The effect of affective response to corporate social irresponsibility on consumer resistance behaviour: validation of a dual-channel model

ABSTRACT

This paper delineates a dual-channel model of consumer resistance towards corporate social irresponsibility. The model assumes a conditional mediational relationship among affective response to corporate social irresponsibility and ethical judgement as drivers of consumers’ inclination towards boycotting, protesting and negative word-of-mouth. Mediation-analysis results are largely in line with the model hypotheses: Affective response to corporate social irresponsibility solely has a significant direct effect on resistance inclination of consumers with a low preference for ethical products. Moreover, contractualistic and moral equity (utilitarian) judgement mediate(s) the effect of affective response on resistance intention for consumers with high (low) ethical product preferences. Based on the empirical findings, the paper presents practical implications and avenues for future research.

KEYWORDS:

• Not for profit marketing
• consumer psychology
• Corporate Social Responsibility (CSR)
• ethics
• consumer boycotts
• moderated-mediation analysis

Introduction

Consumer behaviour research shows that consumers increasingly consider ethical factors when purchasing products or services (e.g. De Pelsmacker, Driesen, & Rayp, 2005; Jägel, Keeling, Reppel, & Gruber, 2012; Vitell, 2015), and current purchasing patterns reflect this presumed change in consumer values. For example, ethical spending in the UK increased six fold between 1999 and 2013 (Ethical Consumer, 2014). Likewise, imports of fair trade produce into the United States have increased more than 10 times over the last 10 years (Fair Trade USA, 2016). Today, bananas, cane sugar, coffee and cocoa are among the most important product categories of fair trade worldwide (TransFair, 2016). In addition to different types of ‘positive’ ethical buying or buycotting behaviour, ethical consumers also take their perceptions and evaluations of corporate wrongdoing into account when making buying decisions, and they are reluctant to purchase products from companies that behave in an unethical manner. This phenomenon has been addressed by various authors (e.g. Kozinets & Handelman, 1998; Sen, Gürhan-Canli, & Morwitz, 2001) and prior research in particular has considered boycotts (Hahn & Albert, 2017; Hoffmann & Müller, 2009; Klein, Smith, & John, 2004; Yuksel & Mryteza, 2009). Even though the findings of econometric research on the effects of boycotts are not conclusive (e.g. Teoh, Welch, & Wazzan, 1999), research assumes a substantial detrimental effect of consumer boycotts on corporate profit margins as well as on corporate image (e.g. Klein et al., 2004). According to an estimate by Ethical Consumer (2017), in 2016, boycotted companies lost more than ₤2.56 billion in revenues in the UK. However, in addition to consumer boycotts, additional different forms of online resistance behaviour (e.g. blogging activities or online petitions) have gained momentum in recent years.

Initiators of consumer boycotts and similar activities of consumer resistance regularly employ low-cost communication measures (Gueterbock, 2004) and affect-inducing imagery and message content (Parry, Jones, Stern, & Robinson, 2013). Moreover, media coverage about corporate wrongdoing contains similar images and information as well. Pro-boycott communication and news media reporting both stimulate concern among consumers, which in turn may increase boycott intention (Hoffmann, 2013b). However, research on emotions as a motivating force in boycotting behaviour is still quite scarce. Previous studies focus, for example, on consumer outrage (Lindenmeier, Schleer, & Pricl, 2012), anger (e.g. Grappi, Romani, & Bagozzi, 2013; Romani, Grappi, & Bagozzi, 2013) and consumer animosity (Ettenson & Klein, 2005). In addition, this stream of research considers rather complex moral emotions based on cognitive appraisal and does not explore less complex, spontaneous affective reactions in depth.

Previous research reveals that cognitive constructs such as perceived cost-benefit ratios (e.g. Klein et al., 2004) or expected boycott success (e.g. Sen et al., 2001) are crucial for boycott motivation. However, researchers have not examined the effect of potential boycotters’ ethical judgement of corporate wrongdoing in detail and remain largely silent with regard to the interplay between affective and cognitive drivers of boycotting behaviour. Considering this gap in research, this study assumes that a mediational relationship exists whereby communication-induced negative affect triggers ethical judgement of corporate wrongdoing that, in turn, influences boycott motivation.

From a market segmentation perspective, this study postulates that the mediation effect of ethical judgement depends on consumers’ membership in specific segments. Klein et al. (2004) find that deciding whether to join a boycott is different for consumers with varying levels of preference for the boycotted product. Similar results with regard to substitutability of boycotted products can be found in work by Sen et al. (2001). Confirming these research findings, initiators of boycott campaigns or other forms of consumer resistance frequently advertise alternative products as substitutes for boycotted products. This study distinguishes between consumers with (without) a preference for these advertised ethical product alternatives. The current paper hypothesizes that individual resistance behaviour is more strongly driven by cognitive (affective) forces for those with a high (low) preference for ethical product alternatives.

In sum, the present study analyses whether ethical judgement mediates the effect of affective response to corporate social irresponsibility (CSI) on boycott, negative word-of-mouth (WOM) and protest intention. This research also analyses whether the preference for ethical product alternatives moderates this presumed mediational relationship. The paper has the following structure: the next section discusses different types of consumer resistance and delineates hypotheses based on the gaps in the research described above. Next, this paper validates the developed model of consumer resistance behaviour. The research methodology is subsequently described, and the study findings are presented and discussed. Finally, practical implications are outlined along with study limitations and avenues for future research.

Consumer resistance

Deng (2012) defines consumer resistance as a negative response to firms’ behaviour. Penaloza and Price (1993) characterise consumer resistance as a strategic response to structures of domination, for example, the free-market system and multinational corporations. In addition, Fournier (1998) emphasises the aim of regaining consumer sovereignty by means of purchasing power. Consumer resistance focuses either on whole markets or on specific products and brands. Penaloza and Price (1993) state that consumer resistance can arise in the form of individual and collective behaviour. Furthermore, Ritson and Dobscha (1999) distinguish between observable actions (e.g. public criticism) and non-observable, silent ones (e.g. different forms of anti-consumption and voluntary simplicity). Hollander and Einwohner (2004) name action and opposition as core elements of consumer resistance. Considering Fournier’s (1998) classification of consumer resistance, this paper focuses on boycotting behaviour, negative WOM and protest behaviours that can be classified as types of so-called active rebellion.

Boycotting behaviour

Garrett (1987) defines consumer boycotts as campaigns by activist groups that aim to prevent market transactions between consumers and companies, which are not legally binding for an individual consumer. These activist organisations try to achieve their goals by motivating consumers to refrain from buying products from one or more companies (Friedman, 1985). Consumers can support boycott campaigns by either not purchasing products from boycotted companies (Klein et al., 2004) or switching to ethical product alternatives (Sen et al., 2001).

Boycott initiators have to rely on an individual consumer’s inclination to participate. Therefore, consumer boycotts exhibit characteristics of individual consumer behaviour (Klein et al., 2004). Perceptions of egregious corporate wrongdoing constitute a major trigger for boycott participation, which may represent a way to vent anger (Hahn & Albert, 2017). In addition to these expressive motives, Klein et al. (2004) and Hoffmann (2013a) highlight that instrumental motives may affect boycott inclination as well. Furthermore, a high preference for the boycotted product may dampen boycott intention (Sen et al., 2001). Consumer boycotting behaviour represents a type of collective or prosocial behaviour because boycott success is contingent on the participation of a large group of consumers (Sen et al., 2001). Therefore, every consumer has an incentive to free ride on the efforts of others, which may result in low participation because individuals may regard their contribution to boycott success and the total boycott participation as too low (John & Klein, 2003).

