Toward a theory of ethical consumer intention formation: re-extending the theory of planned behavior

What drives consumers to purchase or not purchase ethical products remains something of a puzzle for consumer behavior researchers. Existing theory—particularly the widely applied theory of planned behavior (TPB)—cannot fully explain ethical purchase decisions. This article contends that not only is the original TPB limited in its applicability to ethical purchasing contexts but also subsequent modifications to it have been generally unfitting. This study advances the literature by taking a different approach to recharacterize and re-extend the original components of the TPB in order to make it more relevant and effective for explaining and predicting ethical consumer decisions. This new theoretical framework of intention formation features four determinants (attitude, subjective norm, moral identity, and perceived behavior control) and a key moderator (level of confidence), and thereby possesses better explanatory and predictive abilities to understand ethical consumer decisions.


Introduction
Consumer intention to purchase ethical products has been attracting increased attention among ethical consumerism and consumer ethics researchers (Andorfer and Liebe 2012;Papaoikonomou et al. 2011). Ethical products can be defined in different ways (e.g., Bray et al. 2010; Lewis and Potter 2011;Szmigin et al. 2007), but, in general, they are defined as products or brands that contain certain ethical attributes or are associated with particular ethical values and claims, such as issues concerning human rights, the environment, and animal welfare (Doane 2001;Harrison et al. 2005;Prothero et al. 2011). In the literature, the term "ethical" in relation to a product is often interchangeable with "social or ecological responsibility," though strictly speaking, the concept of ethics is different from corporate social responsibility (CSR). In a review of the different origins of the business ethics and CSR movements, Donaldson and Fafaliou (2003) find that ethics refers to general norms of behavior for business, including all aspects of a business's operations that relate to dominant values in a culture and society. Broadly, they argue that "everything [a] business does is ethically relevant" (Donaldson and Fafaliou 2003, p. 94). On the other side, CSR is narrowly defined as the social obligations of corporations. In this sense, CSR is a part of business ethics that focuses on the role and impacts of business in and on society and natural environments. In this article, social responsibility is included in the definition of "ethical," and ethical products encompass a set of embedded choices that consumers may perceive as being morally right or wrong and socially and environmentally responsible or irresponsible. Ethical issues might include the safety, ethicality, and integrity of products themselves as well as the related effects of the manufacturing process and marketing practices and the impact of the product on society and the natural environment. Levitt (1980) conceptualizes a product at three different levels: (1) the core product, which is the functional dimension of a product that provides fundamental or basic benefits to consumers; (2) the actual product, which is the physical dimension of a product that delivers benefits to consumers; and (3) the augmented product, which includes the extra services or benefits used to attract consumers. For Smith (1990), the ethical features associated with a product are part of the augmented product. The added benefits derived from ethical attributes are beyond the existing core, actual, and augmented features of a product and may have central effects on the value of a brand perceived by consumers. Crane (2001) suggests that ethical augmentation can be defined directly at the product level (e.g., product design, inputs, manufacturing processes, and disposal of used products) and the marketing level (e.g., cause-related marketing). It also can be indirectly defined at the level of the corporation that supplies the product (e.g., the ethical reputation of the corporation) and the level of the country where the product is produced (e.g., the protection of human rights and intellectual rights). For Crane (2001), a product might be ethically neutral, but it is considered ethical (or unethical) based on the bundle of ethical augmentations associated with it and any number of ethical issues perceived by consumers as significant.
It is widely believed that consumer intention to choose ethical products will lead to purchase behavior. Therefore, the formation of intention plays a central role in ethical purchase decisions (e.g., Shaw and Clarke 1999;Shaw and Shiu 2002; for reviews, see Fukukawa 2003;Papaoikonomou et al. 2011). The dominant theory used to explain and predict consumer intentions is the theory of planned behavior (TPB; Ajzen 1985Ajzen , 1988; for a direct application to consumer behavior, see Ajzen 2015). This theory was originally constructed in the field of psychology but has been widely applied to ethical consumption in recent years (see, e.g., Follows and Jobber 2000;Hassan et al. 2016;Papaoikonomou et al. 2011). The theory asserts that three variables-attitude, subjective norm, and perceived behavioral control-determine behavioral intentions. This implies that the more favorable the attitude (and social approval and perceived behavioral control) toward purchasing an ethical product, the more likely it is that a consumer will decide to buy it. However, the TPB is designed to explain general behavior, not behavioral decision-making in ethical contexts. Beliefs that are ethically relevant to an individual's attitude and subjective norm are hidden, obscured, or not explicitly expressed in the original TPB. Thus, its application to ethical consumer decisions is incomplete and constrained. While the TPB has been modified with additional variables, such as ethical obligation and self-identity, it still cannot fully explain ethical purchase behaviors (Chatzidakis et al. 2016). Additionally, there remains an attitude-intention gap, which suggests that a consumer's positive beliefs about and favorable evaluations of purchasing an ethical product do not necessarily translate to the consumer's final choice regarding the product (Carrington et al. 2010;Chatzidakis et al. 2006;Shaw and Shiu 2003;Vermeir and Verbeke 2006). Therefore, what determines a consumer's purchase intention remains unclear in the literature, and more in-depth exploration is required.
The aims of the present article are to understand what drives consumers to choose or reject ethical products and to address the issues surrounding the attitude-intention gap in ethical purchase processes. Accordingly, this article critically reexamines the TPB and its existing modifications in relation to ethical consumption and then recharacterizes its key components and re-extends the theory to reflect consumers' complex and sophisticated thinking when making ethical purchase decisions. The key modifications proposed in this study are as follows: (a) highlight attitudinal beliefs about outcomes for others in addition to outcomes for oneself; (b) draw out salient beliefs about high-level societal norms along with close others' norms and make the moral elements of social norms more explicit; (c) include moral identity as an important and independent component of the TPB, emphasizing virtue beliefs as the foundation for moral identity and integrating beliefs about self-identity into moral identity; and (d) introduce level of confidence as a key moderator of the relationship between the independent variables (attitude, subjective norm, and moral identity) and the dependent variable (purchase intention). Level of confidence was not included in the original TPB and has rarely been studied in relation to ethical consumption (except for Vermeir and Verbeke 2008, who only use it as an additional determinant, not as a moderator). This study establishes that level of confidence significantly affects the strength of correlations between the components of attitude, subjective norm, moral identity, and consumer intention to purchase ethical products.
With these substantive modifications in place, the newly extended TPB, renamed the theory of ethical consumer intention formation, is much better able to explain and predict purchase intentions than previous versions of the theory, and it can serve as a specific stand-alone theory in the fields of ethical consumerism and consumer ethics. In addition to providing a general understanding of consumer intentions regarding ethical products, the new theory helps to clarify the underlying causes of the attitude-intention gap in ethical consumption. Overall, the re-extended TPB serves as an integrative model for empirical testing and future research. Accordingly, general propositions for future research are offered in the present article.

