Do Consumers Really Want to Reduce Plastic Usage? Exploring the Determinants of Plastic Avoidance in Food-Related Consumption Decisions

The mass production and use of plastics over the past decades is now posing a threat to the environment due to increasing pollution. The industrial transition towards more sustainable plastic alternatives has already started, but the process will require years to become truly feasible from an industrial and economic standpoint. In the meantime, a significant contribution to limit plastic-related environmental impact could arise from consumers’ daily consumption choices. This study aims at investigating the determinants of consumers’ decision to avoid purchasing single-use plastic eating utensils, food products with multiple plastic packaging, and plastic water bottles while grocery shopping. The results of the Structural Equation Model reveal that environmentand health-related concerns associated with plastics are key drivers of plastic avoidance. Instead, subjective knowledge and the importance attached to the commitment of third parties for tackling the plastic issue only indirectly affect consumer behaviors with regard to plastic avoidance.


Introduction
Over the last seventy years, plastic production has constantly increased [1]. Because of their low cost, durability, ductility, and the possibility of being easily added with other elements to enhance their properties, plastic materials currently find application in a wide range of industries. Plastics are now everywhere and constitute objects that are massively used in our everyday life [2]. Despite their anticipated benefits, the mass use of plastics is now posing a threat to the environment [3,4]. Geyer et al. (2017) reported that, as of 2015, only a limited share of all plastic waste has been recycled or incinerated, while nearly 80% has been disposed in landfills or accumulated in the natural environment [5].
Currently, the packaging industry represents the first industry of destination for plastic materials, absorbing about one-third of the worldwide plastic production [1]. Of this, the packaging destined to the food system represents a sizable share [6]. Plastic food packaging is suitable to preserve foods from deterioration, maintaining their properties and quality, while also extending the product shelf life [7]. In the past years, some health concerns related to the possibility that plastics may release residuals in food have been raised both from consumers and from the scientific world. The case of Bisphenol A represents a well-known example [7]. However, the use of these components is strictly disciplined at the EU level and the allowed residual levels in food products amply guarantee public safety [8]. As such, plastic materials are safely used for food packing, and the projections suggest that their usage is expected to quadruplicate within the next 30 years [6]. Plastic materials are extensively used in the food context also for the production of single-use cutlery, dishes, cups (referred to as plastic eating The present paper is structured as follows: Paragraph 1.1 describes the literature background for the study and the hypotheses development, Section 2 illustrates materials and methods used for the study, Section 3 shows the results obtained, and Section 4 provides with the discussion of results.