Negative WOM

Grappi et al. (2013) characterise negative WOM as a form of sharing emotions with others, which, in addition to emotional expression, aims to harm the reputation of companies that do not behave appropriately. Williams and Buttle (2014) found that two significant determinants of negative WOM are dissatisfaction and external media comment. WOM can occur when consumers learn about and evaluate corporate misconduct, and it depends on the strength to which consumers hold specific ethical virtues (Grappi et al., 2013). WOM is triggered by contempt, anger and disgust (Grappi et al., 2013). In general, it can be assumed that bad-mouthing a product or the company itself is not as costly as withholding consumption. Furthermore, because it may contain valuable information, WOM serves the well-being of a greater group of people (Hennig-Thurau, Walsh, & Walsh, 2003). WOM can also be classified as low-involvement and impulsive behaviour. In line with this, according to Stern (1994), WOM is interactive, ephemeral and spontaneous. Consumers wish to immediately share affective states such as satisfaction, pleasure and sadness by sending WOM (e.g. Nyer, 1997). Those in an angry state of mind are particularly likely to express their feelings by engaging in negative WOM (Wetzer, Zeelenberg, & Pieters, 2007).

Protest behaviour

The term ‘protest behaviour’ subsumes alternative resistance behaviours beyond those previously mentioned. This study includes online actions against corporate misconduct (e.g. blogging or sharing on social media, signing an online petition) as well as cash and time donations to activist groups. By doing this, the current study aims to consider newer and less studied types of consumer resistance (Grappi et al., 2013). Protest behaviours are considered a more complex and radical form of consumer resistance than WOM (Penaloza & Price, 1993). Moreover, similar to WOM, protest behaviour is a manifestation of voice behaviour. Voice behaviour has several functions, namely, providing emotional relief by expressing one’s dissatisfaction, alerting others to important issues and thereby producing public interest, and enhancing entrepreneurial efficiency by providing feedback (Keeley & Graham, 1991). Finally, protest behaviour can take place on an individual or collective level, with the protest behaviours examined in this study reflecting collective behaviour.

Conceptual framework and hypotheses

A dual-channel model of consumer resistance

This study uses the cognitive mediation model (Eveland, Shah, & Kwak, 2003) as a conceptual framework to analyse how the interplay of affective response to CSI, ethical judgement, and preference for ethical products affects consumer inclination to engage in resistance behaviour. Communication research and political science use the cognitive mediation model to explain the news media’s influence on recipients’ cognitive and emotional responses to media exposure (Eveland, 2001). In line with this research, the present paper analyses how affective response to information about corporate wrongdoing impacts consumers’ inclination towards different types of resistance behaviour. The cognitive-mediation model expands the well-known stimulus-organism-response (S-O-R) approach by assuming an O₁-S-O₂-R chain of causation (Eveland et al., 2003). The ‘O₁’ component relates to characteristics of the recipients of communication stimuli, and the ‘S’ component refers to communication stimuli that trigger a psychological reaction in recipients (‘O₂’), which produces behavioural reactions (‘R’).

In this study, consumers’ affective reaction to media footage of allegedly unethical behaviour is considered as an indicator of the model’s ‘S’ component. Consumers’ ethical judgement of corporate wrongdoing and consumer resistance are considered the ‘O₂’ component and the ‘R’ component, respectively. Figure 1 provides an overview of the relationships that are proposed in this study, in which affective response to CSI is assumed to impact consumer resistance behaviour via two channels: a direct and affective channel (i.e. affective response to CSI → behavioural intention) and an indirect and cognitive channel (i.e. affective response to CSI → ethical judgement → behavioural intention). Consumer preferences for ethical product alternatives represent the ‘O₁’ component. The model considers preferences for ethical products as a moderator that determines which of the two channels is more decisive for resistance behaviour.

Figure 1. Dual channel model of consumer resistance. The study considers three dependent variables (see Appendix 1).

The direct effect of affective response: the affective channel

According to Prislin (1996), affective extremity is ‘the extent a person experiences positive or negative feelings and emotions in relation to the attitude object’ (p. 451). This study examines the effects of affective response to CSI based on this definition. Affects are important drivers of behaviour according to Frijda (1986). Gregory-Smith, Smith and Winklhofer (2013) identify emotions as elicitors of ethical consumption behaviour and find that negative emotions discourage (encourage) consumers from making (ethical) unethical decisions. In line with this finding, consistent with Hoffmann (2013b), this study assumes that affective response has a positive effect on consumer resistance intention. Västfjäll and Gärling (2006) observe a strong influence of negative affect on an individual’s motivation to decrease his or her negative feelings, which is consistent with the tension-reduction view of vicarious emotion and resulting motivation (e.g. Batson, Fultz, & Schoenrade, 1987). This theoretical rationale also conforms to Cialdini and Kenrick's (1976) negative-state relief model. This study assumes that affective response to corporate wrongdoing is associated with the motivation to reduce a person’s vicarious distress by engaging in helping behaviour. Hence, considering the delineated dual-channel model, consumers who feel strong negative affect (‘S’) may feel obliged to join a boycott or to perform other resistance behaviours (‘R’). Based on this reasoning, an affective channel of causation is assumed and hypothesis H₁ is as follows:

H₁: The stronger the affective response to CSI, the higher the inclination towards consumer resistance.

The mediating effect of ethical judgement: the cognitive channel

According to Sparks and Pan (2010), ethical judgement is ‘an individual’s personal evaluation of the degree to which some behaviour or course of action is ethical or unethical’ (p. 409). The present study employs the multidimensional ethics scale (MES) developed by Reidenbach and Robin (1990) and extended by Cohen, Pant and Sharp (1993). This extended scale consists of four dimensions – moral equity, relativism, contractualism and utilitarianism – that encompass a broad spectrum of ethical perspectives ranging from deontology to relativism as well as from justice ethics to teleology. According to McMahon and Harvey (2007), deontology states that people are required to follow specific rules; relativism states that moral rules are culturally specific and no universal ethical rules exist; justice ethics assume that equals (unequals) should be treated equally (unequally); and teleology postulates an assessment of actions based on their consequences.

The moral equity dimension of the MES scale comprises justice-related and deontological ethical views, and moral equity norms are derived from what people have learned during their childhood or personality development to constitute right or wrong (Reidenbach & Robin, 1990). The relativist dimension involves ethical rules that are defined by social or cultural systems and learned in subsequent years (Smith & Cooper-Martin, 1997). Contractualism is based on deontological ethics, in which an unspoken social contract to avoid breaking certain ethical rules is assumed to exist between society and its members. Utilitarianism belongs to the teleological or consequentialist ethics and embodies the notion of maximising utility for society (Cohen, Pant, & Sharp, 2001).

