Journal of Sustainable Marketing

Outcomes

Antecedents

The action competence approach (ACA) suggests that both greater education about environmental issues and exposure to greater solutions to environmental problems contribute to pro-environmental action (Varela-Candamio et al., 2018). There are two approaches to consumer environmental education. The first is initiated by the consumer, while the second is initiated by the organization. Therefore, we explore one antecedent related to consumers’ initiatives to learn about sustainability issues online and one antecedent related to organizations’ communication strategies designed to educate consumers about their environmentally friendly brands.

Key Mediators of Green Buying Behavior

Environmental concern refers to the degree to which consumers are disturbed by the state of the natural environment and the quality of the world’s ecosystem (Zheng, Siddik, Masukujjaman, Alam, & Akter, 2020) . Individuals who demonstrate high concern for the environment are more likely to become conscientious consumers and purchase more eco-friendly products. Numerous studies have demonstrated that higher degrees of environmental concern contribute to greater green purchase behavior (Heo & Muralidharan, 2019; Sinnappan & Rahman, 2011).

Perceived consumer effectiveness refers to “the degree to which consumers think their individual actions contribute to solving problems” (Zhuang, Luo, & Riaz, 2021, p. 4). Consumers are more likely to take action if they believe their behavior will make a positive difference in helping to mitigate environmental degradation. Both early research in pro-environmental behavior (Bamberg & Möser, 2007; Hines et al., 1987) and recent ones support this relationship (Heo & Muralidharan, 2019; Stojanova et al., 2021; Zheng et al., 2020) and confirm the important role of perceived consumer effectiveness in green buying behavior models.

The direct effects of environmental concern and perceived consumer effectiveness on pro-environmental behavior have been established in previous studies. Missing in the literature is a more robust examination of their mediating effects in green buying behavior models. In particular, few studies examine their role in supporting environmental education objectives “to build up students’ abilities to act―their action competence―with reference to environmental concerns” (Jensen & Schnack, 1997, p. 163). These mediators are key to ACA as they promote behavior change by raising: 1) concern for the environment, and 2) an individual’s confidence in their ability to provide solutions to environmental problems.

Heo and Muralidharan (2019) tested the mediating roles of environmental concern and perceived consumer effectiveness in the relationship between environmental knowledge and ecologically conscious buying behavior and found that “knowledge is related to behavior partially through concern” only (p. 431). Bucic et al. (2012) tested ethical consumerism among millennials in Australia and Indonesia and concluded that greater knowledge of environmental issues does not necessarily lead to ethical consumption decisions, especially in a developing country such as Indonesia. This does not imply that education about environmental issues does not have an effect on buying behavior. Consumers who are more knowledgeable about environmental issues are more likely to act in a pro-environmental manner. However, we contend that a mediating factor or a combination of mediating factors must influence the relationship between education and pro-environmental behavior. More specifically, environmental education’s effectiveness in eliciting an action will depend on the degree to which the content and message contribute to raise a consumer’s concern for the environment and to demonstrate a consumer’s opportunity to make a positive difference with more eco-friendly purchases. However, we did find support for the role of environmental concern and perceived consumer effectiveness in mediating the relationships between antecedents and green brand image as outcome.

Based on the above discussion and previous research findings, we propose the following mediating hypotheses to clarify the roles of environmental concern and perceived consumer behavior in the model:

H₅: Environmental concern will mediate the relationships between (a) online education and green buying behavior and (b) green brand communication and green buying behavior in Brazil and the U.S.

H₆: Perceived consumer effectiveness will mediate the relationships between (a) online education and green buying behavior and (b) green brand communication and green buying behavior in Brazil and the U.S.

Moderating Role of Culture

 Furthermore, cross-cultural research on pro-environmental behavior suggests that cultural differences will play a role in determining the individual motivators or “guidance on appropriate or beneficial behaviors in society” (Zhuang et al., 2021). In particular,Hofstede (2001)’s cultural dimension of Collectivism/Individualism – the degree to which a society is characterized as group-oriented and values group interests over more individual needs and desires (Hofstede, 2001)– can influence environmentalism.