The TPB and its limitations for explaining ethical product choices
The TPB (Ajzen 1985) is a modified version of the theory of reasoned action (Fishbein and Ajzen 1975), with the concept of perceived behavioral control added to the original components of attitude and subjective norm to explain and predict a person's intention to perform a certain behavior (see Fig. 1). According to the TPB, a person's beliefs about a given behavior will determine their attitude toward the behavior, which in turn forms their intention to perform the behavior. Attitude is evaluative in nature and is based on appraisal of the attributes of one's behavioral beliefs, that is, one's belief that performing a behavior will bring about certain consequences. If a person believes that a behavior has largely desirable consequences, they will form a favorable attitude toward the behavior. Additionally, the TPB posits that a person's normative beliefs will influence their behavioral intention. Normative beliefs concern the likelihood of receiving social approval for a certain behavior and hence the perceived social pressure to perform (or not perform) the behavior. A person's normative beliefs are considered to be strong if he or she is motivated to comply with social norms. Further, in the TPB, control beliefs deal with the perceived ease or difficulty of performing a behavior, such as with respect to the available resources or opportunities or any anticipated impediments. If a person believes that he or she possesses more resources and opportunities, they may anticipate fewer impediments to performing the behavior.
The TPB was developed to explain general human behaviors and not specifically predict behaviors in an ethical context. Consequently, its use for explaining ethical purchase decisions is limited, as many of the salient belief components involved in ethical product choices are obscured or different from those featured in the original version of the theory. Without modification and recharacterization (Bray et al. 2010;Ozcaglar-Toulouse et al. 2006), application of the TPB could blur distinctions that are of interest for research on ethical consumerism and consumer ethics. For instance, the behavioral beliefs and evaluations in the original theory are concerned with the desirable or undesirable outcomes or consequences that a person's behavior brings about for themselves, but such self-oriented behavioral beliefs (Sparks et al. 1995) are limited in the context of ethical product choices as they are unconnected to ethical consequentialism (i.e., the total benefit or harm to others, not just oneself). Similarly, the normative beliefs and evaluations in the original TPB are concerned with social expectations to perform or not perform a behavior within the actor's personal circle (i.e., individuals or groups who are close and important to them, such as family members, friends, and coworkers). However, such social pressures are not specifically conceptualized in relation to the normative ethical dimension, which encompasses general social norms and moral duties that may impact ethical purchase decisions. Overall, the original concepts of attitude and subjective norm in the TPB are too narrow or unclearly defined for application in the wider ethical context and need to be expanded in order to explain the complexity of ethical product choices. Ajzen (2011) later suggests that the TPB's sufficiency assumption (i.e., the constructs are sufficient to fully explain people's intentions and actions) might be invalid, as empirical studies have shown that the predictive power of the theory is limited and around 40% of intention formation cannot be explained by the three variables alone (Armitage and Conner 2001;Notani 1998;Rivis and Sheeran 2003;Schulze and Wittmann 2003). Thus, Ajzen (1991Ajzen ( , 2011 makes it clear that it might be useful to add variables to the original theory. Recognizing that the TPB is designed to be adapted to specific contexts, this article proposes expanding the components of attitude and subjective norm in the original TPB, rather than putting them aside or isolating them in respect to ethical consumer decisions. This approach aligns with the argument that, for an ethical product, ethical attributes are only one aspect of the total product offering and cannot be separated from other benefits or aspects of the product and brand (Crane 2001;Szmigin et al. 2007). For most consumers, ethical considerations are not the sole criteria affecting their ethical product choices, and they tend to fluctuate between considering the relative virtues and vices (Irwin 2015;Williams 2015). It seems reasonable to assume that ethically minded consumers might blend the personal consequences of a product with the consequences for others-or juxtapose or balance all outcomes-when forming their attitudes toward an ethical product. A weakness of the TPB is that such ethical consequences are not included as salient outcome beliefs. Thus, the present research proposes to contribute to the original theory  Fig. 1 The original theory of planned behavior: determinants of intention by making explicit the teleological dimension that emerges in consumers' attitudes alongside considerations of personal consequences. Likewise, while the subjective norm component of the TPB does not preclude consideration of ethical norms in intention formation, the original theory does not emphasize ethical norms in an ethical context. Furthermore, the subjective norm in the original TPB relies largely on perceived social norms perpetuated by close others. However, ethical consumption may involve higher-level, broad social norms, which should be highlighted as salient beliefs in ethical purchase decisions.
The expansion approach proposed in this research is distinct from that utilized in previous work that modified the TPB for use in relation to ethical consumer decisions. For example, Ajzen (1991) recognized that, in ethical contexts, one's personal feeling of moral obligation to perform or refuse to perform a certain behavior would influence one's intention. His solution to the lack of explicit ethical dimensions in the original TPB was simply to add the variable "moral obligation" to the three other variables used to predict intentions. Studies of ethical consumerism and consumer ethics have followed suit and added other variables to the theory, such as ethical obligation (e.g., Eagly and Chaiken 1993;Shaw and Shiu 2002), personal values (Vermeir and Verbeke 2008), and moral norms (Dean et al. 2012), to explain consumers' intentions when choosing ethical products. However, in general, those additional variables overlap with the existing components of attitude and subjective norm. For instance, a person's ethical obligation (i.e., an obligation arising out of a consideration of right and wrong) may be based on their consideration of the effects of an action (e.g., not hurting others) or the act itself (in relation to general moral principles such as fairness and justice). The former corresponds to the concept of attitude (desirable consequences) and the latter to the concept of the subjective norm (i.e., social approval) in the original TPB. Thus, it would be illogical and conceptually unjustifiable to treat ethical obligation as an independent and isolated variable in addition to attitude and subjective norm (for an overview of prior modifications to the TPB, see Table 1).
As shown in Table 1, existing research added self-identity to the original TPB for use in both ethical and non-ethical contexts. Fishbein and Ajzen (2010), among others, suggest that self-identity can be added in non-ethical contexts, whilst Sparks and Shepherd (1992) and others (e.g., Fielding et al. 2008;Shaw et al. 2000) added self-identity to the TPB to explain ethical consumption. While it is justifiable to argue that self-identity can be conceptually distinguished from the components of attitude, subjective norm, and perceived control in the TPB (see, e.g., Eagly and Chaiken 1993), the problem with the current conceptualization of self-identity is that it is separate from ethical concerns (Shaw et al. 2000;Sparks and Guthrie 1998) and thus disconnected from moral identity, which tends to be a key determinant in ethical purchase choices (as discussed later in this article). As a result, the conceptual value of self-identity is limited to explaining ethically related decisions. The present article extends the concept of self-identity into the domain of moral identity and suggests that ethical consumers are more likely to perceive morality as  Thorbjørnsen et al. (2007) Self-identity was added to the TPB as an independent variable in non-ethical contexts.
This modification does not consider whether self-identity is related to moral identity and how it is relevant to ethical decision-making.
The proposed variable contains beliefs about teleological (i.e., consequential) and deontological (i.e., duty-related) elements and evaluations, which overlap with ethically relevant beliefs in the existing components of attitude and subjective norm. Thus, the conceptual independence of ethical obligation is questionable.
Moral norms: Dean et al. (2012) Personal values: Vermeir and Verbeke (2008) Emotions: Moons and De Pelsmacker (2012) Moral norms, personal values, and emotions were suggested as additional independent variables of the TPB.
The same problem observed for ethical obligation was observed for these proposed variables; moral norms, personal values, and emotions tend to broadly overlap with the existing components of attitude and subjective norm.
Self-identity is not explicitly connected to moral identity.
central to their self-identity (Aquino and Reed 2002). Furthermore, this study identifies that ethically conscious consumers favor reaffirmation and development of their individual moral character when facing a choice to buy products with or without ethical attributes, and they often rate themselves as above average in terms of morality and integrity (Mazar et al. 2008;Messick and Bazerman 1996). Thus, it is important to incorporate virtue beliefs in the component of moral identity in the TPB. Yet, morally informed self-identity and virtue development in the realm of moral identity are largely absent in existing modifications of the TPB. The re-extended TPB proposed in this study also emphasizes that consumers' confidence in their beliefs and evaluations is a key moderator of the relationship between the independent variables of attitude, subjective norm, and moral identity and the dependent variable of intention. This article suggests that this causal relationship will be stronger or weaker depending on whether consumers are confident in their outcome beliefs, normative beliefs, and identity beliefs, and their evaluations and judgments of those beliefs. Level of consumer confidence, although an important moderator, was not included in the original TPB or other modifications.
The components of the original and modified TPB and those of the re-extended TPB proposed in this study are compared in Table 2.