Considering the components of the cognitive mediation approach, this study assumes that the effect of negative affect (‘S’) on consumer resistance (‘R’) is mediated by ethical judgement (‘O₂’). Hoffman (1994) proposes that empathy may contribute to moral judgement; however, the precise point where emotions occur and affect moral judgement is extensively debated (Huebner, Dwyer, & Hauser, 2009). Based on the social intuitionist model (Haidt, 2001), this study assumes that affective response to CSI occurs before ethical judgement. In line with this notion, individuals have an instant feeling of approval or disapproval when confronted by a stimulus (e.g. media footage on corporate social irresponsible behaviour) without greater cognitive effort, and the cognitive assessment of the stimulus follows this spontaneously provoked feeling (Greene & Haidt, 2002). This assumption is consistent with Hoffman (2008), who depicts several modes of empathic arousal that involuntarily occur with no or minimal cognitive control. Empirical research also shows that affect directly influences judgement (e.g. Finucane, Alhakami, Slovic, & Johnson, 2000).

In sum, affective response to unethical firm behaviour (‘S’) is expected to be marked by physiological arousal, which motivates recipients to process information about ethical misconduct with greater detail. This information processing is assumed to result in an ethical judgement about the allegedly unethical corporate behaviour (‘O₂’), which in turn affects consumer resistance intentions (‘R’). Considering this rationale, a cognitive channel of causation is presumed, and hypothesis H₂ is stated as follows:

H₂: Ethical judgement of corporate misconduct mediates the effect of affective response to CSI on consumers’ inclination towards resistance.

The moderating effect of preference for ethical product alternatives

The cognitive mediation theory expands the well-known S-O-R approach to an O₁-S-O₂-R model (see Figure 1) by including an ‘O₁’ component. The ‘O₁’ component captures different cultural, cognitive and motivational characteristics of the recipients’ communication or information stimuli (McLeod, Kosicki, & McLeod, 1994). According to Eveland et al. (2003), the ‘O₁’ component helps determine how recipients process the information conveyed in communication stimuli (‘S’). This study assumes that the ‘O₁’ component can moderate the effect of the affective response to information about ethical misconduct on consumer resistance intention, which is consistent with research by Grappi et al. (2013) in which they observe a moderating effect of other-regarding virtues on the relationship between emotions and consumer responses to unethical corporate behaviour. Within the scope of this study, consumer preferences for ethical product alternatives are considered as the ‘O₁’ component. This study assumes that the preference construct encompasses the varying subcultural, cognitive and motivational characteristics of consumers with different levels of ethical conscience.

Sen et al. (2001) show that consumer preference for a boycotted product has a negative effect on willingness to engage in resistance against corporate conduct. From an opposite perspective, the availability of substitutes for a boycotted product should boost boycott intention (Sen et al., 2001). Therefore, boycott campaign managers promote ethical product alternatives to increase boycott inclination. Consistent with Crane’s (2001) reasoning, this study considers the ethicality of product offerings from the perspective of the augmented product concept, which implies that ethical product dimensions augment the core product and its functional features. Fair-trade agreements between producers and retailers and sweatshop-free manufacturing are examples of these ethical augmentations. Consumers regard ethically augmented products as morally superior to commonly available products and consider these products to be an alternative buying option.

In contrast to existing research, this study does not primarily examine the direct effect of consumer preference for ethical products on resistance intention. This study rather analyses whether affective response to CSI has a different effect on consumer’s resistance intentions with different levels of a preference for ethical product alternatives. Prior work has claimed that consumers who have a high (low) preference for ethical products have more (less) likely already processed the ethical issue and already have assessed these types of products several times. This is consistent with the insights of De Pelsmacker, Janssens, Sterckx and Mielants (2006), who suggest that consumers do not buy fairly produced products due to a perceived lack of information. Shaw and Clarke's (1999) study concludes that sufficient information is important for ethical belief formation. Similarly, Valor (2007) emphasises the importance of information for ethical consumer behaviour and therefore highlights the need for analysing the influence of the information processing across different groups of consumers. Building on this and on research on attitude–behaviour consistency (Davidson, Yantis, Norwood, & Montano, 1985), this study assumes that ethically concerned consumers have more stable, and in addition, more distinct attitudes towards and/or evaluations of allegedly unethical corporate conduct. Furthermore, information about ethical misconduct should, in case of a high preference for ethical products, immediately elicit these stable evaluations of corporate conduct, which subsequently affect consumers’ inclination to offer resistance. Conversely, resistance behaviours by consumers with a low preference for ethical products should be more affectively driven than the resistance behaviours of consumers with a high preference. Negative affective responses to unethical corporate behaviour should consequently have a stronger direct effect on these consumers’ behavioural intention. Therefore, hypothesis H₃ is stated as follows:

H₃: Preference for ethical product alternatives moderates the mediation effect of affective response to CSI on consumer resistance intentions. The indirect effect of negative affective response to CSI on resistance intentions is stronger (weaker) for consumers with a high (low) preference for ethical products (H_3a). The direct effect of negative affective response to CSI on resistance intentions is stronger (weaker) for consumers with a low (high) preference for ethical products (H_3b).

Methodology

Study design

This paper considers child labour as an instance of corporate wrongdoing. Furthermore, this study considers chocolate that has been produced under fair circumstances as an ethical product alternative. Fair-trade chocolate qualifies as an ethical product because it is produced without child labour and under favourable working conditions with fair wages. In line with this notion, Nicholls and Opal (2005) classify fair trade as a form of market-driven ethical consumption.

In order to validate the model hypotheses, we conducted a survey study. German residents aged 16 to 70 were identified as the population of interest. For this study, 496 respondents were surveyed, and data collection was accomplished in cooperation with the Qualtrics online panel company in February 2018. To increase the data quality, gender and age as quota criteria were employed to ensure a better representation of the population of interest. Table 1 depicts the distribution of respondents across gender, age and citizenship. The sample slightly underrepresents German men aged 26 to 55. The share of the population without German citizenship status, which actually accounts for approximately 12% of the total German population, is also underrepresented in the sample. Although gender is nearly uniformly distributed across consumers with a low or high preference for fair-trade chocolate, age is not. Subjects aged 26 to 45 tend to consume less fair-trade chocolate than others. This is in line with several market surveys that reveal a higher ethical consciousness among older German consumers. For instance and relating to their product evaluation, sustainability features are more important for German consumers over the age of 45 years than for German consumers aged between 18 and 24 years (WE Communications, n.d.). Furthermore, older consumers in Germany pay more attention to animal welfare and attach more importance to ecological cultivation and organic products (IfD Allensbach, n.d.). Finally, 50–59 year olds are the consumers in Germany that buy the most in organic supermarkets (VuMA, n.d.). In line with these studies, Vecchio and Annunziata (2015) found that the willingness to pay for sustainability-labelled chocolate products slightly increases with age. Finally, and with regard to the citizens of the European Union, people aged 40 years and older show a more environmentally friendly behaviour in their everyday life than people aged between 15 and 39 years (European Commission, n.d.).

Table 1. Sample characteristics.