Hofstede’s Collectivism/Individualist index scores range from zero to one hundred with the U.S. having the highest individualist index score of 91, Brazil is considered a collectivist country with an index score of 38Collectivist people tend to act in concert, motivated by a desire to benefit society as a whole, while individualist people are more motivated by personal factors. Hofstede (2011) note that collectivist people focus more on the "We" –consciousness and stress values of belonging and group harmony. Given that environmental issues affect every society in the world, we anticipate that Brazilian consumers will act in a pro-environmental manner out of concern for the collective well-being of the planet and the world population in general. For example, Muralidharan, Rejón-Guardia, and Xue (2016) demonstrated how in India, a collectivist country, environmental concern affected green buyer behavior more than in the U.S. In addition, collectivist societies are more affected by social and emotional factors. Watanabe, Alfinito, Curvelo, and Hamza (2020) established how emotional values, instead of functional values, motivated Brazilian consumers to purchase organic food. We expect that in Brazil, a collectivist country, consumers will be more motivated to buy green products because of their concern for the environment. Damage to the ecosystem is now a collective problem and, since environmental protection is seen more as a social need (Zhuang et al., 2021), we expect Brazilians to act out of concern for the environment.

Individualist societies, on the other hand, focus more on the "I" – consciousness and stress values of personal autonomy, power and independence (Hofstede, 2011). According to Hofstede, “[…] individualist cultures encourage an independent self, while collectivist cultures encourage an interdependent self.” (Hofstede, 2011 , p. 113) In his measurement of IND and CLV, he explains that individualistic people value personal time, freedom, and personal challenge and that such individuals desire a personal sense of accomplishment as they grow up learning how to look after themselves (Hofstede, 2011) . Given the strong focus of the dimension of consumer effectiveness on the power that a single individual has to make a difference, we anticipate that American consumers will act in a pro-environmental manner out of a desire to do their part as individuals. In addition, individualist people are more motivated by functional factors and the desire to fulfill personal goals (Zhuang et al., 2021). We expect that in the U.S., an individualist country, consumers’ motivations will reflect functional and personal values. As such, a consumer’s individual ability to make a difference will be more important in such an individualistic society. Cognitive and functional in nature, perceived consumer effectiveness relates to a consumer’s individual characteristics (Zhuang et al., 2021). It has been found to affect green behavior in various countries, but, given the discussion above, we expect its effect to be greater in the U.S. than in Brazil. Therefore, we propose the following moderation hypotheses:

 H₇: The impact of environmental concern on green buying behavior will be greater in a collectivist country.

 H₈: The impact of perceived consumer effectiveness on green buying behavior will be greater in an individualist country.

Research Methodology

We followed standard procedures in developing the measures and assessing the reliability and validity of the measurement model (Anderson & Gerbing, 1988; Churchill, 1979). We used the recommended methodology to address cross-cultural sampling, measure development, validity, and data analysis (Malhotra, Agarwal, & Petersen, 1996; Reynolds, Simintiras, & Diamantopoulos, 2003; Steenkamp & Baumgartner, 1998) in order to assure an acceptable level of structural equivalence that allowed for cross-cultural sample comparisons (Hair, Black, Babin, & Anderson, 2010).