Literature Background and Hypotheses Development
Consumer decision to avoid plastic objects, which was the target behavior of this paper, can be conceptualized as a PEB. This term refers to any practice (such as purchasing, use, post-use, management, activism behaviors) that consciously seeks to limit the impact of human activities on the environment [20][21][22]. Pro-environmental behaviors involve a number of different targets, such as reducing greenhouse gas emission, reducing water and energy consumption, recycling, therein including the reduction of plastic use. Accordingly, the hypotheses to be tested in this paper were developed within the PEBs framework.
Many studies in the past have investigated the reasons why people engage (or do not) in several PEBs, approaching this issue from different perspectives. For instance, previous studies in the psychology field focused on the role of individual altruism [22]. Other studies recognized individual love of nature, beliefs and attitudes, childhood experience, lifestyles and habits as possible drivers of PEBs [22][23][24][25][26]. Among others, consumers' environmental and health-related concerns have emerged as key factors in influencing PEBs. Environmental concern relates to the extent to which consumers worry about the possible negative consequences that unsustainable consumption patterns can exert on the environment [22,27,28]. Health concern, instead, relates to possible negative effects for human health. As reported in previous studies, both environmental and health concern are associated with plastic usage. Indeed, consumers perceive plastic litter, waste, and pollution as problematic and threatening both for the terrestrial and the marine environment [29][30][31]. At the same time, plastic pollution is perceived as risky for human health due to possible contamination of air, water, and also food [32][33][34]. In this regard, past findings demonstrated that consumers worry that plastics may release contaminants in food. When so, the perceived quality of the product is reduced and health-related risk perception increases [34][35][36][37][38]. Overall, the results of these studies indicated that the higher the concern, the higher the likelihood of actively engaging in sustainability-oriented behaviors [23,[39][40][41]. Based on such evidence, the first hypothesis of the study was formulated as follows: Hypothesis 1 (H1). The higher the consumer plastic-related concern, the higher the probability that they avoid purchasing single-use plastic eating utensils, plastic water bottles, and food with multiple plastic packaging.
Alongside environmental and health concern, also knowledge plays an important role in driving individual behaviors. Most of past studies on PEBs specifically focused on environmental knowledge (i.e., how much people are knowledgeable about specific environmental issues) and highlighted that it is positively associated with the extent to which people engage in PEBs [16,22,23,[39][40][41]. Their findings indicated that when consumers are knowledgeable about a specific environmental problem, they are also more likely to behave sustainably in that specific regard [21,42].
When dealing with the role of knowledge in driving behaviors it is common to refer to two different constructs, namely the objective knowledge and the subjective knowledge. The former reflects the extent to which an individual accurately and objectively knows the specificities of a certain issue, while the latter represents how much he/she thinks to know about that topic. There is still debate in the literature on whether objective or subjective knowledge mostly affect PEBs [16,43,44]. However, several studies suggested that subjective knowledge is a better predictor of both pro-environmental intentions and behaviors. Such evidence emerged in past research involving different PEBs such as recycling behavior [16], energy conservation behaviors [17], and organic food consumption [18]. On these grounds, the second hypothesis of this work was formulated as follows: Hypothesis 2-1 (H2-1). The higher the consumers' subjective knowledge about plastic-related issues, the higher the probability that they avoid purchasing single-use plastic eating utensils, plastic water bottles, and food with multiple plastic packaging.
As for the role of knowledge, there is another aspect to highlight. While some studies considered environmental knowledge as a pre-requisite for engaging in PEBs [45], others described it as a necessary but not-sufficient condition "per se" to behave sustainably [42]. At present, this latter approach is increasingly accepted. In fact, evidence demonstrate that in many cases environmental knowledge does not directly increase PEBs. Instead, it positively affects the extent to which individuals act sustainably by influencing other variables, such as emotional engagement, values, attitudes, awareness, as well as concern [17,18,21,46].
Based on such evidence, and following the rationale of H1, it was hypothesized that: Hypothesis 2-2 (H2-2). The higher the subjective knowledge about plastic-related issues, the higher the plastic-related concern.
A further peculiar characteristic of PEBs is that they are not just related to an individual dimension, but they are also strongly associated with a social dimension. Firstly, environmental sustainability can only be achieved as the result of a common commitment involving the society at different levels [10]. Not only citizens are called to do their part, but also scientist, policy makers, and industries. For instance, with specific regard to plastics, policy makers are called to set the guidelines and rules for a more sustainable production and consumption system [47][48][49], while consumers should act together in a sustainability-oriented manner with their daily behaviors and consumption decisions. At the same time, the industrial sector must reduce plastic use and provide consumers with innovative products and tools, which can enable them to behave sustainably in their everyday life [47]. In fact, the actions of a single individual can only have a very limited impact in increasing environmental sustainability. Instead, they assume an essential role when other members of the society act in the same direction and share the same commitment for the environmental cause [50]. When so, the collective effort is rewarded with increased social welfare, because the benefits deriving from sustainability-oriented behaviors are shifted on the society as a whole [48]. In such context, the extent to which individuals engage in PEBs is strongly affected by what others do [50][51][52][53]. Past studies demonstrated that people may lose their motivation to engage in PEBs when they perceive that they are the only that behave sustainably. In other words, "Why should I if no-one else does?". The latter motivation has been reported by Prabawa-Sear and Baudains (2011) as a key discouraging factor for undertaking PEBs [54].
Building on such evidence, the following was hypothesized:

Hypothesis 3 (H3).
The importance attributed to the commitment of third parties for tackling plastic-related environmental impact affects the extent to which individuals avoid purchasing single-use plastic eating utensils, plastic water bottles, and food with multiple plastic packaging.