	Gender
	Women		Men		Total
Age (years)	N	(%)	N	(%)	N	(%)
16–25	36		35		71	14.31
26–35	47		41		88	17.74
36–45	44		35		79	15.93
46–55	64		47		111	22.40
56–65	51		54		105	21.17
66–70	22		20		42	8.47
German citizenship
Not specified	1		0		1	0.20
Yes	258		225		483	97.38
No	5		7		12	2.42
Total	264	53.23	232	46.77	496	100.00

Information on the use of child labour and inhumane working conditions at Western African cocoa plantations, as well as on a fictitious non-governmental organisation's campaign, were provided at the beginning of the online survey. Furthermore, subjects were told which leading companies supposedly process cocoa from plantations where children work. The questionnaire highlights that fair-trade chocolate represents an ethical alternative to conventionally produced chocolate and that consumers may participate in a boycott by switching to this substitute; participation in WOM or protest behaviour were also named as options to fight against unethical business practices. Pictures of children suffering on cocoa plantations were included and respondents had to describe the affect they felt spontaneously. Thereafter, the participants responded to questions about the remaining explanatory variables depicted in Figure 1. Moreover, before asking the interviewees about their resistance intentions, subjects were asked a distractor question in order to reduce the effect of a possible common method bias. Finally, interviewees provided sociodemographic information.

At the beginning of the survey, in order to reduce the respondents’ inclination to give socially desirable answers, participants were asked to answer the questions truthfully and the researchers guaranteed respondents’ anonymity. The researchers informed the interviewees that the survey would take about 15 min and that they could freely withdraw from the survey at any time. Furthermore, it was highlighted that the data would only be evaluated for scientific purpose. At the end of the survey, subjects were debriefed and dismissed. In particular, they were told that the study included a hypothetical boycott call.

Measurement

The questionnaire (refer to Appendix 1) included several multi-item scales. Different types of scales were used throughout the questionnaire to reduce a possible common-method bias. Affective response to CSI was measured by a six-item scale adopted from Prislin (1996). An extended Reidenbach and Robin's (1990) MES scale was considered to assess consumer’s ethical evaluation. The measurement instrument includes utilitarian judgement according to Cohen et al.’s (1993) scale extension. Both scales employed seven-point semantic differentials. Three multi-item scales were used to measure boycott intention, WOM intention and protest intention. These instruments were mainly adopted from Klein et al. (2004) and Grappi et al. (2013) and considered seven-point Likert scales. Preference for the ethical product alternative was measured using a single question, asking for the percentage share of fair-trade chocolate in total chocolate consumption.

Based on modification indices and considering covariances between some error terms, three confirmatory factor analyses that include all latent independent variables reveal an acceptable-to-good global fit for the measurement models, one for each of the three dimensions of consumer resistance intention (i.e. dependent variables), namely, boycott intention, WOM intention, and protest intention: χ² = 443.37/464.55/446.50, p < 0.01/0.01/0.01; χ²/df = 2.41/2.53/2.43; GFI = 0.92/0.92/0.92; AGFI = 0.90/0.89/0.89; NFI = 0.96/0.96/0.96; TLI = 0.97/0.97/0.97; CFI = 0.98/0.97/0.97; RMSEA = 0.05/0.06/0.05; SRMR = 0.04/0.04/0.04. All factor loadings are significant and exceed 0.5 (see Appendix 1). Appendix 2 depicts the results of reliability and validity analyses, which reveal internal consistency and convergent validity. Only the average variance extracted (AVE) of protest intention is 0.49 and thus slightly below the threshold of 0.50. However, this study retained all question items in the analysis. Conforming to the Fornell–Larcker criterion (1981), all square roots of the AVE are higher than the correspondent interfactor correlations and thus the measurement shows discriminant validity.

Study results

To verify hypotheses H₁ and H₂, three mediation analyses with multiple mediators were conducted based on Hayes' (2017) SPSS PROCESS 3.0 macro. To test hypothesis H₃, also based on Hayes' (2017) macro, moderated-mediation analyses were conducted. Before the mediation analyses were performed, the data were checked for outliers as well as heteroscedasticity and multicollinearity issues. The results of heteroscedasticity tests revealed problems when WOM and protest behaviour are considered as dependent variables. Thus, heteroscedasticity-consistent standard error estimators are used.

Mediation analysis with multiple mediators

In accordance with hypothesis H₁, the mediation analysis reveals a significant direct effect of affective response to CSI on resistance intentions (Boycott intention: r = 0.2780, S.E. = 0.0953, t-value = 2.92, p < 0.01, LLCI = 0.09 [0.12], ULCI = 0.47 [0.44]; WOM intention: r = 0.4314, S.E. [HC₄] = 0.0970, t-value = 4.45, p < 0.01, LLCI = 0.24 [0.27], ULCI = 0.62 [0.59]; Protest intention: r = 0.3256, S.E. [HC₄] = 0.1034, t-value = 3.15, p < 0.01, LLCI = 0.12 [0.16], ULCI = 0.53 [0.50]). In addition, Table 2 shows the 90% and 95% bootstrap confidence intervals for both the indirect effects and the total effect of affective response to CSI on boycott intention, WOM intention and protest intention. Regarding boycott intention, all indirect effects but one are significant, and pronounced mediation effects for moral equity and utilitarianism evaluations are observed. The indirect effect of the relativism dimension of ethical judgement is not significant for all types of consumer resistance intention. Considering WOM and protest intention, moral equity judgement does not mediate the effect of affective response to CSI. Contractualism appears as the strongest mediator for these two behavioural intentions, and yet contractualism is the weakest mediator of affective response’s effect on boycott intention.

Table 2. Total effect and indirect effects of affective response on consumer resistance intentions (cognitive channel).

Dependent variable	Mediator	Effect	Standard error	Boot LLCI (95%)	Boot ULCI (95%)	Boot LLCI (90%)	Boot ULCI (90%)
Boycott intention	Total effect	0.7953	0.0687	0.66	0.93	0.68	0.91
	Moral equity	0.2106	0.0836	0.04	0.37	0.07	0.34
	Relativism	0.0128	0.0401	−0.07	0.09	−0.05	0.08
	Contractualism	0.1371	0.0725	0.01	0.30	0.03	0.27
	Utilitarianism	0.1568	0.0499	0.07	0.26	0.08	0.24
Negative word of mouth intention	Total effect	0.7998	0.0919	0.62	0.98	0.65	0.95
	Moral equity	0.0875	0.0779	−0.08	0.23	−0.05	0.21
	Relativism	−0.0230	0.0312	−0.09	0.04	−0.08	0.03
	Contractualism	0.1642	0.0595	0.06	0.29	0.07	0.27
	Utilitarianism	0.1396	0.0370	0.07	0.22	0.08	0.20
Protest intention	Total effect	0.6810	0.0849	0.51	0.85	0.54	0.82
	Moral equity	0.0911	0.0826	−0.08	0.25	−0.05	0.22
	Relativism	−0.0178	0.0356	−0.09	0.05	−0.08	0.04
	Contractualism	0.1891	0.0681	0.07	0.33	0.08	0.31
	Utilitarianism	0.0930	0.0387	0.02	0.17	0.03	0.16

Confidence intervals printed in bold represent significant total and indirect effects. Standard errors for total effects of NWOM intention and protest intention are heteroscedasticity-consistent standard error estimators (HC₄); standard errors for indirect effects were estimated by means of bootstrapping. Bootstrap intervals are based on 10,000 bootstrap runs. Boot LLCI: lower-level bootstrap confidence interval; boot ULCI: upper-level bootstrap confidence interval.

In sum, hypotheses H₁ and H₂ are both confirmed. Consumer resistance intentions are thus triggered both directly and indirectly, but with varying degrees of intensity. When comparing direct and indirect effects, it becomes clear that whereas WOM appears to be more affectively driven, boycott and protest reactions are more strongly based on cognitive processes.

