accounts for 8.288% of the variance.

The relationship of the six PCs is summarized in Figure 1. On the horizontal axis, perceived pesticide risk loads in the opposite direction to perceived plant-food safety, pesticide benefits and consumer confidence in certification, and is negatively correlated with these variables. It is also basically orthogonal to official sources of information, which also implies a rather negative relationship.

3.3.2 Predictive variables of participants' perceptions – Logistic regression model

A binary logistic regression analysis was conducted to identify any variables that might predict participants' attitudes towards the safety of plant foods. Specifically, the dependent variable concerns participants' responses to the statement that plant-based foods produced in Greece are as safe as those produced in other EU Member States in terms of pesticide residues. Using a stepwise procedure, background socio-demographic variables and the six principal components previously retained from the PCA were included in the model as potential predictors.

Specificity (percentage of cases correctly negatively predicted) and sensitivity (percentage of cases correctly positively predicted) are the performance parameters of the model. They have values of 0.74 and 0.71 respectively. The overall predictive accuracy of the model, as measured by the AUC value (area under the ROC curve), is 0.79 (Appendix A, Table A3), which is considered a very good model fit.

The regression coefficients for perceived plant-food safety (b = 0.863; p < 0.001), higher education level (b = 0.553; p < 0.001), age group ≥ 45 years (b = 0.423; p < 0.001), official information sources (b = 0.408; p < 0. 001), male gender (b = 0.308; p = 0.010), confidence in traceability (b = 0.231; p < 0.001) and perceived benefits of pesticides (b = 0.228; p < 0.001) are all positive and statistically significant. The results indicate that respondents who report higher levels of plant food safety, are better educated, over 45 years of age, more likely to obtain official information on plant protection products, male, have higher levels of trust in the traceability procedures of plant food production, and perceive greater benefits from plant protection products, have a higher likelihood of accepting the safety of Greek plant food.

The regression coefficients for perceived pesticide risks and pesticide user status are both negative and statistically significant (b = -0.123; p = 0.045 and b = -0.327; p = 0.020, respectively). This suggests that people are less likely to view plant foods from Greece as safer for pesticide residues than those from other EU countries if they have a high perceived risk of pesticide residues in food, and actually use plant protection products whether for professional or recreational purposes. In order to improve the analysis of the influence of pesticide user status on participants' opinions, certain key variables were controlled for. It was found that people who use pesticides reported a greater sense of safety regarding Greek plant foods compared to those who do not use pesticides (Mann-Whitney U, W = 371,479.0; p < 0.001), although this effect size is modest, as indicated by the rank-biserial correlation coefficient of 0.109. Nevertheless, there was a 45.05% probability for pesticide users (n = 788) to agree with the position of reduced monitoring frequency for Greek food compared to the EU. In contrast, individuals who do not use pesticides (n = 1058) had a 32.99% probability of expressing the same attitude (χ² = 27.860; df = 1; p < 0.001). Similarly, it was observed that those who use pesticides had a 66.87% probability of agreeing with the tendency to consume food that they produce themselves. In comparison, a significantly lower percentage of non-users, 33.13%, showed this level of agreement (χ² = 146.420; df = 1; p < 0.001). There is a statistically significant, albeit modest, positive correlation between these two original attitude variables (Spearman's ρ = 0.10; p < 0.001).

According to the odds ratios, the odds of a participant assuming that Greek plant foods are as safe as in other EU Member States change by a factor of 2.46 (95% CI: 2.17 - 2.79) with each unit change in their propensity towards perceived plant food safety, by 1.73 (95% CI: 1.33 - 2.25) if they have a higher education level, by 1.50 (95% CI: 1.21 - 1. 87) if they belong to the age group ≥45 years, by 1.40 (95% CI: 1.22 - 1.61) with each unit shift towards obtaining official information, by 1.30 (95% CI: 1.05 - 1.62) if they are males, by 1.26 (95% CI: 1.12 - 1.42) with each unit increase in their trust in traceability, and likewise 1.23 (95% CI: 1.09 - 1.39) towards perceived benefits of pesticides. In contrast, participants' odds of perceiving Greek plant foods as safe change by a factor of 0.72 (95% CI: 0.55-0.94) if they are pesticide users and by 0.88 (95% CI: 0.78-1.00) for each unit change in perceived pesticide residue risk.

No other background variables or principal components were found to be significant predictors and are therefore not supported in the model.