Data Collection
The data for the analysis were collected by means of a consumer survey carried out in the metropolitan area of Milan (northern Italy) during spring 2019. Before collecting the data, a pilot study was conducted on a small sample of 40 respondents in order to test the appropriateness of the questions and the length of the survey. Consumers were approached face-to-face by the researchers following a random sampling approach. They were verbally informed about the aim of the survey and then invited to respond to the questionnaire. All invited respondents were informed that the questionnaire was anonymous and that the data were used exclusively for research purposes. After agreeing to participate, they received from the researchers a brief explanation about the type of questions involved and the related response scales. Respondents were then allowed to fill the questionnaire autonomously and to deliver it to the researcher after responding. Only people older than 18 years old (i.e., the legal age in Italy) were considered eligible for the survey. Totally, 520 adult consumers were approached. The final sample resulted in 441 complete questionnaires, after screening for incomplete answers. The ad hoc questionnaire formulated for this study was composed by five sections, respectively aimed at investigating plastic-related behaviors and at eliciting respondents' plastic-related concern, subjective knowledge, the importance attributed to the commitment of third parties for tackling the plastic issues. One additional section was dedicated to the main sociodemographic and economic variables. The order of presentation of the sections was randomized to overcome potential ordering bias (integral version of the questionnaire available upon request.).

Analytical Approach
A Structural Equation Modeling (SEM) approach was used to test the hypotheses described in Section 1. The SEM is a multivariate statistical method that is based on the combination of factor analysis and multiple regression analysis. It is extensively applied in many social, educational, and behavioral research fields to simultaneously analyze structural relationships (i) among exogenous measured variables (i.e., indicators) and the related endogenous latent constructs (i.e., variables that do not have observed/manifest realization in the sample population), as well as (ii) among the latent constructs and the dependent variable in the model [55][56][57].
The SEM can be described as in Equation (1): where ξ is a column vector of n exogenous latent variables, and η is a column vector of m latent (endogenous) variables. B is a matrix of coefficients associated with the latent constructs, and ζ the remaining error terms associated with them. The measurement equations take the following form: where y (p×1) and x (q×1) are the column vectors of the p-measured endogenous variables and q-measured exogenous variables, Λy and Λx are the associated factor loadings matrices, and ε and δ are the remaining uncorrelated error terms. Subsequent analytical steps were followed to perform the SEM. The first step consisted in defining the latent constructs to be included in the model. Each latent construct was expressed by a set of related measured variables (or indicators). To define each latent, a factor analysis was performed on the related measured variables. Then, a confirmatory factor analysis was used to select the indicators that better expressed the latent construct. The relationship between the aforementioned constructs was then explored through a mediation analysis aimed at unveiling possible indirect relationship across the latent constructs. The final SEM was performed by taking into account the mediation results.