Moderated-mediation analysis

To validate hypothesis H₃, three moderated-mediation analyses were conducted. First, the results of the moderated regression models with the proposed mediators as dependent variables and consumer preferences for the ethical product as the moderating variable are reported in Table 3.

Table 3. Regression models with the dimensions of ethical judgement regressed on affective response, preference for ethical products, and affective response × preference for ethical products interaction.

Dependent variable	Independent variables	Coefficient	Standard error	t-Value	p-Value	Regression model assessment
Moral equity	Constant	0.0492	0.0288	1.71	<0.10	R^2 = 0.56; F(3, 492) = 210.55; p < 0.01
	AR	0.5418	0.0285	18.98	<0.01
	PREF	0.0027	0.0010	2.80	<0.01
	AR × PREF	−0.0060	0.0010	−6.19	<0.01
Relativism	Constant	0.0170	0.0629	0.27	ns	R^2 = 0.22; F(3, 492) = 46.28; p < 0.01
	AR	0.5799	0.0624	9.29	<0.01
	PREF	0.0057	0.0021	2.67	<0.01
	AR × PREF	−0.0021	0.0021	−0.98	ns
Contractualism	Constant	0.0303	0.0437	0.69	ns	R^2 = 0.39; F(3, 492) = 105.32; p < 0.01
	AR	0.6031	0.0433	13.91	<0.01
	PREF	0.0049	0.0015	3.26	<0.01
	AR × PREF	−0.0037	0.0015	−2.51	<0.05
Utilitarianism	Constant	0.0446	0.0618	0.72	ns	R^2 = 0.23; F(3, 492) = 49.39; p < 0.01
	AR	0.5011	0.0613	8.18	<0.01
	PREF	0.0089	0.0021	4.23	<0.01
	AR × PREF	−0.0055	0.0021	−2.62	<0.01

AR indicates affective response; PREF indicates preference for ethical products. All variables have been mean-centred.

Based on the significant effect of preference for ethical products, consumers with a low preference for ethical products were less critical in judging the alleged unethical behaviour of the chocolate companies. The ‘affective response’ × ‘preference for ethical products’ interaction has a significant negative effect on all dimensions of ethical judgement except relativism. Therefore, affective response to corporate misconduct has a more pronounced effect on the ethical judgement of the consumers with a low preference for ethical products. Hence, in line with hypothesis H₃, affective response to CSI does not have a strong effect on the relationship between preference for ethical products and ethical judgement of the consumers that are inclined towards, for example, fair-trade products. Therefore, this result indicates a more pronounced stability of ethical evaluations among the group of conscious consumers.

Second, resistance intentions are regressed on affective response, the dimensions of the extended MES scale, and five interaction terms (refer to Table 4). With regard to the directions of effects and the significances of the MES dimensions, the same pattern emerges as in the mediation analyses depicted previously (see Table 2). Contractualism and utilitarianism have a positive and direct effect on all three types of consumer resistance intention; moral equity, however, only affects boycott intention significantly. Both the effect of affective response on consumer resistance intentions and the regression coefficient of preference for ethical products are positive and significant. Thus, consumers with a low preference for ethical products and less pronounced felt affect have a lower willingness to resist unethical company behaviour. Interestingly, the regression coefficient of the preference for ethical products is lower and the regression constant is higher when the WOM intention is used as a dependent variable. Thus, WOM intention is not affected as much by an ethical product preference as boycott and protest intentions are. Further, the willingness to engage in WOM is generally higher than the inclination towards boycotting and protesting. This also becomes clear by comparing the means of the dependent variables (M_Boy = 4.7466; M_WOM = 5.3488; M_P = 4.3259). The ‘affective response’ × ‘preference for ethical products’ interaction has a significant negative effect on both boycott intention and protest intention. This result indicates that affective response to CSI has a stronger effect on boycott and protest intention for consumers with a low preference for ethical products, thus confirming hypothesis H_3b. Whereas preference for ethical products amplifies the relationship between moral equity and boycott intention, it solely strengthens the relationship between relativism and protest intention. Furthermore, and surprisingly, preference for ethical products dampens the relationship between utilitarian judgement and boycott intention as well as WOM intention.

Table 4. Consumer resistance regressed on affective response, mediators and preference (interactions).

Dependent variable	Independent variables	Coefficient		Standard error	t-Value	p-Value	Regression model assessment
Boycott intention	Constant	4.8154	0.0705		68.30	<0.01	R^2 = 0.39; F(11, 484) = 28.27; p < 0.01
	AR	0.1889	0.0921		2.05	<0.05
	ME	0.3438	0.1314		2.62	<0.01
	RE	0.0072	0.0571		0.13	ns
	CO	0.1742	0.0888		1.96	<0.10
	UT	0.2159	0.0577		3.74	<0.01
	PREF	0.0158	0.0024		6.52	<0.01
	AR × PREF	−0.0061	0.0031		−1.99	<0.05
	ME × PREF	0.0091	0.0044		2.05	<0.05
	RE × PREF	−0.0003	0.0019		−0.17	ns
	CO × PREF	0.0003	0.0028		0.11	ns
	UT × PREF	−0.0067	0.0019		−3.53	<0.01
Negative word of mouth intention	Constant	5.3949	0.0732		73.68	<0.01	R^2 = 0.45; F(11, 484) = 25.07; p < 0.01
	AR	0.3818	0.1020		3.74	<0.01
	ME	0.1071	0.2076		0.52	ns
	RE	−0.0469	0.0487		−0.96	ns
	CO	0.2218	0.1013		2.19	<0.05
	UT	0.2013	0.0518		3.89	<0.01
	PREF	0.0107	0.0027		4.04	<0.01
	AR × PREF	−0.0015	0.0036		−0.43	ns
	ME × PREF	0.0020	0.0067		0.30	ns
	RE × PREF	−0.0001	0.0017		−0.07	ns
	CO × PREF	−0.0014	0.0042		−0.34	ns
	UT × PREF	−0.0028	0.0016		−1.82	<0.10
Protest intention	Constant	4.3268	0.0698		61.98	<0.01	R^2 = 0.35; F(11, 484) = 19.10; p < 0.01
	AR	0.2383	0.0794		3.00	<0.01
	ME	0.1603	0.1582		1.01	ns
	RE	−0.0243	0.0546		−0.45	ns
	CO	0.2687	0.0929		2.89	<0.01
	UT	0.1006	0.0542		1.85	<0.10
	PREF	0.0156	0.0024		6.57	<0.01
	AR × PREF	−0.0056	0.0028		−2.02	<0.05
	ME × PREF	0.0026	0.0049		0.53	ns
	RE × PREF	0.0037	0.0018		2.06	<0.05
	CO × PREF	0.0008	0.0021		0.38	ns
	UT × PREF	−0.0013	0.0017		−0.78	ns

AR, ME, RE, CO, UT and PREF indicate affective response, moral equity, relativism, contractualism, utilitarianism and preference for ethical products, respectively. All variables have been mean-centred and heteroscedasticity-resistant standard errors (HC₄) were used to determine the significance for the NWOM and the protest intention model.