Population and Sampling Procedure

We collected the data via a self-administered electronic survey using Amazon’s Mechanical Turk (MTurk). Generally, MTurk respondents have been found to be better representatives of the general population than participants from convenient samples, such as college student samples (Berinsky, Huber, & Lenz, 2012; Buhrmester, Kwang, & Gosling, 2011) or professional panels (Berry, Kees, & Burton, 2022; Kees, Berry, Burton, & Sheehan, 2017; Zhang & Gearhart, 2020). However, recent studies have found that data quality from respondents on MTurk and other crowdsourcing platforms has decreased since 2015 (Chmielewski & Kucker, 2020; Kennedy et al., 2020; Peer, Rothschild, Gordon, Evernden, & Damer, 2022). While MTurk participants are more attentive to instructions and in answering survey questions than college students , it is therefore crucial to vet respondents and to include screening criteria to improve data quality (Agley, Xiao, Nolan, & Golzarri-Arroyo, 2022; Berry et al., 2022; Chmielewski & Kucker, 2020; Kees et al., 2017). We consequently required our study’s participants to sign an informed consent question, to answer an attention check question, and to provide a unique code at the end of the survey (Agley et al., 2022; Aguinis, Villamor, & Ramani, 2021). Additional precautions included restricting the country location was restricted to assure that only U.S. or Brazilian respondents had access to the survey in their corresponding languages (English and Portuguese). Finally, participants were required to be classified as MTurk Master Workers, i.e., respondents who have traditionally performed well with a 95% approval rating (Jia, Steelman, & Reich, 2017), in order to assure high quality responses.

We cleaned the dataset and removed incomplete and low quality responses that did not pass the attentive test. To assist participants in answering the survey in a similar manner, and to help reduce invariance and method biases (Podsakoff, Mackenzie, Lee, & Podsakoff, 2003), all respondents received the same instructions. Respondents were instructed to continue only if they had purchased a green/eco-friendly product or service in the past year. The following brief description of what constitutes green products and services was provided:

“Green products are those that cause no or minimum harm to the environment, people or animals. Examples include, but are not limited to organic, non-GMO, recycled, energy efficient and/or free from toxic products. Green services are those provided by companies with sustainable practices such as companies that save energy, practice fair trade; attain their natural resources from environmentally friendly farms and producers, etc. in order to minimize their environmental impact.”

We conducted nonresponse bias tests and found no statistically significant differences between the means of early and late respondents (Armstrong & Overton, 1977). The final samples sizes were as follows: U.S. (n = 195) and Brazil (n = 209). The distribution of the samples is summarized in Table 1.

Measurement Development, Validity, and Reliability

The scales were adapted from existing green buying behavior, green brand communication, pro-environmental and sustainability marketing studies (Ali, 2021; Chen & Deng, 2016; Khalique et al., 2020; Rivera-Camino, 2007; Smith & Brower, 2012; Stojanova et al., 2021; Varela-Candamio et al., 2018; Zheng et al., 2020). The original English version of the survey underwent translation and subsequent back-translation by professional translators well versed in Portuguese to ensure content validity, semantic and conceptual equivalence, and consistency of construct meaning among samples. Advice of bilingual experts was sought on the meaning of each item in both the original and the translated survey versions (Craig & Douglas, 2005; Malhotra et al., 1996) . We pre-tested the surveys with practitioners prior to administering them to participants. No concerns were raised in regard to the meaning of each survey item during the pilot tests, suggesting that the survey questions were equally understood and accepted in both countries. The final survey questions are listed in the Appendix A.

 To reduce the potential of measurement error due to the systematic nature of our questionnaire, the following procedural methods were performed in this study. First, participants were assured that the survey was confidential, voluntary, and conducted solely for academic purposes. Secondly, all items belonging to one variable were grouped together (Mackenzie & Podsakoff, 2012; Podsakoff et al., 2003). We also implemented the CFA approach recommended by Malhotra, Kim, and Patil (2006) by loading all items on one factor. “Method biases are assumed to be substantial if the model fits the data” . The U.S. (CFI = .79; IFI = .79; NFI = .75; RMSEA = .156; RMR = .154) and the Brazil (CFI = .78; IFI = .78; NFI = .974; RMSEA = .137; RMR = .143) poor model fit suggests a lack of common method variance in the study. Finally, we conducted a more rigorous CFA test by examining the model including a variable that is “theoretically unrelated to at least one variable in the study” in order to test for CMV using a partial correlation technique . We used measures of the local identity scale (i.e., My heart mostly belongs to the local community) which is conceptually unrelated and have low or nonsignificant correlations with all other variables in our study. Results of the partial correlation analyses showed that all bivariate correlations remained significant when controlling for the marker variable (Gaskin, 2017; Lindell & Whitney, 2001; Malhotra et al., 2006). Given the acceptable results of these tests, we do not expect common method variance to be a major concern in this study.

