*Article* **Influences of Green Eating Behaviors Underlying the Extended Theory of Planned Behavior: A Study of Market Segmentation and Purchase Intention**

**Sasichakorn Wongsaichia 1,2, Phaninee Naruetharadhol 1,2, Johannes Schrank 1,2, Premchai Phoomsom 1,2, Kanjakhon Sirisoonthonkul 1,2, Vorrapol Paiyasen 1,2, Sedthawut Srichaingwang 1,2 and Chavis Ketkaew 1,2,\***


**Abstract:** Green food has been introduced into the market for several years. Nevertheless, most Thai consumers do not commonly purchase green food in their daily routine. This research article aims to identify the market segments and significant factors affecting green food purchase intention in Thailand based on the theory of planned behavior. It employed a sample of 458 green food consumers in five regions of Thailand. Based on the Food-Related Lifestyle model, we used cluster analysis to classify the market segments. Additionally, we employed a multi-group structural equation modeling technique to explore and compare customers' behaviors in different segments. The results demonstrated two primary market segments for green food products, including (1) non-green consumers and (2) green consumers. The findings indicate that green consumers' self-realization related to environmental issues positively affects their attitude and purchase intention, while nongreen consumers reveal none of these relationships. Surprisingly, social norms related to green food consumption influence non-green consumers' attitudes toward green food more than it does toward green consumers. This research paper enlarges the understanding of Thailand's green food market regarding the market segments (non-green and green consumers). Furthermore, it points out implications on how marketing practitioners may penetrate those segments.

**Keywords:** green food; green labeling; green consumer; food-related lifestyle; food industry

### **1. Introduction**

Green products have played a significant role in the global environment showing that consumers are more aware of the negative impacts on the environment caused by global warming [1]. Green products as eco-friendly goods and the green production process uses eco-friendly technologies that are not disadvantageous to nature. Furthermore, characteristics-wise, a perfect green product should be organically grown, reusable, recyclable, biodegradable, non-toxic, non-animal testing, and use eco-friendly packaging [2]. A green product covers different aspects such as product functionalities, product design, product package, and product value [3]. In the food industry, green food products are produced in an eco-friendly way that does not harm the environment. The definition of green food involves a cleaner process starting from collecting resources, consuming, and decomposing the product [4]. Accordingly, the process of producing green food helps to prevent environmental pollution and enhances ecological advantages. Green food is harmless and includes nutritious food for consumer health. Green food should also be organic and nutritious for humans [5].

Environmental issues are the primary concerns of governments and citizens. Rahman and Reynolds [6] recommended that consumption of green products is strongly influenced

**Citation:** Wongsaichia, S.; Naruetharadhol, P.; Schrank, J.; Phoomsom, P.; Sirisoonthonkul, K.; Paiyasen, V.; Srichaingwang, S.; Ketkaew, C. Influences of Green Eating Behaviors Underlying the Extended Theory of Planned Behavior: A Study of Market Segmentation and Purchase Intention. *Sustainability* **2022**, *14*, 8050. https:// doi.org/10.3390/su14138050

Academic Editors: Riccardo Testa, Giuseppina Migliore, Giorgio Schifani and József Tóth

Received: 8 June 2022 Accepted: 30 June 2022 Published: 1 July 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

by consumer buying behavior. Research shows that consumers with environmental conservation awareness are rapidly increasing, and many consumers expect green products and services from producers globally [7]. Consumers' decisions to purchase the green product depend on a specific group's perception, values, behavior, and the individual's attitude [8]. Since 2017, the green consumer markets worldwide have generated approximately USD 290 billion annually. In addition, 14% of the gross world product (GWP) of a green product represents the eco-tourism market, increasing global environmental awareness [1]. This information shows that the number of green consumers has increased significantly, and the food industry should not ignore the green food market.

In Thailand, since January of 2020, there have been policies to reduce the use of plastic bags in convenience stores, supermarkets, and shopping malls. Jafarzadeh et al. [9] stated that 2020 would be the year of green, including green food, green packaging, and green organizing according to the environmentally friendly trend that has become the popular trend worldwide. Yanakittkul and Aungvaravong [10] also reported that 37% of Thais use only natural and organic products or green products daily. Nevertheless, the green market is new to Thailand, especially green foods. Therefore, a market segmentation study is a prerequisite for marketers to implement marketing strategies successfully. Segmenting customers allows marketers to understand the customers' behaviors deeply. This approach also allows marketers to tailor marketing strategies and deliver products and services in response to the segment's needs. However, few studies on green food and green consumers in Thailand are related to market segmentation and purchase intention. Thus, we considered exploring the consumers' buying behavior and categorizing consumers into segments to fulfill the knowledge gap.

In 2020, Thailand produced 25.37 million tons of waste [11]. Food packaging is one of the main contributors. Tangwanichagapong et al. [12] reported that the amount of all packaging materials has increased, and in particular plastic packaging, which has increased at a rapid rate in Thailand. Packaging waste comprised 22.5% of total municipal solid waste, and plastic was the major type of packaging found in the waste stream (15.8%), followed by glass (3.5%) and paper (3.2%). According to Sawasdee et al. [13], one of the biggest sources of solid waste from the food industry in Thailand is due to discarded non-degradable packaging; hence, green marketing focuses on creating more eco-friendly packaging. Yashasvini and Sundar [14] stated that eco-friendly packaging could reduce waste production and can minimize costs. Many resources are lost in the collection and degradation of plastic. However, eco-friendly packaging is naturally degradable, serves as a recyclable fuel, or is absent altogether. Therefore, it saves resources and investments. Thai authorities have been increasing their efforts to tackle the environmental problem, especially plastic waste and plastic packaging. They aim to reduce packaging and use bio-materials and green packaging instead. The Thailand Single-Use Plastic Reduction Roadmap aims to reduce 50% of packaging waste by 2025 and 55% by 2030. Plastic packaging reduction, increase in product recyclability, and the use of recycled material are the main environmental focuses. Green products are products that are produced in an environmentally friendly production process, while green packages are packages that are harmless to the environment [15]. In Thailand, most firms have expressed their social and environmental responsibility by offering green packaging. Hence, consumer products are becoming more available in recyclable and biodegradable packages. Fangmongkol and Gheewala [16] stated that biodegradable food containers from bagasse have a good environmental performance in Thailand.

Firms in Thailand acknowledged the need for recycling, waste reduction, and sustainable packaging. As an example, PTT Public Company Limited, the largest energy company in Thailand, strengthened its commitment to environmental friendliness by using compostable cups in coffee shops. CP ALL Public Company Limited, the largest operator of retail and wholesale businesses for consumer goods in Thailand, states that all plastic packaging of products under the company's control must be reusable, recyclable, or compostable by 2025. The company supports the use of materials from sustainably managed

renewable resources, such as paper material that is certified by the Forest Stewardship Council (FSC) or by the Program for the Endorsement of Forest Certification Scheme (PEFC). It aims to increase the amount of compostable packaging material, such as the usage of polybutylene succinate coated paper and the replacement of plastic with biodegradable material. It aims for recycled material and increases reusable packaging. ThaiBev Public Company Limited, Thailand's largest beverage company, aims to increase the amount of bio-based and bio-degradable materials used in plastic bottles and plastic bags.

This paper aims to explore the significant relationships among factors such as selfefficacy, environmental concern, utilitarian eating value, perceived price, attitude, perceived behavioral control, subjective norm, and purchase intention associated with consumer buying behavior and marketing segmentation of green foods. The variables are mainly derived from the theory of planned behavior [17]. It helps to predict consumer behavior by exploring subjective norms, perceived behavioral control, attitude, and purchase intention. We added four additional variables, which are utilitarian eating value, environmental concern, perceived price level, and self-efficacy. These variables can help to predict consumers' purchase intention. The cluster analysis and structural equation modeling (SEM) methods were used to explore the market segments and examine the significant relationships of these variables. These approaches are suitable for this research because we are trying to create a multi-factor model to predict the purchase behavior of a cross-sectional sample divided into multiple groups [18,19].

This paper is organized as follows: Section 2 begins with a review of the literature on the existing theories; Section 3 outlines the research methodology involving pilot test and cluster analysis, sample and data gathering, development of measures as well as data analysis and statistical measures; Section 4 presents the analyses and findings; Section 5 discusses the research implications for theory and practice; Section 6 shows the research limitations; Section 7 summarizes this research article by discussing major conclusions drawn from this study.

### **2. Literature Review**

Several research articles related to food marketing utilized employing the SEM approach. Most research results revealed relationships among consumer attitudes and marketing terms interpreted by consumer perceptions and behavioral intention [20–23]. Following those recent papers, we propose that the relationship of variables in this paper could be created using the SEM framework. The following sections provide details regarding theories and the related literature that helped create a structural model and hypotheses for this research.