To provide deeper insights into the moderated-mediation mechanism, the conditional direct effects of affective response to CSI on boycott, WOM and protest intention are estimated for different moderator values. Table 5 presents the results of these calculations and reveals that affective response has a significant direct effect on the consumer resistance intentions of consumers with a low and moderate preference for ethical product alternatives when boycott and protest intention are considered as dependent variables. Hence, hypothesis H_3b is confirmed when boycott or protest intention are considered. By contrast, affect response’s direct effect on WOM intention is not moderated by preference for the ethical product alternative. However, in line with H_3b, at least the size of the conditional regression coefficients decreases with ethical product preference.

Table 5. Conditional direct effect of affective response on boycott intention, negative word of mouth intention, and protest intention (affective channel) at values of the moderator preference for ethical products.

Dependent variable	Moderator	Coefficient	Standard error	t-Value	p-Value	Boot LLCI (95%)	Boot ULCI (95%)	Boot LLCI (90%)	Boot ULCI (90%)
Boycott intention	m − s	0.3700	0.1218	3.04	<0.01	0.13	0.61	0.17	0.57
	m	0.1889	0.0921	2.05	<0.05	0.01	0.37	0.04	0.34
	m + s	0.0079	0.1365	0.06	ns	−0.26	0.28	−0.22	0.23
Negative word of mouth intention	m − s	0.4276	0.1306	3.27	<0.01	0.17	0.68	0.21	0.64
	m	0.3818	0.1020	3.74	<0.01	0.18	0.58	0.21	0.55
	m + s	0.3360	0.1613	2.08	<0.05	0.02	0.65	0.07	0.60
Protest intention	m − s	0.4043	0.1226	3.30	<0.01	0.16	0.65	0.20	0.61
	m	0.2383	0.0794	3.00	<0.01	0.08	0.39	0.11	0.37
	m + s	0.0724	0.1050	0.69	ns	−0.13	0.28	−0.10	0.25

m = average value of the moderator; s = standard deviation. Bootstrap intervals are based on 10,000 bootstrap runs and confidence intervals printed in bold represent significant direct effects. Standard errors for the negative WOM and the protest intention model are HC₄ standard errors. Boot LLCI: lower-level bootstrap confidence interval; boot ULCI: upper-level bootstrap confidence interval.

Table 6 reports the conditional indirect effects of affective response on the three types of consumer resistance intention. Consistent with hypothesis H_3a, moral equity and contractualism judgement mediate the effect of affective response on the boycott intention of consumers with a moderate and high preference for the ethical product. In line with hypothesis H_3a, as well, contractualism mediates the effect of affective response to CSI on protest intention of consumers with a moderate and high preference for the ethical product. Preference for ethical products, however, does not moderate the mediation effect of ethical judgement when WOM intention is considered as a dependent variable. In addition, the moderated-mediation analyses reveal a mediation effect of utilitarian judgement. However, the direction of the found mediation effects is contrary to the one hypothesized. More precisely, the indirect effects of affect response via utilitarianism on all three types of consumer resistance decrease with preference for ethical products.

Table 6. Conditional indirect effects of affective response on consumer resistance intentions (cognitive channel) at values of the moderator preference for ethical products.

Dependent variable	Mediator	Moderator	Effect	Standard error	Boot LLCI (95%)	Boot ULCI (95%)	Boot LLCI (90%)	Boot ULCI (90%)
Boycott intention	ME	m − s	0.0538	0.1225	−0.1825	0.2986	−0.1430	0.2609
		m	0.1863	0.0801	0.0328	0.3463	0.0569	0.3179
		m + s	0.2222	0.1052	0.0355	0.4507	0.0612	0.3979
	RE	m − s	0.0111	0.0557	−0.0899	0.1282	−0.0733	0.1098
		m	0.0042	0.0337	−0.0633	0.0716	−0.0503	0.0589
		m + s	−0.0015	0.0409	−0.0891	0.0778	−0.0719	0.0579
	CO	m − s	0.1179	0.0933	−0.0537	0.3131	−0.0319	0.2719
		m	0.1051	0.0657	0.0060	0.2635	0.0217	0.2333
		m + s	0.0903	0.0977	−0.0029	0.3693	0.0089	0.3142
	UT	m − s	0.2743	0.0634	0.1494	0.3961	0.1687	0.3748
		m	0.1082	0.0354	0.0423	0.1813	0.0528	0.1666
		m + s	0.0063	0.0278	−0.0527	0.0636	−0.0392	0.0503
Negative word of mouth intention	ME	m − s	0.0343	0.1135	−0.1882	0.2592	−0.1497	0.2271
		m	0.0580	0.0735	−0.0953	0.1916	−0.0754	0.1724
		m + s	0.0604	0.0815	−0.1471	0.1824	−0.1026	0.1605
	RE	m − s	−0.0276	0.0447	−0.1227	0.0548	−0.1052	0.0430
		m	−0.0272	0.0273	−0.0885	0.0205	−0.0775	0.0132
		m + s	−0.0263	0.0346	−0.1139	0.0251	−0.0943	0.0166
	CO	m − s	0.1882	0.0820	0.0207	0.3421	0.0428	0.3169
		m	0.1338	0.0531	0.0449	0.2508	0.0575	0.2304
		m + s	0.0886	0.0645	0.0145	0.2602	0.0241	0.2283
	UT	m − s	0.1890	0.0526	0.0951	0.3008	0.1097	0.2817
		m	0.1009	0.0308	0.0481	0.1690	0.0557	0.1553
		m + s	0.0399	0.0280	−0.0011	0.1063	0.0038	0.0907
Protest intention	ME	m − s	0.0602	0.1163	−0.1657	0.2938	−0.1244	0.2578
		m	0.0868	0.0708	−0.0494	0.2310	−0.0197	0.2042
		m + s	0.0859	0.0692	−0.0548	0.2221	−0.0263	0.1949
	RE	m − s	−0.0868	0.0515	−0.1916	0.0118	−0.1730	−0.0050
		m	−0.0141	0.0315	−0.0775	0.0458	−0.0672	0.0369
		m + s	0.0449	0.0406	−0.0216	0.1366	−0.0118	0.1194
	CO	m − s	0.1743	0.0914	−0.0012	0.3600	0.0280	0.3269
		m	0.1620	0.0623	0.0527	0.2960	0.0720	0.2725
		m + s	0.1445	0.0719	0.0375	0.3189	0.0534	0.2828
	UT	m − s	0.0928	0.0530	−0.0132	0.1954	0.0063	0.1799
		m	0.0504	0.0288	−0.0020	0.1108	0.0067	0.1004
		m + s	0.0207	0.0259	−0.0235	0.0825	−0.0151	0.0677

ME, RE, CO and UT, indicate moral equity, relativism, contractualism and utilitarianism, respectively; m = average value of the moderator; s = standard deviation. Bootstrap intervals are based on 10,000 bootstrap runs and confidence intervals printed in bold represent significant indirect effects. Boot LLCI: lower-level bootstrap confidence interval; boot ULCI: upper-level bootstrap confidence interval.