Next, we assessed the internal consistency and validity of the scales within each sample by analyzing the composite reliability for each scale. The Cronbach’s Alpha for each scale for both groups exceeds 0.85, above the .70 threshold, demonstrating excellent properties (Hair et al., 2010). The average variance extracted (AVE) provides initial evidence of convergent validity as measures have an AVE exceeding the recommended threshold of 0.50 (Fornell & Lacker, 1981; Hair et al., 2010). Table 2 summarizes descriptive statistics and the reliability and validity tests for the scales in the study’s measures.

To further assess convergent validity, a confirmatory factor analysis was conducted and the overall fit of the measurement models are acceptable (for the U.S. sample: χ2 ₍₁₉₅₎ = 453.300, p<.01; CFI = .95; IFI = .95; NFI = .91; RMSEA = .083; SRMR = .0458 and for the Brazil sample: χ2 ₍₂₀₉₎ = 397.214, p<.01; CFI = .95; IFI = .95; NFI = .90; RMSEA = .071; SRMR = .0439) (Hair et al., 2010). All values are within acceptable range (Hu & Bentler, 1999). The data provided in the correlation matrix for the two groups (see Table 3), demonstrate evidence of discriminant validity as the square-roots of the constructs’ AVEs exceed the correlations between distinct variables (Churchill, 1979).

In addition, the factor loadings of items onto each construct across both samples exceeded the recommended threshold of 0.70 (Hair et al., 2010), with the exception of one item for PCE in the Brazil sample, which has a loading of .63.

Modification indices were also analyzed, and their low values, combined with the excellent goodness of fit of the measurement models reported above, demonstrate a lack of alternative factor structure, indicating further evidence of configural invariance. Measurement invariance was also investigated for both measurement models using multi-group analysis and analyzing the critical ratios for differences between parameters. Partial metric invariance was established, since at least one item per scale did not significantly differ between the two groups. There was no evidence of sufficient differences in regards to how participants from each group responded to the survey and the measurement models demonstrated an adequate level of universal structure to allow for cross-cultural comparisons of the structural model (Mackenzie, Podsakoff, & Podsakoff, 2011).

Data Results

Both samples were diverse in terms of age, gender, income and education (see Table 1) as expected (Berinsky et al., 2012). There were, however, some differences and similarities between the U.S. and Brazilian samples. Brazilian respondents were younger with nearly 90 percent of participants belonging to generations Y and Z. Baby-boomers were almost not represented and Generation X accounted for about 10 percent of the Brazilian participants. American respondents were older, with slightly over 80 percent of the participants belonging to Generation X and Generation Y, and approximately 16 percent being Baby-boomers. The Brazilian sample also included a larger portion of males, while the U.S. sample was evenly distributed between males and females. Both samples were comparable in terms of income and educational backgrounds. Respondents with a Bachelor’s degree were the largest groups in both U.S. and Brazil. Overall, the samples were adequate representations of each country’s population.

Hypotheses Analyses

We simultaneously estimated the model for each group using multi-group analysis (Byrne, 2004; Hair et al., 2010). We analyzed and statistically compared the path coefficients for the two groups using AMOS 29 structural equation modeling software (Helmke & Levitsky, 2004) and pairwise parameter comparisons to examine significant differences and test the moderation effects on the hypothesized relationships. We tested the invariance in the parameter coefficients for each relationship by calculating the critical ratios for differences between parameters, to assess the z-scores for the difference in each parameter between groups (Byrne, 2004). To test for the mediating effects of environmental concern and perceived consumer effectiveness in the model, we performed bootstrapping (Preacher & Hayes, 2008) analyses with 5000 samples to calculate the indirect effects of the antecedents (online education and green brand communication) on the outcome (green buying behavior). As suggested by Zhao, Lynch, and Chen (2010), we estimated all parameters simultaneously using structural equation modeling to determine the direct and total effects in order to classify the mediation. Complementary mediation is demonstrated when the indirect and direct effects of an antecedent on an outcome are significant and have the same sign. Indirect-only mediation is demonstrated when the indirect effect is significant, but there is no significant direct effect. We report the indirect effects with 95% confidence interval from the bootstrap analyses and the unstandardized regression coefficients for all effects (Zhao et al., 2010). Lastly, we used one-tail test to examine the moderation hypotheses and included age, gender, income and education, as control variables. We found no significant influence of the control variables on both outcomes. Table 4 summarizes the results of hypotheses H₁, H₂, H₃ and H₄ the results of the moderation hypotheses H₇ and H₈ and effects of the control variables and total variance explained for the two outcomes.