### *2.1. Theory of Planned Behavior*

The theory of planned behavior (TPB) consists of attitude, subjective norm, and perceived behavior control [17]. The theory of planned behavior is universally used to predict consumer behavior [24]. Dowd and Burke [25] applied the theory of planned behavior in the previous study regarding food choice. Additionally, prior research adopted TPB to examine green food intention among Asian consumers [26]. Hence, TPB has become a successful theory for predicting and forecasting consumers' buying behavior globally. According to Qi and Ploeger [27], TPB is one widely used framework to explain consumers' food choices. Considerably, in the food industry, TPB could predict approximately 60% of consumer intentions and estimate 50% of fast-food predictions. Qi and Ploeger [28] found that the TPB is useful in predicting consumers' green food purchase intention. Wang and Wang [29] studied the theory of planned behavior to predict the green food and beverage behaviors in protecting the food environment and found that commitment, perceived behavioral control, and perceived knowledge are the most influential factors of green food and beverage.

Nevertheless, the present study enhanced its conceptual framework partly from the extended TPB. This theory suggests that a person's behavioral intention influences one's behavior. The behavioral intention construct is an indicator of one's willingness to perform a given action. Instead, the behavior construct is an individual's observed response in a given situation concerning a given target [30]. This paper assumes that green consumers (behavioral intentions) are more likely to consume green food products (behavior). Additionally, it is hard to measure actual behavior. Hence, the real behavior construct is omitted in this study following Ketkaew et al. [19], Nosi et al. [21], and Watanabe et al. [22].

### 2.1.1. Subjective Norm

A subjective norm refers to a particular behavior influenced by social forces such as their communities, associates, or close family members' friends. It can change an individual's behavior performance [17,23]. Hence, this study suggested that the subjective norm affects intention. Furthermore, various studies found that subjective norms are significantly related to attitude, perceived behavioral control, and purchase intention [31,32]. Perceived behavioral control has been indicated as a significant component of an individual's intentions to purchase green commodities [33]. We, therefore, developed H3, H5, and H8: the subjective norm has a positive influence on the perceived behavioral control, attitude, and purchase intention.

### 2.1.2. Attitude

Ajzen [17,23] revealed that perceived behavior control affects intention. Previous studies recommend that customers' attitudes toward environmentally friendly commodities play a mediating role in the connection between consumption value and purchase intention [7,34]. Empirical studies suggested that attitude and purchase intention are positively correlated [35,36]. Additionally, a prior study on eco-friendly products and green purchase behavior indicated that consumers' attitude positively affects green purchase intention [24]. Furthermore, a recent study revealed that there is a positive and significant relationship between a green attitude and purchasing behavior [37]. Hence, we set up H7, i.e., attitude has a positive influence on purchase intention.

### 2.1.3. Perceived Behavioral Control

Ajzen [17,23] claimed that perceived behavioral control affects intention. Several studies indicated that perceived behavioral control is an essential component of intention [38,39]. Therefore, the consumer's perceived behavioral control variable directly affects purchase intention. Higher perceived behavioral control leads to a higher possibility of purchasing green food products [32], and this relationship is mediated by attitude [40]. Additionally, the previous studies recommended that perceived behavioral control is associated with individual factors such as skill, time, money, and resources [41,42]. Thus, we developed H6 and H9: perceived behavioral control has a positive impact on attitude and purchase intention.

#### 2.1.4. Purchase Intention

Purchase intention refers to consumers' readiness to purchase sustainable products [43]. Prior research said that consumers prefer to perform a behavior to engage when they have a more accepting attitude towards purchase intention [17,23]. Purchase intention represents the possibility of engaging an individual's behavior, which can be affected by perceived behavioral control, subjective norms, and attitude. The positive attitude toward green foods, subjective norms, perceived behavioral control, and self-efficacy are used to predict the possibility of purchasing green foods [44]. Ahmed et al. [45] found that attitude, subjective norms, and perceived behavioral control have positive effects on the purchase intention of young consumers of organic food. Moreover, Liu et al. [46] discovered that attitude plays the most important role in predicting consumers' green purchase intentions.

### *2.2. Utilitarian Eating Value*

Hoffman and Novak [47] defined utilitarian eating value as overall values judgment of functional advantages. Consumers with utilitarian motivation concentrate primarily on instrumental value. Therefore, people with the utilitarian component are generally considered goal-oriented people. Additionally, a previous study suggested that utilitarian value positively influenced subjective norms [48], whereas Ajzen [49] found that attitude and subjective norms affect the intention to execute a specific behavior. Hence, utilitarian value is related to an individual's decision-making process of consuming green food before the actual purchase [50,51]. This leads to H1: utilitarian eating value has a positive influence on the subjective norm.

### *2.3. Perceived Price*

Zagata [52] recommended that perceived price is associated with the construct of perceived behavioral control. In contrast, Al-Swidi et al. [53] suggested that perceived price relates to the construct of attitude. Consumers' perception that organic food is expensive has a positive effect on purchase intention [54]. The higher price of products represents a higher quality and functional benefit [55]. Thus, premium prices expand organic products' attractiveness and influence one's perceived behavioral control. We established H2: The perceived price level has a positive impact on the perceived behavioral control.

### *2.4. Environmental Concerns*

Environmental concern is a significant factor in encouraging the consumer to improve their consumption behavior to be environmentally friendly [28]. Green consumers are aware of using and consuming natural resources, which are limited resources. Hence, environmental concerns can cause green consumption behavior. Prior studies found that environmental concern positively influences the attitude towards green food consumption [56]. Environmental concern is a principal motivational element for purchasing any merchandise, including organic commodities [57]. Moon et al. [58] extended the theory of planned behavior by adding beliefs about the outcome of ecological actions and found that the perceived seriousness of environmental problems is one of the most influential determinants of green purchase intentions. Thus, H4 states that environmental concern has a positive effect on attitude.

### *2.5. Self-Efficacy*

Self-efficacy refers to people's judgments of their competence to arrange and conduct courses of action needed to accomplish designated categories of performances [59]. In commercial terms, self-efficacy implies an individual's evaluation of products [60]. Selfefficacy is based on past behavior or experience [61]. It can identify factors of behavioral intention and can be influenced by demographics, personality traits, and attitudes toward surrounding aspects [17,23]. Theoretically, self-efficacy is a powerful factor that predicts and encourages decision-making of purchase intention [62]. Hence, self-efficacy directly influences purchase intention. Therefore, we created H10: self-efficacy has a positive impact on purchase intention.

Based on the literature review, this research established ten hypotheses and proposed the following conceptual framework. We also proposed that the market segments of green food play a moderation effect in this structural model. The model examined the relationships among factors such as utilitarian eating value, environmental concern, perceived price, subjective norm, perceived behavioral control, self-efficacy, attitude, and purchase intention linked with consumer buying behavior and marketing segmentation of green foods based on the theory of planned behavior. A solid blue line shows the effect of one factor on another factor. A dashed orange line shows the effect of market segments on a factor.

### **3. Research Methodology**

#### *3.1. Pilot Test and Cluster Analysis*

The first purpose of the study concerns the market segmentation for green food products. Thus, we performed the pilot study by collecting data from consumers in five regions in Thailand. It was suggested to have a minimum sample size of 20–30 for the pilot study [63]. We decided to collect data from 60 respondents. More specifically, by employing the food-related lifestyle (FRL) instrument, this study shows consumer groups' existence sharing typical food lifestyles, preferences, and purchases of green food production. The FRL dimensions were established from the 69 statements containing 23 scales with three items each [64]. All items are related on a 7-point Likert-type scale. Ward's hierarchical clustering approach was performed to segment consumers into two groups. Moreover, a *t*-test was executed to identify whether any differences existed between the means of variables that belong to each cluster to determine the number of clusters. Each cluster was named based on the characteristics of the descriptive statistics for each cluster.

#### *3.2. Sampling, Data Collection, and Development of Measures*

Data collection in this research was based on quota sampling. The data were gathered from 500 respondents. The data were collected in front of supermarkets in five regions (north, northeast, central, east, and south), which can be used as a representation of Thailand. There were 100 respondents per region. The cities in the five regions are Chiang Mai in the north, Khon Kaen in the northeast, Bangkok in the center, Pattaya in the east, and Phuket City in the south. The selected supermarkets sell both green and non-green food. The data collection was conducted during the COVID-19 pandemic, but all required health standards were met, including distancing, mask-wearing, hand washing, body temperature screening, etc. It was suggested to have a minimum sample size of 200 for any SEM analysis [65]. In this study, data from 500 respondents were collected. After removing irrelevant data, outliers, and errors, 458 responses were usable. Hence, the rate of invalid samples was 8.4%. The data remained confidential. The data were gathered based on a structured questionnaire. Questionnaires had introductory questions, demographic questions, and questions regarding customer attitudes towards green eating behavior (see more details in Appendix A).

The demographic profiles reveal that most of the participants were female (78.2%); 48.5% of the respondents were aged below 21 years. People aged between 22 and 38 years old were 48% of the total, 2% were aged between 39 and 53 years of age, and the smallest group was those over 54, which accounts for 1.5%. With respect to income, 81.4% have an income less than THB 15,000, 13.8% make THB 15,001–20,000, 1.5% make THB 20,001–25,000, 1.1% earn THB 25,001–30,000, and 2.2% have an income more than THB 30,000. For taste experience, 85.2% have consumed green food, and 14.8% have never purchased green food before; 94.1% will consume green food products in the future, and 5.9% will not buy green food production.