Discussion of study results

Based on the cognitive mediation theory and considering a dual-channel model, the present study sought to analyse whether affective response to CSI drives consumer resistance through a direct channel (affective channel) or an indirect channel (cognitive channel with ethical judgement as mediator). Mediation analyses with multiple mediators reveal that the aroused negative affect has an influence on all three types of consumer resistance via both channels. The found mediation effect of consumers’ ethical evaluations confirms the cognitive mediation theory that originates from communications research and political science within a consumer resistance context. Moreover, these mediational effects are in line with Haidt’s (2001) social intuitionist model. These findings are also consistent with Morwitz and Sen’s (2004) consumer-boycotting research, which shows a mediational effect of consumer attributions on the relationship between boycott communication and boycott willingness. Three of the four considered dimensions of ethical judgement mediate the effect of affective response on boycott intention. The finding that relativist ethical judgement does not act as a mediator may be due to its inherent nature. Because relativism is based on the philosophic viewpoint that universal ethical norms are non-existent (McMahon & Harvey, 2007), relativistic evaluation should be less rigid and thus less decisive for consumer behaviour due to its context-specific and situational nature.

The moderated-mediation analyses show that preference for ethical products moderates the ‘affective response-ethical judgement’ relationship, and this study largely confirms hypothesis H₃. First, in keeping with hypothesis H_3b, for consumers with a low (high) preference for the ethical product, the analysis shows that affective response has (does not have) a direct impact on resistance intentions. Hence, less ethically conscious consumers tend to be more driven by negative feelings due to current messages about corporate misconduct. Second, subjects ethically judge corporate misconduct independent of spontaneously aroused negative feelings when they already have pronounced ethical product preferences. This aligns with research on attitude strength (e.g. Prislin, 1996), which assumes that more information or knowledge about the attitude object makes attitudes less dependent on other factors.

Third, in accordance with hypothesis H_3a, the mediation effect of moral-equity and contractualism judgement is stronger for consumers with a higher preference for ethical products. Thus, the cognitive channel of causation appears to be more decisive for boycott and protest intention among this group of conscious consumers. In contrast to this finding, consumers’ ethical preference does not moderate the mediation effect of ethical judgement when WOM intention is considered as a dependent variable. A possible explanation for this result can be found in the differences between the types of consumer resistance considered in this study. WOM is not that strongly associated with factual actions and consequences in the consumer arena, as opposed to, for example, a boycott in which consumers must actually refrain from buying preferred products. Furthermore, according to Stern (1994), WOM is impulsive and spontaneous in nature. It is therefore not surprising that WOM as an immediate expression of individuals’ dissatisfaction is more strongly driven by the affective response to unethical firm behaviour. By contrast, it appears logical that for boycotting behaviour in particular, cognitive considerations play a central role for consumers with ethical preferences.

Moreover, the analysis reveals a moderating effect of preference for ethical products on the relationship between relativism and protest behaviour. Relativism has a stronger effect on protest inclination when ethical preference is high. Previous research revealed that relativists are more prepared to engage in unethical behaviour (Kish-Gephart, Harrison, & Treviño, 2010). Because protest behaviours can be classified to some degree as aggressive actions, conscious consumers apparently are more inclined towards them when they hold relativistic ethical viewpoints. This may be because consumers with high ethical preferences put less emphasis on compliance with universal standards (i.e. not harming others) when they hold relativistic viewpoints. They are more likely to participate in a protest action based on situational and contextual considerations.

Finally, utilitarianism evaluations mediate the effect of affective response on boycott intention and WOM intention. However, due to the moderation effect’s direction, the empirical results contradict hypothesis H_3a. Consumers with a low preference for ethical products put a stronger weight on the consequences of criticised corporate conduct. Contrary to this, the resistance behaviour of the more ethically conscious consumers is less contingent on utilitarian judgement. The group of more conscious consumers appears similar to Forsyth’s (1980) idealistic person who holds the opinion that no single person should be harmed and that no good purpose justifies unethical means. As less-conscious consumers place more emphasis on functional and utilitarian product characteristics rather than on ethical product augmentations (e.g. fair-trade agreements), it is unsurprising that these consumers’ resistance behaviour is more strongly guided by the teleological ethical assessment of the consequences of criticised corporate behaviour.

Practical implications, study limitations and future research

Because activist groups that call for consumer resistance often are non-profit organisations, they commonly depend on charitable donations (Helmig, Jegers, & Lapsley, 2004) and thus must carefully allocate their scarce resources in order to effectively fulfil their mission (Pallas, Fletcher, & Han, 2017). Based on the present study’s findings, it appears to be instrumental to send calls for action that induce affects in order to encourage consumer resistance behaviour. In particular, communication activities that target individuals with a low preference for ethical products should be emotionally charged. Campaigners could target this group of consumers using communication that includes drastic imagery of suffering entities that are harmed by corporate misconduct. Furthermore, considering the same consumer segment, campaign management should highlight the negative consequences of corporate misconduct. As the moderated-mediation analysis shows, utilitarianism as an ethical philosophy that is based on a consequentialist notion functions as a mediating variable; therefore, negative affect should be transferred to a greater extent into resistance intention when utilitarian judgement is distinctly unfavourable. Communication that conveys information indicating that the criticised corporation breaks an unspoken societal contract and harms the physical integrity of human and other living beings with their business practices should be presented to individuals who hold high ethical preferences. These campaign strategies should result in a more pronounced inclination to boycott as well as to engage in protest actions.

Because boycott targets have little to no influence on consumers’ affective response to campaigns against their allegedly unethical conduct, based on the results of the present study, they can only attempt to influence consumers’ ethical judgement to alleviate the effect of negative affect on consumer resistance intentions. Strategies that counter calls for consumer resistance should therefore try to influence consumers’ ethical judgement. Considering the segment of ethically conscious consumers, crisis communication could include messages that credibly show the need for the corporate behaviour and demonstrate that a corporation does not violate unwritten societal contracts. These communication strategies could improve moral equity and contractualistic judgement and subsequently attenuate the effect of negative affect on boycott and protest intention. By contrast, affected companies should explain to less ethically conscious consumers that their criticised behaviour may also have positive consequences. For example, they could highlight that it may be better to employ people in third-world countries at low wages instead of leaving them without work. In addition, the negative consequences of the protest actions (e.g. loss of domestic jobs) could be highlighted in crisis communication. The resulting more favourable utilitarian judgement should demotivate consumer resistance. Finally, manufacturers and retailers of ethical product alternatives could consider message appeals that activate moral equity norms, contractualism evaluations and in certain circumstances empathic affects if they plan to appeal to the segment of morally conscious consumers.

Any interpretation of this study’s results must consider its limitations. First, child labour and inhumane working conditions at Western African cocoa plantations as research objects represent specific subject areas. Thus, the study results may not exhibit external validity. Replication studies focusing on other types of corporate misconduct and ethical product substitutes could enhance the external validity of the present study’s findings. Second, this study applies a survey research design, and internal validity may thus be limited. Third, instead of focusing on actual behaviour, individual willingness to participate in consumer resistance behaviours has been considered. According to Sheeran (2002), individuals’ attitudes towards a behaviour do not perfectly translate into actual behaviour. In particular, research on ethical and environmentally friendly consumption behaviour reveals that consumers do not always ‘walk their talk’ (e.g. Carrington, Neville, & Whitwell, 2010). However, Godin, Conner and Sheeran (2005) show that the consideration of moral norms helps bridge the gap between behavioural intent and actual behaviour. Thus, the attitude-behaviour gap may not be as pronounced in this study because respondents’ ethical judgement is based on internalised moral norms.