Teleological evaluative schema: Outcome beliefs and evaluations
In line with normative ethical theories in moral philosophy, Hunt and Vitell (1986) propose two ethical evaluative schemas that can be used in marketing ethics to evaluate ethical beliefs and possible alternative actions: teleological evaluation and deontological evaluation. Teleological evaluation focuses on the consequences of an action. An individual calculates how much good and harm an action will produce, and an action is considered ethically right if the total good is greater than the total harm incurred for the largest number of people. Accordingly, purchasing a product is considered the most ethical choice if it brings about greater benefits and less harm than purchasing any other product. For example, purchasing an electric car may produce benefits, such as energy savings, reduced environmental pollution, and improved health, which are greater than any negative consequences, such as potential battery pollution and power shortages. Hunt and Vitell (1986) provide four criteria for evaluating the outcomes of an action: (1) the perceived consequences for various stakeholder groups, (2) the probability that each consequence will occur for each stakeholder group, (3) the desirability or undesirability of each consequence, and (4) the importance of each stakeholder group.
The consequentialist connotation of teleology is represented in the TPB through the conceptual construction of attitude. According to Fishbein and Ajzen (1975), a person's attitude toward a behavior is determined by their beliefs about and evaluation of the possible consequences of that behavior. If the behavior would bring about desirable consequences that are perceived to be important by the individual, he or she is more likely to form a positive and favorable attitude toward performing the behavior. The component of attitude in the TPB involves general beliefs about the outcomes desired by, and usually for the benefit of, the individual (Sparks et al. 1995). Although Fishbein and Ajzen (1975) suggest that an individual's attitude is a complex combination of feelings, convictions, motivations, and expectations, ethical consequences are not explicitly included and so tend to be obscured in the conceptualization of the component. For this reason, the original TPB appears weak when applied in relation to ethical consumer decisions. Subsequent research has demonstrated that ethical concerns and obligations are influential in the Table 2 A comparison of components of the original TPB, modified TPB, and re-extended TPB formation of intention in ethical contexts (e.g., Beck and Ajzen 1991;Conner and Armitage 1998;Parker et al. 1995). Therefore, it is important to extend the concept of attitude to incorporate, or make explicit, the consequences of a purchase for others as salient beliefs for understanding ethical product choices.
In practice, consumers face a complex array of beliefs about the possible outcomes of purchasing an ethical product for the self and for others. It is likely that ethically minded consumers will evaluate and weigh those two kinds of outcome beliefs as part of a comprehensive calculation, rather than treating them in isolation, when making a purchase decision (Follows and Jobber 2000;Green and Peloza 2011;Öberseder et al. 2011). The findings of social psychology studies suggest that, when engaging in behaviors with a moral or ethical component, people tend to adhere to internalized moral rules (i.e., personal norms) when approving or disapproving of a behavior (Parker et al. 1995;Schwartz 1977). Thus, the inseparability and internalization of ethical consequences relate to a consumer's attitude towards ethical product choices, justifying the conceptual development of attitude in a way that recharacterizes, rather than substitutes, the attitude component of the TPB. This article contends that this approach is more logical and reasonable than other existing approaches to modifying the TPB, which simply isolate ethical obligation from attitude as an independent determinant of intention formation. As noted in Table 1, the conceptual overlap problem is neglected in prior modifications of the TPB.
Also, the original construction of attitude in the TPB does not consider the impact of brand credibility on the formation of outcome beliefs. This is because the TPB was developed in an attempt to move away from previous models of consumer studies that dealt with product utility and overall evaluation (i.e., the multi-attribute decision model and the subjective expected utility model) and focus instead on the specific consumer behavior of interest (Ajzen 2015). This shift in emphasis toward factors other than a product's attributes offers many analytical advantages. However, ignoring the product itselfits features, function, style, and aesthetics-as well as the quality of the product and the credibility of product claims significantly weakens the formation of beliefs about the possible outcomes of purchasing a product. When making product choices, consumers are naturally concerned about whether a desired brand can truly deliver the features and functions promised by the producer, whether it can really produce the anticipated benefits and outcomes, and whether the producer has the ability, expertise, and willingness to fulfil its promises (Erdem and Swait 1998;Keller and Aaker 1998;Sternthal and Craig 1982). Credible and trustworthy product claims can enhance consumers' beliefs about a product, which are subsequently translated into outcome beliefs. When consumers are evaluating a number of brands and information is costly to obtain or difficult to process, credible brands become a signal of product quality and trustworthiness (Wernerfelt 1988). The impact of brand credibility on consumer beliefs is normally established via three mechanisms: perceived quality, perceived risk, and saved information costs (Erdem and Swait 2004). A credible brand with higher perceived quality, lower perceived risk, and lower information costs will be evaluated more favorably by consumers (e.g., Aaker 1991;Erdem and Swait 1998), increasing the likelihood that the brand will be considered and chosen in a purchase decision (Erdem and Swait 2004;Maathuis et al. 2004;Swait and Erdem 2007).
Deontological evaluative schema: Norm beliefs and evaluations Hunt and Vitell's (1986) deontological evaluation schema emphasizes the nature of action, regardless of its consequences. An individual evaluates the inherent rightness or wrongness of an action (implied by alternative actions) against a set of predetermined ethical norms that represent personal values or rules of behavior. These norms include general or issuespecific ethical beliefs related to moral principles such as integrity, fairness, justice, and responsibility (e.g., "The purchase of a 'green' car is part of my duty to protect my family's health and the surrounding environment"). The component of the subjective norm in the original TPB is compatible with the deontological schema of evaluating ethical decisions because its conceptualization of the subjective norm-defined as a socially expected mode of conduct or perceived social pressure (Ajzen 1991)-does not distinguish between social and moral norms or between moral considerations and norm beliefs. Furthermore, social norms often overlap or are intertwined with moral norms (Copp 1995;Owens 2012;Velleman 2013;Wong 2006). For Hunt and Vitell (1986), "social norms" encompass organizational norms, an industry's ethical climate, a society's moral standards, and traditional and cultural values, all of which influence people's moral perceptions. Therefore, it would be difficult to determine whether social approval of a behavior is based on social norms or moral norms, particularly in regard to ethical product choices, in which social influences are often based on moral grounds.
The concept of the subjective norm in the TPB concerns the perceived norms of those who are close and important to the decision-maker. Such norm beliefs and evaluations are too narrow to apply in relation to ethical product choices, which may involve much wider, higher-level norms, such as societal norms. Inclusion of societal norms as particularly salient beliefs in the conceptualization of the subjective norm proposed here is an important contribution to the TPB; in the original conceptualization, societal norms were little visible in ethical decisions. Like the proposed re-extension of attitude, this represents an internalization approach to advancing the concept of the subjective norm. This approach is distinct from existing modifications to the TPB, which separated the deontological aspect of moral obligation from the subjective norm.
Individually perceived norms should reflect the influence of social and societal norms, and they may include both socially injunctive and descriptive norms. Socially injunctive norms are one's perception of what other people think one should do, while socially descriptive norms are one's perception of what other people actually do. The former motivate action by signifying potential social rewards or punishments for a behavior, and the latter motivate action by providing evidence about what actions are likely to be effective and appropriate (Cialdini et al. 1991). As the subjective norm variable in the TPB is only a socially injunctive norm, Fishbein and Ajzen (2010) added descriptive norms to the normative component.
In sum, after performing deontological evaluation of all product alternatives, a consumer will comprehend that choosing a particular product would be the most ethical action and better than purchasing any other product if the purchase conforms to ethical and social norms, including the norms of close others, general societal norms, and socially injunctive and descriptive norms.