In order to assess customer attitudes toward green eating behavior, the research methods used were data collection via a survey using questionnaires and data analysis using quantitative methods. Leung [66] stated that quantitative research is accomplished according to primary numerical data and statistical interpretations under a reductionist, logical, and rigidly objective paradigm. Hence, this study used a questionnaire to identify the main factors that affect green food purchase intention. Bell and Bryman [67] argued that quantitative research involves the collection of numerical data and presentation of the relationship between theory and research as deductive. In this paper, a survey was used to perform data collection of customers in Thailand.

The collected data were information from Thai customers based on a questionnaire survey that was conducted in front of supermarkets in five regions of Thailand. The survey questionnaire followed eight identified factors that affect green food purchasing, i.e., self-efficacy, environmental concern, utilitarian eating value, perceived price, attitude, perceived behavioral control, subjective norm, and purchase intention. These eight indicators were assessed by a total of 34 questions. The questionnaire was divided into three sections. The first section comprised introductory questions that identify regular and potential buyers of green food. The second section consisted of demographic profile questions in the form of multiple-choice questions, including gender, age, income, and family size. The demographic profiles were also used as a nominal variable to classify the scale. In the third section, the survey provided a linear scale of the eight indicators to allow individual participants to assess their views. The linear scale was composed of seven levels of agreement (1 = strongly disagree, 2 = disagree, 3 = somewhat disagree, 4 = neutral, 5 = somewhat agree, 6 = agree, and 7 = strongly agree). The second factor (environmental concern) targets the consumers' behavior towards green packaging and the reduction in plastics. Consumers with a large environmental concern are likely to aim for green food and green packaging.

### *3.3. Data Analysis and Statistical Measures*

Before scrutinizing the data, we addressed common method variance (CMV) in this study. CMV occurs when variables in the same model are estimated employing the same technique or derived from the same source. The findings have systematic error variances among those variables and might have biased the assessed relationships [68]. This study gathered the data, including dependent and independent variables from the same respondents, thus exhibiting a CMV risk. We applied Harman's single factor test following Podsakoff et al. [68]. The results disclosed the cumulative variance of 49.835 percent (less than the 50% threshold), which further assured the absence of CMV.

The study's data analysis used the structural equation modeling (SEM) method. SEM encompasses such diverse statistical techniques as path analysis, confirmatory factor analysis (CFA), and causal modeling with latent variables. SEM was executed to estimate the model's estimation in two steps [69]. Step 1 validates the CFA model to measure each indicator's relationship and its variable, whether valid and reliable. This step requires appraising the goodness of fit (GOF), convergent validity, and discriminant validity. As for the GOF and convergent validity conditions, the designated thresholds included CMIN/df < 3.00, CFI > 0.90, RMSEA < 0.10, AVE > 0.50, and CR > 0.70 [17]. As for the discriminant validity condition, this paper studied issues of multicollinearity and the identity matrix of the indicator variables. The study used Pearson's moment correlations with the threshold <0.80 to check multicollinearity [70]. The Kaiser–Mayer–Olkin (KMO) and Bartlett's sphericity tests were employed to check an identity matrix [71]. These criteria were all satisfied. Step 2 evaluates the structural model to measure whether the entire structure is reliable, including the estimation of GOF. The designated fit indices thresholds were CMIN/df < 3.00, CFI > 0.90, and RMSEA < 0.10. In step 3, to examine the segment's moderating effect on the structural relationship, we conducted a multi-group moderation analysis [72]. This step performs a measurement invariance (MI) analysis utilizing the segment as a moderator dividing the sample into two groups (non-green consumer and green consumer) and then performing a z-test for the difference between the two groups' factor loadings. A z-test was used for the multi-group analysis in SEM [20,73,74]. The results of the statistical analysis are discussed in the next section.

### **4. Result of the Study**

#### *4.1. Pilot Study and Market Segmentation*

The *t*-test result showed questions that were significant at <0.05. The target was classified into segments by analyzing segments of FRL questions and assessing the segments through *t*-tests. The findings revealed that there are two segments, including (1) non-green consumers and (2) green consumers. These names follow the characteristics of each cluster inferred from the descriptive characteristics. The test results demonstrated significant differences between the means of the FRL scores of the segment 1 and 2 consumers, with all the *p*-values below 0.01. The overall means of segment 1 ranged from 2.65 to 3.65, and segment 2 ranged from 4.14 to 5.75. Segment 1 comprised those who do not care about reading food labels. Their decision to purchase food depends on their preferences; they are pleased with inexpensive food without regard to its nutritional value or environmental friendliness. People in segment 2 are typically concerned with food label information and base their food consumption decisions on criteria such as price, food nutrition, and environmentally friendly or "green" food products.

According to Table 1, there was a sample size of 100 in segment 1. The sample consists of 41 males and 105 females. Most of the participants were age below 21 years old and were college students, and had earned less than THB 15,000 per month. In segment 2, there was a sample size of 358. The composition of sample size consists of 59 males and 253 females. Most of the participants were aged between 22 and 38 years old and were college students, earning less than THB 15,000 per month.


**Table 1.** Descriptive statistics for demographic profile.

Source. Data adapted from authors (2022).

There are two primary steps to perform a statistical test on structural equation modeling (SEM): measurement and structural models [69].

### *4.2. Measurement Model*

The measurement model was examined using CFA. The model was determined for internal consistency, reliability, convergent validity, and discriminant validity in this context. All constructs were connected with covariances to perform CFA [17]. The indicator must involve each construct before testing. In order to enhance the goodness of fit (GOF) relationship, we allowed covariances among errors within the same construct.

### 4.2.1. The Goodness of Fit (GOF)

Table 2 illustrates the GOF measures and their thresholds. The results were acceptable in that all the measures passed the required threshold. CMIN/df (2.649), Tucker–Lewis index (TLI; 0.944), comparative fit index (CFI; 0.951), incremental fit index (IFI; 0.951), and root mean square error of approximation (RMSEA; 0.060) passed the designated threshold.


**Table 2.** The Goodness of Fit of Measurement Model.

Source. Data adapted from authors (2022). Note. TLI = Tucker–Lewis index; CFI = comparative fit index; IFI = incremental fit index; RMSEA = root mean square error approximation.

#### 4.2.2. Convergent Validity

Convergent validity was estimated by comparing the model results with the fit index threshold. The average variance extracted (AVE) [75] and composite reliability (CR) [17] were determined. The thresholds for AVE and CR are 0.50 and 0.70, respectively. Table 3 shows the suggested thresholds of the convergent validity measures, and the calculated indicators are as follows.

**Table 3.** Convergent validity.


Source. Data adapted from authors (2022). Note. AVE = average variance extracted; CR = composite validity. \*\*\* significant at <0.001.

Table 3 shows the SE (self-efficacy), EC (environmental concern), UT (utilitarian eating value), PP (perceived price), AT (attitude), PC (perceived behavioral control), SN (subjective norms), and PI (purchase intention) constructs nicely passed the convergent validity criteria when comparing the calculated measures with their thresholds.

### 4.2.3. Discriminant Validity

Discriminant validity is the level to which two or more theoretically similar constructs are different. This is assessed by comparing the square root AVEs (on diagonal) with the correlations of the related matrices [74]. According to Table 4, each AVE's square root was higher than the off-diagonal correlation coefficients, recommending that all constructs could theoretically measure the different constructs, and this result was acceptable.


**Table 4.** Discriminant validity.

Source. Data adapted from authors (2022).

### *4.3. Primary Structural Model*

After examining the measurement model, we connected all the constructs to develop the structural model according to the purpose model in Figure 1. Furthermore, we studied the variables via the main objective structural model. The results of most of the goodness of fit (GOF) criteria show how constructs support each other. All GOF indices were satisfied with the thresholds of [76] (see Table 5).

**Figure 1.** Proposed Model. Source: Figure created by authors (2022).


**Table 5.** The GOF of the Structural Model (SEM).

Source. Data adapted from authors (2022).

According to Table 6 and Figure 2, the structural model's test results supported H1 to H8 and H10 at the significant level of 0.001 or less, whereas H9 was not supported. The relationships among the constructs were highly significant in statistics. The researchers established the analysis by considering the following constructs: utilitarian eating value, perceived price, subjective norm, environmental concern, perceived behavioral control, attitude, self-efficacy, and purchase intention adapted to the theory of planned behavior [17,23]. H1 was supported first, indicating that the utilitarian eating value had positive influences on subjective norms with a standardized factor loading of 0.695.

**Figure 2.** The structural model. Source. Figure created by authors (2022).


**Table 6.** Test results from the structural model.

Source. Data from this study (2022). Note: \*\*\* Significant at <0.001, \*\* Significant at <0.01.

H2 was supported, which predicts that perceived prices had a significant effect on perceived behavioral control with a standardized factor loading of 0.567.

H3 was also supported, recommending that subjective norms directly affected perceived behavioral control with a standardized factor loading of 0.383. H4 predicted that environmental concern significantly influences consumers' attitudes toward green food products; it was also supported (standardized estimate = 0.192). This study's findings recommended that Thai consumers are aware of environmental defense issues and obtain their responsibilities towards environmental defense. Therefore, customers with pro-environmental behavior have a positive attitude towards green food production.