Table 5 summarizes the results of the mediation hypotheses (H_{5 a,b} and H_6a,b).

H₁, H₂, H₃ and H₄ test the direct effects of educational factors on green buying behavior and green brand image. As shown in Table 4, H₁ is fully supported. Online education has a positive, direct and significant effect on green buying behavior in the U.S. and in Brazil. Hypothesis H₂ is not supported. The direct effect of online education on green brand image in both country samples is not statistically significant. H₃ and H₄ are fully supported, confirming the positive, direct and significant effects of green brand communication on green buying behavior and green brand image for both groups.

H_{5 a,b} and H_6a,b test the mediating effects of environmental concern and perceived consumer effectiveness on the relationship between educational factors and green buying behavior. As shown in Table 5, we found support for the role of environmental concern in mediating the relationship between both education factors (online education and green brand communication) and green buying behavior in the U.S. and in Brazil, supporting H_{5 a} and H_5b. Similarly, perceived consumer effectiveness mediates all relationships in the model with the exception of online education and green buying behavior in Brazil. Results show that, in that country, there is a direct-only relationship between the antecedent and outcome with perceived consumer effectiveness tested as mediator (H_{6 a} was supported in the U.S., but not in Brazil, while H_6b was supported in both countries). It is worth noting that both antecedents have positive and significant total effects on green buying behavior; albeit, the total effect of green brand communication on green buying behavior is greater in both countries.

For the moderation hypotheses, H₇ and H₈ were also supported, demonstrating that the effect of environmental concern on green buying behavior is statistically and significantly greater in Brazil, while the effect of perceived consumer effectiveness on green buying behavior is statistically and significantly greater in the U.S.

The overall fit of the structural model indicates the model has a good fit to the data: χ2 ₍₁₉₅₎ = 515.116, p<.01; CFI = .94; IFI = .94; NFI = .90; RMSEA = .089; SRMR = .0462 and for the Brazil sample: χ2 ₍₂₀₉₎ = 442.702, p<.01; CFI = .95; IFI = .95; NFI = .90; RMSEA = .077; SRMR = .0441). Sizable values of R² of the endogenous variable suggest sufficient explanatory and predictive power. Combined, the direct and indirect effects of both education factors and mediators explained 89% of variation in green buying behavior in the U.S. and 82% in Brazil. The direct effects of the sustainability education factors combined explained 45% of variation in green brand image in the U.S. and 52% in Brazil. Next, we discuss the implications of the study, theoretical and practical contributions, limitations and possible future research opportunities.

Discussion

A growing number of marketing scholars and practitioners now realize the detrimental effects of marketing practices that encourage unsustainable consumption and threaten the ability of the planet to survive for future generations to come. Therefore, sustainability marketing must become a ‘pressing issue’ that guides marketing research and practice (Mcdonagh & Prothero, 2014) and that contributes to solve the world’s biggest threats such as climate change, pollution and the depletion of Earth’s natural resources. In addition, global attitude towards purchasing sustainable products has increased in general among individuals from all generations, especially in emerging markets (Muralidharan et al., 2016; Simon-Kucher and Partners, 2021). Individual consumers play a major role in driving sustainability, thus the need to understand green consumerism will continue to grow (Kumar & Polonsky, 2017). As the green market segment expands, so does the need for marketers to effectively “integrate sustainability into branding” to appeal to this growing market and gain competitive advantage (Kumar & Christodoulopoulou, 2014).