3.4. Latent class analysis of the respondents

The Latent class analysis (LCA) was based on the original perception variables relating to the sources of information used by participants to obtain information on pesticide-related issues of general concern. The aim of the analysis was to determine the influence of the public's sources of information on agrochemicals and food safety issues on their perception of the safety of Greek food (RQ3). The analysis included eight original ordinal variables relating to official web pages, public agency bulletins, scientific periodicals, agronomists (official sources of information), television/radio, electronic and print press, and social media (general sources of information).

Based on previous findings by Simoglou and Roditakis (2022) [56], a two-class solution was chosen for further analysis. Class 1 (n = 871), with a prevalence probability of 0.468, comprised respondents who were less likely to agree with the research question (RQ1), as the median of the response distribution was 3 (IQR: 1), corresponding to neither disagreement nor agreement at the level of central tendency. In contrast, the median in class 2 responses (n = 975), which has a prevalence probability of 0.532, was 4 (IQR: 1), indicating a level of partial agreement with the research question. The median difference is statistically significant (Mann-Whitney U, W = 350,100.5; p < 0.001).

To further characterize the two groups of participants, the Mann-Whitney U test and the chi-squared test of association were performed, taking into account the focus variables (principal components, or original opinion variables) (Table 1) and the background variables (demographics) (Table 2), respectively. According to the results, class 1 consists of participants with a higher intensity of perceived pesticide residue risks (p < 0.001), a higher proportion of females (p < 0.001), residents of areas of northern Greece (p < 0.001), non-pesticide users (p < 0.001), civil servants and university students (p < 0.001), and secondary school graduates (p = 0.011). Class 2 is characterized by more frequent use of official and general sources of information on pesticides (p < 0.001), higher perceived benefits from their use (p < 0.001), higher confidence in the role of traceability in food safety (p < 0.001) and higher perceived safety of plant foods (p = 0.001). They are more likely to be male (p < 0.001), to live in southern Greece (p < 0.001), to be farmers (p < 0.001), to be employed in the private sector (p < 0.001), to be self-employed (p < 0.001), to be retired (p < 0.001) or unemployed (p < 0.001) and, finally, to be university graduates (p = 0.011). It is important to note that there was no statistically significant difference between the two groups in terms of the participants' high level of concern about their safety from pesticide residues in food (p = 0.588). Similarly, no statistically significant difference was observed in their level of neutrality towards the statement that Greek food is less frequently tested for residues compared to other European Union Member States (p = 0.760). The age of the participants in both classes did not show a statistically significant difference (p = 0.738).

5. Conclusions

The aim of this study was to provide an in-depth understanding of the factors that influence Greek consumers' views on the safety of Greek plant-based foods compared to those produced in other EU Member States, with a particular reference to concerns about pesticide residues. Concurrently, the main focus of this study is to examine the influence of pesticide-related information on the perceptions and attitudes of Greek consumers towards food safety. The study found that Greek consumers have a balance between two conflicting perspectives on the safety of Greek food. Half of the respondents believe that Greek food has a level of safety comparable to other European foods, while the other half hold the opposite view. Several predictive factors have been identified that influence the dichotomous nature of Greek consumer behaviour. These factors include perceived safety of plant foods, higher education level, age, gender, perceived contribution of traceability to food safety, perceived benefits of pesticide use, perceived risk of pesticide residues in food, pesticide user status, and information obtained from authoritative-official sources. Obtaining information about pesticides and related food safety concerns emerges as a crucial determinant in shaping consumer perspectives. Receiving authoritative-official information was found to be associated with a reduction in risk perception and improved scores in focus-opinion variables, and a positive response to the research question on Greek food safety.

The results of our study suggest potential implications for the importance of providing accurate and timely information to the public. First, it is of utmost importance that administrative and regulatory institutions prioritize transparency in their risk communication strategies. It is highly important to give careful consideration to the impact of socio-demographic and other intrinsic attributes on individuals' perceptions of risk. By doing so, regulators can effectively build trust and credibility with consumers. It is also essential to ensure the efficient dissemination of official information across multiple platforms, including websites, social media platforms and public awareness campaigns. However, it is crucial to prioritize the maintenance of direct communication channels between public institutions and consumers. Such mechanisms, like agricultural warning services, or peripheral agriculture directorates should be further developed to take on an expanded role in informing the general public about food safety issues. Ensuring effective risk communication channels between public institutions and consumers, bridging the gap between experts and the general public, is essential to address concerns and inquiries and ultimately promote transparency and accountability. In addition, tailoring risk communication strategies to different socio-demographic sub-groups can ensure the effective delivery of information in a way that is understandable and relevant to a wide range of people, thereby promoting greater uptake and engagement. In addition, collaboration with other stakeholders such as consumer and producer organizations can further strengthen these communication initiatives.