H5 was also supported, implying that the subjective norm directly affects consumers' attitudes toward green food consumption with a standardized factor loading of 0.568. Further, H6 was supported, which suggested that perceived behavioral control positively impacts consumers' attitudes toward green food products with a standardized factor loading of 0.268. H7, regarding the positive impact of consumers' attitudes on their purchase intention for green food consumption, was supported (standardized estimate = 0.516). H8 was also supported and confirmed that subjective norms significantly influenced green food purchase intention with a standardized factor loading of 0.373.

H9 was rejected, which stated that perceived behavioral control is not influenced by purchase intention in consumer buying behavior in the green food industry. Finally, H10 was supported, claiming that self-efficacy positively affects purchase intention for green food production with a standardized factor loading of 0.107.

#### *4.4. Multigroup Moderation Analysis (MGA)*

### 4.4.1. Measurement Invariance

Measurement variance (MI) is a method to demonstrate the difference between two groups of the measurement model, whether different or not [72]. Multigroup analysis helps us understand the constructs of questionnaires in the same way by assessing the responses between two groups (non-green consumers and green consumers). According to the measurement model (CFA), the multi-group analysis reveals the following: (1) configural invariance (unconstrained model), (2) metric invariance (equal factor loading), and (3) scalar invariance (equal intercept). If only configural invariance and metric invariance are satisfied, then partial MI is supported, allowing one to compare factor loadings between two groups. Nevertheless, if partial MI detains and scalar invariance is accepted, then full MI is formed, which lets us compare factor loadings between them. Table 7 exhibits the assessment of MI successively performed after the CFA model.


**Table 7.** Measurement Invariance.

Source. Data from this study (2022).

According to Table 7, the CMIN/*df* values of configural invariance, metric invariance, and scalar invariance passe the threshold of <3.00. Other fit indices such as TLI, CFI, IFI, and RMSEA of configural invariance, metric invariance, and scalar invariance are considered also pass the thresholds of >0.90, >0.90, >0.90, and <0.10, respectively. Therefore, full MI was established, allowing us to conduct further analysis in the next section.

### 4.4.2. Z-Test for Loading Differences

We next used critical ratio difference to gather z-test results by comparing factors loading between two segments (1. non-green consumers and 2. green consumers) from structural models [19]. In the multi-group analysis, we used the pairwise parameter comparison to estimate each parameter's critical ratios' difference to test differences in statistically significant. The factor loadings are significantly different between two segments (1. non-green consumers and 2. green consumers) when the value of the critical ratio is more than the threshold of 1.96. The paths of H1, H2, H3, H5, H6, H7, and H8 were statistically significant for non-green consumers. The paths of H1, H2, H3, H4, H5, H6, H7, and H8 were statistically significant for green consumers.

Table 8 and Figure 3 demonstrate that the paths of H1, H2, H3, H4, H5, H6, H7, and H8 for both segments are statistically significant (see the stars), which is in line with the results shown in Table 6. The paths of H9 and H10 for both segments were not statistically significant; they are also consistent with the findings in Table 6. However, only two path differences exist in H4 and H5 (see the stars under the critical ratio difference column).

**Table 8.** Test result from loading differences. (N = 458, Non-green Consumers = 146, Green Consumers = 312).


Source. Data adapted from authors (2022). Note: \*\*\* *p* value < 0.001, \*\* *p* value < 0.01, \* *p* value < 0.05.

The critical ratio value of H4 is slightly greater than the threshold, suggesting that segment 1 and segment 2 have different perspectives on environmental concerns and green attitudes. This result is consistent with the existing literature [28,56]. Additionally, environmental concerns do not affect (loading = 0.046 and insignificant) attitudes toward green labeling products for the non-green consumer segment because they do not think that

environmental issues are caused by human consumption. Some of them are unnoticeable environmental issues. Thus, a non-green consumer who is unaware of an environmental problem will not have a good attitude toward green food. In contrast, segment two (green consumers) weigh environmental concerns as particularly important and are willing to improve their consumption actions. They attempt to find the resolution of environmental issues. Hence, a green consumer deeply concerned about the environment will have a positive attitude toward green labeling.

**Figure 3.** Moderation Effects and Structural Model. Source. Figure created by authors (2022).

H5 demonstrates a high level of critical ratio at |−3.643|, which means that the paths of segments 1 and 2 are significantly different. According to the standardized loadings, non-green consumers' subjective norms impact their attitudes more than green consumers. Non-green consumers are more likely to consume any food regardless of environmental attitude toward green food production because their consumption choice is influenced mainly by friends. However, green consumers constantly consume green labeling as usual. This is sometimes due to their environmental awareness—communities can impact dietary choices. This finding is in line with the existing literature [32].

### **5. Discussion**

We found that the utilitarian eating value had positive influences on the subjective norm, consistent with previous studies regarding green food production [48]. The results imply consumers prefer functional attributes of green food products concerned with environmental friendliness and would like to receive social acceptance when making food decisions before purchasing. Customers feel more pressure from other peers to purchase green food products. They may become more engaged in purchasing green food products [51]. We found support that perceived prices had a significant effect on perceived behavioral control. This confirms previous research findings on the positive influence of price on perceived behavioral control [5,52]. The premium price increases the perceived behavioral control and purchase intention for green food. Consumers believe that the higher price of green products represents a higher quality and functional benefit [55]. We discovered that subjective norms directly affected perceived behavioral control. This result suggested that perceived social pressure from others can form a consumer to act an eco-friendly behavior. This behavior relates to an environmentally friendly lifestyle in their consumption pattern of green food products. We found that environmental concerns significantly influence consumers' attitudes toward green food products. The findings are consistent with previous studies by Zhu et al. [56]. We identified that the subjective norm directly affects consumers' attitudes toward green food consumption. This finding revealed that others' perceived social pressure could establish an individual's attitude toward eco-friendly food consumption. Furthermore, we found that perceived behavioral control positively impacts consumers' attitudes toward green food products. This finding implied that the perceived behavioral control in eco-friendly lifestyle increased and attitude toward green food products became more positive [40]. Additionally, we detected the positive impact of consumers' attitudes on their purchase intention for green food consumption. Thus, buyers with a positive attitude toward eco-friendly food packaging are more willing to purchase those products [24]. We found evidence that subjective norms significantly influenced green food purchase intention. The results indicated that subjective norm emerged as the strongest among the other significant factors of the purchase intention of eco-friendly packaged products. This reflects that Thai consumers received peer pressure from others about the environmental protection issue. Thus, consumers desire social acceptance and moral responsibility towards the environment, which influences their food purchasing choices. We discovered that perceived behavioral control is not influenced by purchase intention in consumer buying behavior in the green food industry. It contradicts the theory of planned behavior hypothesis proposed by Ajzen [17,23], which implied that purchase intention was not dependent on the consumer's perceived behavioral control. We identified that self-efficacy positively affects purchase intention for green food production. This result revealed that self-efficacy in green food consumption might encourage decision-making of purchase intention in green food production.

### **6. Research Implications**

The following suggestions were presented to three main stakeholders: producers, consumers, and policymakers. Green labeling is an essential tool to disclose goods and services' environmental and social performance from a green-producer viewpoint. Thus, a producer can use green labeling as a benchmark for the enhancement and competitiveness of their products. Green food manufacturing companies and green packaging producers should create green labeling merchandise because it can influence the purchase decisions of consumers who are genuinely concerned about environmental issues. Marketers and research and development (R&D) teams would directly benefit from this research by receiving and understanding consumers' buying behavior and consumer types in the potential market for green food products and green packaging.

Moreover, this research can help consumers understand more about green food and green labeling because green labeling offers consumers information regarding the green components of the products. This information is a form of increased quality evaluation of goods and services. The green consumer can use green labeling as an essential factor in making a purchasing decision. With the help of green labeling, consumers can target to purchase green food and green packages.

Furthermore, this research can be a practical tool for estimating and improving Thailand's sustainable foods and packages production from a policymaker's perspective. Policymakers must be aware of the importance of green labeling and other green food/package production by using green labeling as a complementary tool to generate food producers' motivations to produce eco-friendly food products and green packages. Moreover, Thai policymakers should create an environmental awareness campaign to inform Thai consumers of the benefits of consuming green labeling food. These policies can encourage them to produce and purchase more green foods and packages.

In addition to the described practical implications, there are theoretical implications. This research revealed relationships among consumer attitudes and marketing terms interpreted by consumer perceptions and behavioral intention. The utilitarian eating value has a positive effect on subjective norms. Perceived prices have a significant effect on perceived behavioral control. Subjective norms have a positive impact on perceived behavioral control. Environmental concerns, the subjective norm, and the perceived behavioral control significantly affect consumers' attitudes towards green food. Consumers' attitudes, the subjective norm, and self-efficacy have a positive impact on the purchase intention for green food consumption. These research findings provide evidence for the theory of planned behavior. The subjective norm and attitude of consumers can be used to predict consumer behavior towards the purchase intention of green food.