Measures
The questionnaire used for the data collection was developed to measure the latent constructs included in the model, named PL_BEHAV, representing the dependent variable of the SEM, PL_CON, SUBJ_KN, and SHRD_COMM. The questionnaire also included questions to elicit the sociodemographic and economic characteristics of the sample. The latter are reported in Table 1, in comparison with the available sociodemographic data of the Milan population, as provided by the Italian Institute of Statistics (2019) [58]. As illustrated, while the distribution by gender of the sample population reflects the Milan population data, some differences emerge with regard to age. The age class 25-34 years old was overrepresented in our sample, while people older than 65 years old were remarkably underrepresented. All the latent constructs were measured by means of multiple items taken or adapted from previous studies with validated scales. The explanatory factor analysis was performed on all these items and the Kaiser-Meyer-Olkin (KMO) and Bartlett's sphere test was used to test the reliability of the results. The coefficients showed that the indicators of the latent constructs were unrelated and, therefore, suitable to describe our model (KMO = 0.756; significance of Bartlett's test = 0.000). Afterwards, the confirmatory factor analysis was performed in order to test the goodness of the constructs. The resulting model fit were slightly below the acceptable values. Thus, the worst-performing items were dropped in a second CFA, and this resulted in a better model with largely acceptable fitting measures (χ 2 /df < 3; TLI > 0.9; CFI > 0.9; RMSEA< 0.08) [59].
The dependent variable (PL_BEHAV) aimed at capturing the the extent to which respondents consciously avoid plastics while grocery shopping, with a focus on food and drink packaging. In detail, the dependent variable in the model was represented by an index resulting from the summation of the scores assigned to three questions. The latter asked respondents whether they generally avoid purchasing (i) single-use plastic eating utensils (e.g., plastic cutlery, plates, cups, etc.), (ii) food products with multiple plastic packaging, and (iii) plastic water bottles. These objects were chosen as they represent the most ubiquitous plastic objects in the environment, they are massively used and easily discarded in incorrect manners, and their market is expected to remain stable or even to grow in the next years [6,10]. Questions were reported in the form "When you go get groceries, do you generally avoid buying . . . ?" with "Yes/No" dummy response alternatives. These three items were formulated based on a group of questions included in the Special Eurobarometer 468 on the "Attitude of European citizens towards the environment" [60]. The resulting index ranged from 0 to 3, with higher scores associated with more sustainable behaviors (i.e., higher plastic avoidance).
In order to test the hypotheses formulated with regard to the effect of individual plastic-related concern, subjective knowledge, and the importance attached to the commitment of third parties for tackling plastic issues, the following exogenous latent constructs were considered. The first latent variable (PL_CON) aimed at reflecting consumers' level of concern about the effects of plastic contamination on the environment as well as on health. While past studies on PEBs have focused exclusively on environmental concern, this study also considered the health dimension. Indeed, as explained in Section 2, when plastic packaging is put in contact with food consumers might worry about the release of possible contaminants in the product. Hence, PL_CON was expressed by three items, measured on a scale ranging from 1 (= strongly disagree), to 7 (= strongly agree), adapted from the Eurobarometer 468 survey [60]. Furthermore, when dealing with the role of environmental concern on PEBs, most of past studies used unspecific measures to be associated with specific behaviors [61][62][63]. This approach implies not accounting for some peculiar determinants that can be strictly related to the specific behavior under study, and not with others. Bamberg (2003) discussed this issue, suggesting that context-specific measures of environmental concern that closely relate to the PEB under study could provide more accurate results [64]. In fact, it is quite frequent to observe inconsistency between individual self-reported environmental concern and concrete behaviors. This can be, at least in part, attributed to the way in which environmental concern is elicited [21,23]. For this reason, in this paper the latent PL_CON was measured with indicators that are directly associated with plastics.
The second latent construct in the model (SUBJ_KN) was expressed by three items adapted from Pieniak et al. (2010) [18]. The items aimed at eliciting respondents' subjective knowledge about plastic materials, their usage and sorting, and at capturing individual perceived level of plastic-related subjective knowledge with respect to other people's [18]. Also in this case, respondents stated their level of agreement to each item on a Likert scale from 1 (strongly disagree) to 7 (strongly agree).
The latent construct SHRD_COMM represents the importance attached by consumers to the commitment of third parties for tackling the plastic issue. This latent was expressed by means of four items adapted from the Eurobarometer 468 (2017) [60]. Such items aimed at measuring the importance that respondents attributed to the actions of producers, authorities, and other citizens for tackling the plastic environmental issue (on a Likert scale from 1 = not at all important, to 5 = very important). All latent constructs as well as their indicators are illustrated in Table 2.