Control beliefs and evaluations
It is worth noting that perceived behavioral control is another valid variable that can be adapted to explain consumer purchase intentions. In order to purchase a product, the consumer must have the necessary resources (e.g., time, money, required facilities) to buy the product (i.e., the ability to successfully perform the behavior, which corresponds to the idea of selfefficacy; see Ajzen 1985;Bandura 1977). It is also necessary that the consumer obtains essential information regarding the availability (i.e., opportunity) of the product in the market and anticipates any possible risks or obstacles associated with the purchase. It is unlikely that a person would intend to perform a behavior that is out of their control and practically inaccessible (this relates to the concept of controllability; Ajzen 1985;Sheeran 2002). That is, once a consumer perceives any control problem, whether due to changes in personal circumstances or external factors, that makes a purchase unfeasible, they will decide to not purchase the product. Therefore, perceived behavioral control is an apparent component of the formation of ethical purchase intention; if a consumer perceives a high level of control over the purchase, he or she is more likely to purchase a product (Carrington et al. 2010).

Summary
To summarize this section, three propositions are offered: P1: Outcome beliefs (both self-and other-oriented) and evaluations of these beliefs in regard to purchasing an ethical brand will affect a consumer's attitude toward purchasing from the brand and, consequently, their purchase intentions. Positive beliefs and evaluations will affect a consumer's attitude, and thus intentions, positively.
P2: Normative beliefs (of both close others and society) and evaluations of these beliefs in regard to purchasing from an ethical brand will affect a consumer's subjective norm regarding purchasing from the brand and, consequently, their purchase intentions. Positive beliefs and evaluations will affect a consumer's perceived norms, and thus intentions, positively.
P3: Control beliefs and evaluations of these beliefs in regard to purchasing from an ethical brand will affect a consumer's perceived control over purchasing from the brand and, consequently, their purchase intentions. Positive beliefs and evaluations will affect a consumer's perceived control, and thus intentions, positively.
With the recharacterized components of attitude and subjective norm, the revised TPB should be more compatible with ethical decision-making and better explain ethical product choices. However, two limitations still constrain the explanatory and predictive power of the model. First, the original theory does not acknowledge the critical role of moral identity in the formation of intentions, and although Fishbein and Ajzen (2010) later incorporate self-identity into the theory, self-identity is applied only in non-ethical contexts. Second, the TPB treats consumer confidence as a background factor (i.e., a secondary characteristic of personality disposition; Ajzen 2005), not as a central component, and its important moderating effects on the relationships between attitude, subjective norm, and behavioral intentions are overlooked. In the following sections, these two issues are addressed, and further improvements to the TPB are proposed so that it can be successfully applied to understand ethical purchase decisions.

Moral identity in the formation of ethical purchase intentions
While teleology and deontology provide useful bases for moral evaluation of individual decisions, as discussed above, both approaches fail to acknowledge the crucial role of personality traits and identity in a person's moral concern and reasoning related to the act of consumption (Belk 1988;Friedman 1994;Garcia-Ruiz and Rodriguez-Lluesma 2014;Melé 2009). The personality traits of a moral agent are the central element of virtue ethics. People with good character traits ("virtues") will behave in line with high moral standards, whereas people with bad character traits ("vices") tend to behave in an immoral or evil way. The concept of moral character in ethics is closely connected to moral identity, a term used in psychology that refers to the degree to which morality is perceived as essential to a person's sense of self (Blasi 1984). For Blasi (1995), moral identity is an important part of moral integrity and reaches a climax of highest development in one's overall moral character. Moral identity is built on and organized around a set of moral character traits (Weaver 2006). In other words, it is a psychological realization of one's virtues.
In the present article, moral identity is proposed to be a determinant of an individual's intentions to purchase an ethical product and is introduced to the TPB to explain and predict ethical consumer decisions in combination with attitude, subjective norm, and perceived behavioral control. Two kinds of interrelated salient beliefs are important within the domain of moral identity: (1) beliefs about the reaffirmation and development of underlying virtues and (2) beliefs about the strengthening, extension, and promotion of impressive and symbolic self-identity. Both influence the formation of ethical product choices. Self-identity, which the literature already recognizes as an independent variable in ethical decisions, is a kind of salient beliefs within the component of moral identity. This is because self-identity in ethical consumption inevitably involves ethical considerations (Sparks and Shepherd 1992;Sparks and Guthrie 1998) and, therefore, separating selfidentity from moral values and moral identity is conceptually deficient and defective.

Virtue beliefs and evaluations
Building upon MacIntyre's (1994MacIntyre's ( , 2007 work, Garcia-Ruiz and Rodriguez-Lluesma (2014) outline a conceptual framework of virtue ethics for ethical analysis of consumption. They posit that, when engaging in consumption practices, consumers may advance virtues that make themselves ideal both as consumers and moral agents. That is, consumption serves as a context in which consumers' virtues can be developed to enable them to pursue goods in consumption practices, their individual lives, and their communities. For MacIntyre (1999,2007), goods are reasons for acting. External goods are the effects resulting from or attached to practices, such as prestige, status, and wealth. Internal goods arise intrinsically from practices and include specific achievements, inner satisfaction, and standards of excellence. Possession and exercise of virtues enables people to achieve internal goods. Garcia-Ruiz and Rodriguez-Lluesma (2014) elaborate on this perspective, contending that certain virtues, such as constancy, humility, simplicity, frugality, temperance, and austerity, help consumers to pursue the internal goods in the practice of reference through consumption. The authors assert that the central virtue of consumption is moderation, or proportionate use of resources to avoid the vicious extremes o f o v e r c o n s u m p t i o n a n d u n d e r c o n s u m p t i o n .
Overconsumption can be harmful to individual health, social wellbeing, and environmental sustainability, while underconsumption can limit one's ability to thrive and lead to health hazards. According to this reasoning, moderation is a golden mean between excess (i.e., prodigality) and deficiency (i.e., stinginess) and represents excellence in consumption. Furthermore, moderation is individually varied and contextually dependent, and so requires the virtue of practical wisdom to judge actions and choices in consumption and attain the balance between consumption practices and achieving goods internal to the practice.
Moreover, Garcia-Ruiz and Rodriguez-Lluesma (2014) maintain that virtues are needed to achieve good within a consumer's life as a whole. Practices with internal goods may conflict with one another in terms of an individual's purposes, schedules, or priorities. Virtues enable people to arrange goods orderly and properly while pursuing an ultimate life goal by encountering and coping with various harms, dangers, temptations and distractions, thereby increasing one's self-knowledge and knowledge of good (MacIntyre 2007). When selecting consumption goods, virtues such as kindness, caring, compassion, tolerance, and persistence are important for consumers to prioritize moral goods, create a coherent life narrative, and make fair choices regarding goods (Garcia-Ruiz and Rodriguez-Lluesma 2014). In Miller's (1998) ethnographic studies, for example, provisioning (i.e., the act of buying goods) is a consumption practice for sustaining family life. In this case, choosing goods is not an individualistic behavior, but primarily an expression of the relationship between the shopper and their family members. Shopping for others is done not only to please others but also express love and care for them. Conversely, selfishness, mindlessness, or impulsiveness-the opposite of virtues-could lead a consumer to make a poor choice regarding goods for others, even though the shopping action might be motivated by affection.
For Garcia-Ruiz and Rodriguez-Lluesma (2014), virtues also relate to the community and moral tradition in which consumers are embedded. Moderate and appropriate consumption decisions not only preserve the material resources of the community but also optimize its moral configuration by promoting internal goods and keeping external goods at a reasonable level. The voluntary simplicity movement, for example, promotes the good of the whole community by opposing the overly materialistic lifestyle that is prevalent in modern society, by increasing one's happiness and healthiness, and by promoting harmonious relationships between members of the community (Elgin 2010;Johnson 2004;Schor 1992).
In sum, when consumers make purchase decisions, the praiseworthy character traits they possess-whether naturally inherited, such as sympathy and benevolence (Hume 1739), or developed through consumption practices, such as simplicity and moderation (Garcia-Ruiz and Rodriguez-Lluesma 2014; Grigsby 2004)-provide good reasons to choose ethical products over non-ethical products and to choose a particular ethical product over others. If consumers believe that purchasing an ethical product will maintain, evoke, or nurture their virtues, and thereby contribute to their moral identity, they will most likely choose that product. The virtue-dominated moral reasoning behind product choices will help consumers determine what is good for them and how they can realize this good in order to be a good person. Ethical purchases and consumption are practices in which virtues enable and/or are developed for individuals to pursue the good that is intrinsic to consumption-related practices (e.g., sustainability and standards of excellence), the good associated with unity of their life as a whole, and the good for the communities in which they live.

