Introduction

Consumers consider a variety of factors when deciding where to allocate their discretionary income, such as price, quality, product features, and benefits. One area that is becoming important in the decision-making of consumers is the ethical behaviour of organisations. Consumers have demonstrated an awareness of issues that may affect their consumption behaviour. Typical issues include the use of child labour in developing countries, the global environmental impact of production (and consumption), animal testing and welfare, employment labour standards, human rights, use of child labour, and health and safety issues (Racela 2012).

While traditional forms of punishment have involved calls for boycotts and protests, the prevalence of the internet and social media has created new ways for consumers to communicate and express their judgements regarding the behaviours of firms (Kozinets 2002; Li 2019). Forms of social media such as Facebook and Twitter provide a fast and easy way for a consumer to express dissatisfaction about a firm or attempt to share negative word of mouth.

The internet has facilitated consumer punishing behaviours, particularly in terms of attempts to damage reputation or gather support through electronic media, such as online discussion forums, electronic bulletin board systems, newsgroups, blogs, review sites, and social networking sites (Goldsmith 2006; Li 2019). A body of literature has demonstrated how third parties react to injustices committed against others by attempting to restore justice (e.g. Skarlicki and Kulik 2005). Building upon this literature, we examine the behaviour of consumers who undertake punitive action towards the company, despite the fact they may not have purchased goods and services from the organisation.

A number of untested conceptual models (O'Reilly and Aquino 2011; Skarlicki and Kulik 2005) and some partial empirical models (Gregoire and Fisher 2006, 2008) have been developed in the literature. The model draws together a number of previous hypotheses which have only been partially examined. In part the model examines the determinants of consumer punishment behaviours by examining the potential impact of key constructs including perceived injustice, consumer-company identification, identification with the victim, and consumer involvement. The impact of consumer involvement on punishing behaviours has not previously been tested. A number of ‘real-life’ ethical scenarios are employed, and structural equation modelling (SEM) is employed to test the posited hypotheses. The scenarios relate to actions of multinational organisations which have caused either physical, economic, or psychological harm. The approach assesses consumers’ retaliations to these adverse events as third parties.

This paper makes a number of contributions to our understanding of how consumers as third parties may punish firms for perceived injustice towards other stakeholders. Most organisational justice research has focussed on the justice evaluations of individuals directly involved in justice situations. Yet, researchers have already acknowledged the role of individuals as observers or third parties in others’ justice situations (e.g. Lind et al. 1998). As suggested by Miles and Naumann (2003), this study has controlled for the identity of the third party that being consumers. This study contributes to the literature by showing that consumers as third parties can and will make judgements regarding justice situations and subsequently act to restore justice between the parties.

A contribution has been made through the use of the consumer-company identification scale developed by Hildebrand et al. (2010). This study provided support for the use of this scale in the consumer setting and suggests that the scale has good indicators of reliability and validity as a six-item construct. This is important, as the existing scales in the literature seem better suited to organisational identification from within the firm, rather than how a consumer identifies with a company. The scale of Hildebrand et al. (2010) was not employed in other studies, so this study provides support for the use of this scale in future research tapping into the consumer-company identification construct. Further, the notion that the consumer-company identification relation with punishment may be mediated by perceived injustice is examined.

Moreover, a contribution is made to the literature through the use of scenarios as the method of data collection in this study. The literature indicates that scenarios are suitable methods for use in a range of disciplines and the importance of creating a sense of realism in scenarios should be considered (Alexander and Becker 1978; Randall and Gibson 1990; Weber 1992). Existing scenarios available in the literature are often limited to particular contexts, and there was need to create scenarios that were specific for the research issues under investigation in this study. This study also provides support for the use of scenarios as an appropriate means of tapping into consumers’ judgements and decision-making.