Reliability and Validity of the Model
The validity of the constructs was investigated through the values of Composite Reliability (CR), Cronbach's Alpha, and Average Variance Extracted (AVE) [57]. Results are showed in Table 3. Almost all constructs measures fitted the required levels (CR > 0.7; Cronbach's Alpha > 0.7; AVE > 0.5), indicating an overall good convergence validity [62]. The low AVE value associated with the shared commitment latent variable could be expected, as this issue is relatively new in the literature. However, we retained the latent variable because of the overall good fitting of other measurements and of the whole CFA [65]. Moreover, the overall goodness-of-fit measures of the model were highly acceptable, thus suggesting that the relationships tested were significant and well explain our dependent variable.

Mediation Analysis
The mediation analysis was conducted to unveil possible indirect relationship across the latent constructs. Indeed, past evidence showed that concern, knowledge and the commitment of third parties for the environmental cause, can have an indirect effect on PEBs [21,28,40,46,66]. The mediation analysis was conducted following the procedure proposed by Baron and Kenny (1986) and further adopted also in Elgaaied (2012) [67,68]. The resulting paths and coefficients of the model are reported in Figures 1 and 2. The results in Figure 1 show that the relationship between plastic-related SUBJ_KNOW and PL_BEHAV was fully mediated by PL_CON. This is in line with past findings indicating that knowledge does not directly affect PEBs, but positively influences the extent to which consumers engage in environmentally friendly behavior by increasing concern [21,46]. Similarly, and in line with past literature [28,66], the results in Figure 2 show that also SHRD_COMM did not directly increase the extent to which consumers consciously avoid plastics while grocery shopping, Instead, it indirectly affected PL_BEHAV by increasing PL_CON. As such, the results of the mediation analysis indicated that H2-1 (i.e., the higher the consumers' subjective knowledge about plastic-related issues, the higher the probability that they avoid purchasing single-use plastic eating utensils, plastic water bottles, and food with plastic packaging) was rejected, while H2-2 (i.e., the higher the subjective knowledge about plastic-related issues, the higher the plastic-related concern) was confirmed.

Structural Equation Model
Accounting for the results of the mediation analysis the Structural Equation Model was performed as described in Figure 3. The resulting coefficients and goodness-of-fit indices are displayed in Table 4. The results showed a positive and significant association between PL_CON and PL_BEHAV (β = 0.212, p < 0.001). To explain, this positive relationship indicates that when consumers are concerned about the possible negative effects of plastic contamination on the environment as well as on their health, they are more likely to avoid purchasing single-use plastic objects, plastic water bottles, and products with multiple plastic packaging. This evidence confirmed H1, that is the higher the environmental concern, the higher the probability to avoid purchasing single-use plastic objects while grocery shopping.
The coefficients obtained also showed a positive relationship between SUBJ_KN and PL_CON (β = 0.253, p < 0.001). Indeed, this result reveals that consumers exhibiting a higher level of subjective knowledge on the issue of plastic pollution and on how to act to diminish the impact of plastics on the environment were also more concerned about the negative consequences of plastics on the environment and on human health. This outcome of the model fully confirmed H2-2, which stated that the higher the subjective knowledge about plastic-related issues, the higher the level of plastic-related concern.
Lastly, the relationship between SHRD_COMM and PL_CON resulted to be positive (0.553) and significant (p < 0.001). This finding reveals that consumers who attributed more importance to the collective commitment against plastic pollution (with other citizens, with the firms, and the policy makers) were also more concerned about the adverse effects of plastic over-use. This latter evidence partly confirmed H3. In fact individual perception of others' commitment in reducing the environmental impact of plastic only indirectly affected the probability that consumers avoid purchasing single-use plastic eating utensils, plastic water bottles and food with plastic packaging, by increasing plastic-related concern. Notes: n.s. = non-significant; significance at *** p < 0.001.