Self-identity beliefs and evaluations
Self-identity refers to the way one looks at oneself and one's relationship to the world (Weinreich 2003). It involves the individual construction and maintenance of selfhood (i.e., one is a distinct entity with one's own goals, rights, responsibilities, and boundaries, defined by questions such as "Who am I?" and "Who do I want to be?"), self-image ("Who do I look like in the eyes of myself and others?"), and self-esteem (i.e., self-evaluation and appraisal of one's worth, which leads to respect, admiration, and self-confidence; Oyserman et al. 2012). Self-identity was added to the TPB as an independent component by Fishbein and Ajzen (2010), among others, in non-ethical contexts. For them, self-identity is unrelated to morality and is thus conceptually separate from moral obligation. However, prior studies suggest that, in many circumstances, self-identity involves ethical considerations (e.g., Sparks and Shepherd 1992). For example, vegetarianism is commonly associated with ethical values (Shaw et al. 2000). Furthermore, research has shown that the highest level of selfunderstanding (i.e., self-identity or self-concept) implies a moral point of view on each domain of self (i.e., physical, active, social, and psychological; Damon and Hart 1982). Thus, morality is inferred to be an essential part of selfidentity (Strohminger and Nichols 2014) and an outcome of self-development (Lapsley 2005). In the context of ethical product purchases and consumption, self-identity is naturally related to moral values and obligations. For example, the proenvironmental identity of the green consumer is based on the ethical considerations of environmental protection and sustainability (Winston 2011;Zuckerman and Reis 1978).
As described above, virtues are moral character traits and qualities that are intrinsic to and/or further developed by many individuals. They tend to be universal, persistent, independent of context, and fundamental to individual moral behaviors. Self-identity, on the other hand, refers to how a person is defined, evaluated, and recognized by themselves and others in a social context. A person's self-identity is unique to them, as an individual that is indicatively and symbolically distinguishable from others in a society. In a moral sense, self-identity is based on moral character and is the realization and application of virtues in different contexts. Hence, selfidentity tends to be contextually dependent, multi-faceted, and relatively dynamic and changeable. While constructed within the context of society, self-identity is related to social identity (i.e., one's sense of belonging to the social world). Social identification occurs when an individual categorizes people into groups and then adopts the identity of the particular group to which they belong. Self-esteem is maintained and enhanced by comparing oneself with out-groups (i.e., groups to which one does not belong; Tajfel and Turner 1986).
In the psychological and sociological literature, selfidentity is viewed as an important determinant of behavioral intentions (Smith et al. 2008). It is especially important for understanding consumer intentions because consumption symbolically signifies a person's unique values, identity, social status, and personality (Bourdieu 1984;Dibley and Baker 2001;Fischler 1988;Piacentini and Mailer 2004;Wright et al. 1992). In the field of ethical consumerism, prior studies have found that self-identity has predictive power regarding ethical purchase intentions even greater than that of attitude, subjective norm, and perceived control (e.g., Granberg and Holmberg 1990;Ozcaglar-Toulouse et al. 2006;Shaw et al. 2000;Sparks and Guthrie 1998;Sparks and Shepherd 1992). Consumers are most likely to favor products if they believe that the products will maintain, reaffirm, extend, and promote their personal and social identities in relation to morality (Ahuvia 2005;Arnould and Thompson 2005;Belk 1988). A "green consumer," for instance, is more likely to buy green products such as organic food, an electric car, and environmentally friendly household products, even during a recession, than those who do not identify as green consumers (Sparks and Shepherd 1992;Winston 2011). For members of specific groups, like voluntary simplifiers, downshifters, followers of the slow food movement, and green activists, product evaluations will be strictly based on personal and social identification with their group (Lorenzen 2012). Affiliated individuals (as well as some non-affiliated consumers) tend to engage in ethical consumption to express their feelings, beliefs, motivations, and obligations regarding certain lifestyle goals (e.g., a passion for clean, green energy and a green lifestyle; Garcia-Ruiz and Rodriguez-Lluesma 2014).
Discussions in this section on moral identity lead to the development of the fourth proposition here:

P4: Moral identity beliefs (both virtue and self-identity beliefs)
and evaluations of these beliefs in regard to purchasing from an ethical brand will affect a consumer's perceived moral identity regarding purchasing from the brand and, consequently, their purchase intentions. Positive beliefs and evaluations will affect a consumer's perceived moral identity, and thus intentions, positively.