### **7. Limitations of the Study**

Thus, this study's information is insufficient to support the generalized market because we only focused on green food products, which are particular in the market compared with general food. Future research may apply other antecedent variables to the current structural model, such as hedonic eating value, to understand consumers' experiences. Moreover, it may change the consumer segment's moderator to a more varied segment such as age.

### **8. Conclusions**

Environmental awareness and consumer behavior have changed dramatically in recent years. Consumers have raised their environmental awareness and adjusted their consumption behavior to reduce overall environmental impacts by using more eco-friendly products and services. Thus, we analyzed market segmentation by collecting data based on the food-related lifestyle criteria and performing cluster analysis. Consequently, we found consumers were divided into (1) non-green consumers and (2) green consumers. Moreover, this article aimed to examine the significant relationships among factors such as self-efficacy, environmental concern, utilitarian eating value, perceived price, attitude, perceived behavioral control, subjective norm, and purchase intention associated with consumer buying behavior and marketing segmentation of green foods. The hypothesized relationship was analyzed using a structural equation modeling (SEM) technique. This study formed ten hypotheses, as previously explained. We performed quantitative research based on structured questionnaires with 458 valid respondents consuming green food in Thailand. Most of the hypothesis test results supported the previously formed hypotheses except for H9, which concluded that perceived behavioral control was not related to their purchase intention of green food. Additionally, the multi-group analysis suggested that green consumers make their purchase decision of green foods based on their perception of environmental issues, whereas non-green consumers demonstrate no effects.

**Author Contributions:** Conceptualization, C.K.; methodology, P.N.; formal analysis, S.W.; resources, P.N.; data curation, P.P., K.S., V.P. and S.S.; writing—original draft preparation, S.W.; writing—review and editing, J.S.; supervision, C.K. and P.N.; project administration, C.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This project is primarily funded by Research and Graduate Studies, Khon Kaen University, Thailand (Ref. RP65).

**Institutional Review Board Statement:** Khon Kaen University Ethics Committee for Human Research, Khon Kaen University, Khon Kaen, Thailand, has made an agreement that this study has met the criteria of the Exemption Determination Regulations on 11 November 2021 (HE643227).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** Not available.

**Acknowledgments:** We would like to thank the International College and the Center for Sustainable Innovation and Society, Khon Kaen University, Thailand, for providing research facilities.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **Appendix A. Questionnaire**

*Appendix A.1. Introductory Questions*


*Appendix A.2. Demographic Data of Respondents*


#### *Appendix A.3. Customer Attitudes*

	- 7.1. Self—Efficacy [59,62] Do you trust farmers to grow a green plant for green food? Do you trust the procession of a producer to produce green food? Do you trust the government to manage green food policies? Do you trust the green food certificate from the certificate authority? Do you strongly trust green food?
	- 7.2. Environmental concern [28] Do environmental issues impact your purchasing decision on green food? Does your knowledge of environmental issues impact your purchasing decision on green food? Does your realization of environmental issues impact your purchasing decision on green food? Does the threat of environmental issues impact your purchasing decision on green food? Do the government policies about environmental issues affect your responsibility to the environment?
	- 7.3. Utilitarian eating value [50,51] Is the price of green food reasonable? Do you rather consume only food that you had before and you know it is tasty? Does the food portion of green food can supply your hunger (per meal)? Do you like a variety of food recipes? Do you like a variety of green food recipes?
	- 7.4. Perceived price [52,53] Is the price of green food expensive? Is the price of green food reasonable? Is green food more expensive than normal food? Is the price of green food higher than you expected?
	- 7.5. Attitude [17,23] Does buying green food benefit your purchasing decision? Do you buy green food for your safeness? Do you demand to buy green food? Do you buy green food for a better quality of life? Are you interested to buy green food?
	- 7.6. Perceived behavioral control [17,23] Does it depend on your decision whether you buy green food or not? Do you believe that you could buy green food whenever you want? Do you have enough money, time, and a chance to buy green food?

### **References**


**Qiuqin Zheng 1, Xiaoting Wen 1, Xintian Xiu 2, Xiaoke Yang <sup>3</sup> and Qiuhua Chen 1,\***


**\*** Correspondence: 000q091001@fafu.edu.cn

**Abstract:** Chemical pesticides are a serious impediment to agricultural sustainability. A large-scale reduction in their use to secure food supplies requires more innovative and flexible production systems. Pesticide-free production standards bring together the strengths of all participants in the food value chain and could be the catalyst for this transition. Using a choice experiment approach and green tea as an example, this study investigated consumers' preferences for organic and pesticide-free labels. According to the findings, organic and pesticide-free labels and brands are all major factors that affect consumers' purchase decisions. Consumers are more willing to pay for organic labels than pesticide-free labels. There is a substitution effect between organic labels and pesticide-free labels. Complementary effects exist between organic labels and national brands, pesticide-free labels, and national brands. Consumer trust has an impact on consumers' choice of organic labels and pesticidefree labels. The use of pesticide-free labels is an alternate approach for small- and medium-sized businesses in a specific market to lower the cost of organic certification.

**Keywords:** organic labels; pesticide-free; choice experiment method; willingness to pay; green tea; consumer preference; eco-label

### **1. Introduction**

A critical attribute of food safety is pesticide residue [1]. Using chemical pesticides can significantly increase food production and improve agricultural efficiency, but it also causes damage to the natural ecological environment and the quality and safety of agricultural products [2]. The overuse of chemical pesticides can lead to the rapid growth of resistance in target pests, as well as serious impacts on non-target organisms, for example, endocrine disorders in rats, birds, and fish [3]. Pesticide residues can spread throughout the environment, contaminating different ecosystems and damaging food and water resources. Examples include high nitrate levels in groundwater, reduced soil fertility, and increased greenhouse gas emissions [4]. Chemical pesticides are considered to be one of the most prominent barriers to agricultural sustainability [3]. Pesticide risk reduction is at the top of many countries' policy agendas, but most have failed to meet their targets [5]. Existing policies often fail to promote widespread adoption of pesticide-free production practices due to the lack of cost-effective alternatives [6]. The vigorous development of organic agriculture is one approach to addressing the problem of agricultural products quality and safety [7]. Organic certification requires attributes such as no chemically synthesized fertilizers, pesticides, growth regulators, or other substances, and no pesticide residue, growth hormones, or genetic engineering [8].

In China, the organic food market is rapidly expanding and has reached a considerable size. Nonetheless, the share of available organic food remains small [9]. According to the Global Organic Agriculture Statistical Yearbook 2020, global sales of organic food and drinks exceeded EUR 95 billion in 2018. Of this, China's organic food sales were EUR

**Citation:** Zheng, Q.; Wen, X.; Xiu, X.; Yang, X.; Chen, Q. Can the Part Replace the Whole? A Choice Experiment on Organic and Pesticide-Free Labels. *Foods* **2022**, *11*, 2564. https://doi.org/10.3390/ foods11172564

Academic Editors: Riccardo Testa, Giuseppina Migliore, Giorgio Schifani and József Tóth

Received: 8 July 2022 Accepted: 20 August 2022 Published: 24 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

<sup>1</sup> College of Economics and Management, Fujian Agriculture and Forestry University, Fuzhou 350002, China

8.1 billion, accounting for only 8.3% [10]. Organic farming production in China necessitates a 3-year conversion period and increased labor expenditure [11]. Despite the potential premium, organic agricultural products incur higher production costs than conventional agricultural products and require significant investment, which many Chinese small- and medium-sized businesses (SMEs) cannot afford [12]. For consumers, the high cost of organic production leads to higher prices for organic agricultural products, which has hindered many consumers from buying [13].

Large-scale reductions in pesticide use in the context of unfavorable food production require more innovative and flexible systems to complement organic farming [14]. Pesticide-free production standards, which combine the strengths of all food value chain players, may be the cornerstone of this shift [15]. In Switzerland, the IP-SUISSE producer organization is introducing a nonorganic, private–public standard for pesticide-free wheat production [15]. Studies have demonstrated that the pesticide-free attribute is the most important aspect of consumer interest when purchasing organic produce [16,17]. The study by Britwum, et al. [18] on consumers' perceptions regarding the desired attributes of organic produce found that consumers place the highest importance and confidence in the "free of growth hormones" and "free of synthetic pesticides" aspects of organic labeling. For Chinese consumers, purchasing organic agricultural products is motivated more by concerns regarding food safety and personal health and less by environmental protection [19,20]. Generally, institutional pesticide-free certification is less difficult and less costly to achieve than certified organic labeling. Do consumers prefer separate pesticide-free information? If consumers are willing to pay for separate pesticide-free information, SMEs can use such certification without assuming the prohibitive expenditure of converting to organic operations. For SMEs, pesticide-free information could offer a strategic alternative to give farmers a competitive advantage. Consumers will then be able to buy healthy and safe products at a lower cost. Hence, investigating consumers' preferences and willingness to pay (WTP) for organic labels and pesticide-free information will directly affect agricultural certification decision-making.