Discussion
This paper aimed at investigating which could be the drivers of consumers' decision to avoid plastic usage with a specific focus on the food context. Using a Structural Equation Modeling approach, we explored the role of plastic-related concern, subjective knowledge, and the importance attributed to the commitment of third parties for tackling plastic environmental issues as possible drivers of plastics avoidance. In detail, the interest was centered on consumers' decision to avoid purchasing single-use plastic eating utensils, food with multiple plastic packaging, and plastic water bottles while grocery shopping.
Overall, the results highlighted that all the latent constructs included in the analysis affected consumers' plastic avoidance, at least in the specific context of our analysis. In detail, plastic-related concern was found to directly drive plastic avoidance, which confirmed the first hypothesis of the study. This result is in line with past literature showing that environmental concern is a key behavioral determinant of many PEBs [29][30][31]. In this regard, our results contributed to extend current evidence by also considering health-related concern. In fact, consumers sometimes associate plastic materials with potential health risks. This is due to the possibility that plastics may release chemical contaminants in the environment or the product that they are in contact with [34][35][36]. Such concern is especially high when plastics are in direct contact with food [37]. Our findings suggested that both environmental and health related concerns should be accounted for when analyzing consumer plastic-related behaviors.
The results indicated that also plastic-related subjective knowledge (i.e., subjective knowledge about plastic materials and their recycling) was able to influence the extent to which consumers avoid purchasing plastic items while grocery shopping, although indirectly. Indeed, the observed relationship between subjective knowledge and plastic avoidance was fully mediated by plastic concern. In other words, high plastic-related subjective knowledge increased plastic concern (both environmentand health-related), which in turn increased the probability that consumers avoided plastic items. This finding is in line with previous research stressing that environmental knowledge does not directly increase PEBs, but contributes to drive sustainability-oriented behaviors by influencing other variables, including concern [17,18,21,46]. This approach has become increasingly accepted in the literature on PEBs [23,50] and was further confirmed by the evidence of this study. Another key finding of this paper regards the importance that consumers attributed to others' commitment for tackling plastic-related environmental issues (i.e., shared commitment). Similar to subjective knowledge, shared commitment did not directly drive plastic behaviors, but influenced the extent to which consumers avoid plastic food packaging, single-use eating utensils, and water bottles by increasing concern. This result remarks that studies on PEBs, including plastic avoidance, should not limit the analysis to the individual behaviors. As suggested in previous studies, the extent to which consumers decide to behave sustainably is strongly affected by what others do in the same regard [50][51][52][53]. If people perceive that the environmental cause is not shared by the other members of the society, they may feel like their efforts are useless, thus losing their motivation to engage in PEBs [51]. Our analysis highlighted that, with specific regard to plastics, consumers attach high importance not only to other citizens' behaviors, but also to what the firms and the politics do.
It must be acknowledged that our study suffered from some limitations. Firstly, the survey was based on stated data, and it is possible that participants over/underestimated their responses. Secondly, the data might have been affected by social-desirability bias. However, given that respondents were asked to autonomously fill the questionnaire, we expect this bias to be limited. The third limitation regards the distribution of the sociodemographic variables in our sample, especially with regard to age. Indeed, while the age class 25-34 was overrepresented in our sample, the age class of people older than 65 years old was remarkably underrepresented. To some extent, this might have affected the results. Indeed, young adults are known to be more sensitive to the environmental sustainability, while older consumers are typically less attentive towards this issue [69,70]. Furthermore, the distribution of age classes in our sample did not reflect the Milan population, which implies that the results cannot be generalized.
Although these limitations did not allow us to derive definitive conclusions and to provide robust policy implications, we believe that this study offers some useful hints for developing future consumer-based studies on plastics. Essentially, the results stressed that plastic-related PEBs are mainly and directly driven by environmental and health concern. Therefore, future studies should further investigate which variables can determine an increase/decrease in consumers' concern with regard to plastic issues. Moreover, the results seem to indicate that policy measures geared at increasing the extent to which consumers worry about the consequences of plastic contamination (e.g., by increasing plastic specific knowledge) might be effective in leveraging a sustainability-oriented behavioral change.
Our findings indicated that also plastic-related subjective knowledge and shared commitment play a role in this regard. However, their effect seems weaker and strictly related to other variables. Future studies should broaden the scope of our analysis in order to clarify how and to what extent these factors are associated with concern, eventually considering other individual characteristics and behavioral determinants.