The role of confidence in predicting ethical purchase intentions
The concept of confidence in the original TPB and its antecedent is mentioned in relation to the strength of beliefs (Fishbein and Ajzen 1975) or is regarded as a background factor that affects the belief strength (Ajzen 2015). For Fishbein and Ajzen (1975), belief formation for attitude is composed of two elements: the readily accessible memory (or stored information) about possible outcomes from a behavior in question, and the subjective probability that the outcomes may occur. Belief strength refers to the perceived likelihood of the association between a behavior and its consequences. The notion of confidence is used here as perceived certainty regarding the outcome occurrence. More confidence in the beliefs would lead to more certainty of the attitude. Thus, confidence in the TPB is treated as an input in the predictors rather than as a moderator. Ajzen (2005) acknowledges the moderating role of confidence in attitude strength, as this was proven empirically by other studies. But he disregards it (along with other moderating variables) as being useful in predicting specific actions because of the complexity of interactions among various moderators. Confidence is rarely explored in ethical consumption, and the only existing study to do so (Vermeir and Verbeke 2008) added it to the TPB as a determinant instead of as a moderator. The present article proposes that, as a moderator, the level of confidence plays a more significant role in explaining the relationship between attitude (and subjective norm and moral identity) and intention to purchase ethical (vs. non-ethical) products. Confidence as a moderator variable was traditionally used to segment samples (i.e., divide a heterogeneous pool of subjects into more homogeneous groups; Ghiselli 1963). Following this approach, consumers in general (e.g., the entire sample in an empirical study) can be divided into two different groups: high confidence group and low confidence group. The high confidence group implies that consumers hold a high level of confidence in their beliefs, evaluations, and overall judgments on, or responses to, ethical products, ethical claims, and possible ethical purchases. By contrast, consumers in the low confidence group have a low level of confidence in their overall attitudes, and their perceptions of subjective norm and moral identity toward an ethical brand and purchase. The difference between confidence as an input in the predictors (e.g., affecting the belief strength in the formation of attitude) and as a moderator (e.g., affecting the strength of the attitudeintention correlation) is that the former treats confidence as a generic term in the strength of beliefs (e.g., a certain information choice due to high or low confidence in different information sources) regardless of different samples of subjects, whilst the latter associates the level of confidence specifically with different segment samples (i.e., different consumer groups in the aggregate sample). The moderating effect of confidence suggests that this generally positive relationship between the predictors (attitude, subjective norm, and moral identity) and consumer intention to purchase an ethical product tends to be stronger among consumers who are more confident about their beliefs and judgments regarding the ethical purchase in question, but weaker among consumers who are less confident about their beliefs and judgments. In other words, attitude, subjective norm, and moral identity have a greater impact on confident consumers and a lesser impact on less confident consumers.
Level of confidence is important for understanding the relationship between attitude (and the subjective norm and moral identity) and purchase intention because many consumers lack confidence about their own beliefs and judgements in relation to ethical product purchases, even though they have a generally favorable attitude toward ethical products. There are three potential reasons for this lack of confidence. First, consumers often do not have sufficient knowledge about the ethical products available on the market and are ill-informed about the claims made regarding those products. This information deficiency generally occurs for two reasons. The first is that the information provided by suppliers is insufficient, inadequate, or ineffectively communicated. In a survey of the world's 15 largest markets in 2013, the Reputation Institute found that most consumers did not have information or knowledge about companies' CSR activities and thus were unsure about the trustworthiness of the companies and their CSR claims (Rogers 2013). The companies were not communicating clearly with consumers and had not conducted adequate research regarding stakeholders' views on and demands for CSR (Rogers 2013). The second cause of information deficiency is that ethical products tend to be complicated by multiple product and ethical claims, and consumers have little time, energy, and expertise with which to obtain proper information and carefully examine such claims. Consequently, many existing ethical products have done a poor job of attracting consumers' attention (Beckmann 2007;Smith et al. 2008). For example, despite the popularity of the Fairtrade brand, many consumers still do not understand how purchasing Fairtrade products will help to address deep-rooted injustices in the global economic system, and this lack of communication and transparency often diminishes consumers' willingness to choose products with Fairtrade labels. Once they are fully informed and understand, however, consumers' interest in Fairtrade products increases significantly (White et al. 2012).
Second, people tend to be less confident in their own judgments regarding complicated actions as their belief formations and evaluations are spontaneous and subjective, may be unconsciously biased and irrational, and do not necessarily reflect the reality of objects (Ajzen 1985(Ajzen , 1988(Ajzen , 1991. Hunt and Vitell (1986) also point out that there are no specific information-processing rules for teleological evaluation, and the overall result is the evaluator's beliefs about the relative good (vs. harm) produced by each option. Thus, evaluations of beliefs about a possible ethical purchase are beliefs as well. Beliefs are concerned with the perceived likelihood of uncertain events (Ajzen 1985). Decisions based on personal feelings and estimations may call into question the quality and reliability of these decisions, and sometimes may lead to systematic errors (Tversky and Kahneman 1974).
Lack of confidence in one's own evaluations and judgments is well documented in psychology, economics, and consumer studies. Studies have shown that individuals almost always feel uncertain and insecure about their own judgments and decisions when facing multiple choices (Iyengar 2011), or complex and difficult tasks (Brooks et al. 2015). The theory of bounded rationality explains that rational human decisions are constrained by limited available time, insufficient information, confined cognitive abilities, and complex circumstances and uncertain environments (Simon 1976). The limitations of rationality threaten to eliminate human confidence in decisions and choices (Buchanan and O'Connell 2006). In ethical consumption studies, it has been demonstrated that a consumer's level of confidence is usually lower when evaluating and choosing products with complicated ethical features and ambiguous ethical claims (vs. products without such attributes and claims; e.g., Robinson and Smith 2002;White et al. 2012). For example, the meaning of "sustainable food" or "sustainably produced food" is poorly understood by the general population, not only because of the conceptual ambiguity of "sustainability" itself but also because of the lack of scientific consensus on the specific principles associated with sustainable agriculture (Robinson and Smith 2002). Various studies indicate that consumers are often unresponsive and may even have negative responses toward ethical claims because they lack information and confidence in judging the claimed ethical values and benefits of products, such as the link between Fairtrade products and justice (Auger and Devinney 2007;Carrington et al. 2010;Nicholls and Lee 2006). Consumers' responses to ethical features are also affected by the nature of the products. For instance, if Fairtrade products (e.g., tea, coffee, chocolate) are positioned as non-luxurious products with few indulgent qualities, they are less appealing to justice-conscious ethical consumers, but if the same products are positioned as luxurious, they become more attractive to ethical consumers (White et al. 2012).
Third, consumers' confidence in ethical beliefs and evaluations can be further undermined by psychological factors or biases related to skepticism and cynicism, which are prevalent in today's business environment. Skepticism refers to a person's tendency to doubt and question information, the substance of communications, or the truth or genuineness of a behavior (Fein et al. 1990;Foreh and Grier 2003;Kanter and Mirvis 1989). Cynicism is the persistent disbelief in, and distrust of, ethical and social values expressed by companies. In the present context, it involves negative responses to CSR efforts. Cynicism is a personality trait or habitual disposition related to the way in which one views and thinks about the world (e.g., always disbelieving and distrusting human sincerity and integrity), while skepticism is a cognitive response to a communication input. Skeptics doubt the content of messages, while cynics tend to distrust conveyed messages as well as the deeper motives behind communications (Kanter and Mirvis 1989). Both skepticism and cynicism are key factors blocking the formation of ethical purchase intentions. For instance, a lack of trust provokes cynicism and skepticism toward any CSR efforts, genuine or not (e.g., Becker-Olsen et al. 2006;Vlachos et al. 2009). Even though many companies actively engage in CSR activities, consumers and the general public often have little confidence in them due to the frequency of corporate social irresponsibility, including corporate fraud, scandal, failure, abuse of power, and environmental disasters (see Sun et al. 2010;Tench et al. 2012). CSR initiatives and programs are often regarded as tools for public relations and greenwashing or simply window dressing (Becker-Olsen et al. 2006;Vlachos et al. 2009). This skepticism and cynicism cause consumers to perceive ethical products as being of lower value and higher risk. Recent studies on ethical consumerism have confirmed these negative effects on consumer confidence. In their investigation into teenagers' attitudes toward Fairtrade products, Nicholls and Lee (2006) find that the children commonly disbelieved the information they were given regarding the ethical issues (e.g., serious consequences arising from the working conditions of banana growers in developing countries). Shaw and Clarke (1999) find that consumers are, in general, skeptical about the labeling and advertising of ethical products, which significantly affects their formation and evaluation of ethical beliefs. Bray et al. (2010) also report that even if consumers have expressed a desire to consume ethically, skepticism may deter them from buying ethical products.
The above discussion demonstrates that consumer confidence significantly affects the formation of ethical purchase intentions. This effect is largely reflected in its moderating function regarding the relationships between the independent variables of attitude, subjective norm, and moral identity and the dependent variable of purchase intention because the level of confidence varies among consumers: some are confident in their beliefs and judgements about an ethical purchase while others are not, or some are more (or less) confident than others. The higher the level of consumer confidence, the stronger will be the correlation between attitude, subjective norm, and moral identity and purchase intention (and vice versa). In other words, the components of attitude, subjective norm, and moral identity are predictive of intentions when the level of consumer confidence is high, but not predictive of intentions when the level of confidence is low. An early empirical study on consumer confidence by Bennett and Harrell (1975) indicates that Fishbein's (1967) behavioral intention model (the antecedent of the TPB) demonstrates more predictive power regarding consumer attitudes and purchase intentions when consumers are confident in their beliefs and/or their judgments of their beliefs concerning the outcomes of purchasing a product. In Bennett and Harrell's (1975) work, consumers were segmented based on their level of confidence in their beliefs and evaluations. The behavioral intention model performed better for consumers who were more confident (vs. less confident) in their beliefs and their ability to judge their beliefs.
The perceived importance of confidence level also varies along a continuum of behavioral decisions and information processing (Ajzen and Sexton 1999). Confidence level does not matter much for less important decisions in ethical purchases, such as for small, less valuable items for which little effort is needed to process information. However, for important decisions-for example, large and valuable items with significant purchase consequences-consumers' level of confidence becomes more crucial, and it is necessary to perform information processing and carefully consider the potential purchase consequences, social expectations, realization of moral identity, and purchase barriers. The nature of importance of purchase decisions is closely related to the moderating effect of confidence level.
P5: Consumers' level of confidence in outcome beliefs, norm beliefs, and moral identity beliefs and their judgments in association with the ethical product purchase in question will affect the strength of the relationships between attitude, subjective norm, and moral identity and purchase intention. Attitude, subjective norm, and moral identity predict intentions when consumers' level of confidence is high, but not when it is low.
Theoretical validity, novelty, and explanatory ability Fishbein and Ajzen (2010) recommend that predictors should be added to the TPB with caution. They advise that any proposed addition should (a) be behavior-specific, (b) act as a causal factor in relation to intention or action, (c) be conceptually independent, (d) be applicable to a wide range of behaviors, and (e) be able to consistently improve the prediction of intentions or behaviors. The variable of moral identity proposed in this study fulfils these criteria. Self-identity, which is treated as a kind of core salient beliefs in the re-extended TPB, undoubtedly meets the criteria, as Fishbein and Ajzen (2010) already accepted it as an additional variable (albeit in nonethical contexts). The virtue beliefs, as the foundation of moral identity, need to be highlighted in order to understand ethical behavior, which, like self-identity, are conceptually distinguishable from other salient beliefs in the TPB. Also, Ajzen (2005) acknowledges that personality traits are different from attitudes. The re-extended TPB proposed in this article is depicted in Fig. 2. The new model confirms that the original TPB, as a general theory of volitional behavior, can be flexibly adapted to specific contexts, such as ethical product purchase decisions. However, its overly generalized input beliefs, which lack specific ethical connotations, need to be recharacterized so that the theory can be applied to studies of ethical consumption. As ethical purchase behavior involves additional and more complex considerations of ethical attributes than normal purchase behavior, morally relevant and salient beliefs related to the original components of attitude and subjective norm, including ethical consequences and moral norms, need to be emphasized or made explicit and should not be repressed or rendered conceptually obscure. Moral identity, which is absent in the original TPB, needs to be separate from the very general components of attitude and subjective norm. Otherwise, it will "blur distinctions that are of interest" between beliefs (Ajzen 1991, p. 199) in studies of ethical consumer behavior.
The novelty and contribution of the re-extended theory lie in its recharacterization and expansion of the original TPB. It enriches the original beliefs and evaluations regarding attitude and subjective norm and includes a new component, moral identity, which is supported by the moral foundation of virtues and extension of the existing concept of self-identity in an ethical context. The distinguishing features of the reextended theory in comparison with the original TPB and its prior modifications are shown in Table 2 (above).
The re-extended theory in this study has a greater explanatory and predicative ability than the original and previously modified versions of TPB. For instance, it helps to solve the attitude-intention gap observed in studies of ethical consumption. The theory of reasoned action (Fishbein and Ajzen 1975) and the TPB (Ajzen 1985) claim that once a person's beliefs are formed and evaluated, their attitudes toward a particular behavior (along with the subjective norm and perceived behavioral control) will be automatically and consistently based on their beliefs, which directly translate into their intention to perform a behavior. However, such "reasoned" logic is challenged by ethical consumption studies revealing that, for most consumers, favorable attitudes toward ethical products do not translate into purchase decisions and actions (for reviews, see Fukukawa 2003;Papaoikonomou et al. 2011;Vitell 2015). This phenomenon is referred to as the attitude-intention gap or attitude-behavior gap in the literature (see Chatzidakis et al. 2006;Kuokkanen and Sun 2016).
Consumers' favorable attitudes toward ethical products can be either inflated or genuine. An inflated attitude is described as a consumer's tendency to express in a survey or in public a positive attitude or intention, even when it does not represent their true thoughts and subsequent actions. In such a case, the attitude-behavior gap may be exaggerated due to inflated measures of attitude and intention (Auger and Devinney 2007;Carrigan and Attalla 2001). Conversely, there are ethically minded consumers whose genuine attitudes or intentions may not translate into a final purchase due to other impacts and constraints. This represents a real attitude-behavior or attitude-intention gap (Carrington et al. 2010;Shaw and Shiu 2003). Studies have found that positive attitudes toward ethical products do not necessarily lead to intention to buy the products (e.g., Grunert and Juhl 1995;Shaw et al. 2000;Vermeir and Verbeke 2006;White et al. 2012). These findings devalue the TPB's assumption of attitude-intention consistency and discourage application of the original theory in ethical consumer studies. Even with previous additions to the TPB, "a substantial amount of information remains unexplained" with regard to the attitude-intention gap in ethical purchases (Ozcaglar-Toulouse et al. 2006, p. 512; see also Chatzidakis et al. 2016). Ozcaglar-Toulouse et al. (2006) and others have called for a better understanding of consumer intentions, but research in this area has remained underdeveloped to date.
The new constructs in the re-extended TPB are helpful for addressing the attitude-intention gap. A favorable attitude toward an ethical product purchase does not guarantee a consumer's intention to purchase the ethical product if the wider societal norms do not explicitly or implicitly encourage or pressure them into this attitude. For example, many consumers may care about animal welfare and generally favor products that are not associated with animal testing. However, they may not choose those products because such a social norm concerning animal testing has not been established or because the general public believes that animal testing is necessary to ensure products' safety. As discussed previously, social norms, particularly general societal norms (e.g., moral values, traditions, and cultures), have a strong influence on consumer behavior. A lack of such norms is more likely to result in an attitude-intention gap in ethical purchase and consumption.
The construct of moral identity in the re-extended theory is also critical for understanding the attitude-intention gap. Purchasing an ethical product has strong symbolic meaning for highly moral-minded consumers; it reaffirms, extends, and promotes their personal and social identities in a moral context. For such consumers, if an ethical product does not specifically express or induce their feelings, beliefs, and motivations and conform to their mindset and predefined moral selfidentity (e.g., moral leader, care exemplar, righteous man, green activist, human rights defender, animal rights activist), they are most likely to not choose the product, even though they may exhibit a generally positive attitude toward ethical product choices. Thus, an absence of moral identity in models aiming to predict ethical consumer decisions adds further obfuscation regarding the attitude-intention gap.
Moreover, the level of confidence remains key for understanding the attitude-intention gap. Genuine intentions to