A series of studies have been conducted on consumers' preference, WTP, and the influencing factors of organic labels [13,21–24]. Regarding how consumers perceive pesticidefree attributes, scholars believe that previous research has not been systematic and in-depth enough [18,25]. Bernard and Bernard [26] examined the WTP for two core attributes of organic labeling (pesticide-free and non-GMO), finding that consumers were willing to pay for the pesticide-free information. By contrast, Edenbrandt [25] surveyed Danish consumers and found that pesticide-free information was less important to consumers than the organic label, indicating that Danish consumers preferred to buy organic produce. These contradictory findings warrant further investigation.

Tea is one of the three most recognized drinks worldwide. China is the largest teaproducing country and a major tea-consuming and exporting country in the world [27]. Green tea production accounted for 61.70% of the total tea production in 2020. The export volume of green tea is 293,400 tons, accounting for 84.1% of China's total tea exports [28]. With consumers' increasing concerns regarding the quality of life and the rising threshold of international trade in tea, the production of organic green tea represents an important approach for enhancing the competitiveness of green tea in China, promoting green tea export, and expanding domestic demand for green tea. Existing literature focuses on the organic consumption behavior of milk [29–31], rice [32], and other crops [33,34], but there are fewer studies on the organic consumption behavior of tea [35]. Thus, green tea was chosen as the experimental subject in this study.

The choice experiment (CE) method can estimate consumers' preferences for different product attributes and assess the relationships between attributes. It avoids the limitations of the contingent valuation method that can only measure a single attribute of a product [35]. Based on the above background, taking green tea as an example, this study applies the CE method to analyze the following questions: (1) Under current conditions, do Chinese consumers have a preference and WTP for organic and pesticide-free labels? (2) Are

pesticide-free labels valid in comparison with organic labels? (3) What are the factors that influence consumers' WTP for organic and pesticide-free labels? This study can provide valuable information for market expansion and marketing of organic agricultural products and also reduces the degree of information asymmetry between SMEs and consumers, providing a reference for SME producers to control production costs.

### **2. Materials and Methods**

#### *2.1. Attribute Selection*

The CE method is widely used to measure product preferences and is an excellent approach for estimating multiple attributes. Attribute selection is the basis for determining the validity and precision of the results [36]. Previous studies have shown that food safety attributes and brands are crucial to consumers' preferences in green tea [34]. This study assumes that green tea is a collection of organic labels, pesticide-free labels, brands, and prices. Table 1 presents the attributes and levels.

**Table 1.** Green tea attributes and respective levels.


Organic labels are widely evident in the real market. Tea companies use organic logos in product packaging to distinguish products from conventionally produced teas. There are currently no certified pesticide-free labels in tea packaging, and only some e-commerce tea companies present reports confirming pesticide-free status on product details pages. To highlight the pesticide-free characteristic and facilitate respondents' understanding, this study used a simplified logo to represent pesticide-free status, referring to Grebitus, et al. [37]. The pesticide-free label used in this study refers to the green tea are grown without chemical pesticides, herbicides, or synthetic fertilizers.

The brand is also an important factor in consumer decision-making. The brand is a "search attribute" that serves as an extrinsic factor to signal and enhance consumers' trust [38]; thus, consumers are willing to pay a higher price premium for preferred brands [34,39]. The cultivation and promotion of brand identity can motivate green tea producers to improve and optimize product quality. From the perspective of SME tea producers, branding should be vigorously established and promoted. Generally speaking, national brands are considered to have higher quality and safety than regional brands [35,40]; however, different tea drinking habits exist in different regions, and the effect of teas' origins is extremely prominent [41,42]. Hence, regional brands may be more easily accepted by local consumers [35]. Previous studies have conducted investigations regarding geographical indications or origins [42], but few studies have analyzed both national and regional brands.

Price is one of the most significant factors in consumers' purchase decisions. To set realistic price levels, this study averaged the prices of the top 50 bulk green teas sold on Taobao. Given the considerable premium for green tea in gift boxes, only green tea in bags is used. It should be noted that the green tea set up in this study does not exactly exist in the real market. Generic green tea can be considered as the lowest level of hypothetical green tea varieties [43]. Therefore, the final average price of green tea was set at 101 RMB/500 g, and the other three levels are set at 10%, 15%, and 20% higher.

### *2.2. Experimental Design*

According to the settings in Table 1, a total of 2 × 2 × 3 × 4 = 48 dummy scenarios are generated in the full-factorial experimental design. If each choice set contains two different green tea profiles, respondents will face 2256 choices. Considering the cost and feasibility, this study applied a D-optimal design, as it can ensure validity (D-efficiency) while reducing the asymptotic standard error among attributes [44]. After D-optimal using Negene 1.0 software, a final set of 36 options was randomly generated with a D-efficiency of 93.73%, a D-error of 0.089, and an A-error of 0.103.

According to Kessels et al. [45], due to consumers' limited information load capacity, the number of consumer choices is appropriate at eight. Thus, 36 choice sets were assigned to six versions of the questionnaire, and each version of the questionnaire contained six choice sets. Following Wu, et al. [46], "neither option A nor option B" was included to simulate purchase circumstances more realistically. Hence, each choice set contained two virtual green tea product sets and one "neither option A nor option B." Figure 1 shows an example of the choice set.

**Figure 1.** Example of a choice set. Note: (**A**–**C**) in the figure means the alternative in the choice set.

Several studies have argued that trust would affect consumers' preferences [47]. Low trust is associated with lower ratings of the label itself, which further reduces purchase intention [48]. Two kinds of labels were set in this study. Referencing Wu [49], this study established items of consumers' trust in organic labels and pesticide-free labels. These items were scored using a five-point Likert scale from 1 for "absolutely disagree" to 5 for "absolutely agree." Table 2 presents the detailed items.


**Table 2.** Characteristics of consumer trust in organic label and pesticide-free label.

Notes: SD = standard deviation.

### *2.3. Sample Size Determination*

The rule of thumb is usually used to calculate the required sample size. The minimum sample size is determined by a combination of three factors: the number of choice sets (*t*), the number of alternatives (*a*), and the maximum number of levels (*c*)of the attribute [50,51].

$$N > \frac{500 \times c}{t \times a} \tag{1}$$

Hence, the minimum sample size of this study is 500 × 4 ÷ 6 ÷ 3 = 111. Furthermore, according to Yamane (1967) [52,53], the minimum sample size in the study should be:

$$m = \frac{Z^2 p(1-p)}{e^2} = \frac{1.96^2 \times 0.5 \times (1-0.5)}{0.05^2} = 384.16\tag{2}$$

where *Z* is the significance level of 95%, the value of the distribution table *Z* = 1.96, *p* is the estimate of the correct prediction of *n* for *p* = 0.5, *e* is the sampling error allowed with +/−0.05 (5%). It is noted that the sample size calculated according to the formula is the minimum sample size suggested due to the requirement for stability of the utility estimates. In the actual research situation, the required sample size is larger than the minimum value.

#### *2.4. Data Collection*

For respondent selection, actual consumers of the product should be selected as the target, as only respondents who are familiar with green tea will be concerned about the various attributes [54]. According to Determann, et al. [55], no significant difference was found between online and offline surveys for consumers' preferences in CE; hence, this study used an online survey.

We chose the Questionnaire Star platform (a professional online survey company) to conduct the online survey. The Questionnaire Star sample base is widely sourced and covers a wide range of consumer groups of different ages, occupations and income levels. It is widely used in consumer preference research [56]. As a commissioned network survey, the respondents are generally randomly selected by the commissioned company in its sample database through the network system. To ensure that the respondents identified by random selection met the requirements of this study, the following controls were also conducted in this study. (1) By setting the sample filter question before the formal questionnaire responses: "Have you purchased green tea in the last year?" (2) Screening of targets by age information in the sample pool. This ensures that the participants in the choice experiment survey are real consumers who are at least 18 years old and have had experience in purchasing green tea. Additionally, this study set a validation question [57], "Please select the 'red' option from the following options." Respondents who chose another color were direct to the end of the surveys. A total of 430 valid questionnaires were returned, and Stata 16.0 was used to calculate the final questionnaire data.

The questionnaire consisted of three parts: (a) consumers' trust; (b) comparing alternatives in CE; (c) respondents' socio-demographic characteristics. Given that CE is a hypothetical experiment, hypothetical bias may be present. Referencing Tonsor and Shupp [49], this study presented a brief introduction to respondents, using pictorial examples and textual descriptions of organic labels and pesticide-free labels. After this, two multiple-choice questions were set in this study: "Which of the following characteristics does the organic label contain?", "Which of the following characteristics does the pesticide residue-free label contain?" Only those who choose both correct questions are considered valid. This ensures that respondents understand the meaning of organic and pesticide-free labels before conducting the CE.