 This study sheds light on how companies can design marketing communication strategies and educate a global audience about their sustainable practices, thus improving their green brand image and promoting the consumption of their green products and services. It has numerous theoretical and managerial contributions, which are discussed next.

Theoretical Contributions

 The first contribution of this research relates to needed insight on the significance of ACA for green buyer behavior theory. The importance of sustainability education, whether initiated by the consumer or by the organization, in contributing to promote a brand’s green image also lends support to the relevance of ACA in branding theory. In addition, the interplay of exogenous factors in influencing green behavior is complex. This study reveals that a combination of mediators (environmental concern and perceived consumer effectiveness) are essential factors in increasing the effectiveness of sustainability educational factors in promoting green buying behavior. Lastly, the moderating effects of culture (collectivism versus individualism) is found to be significant in this study, specifically in its role in enhancing the relationship between environmental concern and green buying behavior in a collectivist country and perceived consumer effectiveness and green buying behavior in an individualist country.

Managerial Contributions

First, green brand communication significantly contributes to build the brand’s green image in the U.S. and Brazil alike. Companies should invest heavily in building a green brand image by designing communication to focus on overall green brand initiatives and green products or services offered. In addition, green brand communication is also the strongest determinant of green buying behavior in both countries. Together, these results suggest that consumers prefer to buy from brands that are advertised as environmentally friendly and that demonstrate their eco-friendliness with ecolabels. Brands can use, for example, reputable ecolabels such as “USDA Organic” in the U.S. and “Producto Orgânico Brasil” in Brazil. It is important to note that green brand communication needs to be done in a manner that is not perceived as ‘greenwashing’ (Ali, 2021). Communication about a brand’s products can have a positive impact on the brand’s image, increase the consumer’s perception of their ability to make a difference with their behavior, and directly contribute to green buying behavior.

Second, online education contributes to increase a buyer’s green buying behavior indirectly via environmental concern in both countries and via perceived consumer effectiveness in the U.S. Since online education is initiated by consumers, managers can produce and publish online content that speaks to their green initiatives on various online media platform (e.g., website and social media sites). For example, companies can increase engagement and consumer knowledge about their green products and services via blogs, social media contests, and encouraging consumer reviews. Consumers who are seeking information about environmental issues and green products and services online represent a large segment of the market that can be directly targeted with messages that highlight both a brand’s environmental initiatives and their ability as individuals to make a difference by purchasing the brand’s green products and service.

Third, the role of culture in determining what triggers consumers to buy green is highlighted in this study. In the U.S., an individualist country, consumers are more likely to act if they know that their choices can effectively make a difference. In Brazil, a collectivist country, consumers are more likely to act out of concern for the natural environment. Consequently, marketers should design content that emphasizes consumer’s effectiveness in individualist countries and environmental concern in more collectivist countries as strategies to promote pro-environmental behavior. Nonetheless, consumers in both countries want to know how precisely their behavior contributes to solving the world’s biggest problems. Marketing campaigns that demonstrate how purchasing green products or services can make a positive difference on specific issues such as climate change and, concurrently, raise consumers’ concern for the environment, can be more effective. Therefore, in communicating sustainable practices to consumers, marketers should design content that highlight the positive aspects of their sustainable products and their value and benefits. Advertising campaigns, for example, can include eco-friendly claims, labels, and promotional and educational environmental content suitable for online posting.

In summary, consumers are increasingly more concerned about the environment and want to make sustainable purchase choices. That is good news for practitioners that practice sustainability and focus on green branding. These findings, coupled with the importance of sustainability education to increase buyers’ green buying behavior suggest that businesses and organizations would benefit from a robust sustainability campaign delivered via online platforms. Social media campaigns are effective tools to educate consumers about an organization’s sustainable practices and are extremely suitable for sharing with others. Organizations that design online campaigns with incentives for sharing can benefit from all angles, directly and indirectly, through customers’ increased knowledge, concern for the environment and attitude towards the purchase of green products.