Attitude
Outcome beliefs (self)

Subjective Norm
Norm beliefs (close)

Teleological evaluation
Deontological evaluation

Confidence Level
Intention Virtuous evaluation Fig. 2 The re-extended theory of planned behavior (new elements  are bolded and italicized) choose ethical products will be stronger for those consumers who are more confident in their favored attitude toward purchasing an ethical product and more confident in their perceived moral identity associated with, and social approval of, the purchase. Conversely, weak confidence will elicit little or no intention to purchase, even if consumers show somewhat favorable attitudes. This finding helps to explain the persistence of the attitude-intention gap in ethical consumer decisions. Indeed, confidence, as a moderator variable, is a key part of the explanatory power of the re-extended TPB for resolving the attitude-intention discrepancy, as the model only works for those who are highly confident in their beliefs and judgements in association with the possible ethical purchase. Certainly, factors other than these predictors and the moderator in the proposed model may also affect the formation of consumer intention to purchase ethical products. Ajzen (2011Ajzen ( , 2015 identifies a host of background factors, such as personality, education, age, gender, and income, but posits that they only influence intention indirectly through one or more determinants or the formation of beliefs. For example, individual differences may affect the relative weights of different predictors in the TPB, as several studies have shown (e.g., Sheeran et al. 1999;Trafimow and Finlay 1996). The present study takes the position of Ajzen and others, positing that background factors are different from key determinants in their influence on ethical consumer intentions. Price is also an important factor influencing ethical consumer purchase intentions, but it is beyond the scope of the current article. Instead, it is assumed that consumers are willing to pay an average premium of around 20% for a risk-free ethical product over the same non-ethical product (e.g., Ferreira et al. 2010). Other things being equal, consumers will be more likely to purchase ethical products if their prices remain competitive or affordable.