### *2.5. Models*

Based on the consumer utility theory proposed by Lancaster (1966), the utility perceived by consumers from a product does not come from the product itself but from its

attributes; thus, in the discrete choice model, the utility obtained by consumer *n* for choice *i* is expressed as follows:

$$\mathcal{U}\_{\rm ni} = V\_{\rm ni}(\beta\_n) + \varepsilon\_{\rm ni} = \delta(\text{ASC}) + \alpha\_n(X\_i) + \gamma\_n(-P\_i) + \varepsilon\_{\rm ni} \tag{3}$$

where *Unit* is the utility obtained by consumer *n* from choice *i* in choice set *t*, *Vnit* (*βn*) is the observable utility of parameter *βn*, and *εni* represents a random error. *Vnit (βn)* consists of three parts. ASC is the specific choice constant. When ASC is 1, it indicates that the respondent chooses the "opt-out" option. *Xi* is the factor that affects the observable utility *Vnit*, which includes the product attributes and the respondent's characteristics *n*. *Pi* is the retention utility, which represents the premium paid for a change in *Xi. β<sup>n</sup>* = (*δ, αn, γn*) is a vector of parameters reflecting respondents' ASC preferences and other attributes.

In this study, the main effect of the attributes was determined using Equation (4). Organic label (ORG), pesticide-free label (PEST), regional brand (RGB), and national brand (NAB) were the categorical variables, and the "none" label was used as the baseline. Price was the metric variable in accordance with the four price levels designated in the experiment. The utility function model is expressed by Equation (4):

$$\mathbf{U}\_{\rm nit} = A \mathbf{S} \mathbf{C} + \beta\_1 \mathbf{Price}\_{\rm nit} + \beta\_{2\mathbf{n}} ORG\_{\rm ni} + \beta\_{3\mathbf{n}} PEST\_{\rm ni} + \beta\_{4\mathbf{n}} RGB\_{\rm ni} + \beta\_{5\mathbf{n}} NAB\_{\rm ni} + \varepsilon\_{\rm ni} \tag{4}$$

where ASC is the "opt-out" option and the coefficients from *β*<sup>1</sup> to *β*5*<sup>n</sup>* are the parameter vectors of the attributes estimated.

For the interaction effects of the attributes, organic trust (OTRU) and pesticide-free trust (PTRU) were the explanatory variables representing consumer trust in organic labels and pesticide-free labels, respectively. Indices of these two attitudinal variables were created by the mean values of the item scores. The utility function with interaction is expressed by Equation (5):

$$\begin{array}{l} \mathbf{U}\_{\text{nit}} = A \mathbf{SC} + \beta\_1 \mathbf{Price}\_{\text{nit}} + \beta\_{2n} \mathbf{ORG}\_{\text{nit}} + \beta\_{3n} \mathbf{PEST}\_{\text{nit}} + \beta\_{4n} \mathbf{RGR}\_{\text{nit}} + \beta\_{5n} \mathbf{NAR}\_{\text{nit}} + \varepsilon\_{\text{nit}} \\ + \beta\_{6n} (\mathbf{ORG}\_{\text{nit}} \times \mathbf{OTRL}\_{\text{n}}) + \beta\_{7n} (\mathbf{PEST}\_{\text{nit}} \times \mathbf{OTRL}\_{\text{n}}) + \beta\_{8n} (\mathbf{RGR}\_{\text{nit}} \times \mathbf{OTRL}\_{\text{n}}) \\ + \beta\_{9n} (\mathbf{NAR}\_{\text{nit}} \times \mathbf{OTRL}\_{\text{n}}) + \beta\_{10n} (\mathbf{ORG}\_{\text{nit}} \times \mathbf{PTRL}\_{\text{n}}) + \beta\_{11n} (\mathbf{PEST}\_{\text{nit}} \times \mathbf{PTRL}\_{\text{n}}) \\ + \beta\_{12n} (\mathbf{RGR}\_{\text{nit}} \times \mathbf{PTRL}\_{\text{n}}) + \beta\_{13n} (\mathbf{NAB}\_{\text{nit}} \times \mathbf{PTRL}\_{\text{n}}) + \varepsilon\_{\text{nit}} \end{array} (5)$$

Consumer *n*'s WTP for attribute *x* is estimated by Equation (6):

$$\mathcal{WTP}\_n = \beta\_{nx} / \beta\_{np} \tag{6}$$

### **3. Results**

### *3.1. Socio-Demographics of Consumers*

Table 3 presents the socio-demographics of the respondents. Among the final sample of 430, there was a slightly higher number of female respondents (54.46%) than male ones (45.54%). This is consistent with some previous studies wherein females are the primary household buyers [58]. Respondents aged 25–34 years hold the largest share (59.90%), followed by those aged 35–44 years (16.34%). Although middle-aged consumers are the main buyers of green tea, the rise of younger consumers cannot be ignored. The married samples were predominant, and most of them had some college or a bachelor's degree. Respondents with a monthly household income of 14,000 RMB and above occupied the largest proportion (30.94%), followed by those with 10,000–11,999 RMB and 12,000–13,999 RMB monthly household income. The higher monthly income and education may be because the study targeted consumers who had purchased green tea. According to Chen, et al. [59], tea consumption is positively correlated with consumers' income. Almost all of the respondents had more than three people living together. Additionally, 70.3% and 54.21% of the respondents had children aged 12 and below and elderly aged 65 and above, respectively. In terms of tea consumption frequency, the percentage of respondents who purchased green tea once every 1–2 months was 68.56%.


**Table 3.** Sociodemographic characteristics of the sample (*n* = 430).

#### *3.2. Main Effect*

Using the mixed logit model, this study set price and its cross terms as fixed parameters, and other attribute variables are set as random parameters. The log-likelihood values of the mixed logit model (−1629.2003 and −1619.7091) indicate that the regression results are generally significant.

Table 4 presents the results of the mixed logit model. In the main effects model, the parameters of the selected attributes are regressed to elicit the consumer preferences for attributes of the organic label, pesticide-free label, regional brand, and national brand. The results of the model estimation show a log-likelihood of −1629.2003, and the regression results are generally significant. The specific alternative constant ASC is significantly negative at the 1% level, indicating that choosing "neither A nor B" has a negative effect on consumer utility when compared with the combination of green tea attributes offered in the study. All of the green tea attribute combination options offered in this study could increase consumer utility. Price is negative and significant at the 5% level, indicating that consumers prefer lower-priced products. The higher the price of green tea, the more negatively it affects consumer utility. The three organic, pesticide-free, and national brand labels are significantly positive at the 1% level, indicating that consumers hold a positive preference for these three labels. The parameter estimation of different labels reveals that consumers have the highest preference for the organic label (1.282), followed by pesticide-free label (0.662) and national brand (0.459).


**Table 4.** Results of the mixed logit model.

Notes: \* and \*\*\* indicate significance at the 10% and 1% levels, respectively. ASC = opt-out option; ORG = organic label; PEST = pesticide-free label; RGB = regional brand; NAB = national brand; SD = standard deviation.

In the main effect with the interaction model, the variable "ORG × PEST" is significantly negative at the 10% level, indicating that there is a substitution effect between the organic label and pesticide-free label. The variables "ORG × NAB" and "PEST × NAB" are significantly positive at the 10% level and the 1% level, respectively. When the organic label or the pesticide-free label is attached to the national brand, consumers' utility is enhanced.

#### *3.3. Main Effect with Interaction in Trust*

This section investigates the conjoint effect of trust in the organic and pesticide-free with the given attributes. Two averaged indices in Table 2 were used in a conjoint regression. The results are shown in Table 5.

**Table 5.** Main effect with interaction in trust.


Notes: \*, \*\*, and \*\*\* indicate significance at the 10%, 5%, and 1% levels, respectively. OTRU = Organic trust; PTRU = Pesticide-free trust; ASC = opt-out option; ORG = organic label; PEST = pesticide-free label; RGB = regional brand; NAB = national brand; SD = standard deviation.

The interaction term between organic trust and the organic label and regional brand is significantly positive. This indicates that the more consumers show trust in organic labels, the more they prefer organic labels and regional brands.

The interaction term between pesticide-free trust and the organic label, the pesticidefree label is significantly positive. This indicates that those who trust in pesticide-free will prefer organic labels too. Pesticide-free is an important attribute of organic labels. The interaction term between pesticide-free trust and the regional brand is significantly negative.

### *3.4. Heterogeneity Analysis Considering Other Consumer Factors*

Heterogeneity exists in consumer preferences for organic and pesticide-free labels. To analyze the sources of heterogeneity, interaction terms of socio-demographics and consumption habits with each attribute of green tea were introduced in the model. Table 6 presents the results.


**Table 6.** Heterogeneity analysis considering socio-demographics and consumption habits.


**Table 6.** *Cont.*

Note: \*, \*\*, and \*\*\* indicate significance at the 10%, 5%, and 1% levels, respectively. ASC = opt-out option; ORG = organic label; PEST = pesticide-free label; RGB = regional brand; NAB = national brand; SD = standard deviation.

Considering socio-demographics, sex, household size, and income have a significant impact on the preference for organic labels. The "education × PEST" variable is significantly positive, while the "elder × PEST" variable is significantly negative. This implies that green tea with a pesticide-free label could attenuate the utility of consumers with elderly people over 65 years of age at home. The "income × RGB" and "children × RGB" are significantly positive, indicating that higher income consumers and those who with children under 12 years of age at home are more likely to buy green tea with a regional brand. Conversely, female consumers are more likely to buy green tea from a national brand. In addition, females, older, bigger household sizes, and consumers with children under 12 years of age at home are rather to choose the opt-out option. They might tend to keep the status quo.