This study provides several avenues for future research. First, the delineated model of consumer resistance behaviour is quite parsimonious and explains only a portion of the entire issue. The model can be either enhanced or modified, and future research could consider, for example, criticised companies’ corporate image as additional moderating variable. Furthermore and as another example, future research could consider sociodemographic variables as alternative explanatory or moderating variables. For instance, one could hypothesize that protesting behaviour is more pronounced among younger people rather than among older people. Second, it would be interesting to analyse whether other categories of consumer resistance such as avoidance and minimisation behaviour (Fournier, 1998), or prosocial behaviour such as volunteering can be explained with this paper’s model and whether the relationships remain stable in the light of these other areas of interest. In addition, future studies could analyse whether the delineated model is useful in explaining resistance to smart products (Mani & Chouk, 2017). Third, to increase the internal validity of the findings, future studies could consider experimental designs. These experimental studies could consider different consumer resistance appeals (e.g. emotional appeals versus moral appeals) and analyse whether they persuade consumers via an affective and/or cognitive channel. Fourth, to help close the attitude-behaviour gap, future tests of the delineated model should consider data on actual consumption behaviour (e.g. data obtained from retail scanner panels). Fifth, researchers could conduct longitudinal studies or consider repeated measure designs and analyse how repeated exposure to calls for action impacts resistance inclination over a longer period of time. For instance, these studies could investigate whether consumers with a high preference for ethical substitutes gradually develop reluctant attitudes towards the initiators’ consumer-resistance campaigns. Sixth, future research could consider consumers in other global settings to ferret out cultural norms that may contribute to the examined resistance behaviours and non-behaviours among various consumer groups. For instance, further research could investigate whether ‘cultural personalities’ are in place that encourage consumers to be more or less inclined to aggressive resistance behaviours (e.g. protests or boycotts) vs. more passive approaches (e.g. as negative WOM). Finally, further research could consider categories of products which satisfy more basic human needs than fair trade products. For example, access to clean drinking water is an issue in several parts of the world. Empirical studies could thus analyse how corporate irresponsibility in terms of polluting water sources affects the resistance behaviour of consumers that are directly affected by water pollution.

Acknowledgements

The authors acknowledge that this work was not funded by a grant from a third party.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

Stefan Trautwein

Stefan Trautwein is working at the Chair of Public and Nonprofit Management, University of Freiburg, Germany. Stefan holds a master degree in Public and Non-Profit Management from the University of Freiburg, Germany. His research interests lie in the fields of business ethics, corporate social responsibility, and environmental marketing. ORCID: 0000-0002-2464-819X.

Jörg Lindenmeier

Jörg Lindenmeier is a full professor at the Chair of Public and Nonprofit Management, University of Freiburg, Germany. His research is focused on consumer and prosocial behaviour. His research addresses ethical consumption, donation behaviour, and innovation acceptance. Jörg has published articles in the Journal of Business Research, Journal of Macromarketing, Nonprofit and Voluntary Sector Quarterly, Journal of Nonprofit and Public Sector Marketing, Tourism Management, and Transportation Research, Part A: Policy and Practice among others. ORCID: 0000-0001-6736-7067.

Appendices Appendix 1. Questionnaire

Question items and factor loadings (in brackets)

Table

Affective response to CSI (VIF = 2.275)
	How do you feel when thinking about the inhumane working conditions and child labour on cocoa plantations in Western Africa?
AR_1	Glad – angry (0.89/0.89/0.89)
AR_2	Happy – unhappy (0.89/0.88/0.89)
AR_3	Good – bad (0.90/0.90/0.90)
AR_4	Calm – anxious (0.90/0.90/0.90)
AR_5	Pleased – displeased (0.93/0.93/0.93)
AR_6	Not at all disgusted – disgusted (0.80/0.80/0.80)
Moral equity (VIF = 3.399)
	It is … that chocolate producers accept inhumane working conditions and child labour.
ME_1	Just – unjust (0.97/0.97/0.97)
ME_2	Fair – unfair (0.96/0.96/0.96)
ME_3	Acceptable to my family – not acceptable to my family (0.82/0.82/0.82)
ME_4	Morally right – not morally right (0.90/0.90/0.90)
Relativism (VIF = 1.723)
	It is … that chocolate producers tolerate inhumane working conditions and child labour.
RE_1	Culturally acceptable – culturally unacceptable (0.80/0.80/0.80)
RE_2	Always rather acceptable for most people – always rather unacceptable for most people (0.85/0.85/0.85)
RE_3	Always rather acceptable for my friends and acquaintances – always rather unacceptable for my friends and acquaintances (0.82/0.82/0.82)
Contractualism (VIF = 2.575)
	The behaviour of the criticised chocolate producers…
CO_1	Does not violate an unwritten contract – violates an unwritten contract (0.94/0.94/0.94)
CO_2	Does not violate an unspoken promise – violates an unspoken promise (0.94/0.94/0.94)
CO_3	Does not violate moral rules – violates moral rules (0.68/0.68/0.68)
Utilitarianism (VIF = 1.718)
	The behaviour of the criticised chocolate producers…
UT_1	Maximises benefits while minimises harm – minimises benefits while maximises harm (0.84/0.84/0.85)
UT_2	Leads to the greatest good for the greatest number – leads to the least good for the greatest number (0.85/0.85/0.85)
UT_3	Is rather advantageous to all – is rather disadvantageous to all (0.71/0.72/0.71)
Boycott intention
BI_1	I am boycotting the products of the criticised chocolate producers. (0.94)
BI_2	I am sure that I will actually boycott the products of the criticised chocolate producers. (0.94)
BI_3	I am not boycotting the products of the criticised chocolate producers. (0.84)
Negative word of mouth intention
	I can well imagine…
WOMI_1	Saying negative things about the chocolate manufacturers criticised to friends, relatives, and other people. (0.79)
WOMI_2	Recommending that my friends, relatives, and other people do not buy products of the chocolate manufacturers criticised. (0.87)
WOMI_3	Discrediting the chocolate manufacturers criticised with my friends, relatives, and other people. (0.88)
Protest intention
	Furthermore, I can well imagine…
PI_1	Participating in actions of resistance on the internet against the chocolate manufacturers criticised (e.g. blogging or posting on social networks). (0.74)
PI_2	Signing an online petition directed towards the chocolate manufacturers criticised, addressed to political decision makers in Berlin and Brussels (0.80)
PI_3	Sending money to an organisation that devoted itself to fighting abusive labour conditions and child work on cocoa plantations in Western Africa. (0.53)
Preference for ethical products (VIF = 1.183)
	What is the share of fair-trade chocolate when you buy chocolate products?
PREF	My personal share of fair trade chocolate is … %.

The factor loadings are reported in an ordered sequence and, in each case, refer to the boycott model, the negative WOM model, and the protest model.

Appendix 2. Descriptive statistics, reliability statistics, and validity statistics

Table

	Mean value	Standard deviation	Cronbach’s alpha	Composite reliability	Average variance extracted	Fornell–Larcker ratio
Affective response to CSI	6.3061	1.0841	0.95	0.96	0.77	0.81
Moral equity	6.5726	0.9284	0.95	0.95	0.83	0.77
Relativism	5.5780	1.5206	0.86	0.86	0.67	0.79
Contractualism	6.1714	1.1956	0.88	0.89	0.74	0.77
Utilitarianism	5.4422	1.5044	0.83	0.85	0.66	0.78