Managerial implications
This study's re-extended TPB, which is proposed for use in explaining ethical purchase decisions, has profound implications for managers and marketers. For instance, building trust is crucial for mitigating consumers' skepticism and cynicism toward ethical products, and ethical brand credibility plays a key role in gaining customers' trust and loyalty. Moral identity is particularly important for developing ethical products that suit niche markets with strong preferences for certain moral identities. Managers need to pay attention to general social norms and cultures to create ethical products and marketing strategies. Drawing from the ability of confidence level to strengthen consumer intentions, a more prominent implication is that managers should rethink approaches to marketing ethical products.
The effectiveness of marketing approaches is notably related to which information sources provided to consumers are perceived as more credible and convincing. Fishbein and Ajzen's (1975) typology of information sources (i.e., direct observation, inferencing, and outside sources) can be reclassified into the following three categories to better understand information reliability: (1) information from consumer observations, (2) information from companies, and (3) information from third parties. In ethical purchase decisions, consumers rely less on information provided by manufacturers because of widespread skepticism and cynicism, unless trust has been built between the manufacturer and consumers (Helm 2004). If consumers are not confident about their own information and judgments (through observation and inferencing) due to the subjective errors and biases described previously, the only other way to boost their confidence in purchase decisions is through information from credible third parties and acceptance of the outside source's beliefs (Fishbein and Ajzen 1975).
This implies that the traditional company-driven marketing approach, which provides information on consumers through advertising and self-promotion, may be less helpful for facilitating ethical purchasing. Increasingly, consumers tend to find their own ways of determining the true value and trustworthiness of ethical products, which boosts their confidence in their purchase decisions. This is what has been called "consumer-driven marketing" (Court et al. 2009), in which consumers actively obtain information on their own rather than relying on the information supplied by companies in their purchase decision-making processes. Companies should therefore shift from company-driven to consumer-driven marketing and pay more attention to the increasingly influential channels for social information, such as social proof (i.e., informational social influence; Cialdini and Goldstein 2004), expert advice, word-of-mouth communication, and customer reviews. This is not to suggest that companies should manipulate those sources to promote ethical products, but they should recognize the importance of consumer confidence as a frame of reference for formulating and implementing marketing strategies. Companies should focus not only on selling products but also shaping and facilitating consumers' decision-making processes (Edelman and Singer 2015). For example, they should examine how to empower rather than control consumers in their purchase decisions by assisting them with everything they need to make a decision; how to lead rather than follow consumer purchase-decision processes by managing customers' purchase data and anticipating purchase patterns; how to personalize rather than generalize consumers' purchase experiences by enhancing their personal values and satisfaction; and how to save rather than increase consumers' information costs (i.e., time, energy, money, and psychological costs expended to search for information; Hanzaee and Taghipourian 2012). Overall, companies should focus more on the ethical purchase decision-making process than on ethical products themselves, and they should nurture and support the formation of ethical purchase intentions. A true competitive advantage may rely on a company's ability to facilitate the formation of positive and favorable beliefs and evaluations about ethical product purchases and enhance consumers' level of confidence, and thus the relationship between attitudes and intentions.

Conclusions and future research
Research on consumer intention for ethical purchasing has drawn heavily on the TPB, but the model is limited in its applicability to ethical purchase decisions. Prior studies have attempted to modify the TPB by adding more ethics-related variables, such as ethical obligation, to complement the original three components. However, the additional variables either overlapped with the original components or were proven to be less adequate for explaining the formation of intention in ethical product purchases. The present study contributes to the literature by recharacterizing the theory's original components and re-extending the TPB to make it more relevant, specific, and useful for explaining and predicting ethical product purchase decisions. The remodeled theory of intention formation in ethical purchase decisions consists of four components: revised conceptualizations of attitude and subjective norm, the newly added component of moral identity, and the adapted component of perceived behavioral control. The new model also highlights level of confidence as a key moderator of the relationship between the independent variables (attitude, subjective norm, and moral identity) and the dependent variable (intention). The new theory suggests that a consumer's intention regarding an ethical product purchase is likely to be a function of four kinds of evaluations and judgments: teleological evaluation, deontological evaluation, virtue evaluation, and purchase control evaluation. Purchasing an electric car can be used as an example. Based on their beliefs and evaluations, a consumer may conclude that purchasing a specific brand of electric car (vs. other brands) will better conform to social and societal norms; result in good overall utility for themselves, their family, the community, and the environment; reaffirm and enhance their virtues and morally based selfidentity and esteem; and match their purchase abilities and resources. More importantly, the consumer may find that the product is well designed, all the technical features and embedded ethical attributes are satisfactory, the product and ethical claims are genuine, and the brand is trustworthy or well recognized. The product price is competitive with other green cars and has a reasonable premium compared to non-green cars. The purchaser should also have strong confidence in all the beliefs and evaluations regarding the car purchase. Given these terms and conditions, the consumer will most likely have a strong intention to purchase that specific electric car, rather than any other electric or non-green car.
It should be noted that consumers' evaluations and judgments regarding possible ethical purchases are not always ideal, optimum, or linear; in practice, they are often dynamic and fluctuating from time to time and in different contexts. Thus, the four components of the new theory may not have equal weight in the formation of purchase intentions; sometimes only one or two components are important in specific circumstances. For example, a consumer is most unlikely to purchase a product if its perceived ethical credibility is very low, even if all other beliefs regarding the purchase are very positive. Or, even though a consumer does not see any clear moral benefits of purchasing a product, they may still intend to purchase it because many others have done so. These contingent factors and other background considerations, however, do not affect the validity of the re-extended model as a general theoretical framework for explaining and predicting ethical purchase intentions.
Indeed, this study postulates that the new theory is more promising than the original TPB and previous modifications of the model for explaining and predicting ethical purchase intentions and for understanding the well-known attitudeintention gap in ethical consumption. This assertion is built on the model's content validity, which was determined by logical reasoning, observation, and empirical evidence from a variety of studies. The new theory will need to be validated and tested in future research. Apart from theoretical testing, future research could pay attention to the relative weighting of the components (i.e., whether all components are equally important, or some have more weight than others, in ethical purchase intentions, under what conditions, and in what contexts). Cross-cultural studies might be undertaken to help identify and compare the differences in beliefs formed by consumers in different societies and markets. Furthermore, the concept of an "ethical product" is broadly and generally defined in this study, but future research could classify such products into different categories (e.g., eco-friendly, prosocial, moral, charitable) to investigate which kind of ethical products consumers may particularly favor and associate with feelings of moral obligation and identity, and for what reasons they do so. It also will be valuable to study whether products with internal ethical features (e.g., electric cars) and those with only external ethical claims (e.g., cause-related marketing) are equally attractive to consumers. Clarification of this will likely have important implications for managers and marketers (e.g., how to produce and promote ethical products).
Moreover, an ethical product often features a mixture of ethical and non-ethical attributes. The division between ethical and non-ethical salient beliefs (e.g., self-interested vs. otheroriented outcome beliefs) used in this study is conceptually valid and analytically useful, but in practice it might be hard to separate them. For example, it is not easy to isolate the ethical and non-ethical features of an electric car, and thus different beliefs about the outcomes. Yet, it is unclear whether consumers merge or separate the two different beliefs in purchase practices and, if they do so, how. The motivations behind ethical product purchases deserve further study, as most consumers are not intransigent moral actors; only 16% of Americans are committed "super greens," with 66% being "middle greens" (Williams 2015). That is, mainstream consumers are aware of and concerned about CSR issues and appreciate ethical values, but may not translate their beliefs and attitudes into intentions and actions. How ethical brands can appeal to and convince those mainstream consumers is thus a key factor affecting business success.
Finally, the re-extended and additional beliefs and components proposed in this study are not necessarily exhaustive. Other salient beliefs and variables can be added if doing so is justifiable. As Ajzen (1991) indicates, the TPB remains open to the inclusion of additional predictors. This study takes the same position for the newly extended theory in ethical consumer decisions.