### *3.5. Willingness to Pay*

WTP can directly reflect the change in consumer utility when each attribute changes. The Hierarchical Bayes (HB) approach [60,61] was introduced in this study. Estimations were computed in Stata 16.0 using the command Bayesmixedlogitwtp developed by Baker [62]. Some studies have already used HB to estimate discrete choice models [63,64]. Table 7 shows the results.


**Table 7.** Estimated WTP: mean coefficients in 0.01 RMB.

Notes: \* and \*\*\* indicate significance at the 10% and 1% levels, respectively. WTP = Willingness to pay; ASC = opt-out option; ORG = organic label; PEST = pesticide-free label; RGB = regional brand; NAB = national brand; SD = standard deviation.

In terms of magnitudes, Chinese consumers have highly valued the organic label, with a mean WTP of 148.9 RMB/500 g among all attributes. Chinese consumers also showed a positive preference for the pesticide-free label with a mean WTP of 87.1 RMB/500 g. The reason may be that compared to pesticide-free labels, organic labels include not only food safety attributes (e.g., "no pesticide residue") but also environmental value attributes (e.g., "good for biodiversity" and "low pollution") [65]. In addition, the mean WTP for a national brand is 40.6 RMB/500 g.

Relative to the market price (101 RMB/500 g), the premium for the organic label reached 47.43%. In real life, the price premium of organic green tea over conventional green tea is approximately 50%, indicating that the WTP for organic green tea must be further improved.

### **4. Discussion**

Chinese consumers' demand and preference for safer food have increased significantly because of health concerns [19]. This study confirms that both organic and pesticide-free labels can increase Chinese consumers' perceived utility. This finding is consistent with other studies [66,67], i.e., Chinese consumers have a positive preference for organic food. Organic labels contain not only health and safety attributes but also eco-attributes, such as being environmentally friendly. As society evolves and consumer environmental awareness rises, a growing number of Chinese consumers are motivated by environmental beliefs when buying organic products [68]. Researchers have compared consumer preferences for organic and pesticide-free labels in previous studies. Bernard and Bernard [26] examined consumers' preferences and WTP for organic, pesticide-free, non-GMO, and general products. They found no significant difference in consumer preferences between the organic label as a whole and its parts, and a strong substitution relationship between the whole and its parts. Consumers' WTP for the organic label as a whole is found to be greater than the WTP for each part individually. Grebitus, Peschel, and Hughner [37] examined U.S. consumers' preferences and WTP for pesticide-free labels using Medjool dates, finding that U.S. consumers had positive preferences for pesticide-free labels and were willing to pay more. By contrast, Edenbrandt [25] used rye bread as a subject, asserting that the pesticide-free label was not valuable and that people would only buy organic bread. This study demonstrates that the pesticide-free label is considered valuable on its own by Chinese consumers. The possible reason for this result is health concerns. Roos and Tjarnemo [69] noted that consumers were more concerned with attributes related to personal interests than other long-term benefits. Thogersen, et al. [70] confirmed that the positive attitude of Chinese consumers toward organic food is primarily motivated by consumers' concerns regarding the health value of organic food. Farias [71] demonstrated that the level of information on pesticide-free labels affected consumer preferences. As Chinese consumers become increasingly concerned about food quality and safety and health benefits, the pesticide-free label presents pesticide-related information more directly and visibly than the organic label, so that consumers have a clearer understanding of the quality and value of pesticide-free products. To sum up, both organic and pesticide-free labels have heterogeneous consumer groups and should be targeted to build markets according to their different attributes.

In real life, merchants will attach labels or additional features to goods to enhance the utility of the product itself and further gain more profits [72]. However, there is no unanimous conclusion in the academic community as to whether multiple labels necessarily enhance the utility of a product. Wang, et al. [73] proposed that consumers have a higher willingness to pay for food with both organic food and drug-free labels than organic food alone. The reason is that the more labels a food has, the more likely consumers believe the food is safer. The same idea also appears in Gabaix and Laibson [66,74] and Bertini, et al. [75] who propose that based on the quantity effect, consumers always perceive products with more attributes as superior to fewer attributes. However, Meas, et al. [76] proposed that whether more or fewer labels are better is not in the quantity but in the interaction between labels. He classified the interactions of labels into complementary effects and substitution effects. Several previous studies have shown a strong substitution effect between organic and pesticide-free labels [26]. The finding of this study is consistent with them. The organic label also contains the attribute of no pesticide residues, and there is a partial overlap in reflecting the value of the product; therefore, the overall value estimate for both labels will be less than the sum of the value estimates for the individual labels. Therefore, both labels need to be examined carefully and labeling decisions should not be based solely on the cost-benefit profile of a single label. In addition, this study also found a significant positive interaction effect between national brands and both organic and pesticide-free labels, showing strong complementary effects. According to Parguel et al. [77], brands can also act as a quality signal, and a high level of brand equity can represent a high level of product quality. National brands have higher visibility and better brand images than regional brands, and they can reflect the food quality from another perspective. When they are put together with the organic labels or pesticide-free labels, it does produce a one-plus-one effect. Compared with weak brands, strong brands are more likely to benefit from organic or pesticide-free labels. Therefore, well-known Chinese tea companies are encouraged to participate in organic label certification and to develop organic agriculture.

Consumers' trust in labeling is also a new issue in the area of study [78,79]. The interaction terms demonstrated consumer trust has a positive effect on enhancing label preferences. This finding is consistent with those of studies [32,80]. In an earlier study, Yin, et al. [81] revealed a large level of consumer distrust in organic labels; however, in recent years, with the continuous promotion of the Chinese government and the market, consumer perceptions of organic labels have increased significantly. There is also a deeper understanding and awareness of the connotations of organic labels, which also drives consumer preference for pesticide-free labels. This study also examined the role of sociodemographics in choice. Age, marriage, and green tea purchase frequency had almost no effect on the purchase of green tea. Consumers who were female, had high income, had a large household size, and had elderly above 65 years old at home were more likely to purchase organic green tea. Those with higher education were more willing to purchase pesticide-free green tea. Females, older, larger household sizes, and consumers with children under 12 years old in the household were more likely to maintain the status quo. However, socio-demographics alone are not sufficient to explain the differences in consumer behavior and more intrinsic factors such as consumer psychology should be considered [82].

This study has some research limitations. First, the CE method used provides consumers with a given product profile, and consumers who are not price sensitive may bias the results, which can be further demonstrated in the future by incorporating methods such as random Nth-price auction experiments. Second, China is the largest tea-producing country, with significant tea export and trade. To meet the expectations of different countries, tea producers will often put organic labels of other countries on their packaging, such as the EU, Japan, or Brazil; hence, the type of label preferred by consumers is also a potential consideration for future study.

#### **5. Conclusions**

This study focused on consumer preferences for organic labels and pesticide-free labels among Chinese consumers. The research chose green tea, a real product in the organic market to conduct the CE. It was confirmed that Chinese consumers have preferences for organic labels, pesticide-free labels, regional brands, and national brands. The highest premium for selected attributes was about 39.83% for organic labels, followed by pesticidefree labels (20.58%), and national brands (14.26%). In addition, this study also confirmed a substitution effect between the organic labels and pesticide-free labels; a complementary effect between organic labels and national brands, pesticide-free labels, and national brands. Trust was considered and found that consumers with higher scores in trust preferred green tea with organic labels or the regional brand. The socio-demographics were used to analyze the heterogeneity in consumer preferences. Female and consumers with higher income prefer organic green tea, and consumers with higher education prefer green tea without pesticide residues. Household size and whether there are elderly above 65 or children under 12 in the family also affect the preference. Conversely, age, marriage, and green tea purchase frequency have almost no effect on green tea purchase.

The findings of this paper yield several practical insights. First, considering that the pesticide-free label is not currently in use, such labeling may offer a viable alternative to effectively reduce the costs paid by SMEs for organic certification. For marketers, knowing consumers' preferences for pesticide-free attributes can also improve marketing strategy. For example, in certain markets, product packaging may consider using a pesticide-free label. Second, consumers have shown a highly positive preference for organic green tea, especially for when the organic label is placed alongside a national brand. Tea producers of well-known brands are encouraged to shift to sustainable production and organic certification to generate profits. Finally, trust is something that can contribute to the growth of organic green tea consumption. The government should adopt a responsible attitude and strengthen monitoring efforts to reduce food scandals, thus increasing consumer trust in organic food.

**Author Contributions:** Conceptualization, Q.Z.; methodology, Q.Z. and X.Y.; software, Q.Z. and X.Y.; validation, Q.Z., X.W., X.X., X.Y. and Q.C.; formal analysis, Q.Z.; investigation, X.W. and X.X.; data curation, X.X.; writing—original draft preparation, Q.Z.; writing—review and editing, X.Y. and Q.C.; funding acquisition, X.X. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Social Science Foundation of China, grant number 19BJY048.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** We would like to thank the anonymous referees for commenting on this paper.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland www.mdpi.com

MDPI Books Editorial Office E-mail: books@mdpi.com www.mdpi.com/books

Disclaimer/Publisher's Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Academic Open Access Publishing

mdpi.com ISBN 978-3-0365-9341-8