The rest of the paper is organised as follows. In the next section, we present the context of the paper, then the conceptual framework for our study by describing the key variables, and their hypothesised relationships. This is followed by an outline of the methods employed and analysis of results. The paper concludes with a discussion of the findings and their implications.

Context

Consumer involvement in the marketplace has evolved from a somewhat passive role of buying and consuming to having the potential for increased engagement with firms and other consumers. The potential impact of consumers has become a greater consideration of firms, who recognise the potential for consumers to attempt to punish for ethically questionable business incidents. Significant displays of consumer punishment have involved a number of firms over recent decades, including Nike, Shell, Nestlé, Proctor and Gamble, Apple, Volkswagen, and many others.

Research into consumer behaviour has branched into areas such as ethical consumerism, marketing ethics, socially responsible consumption, and consumer ethics (Johnston 2008). Research has found that some consumers are willing to pay more for socially responsible goods, while others are more likely to boycott those they view as being unethical or irresponsible (Auger et al. 2003).

This paper takes the perspective of the consumer as a third-party observer in an ethically questionable situation. Of interest is how they will make judgements which will possibly lead to punishing behaviours. Research has demonstrated that individuals can become outraged and motivated to restore justice on behalf of others (Fehr and Fischbacher 2004a; Folger 2001). This study specifically focusses on the relationship between consumers, organisations, and stakeholders. The third-party perspective of the consumer is important and distinct from other stakeholder groups as the consumer is typically considered to be self-interested and motivated by personal gain. Drawing predominantly on theories of social identity and organisational identification, injustice, retributive justice, and consumer behaviour, the following conceptual framework builds on existing knowledge.

Conceptual framework

Punishment behaviours in ethical situations can be represented in a variety of responses from consumers, such as boycotting, lobbying, spreading negative word of mouth, vandalism, stealing, personal attacks, computer-mediated communication with the firm or other current or prospective consumers, complaining to third parties, and so on (Goldsmith 2006).

Research has indicated that when a third party perceives an injustice to have occurred, they may respond in a variety of ways (Greenberg 2001). Some might try to raise awareness of the injustice and attempt to gain support in acting against the organisation. Others might attempt to punish the organisation on their own (e.g. Turillo et al. 2002). It is also possible that third parties will do nothing, even when they perceive that an injustice has occurred (O'Reilly and Aquino 2011).

Punishment and punishing behaviours

‘Punishment’ is defined as ‘a negative sanction intentionally applied to someone [an individual or organisation] who is perceived to have violated a law, a rule, a norm, or an expectation’ (Lerner and Lerner 1981, p. 146). Punishing behaviours (Gregoire and Fisher 2006) should be broadly interpreted to include any behaviour that the individual considers to cause some form of harm or negative impact, with the intention to restore balance or justice. It may take the form of a deprivation or an unpleasant experience, either of which may be physical, social, or psychological. This definition is robust enough to incorporate the dynamics involved in both legal punishment reactions and in those reactions evoked in most non-legal settings.

Consumer-company identification and punishing behaviours

Research into the relationships between consumers and organisations has progressed through to the concept of consumer-company identification. Consumers can identify with organisations that they believe they share common traits with and that provide a sense of self-enhancement (Ahearne et al. 2005; Bhattacharya et al. 1995; Sen and Bhattacharya 2001). Bhattacharya and Sen (2003, p. 76), defined consumer-company identification (C-C identification) as the ‘primary psychological substrate for the kind of deep, committed and meaningful relationships that marketers are increasingly trying to build with their customers’. Drawing on theories of social identity and organisational identification, Bhattacharya and Sen (2003) and Ahearne et al. (2005) have provided a comprehensive and logical expression of both the conditions in which consumers are likely to identify or feel a sense of belonging with a company and the bases of such identification. Bhattacharya and Sen (2003) conceptualise consumers’ identification with a company as an active, selective, and volitional act motivated by the satisfaction of one or more self-definitional needs.

Consumer-company identification is considered to impact on the use punishing behaviours. Research has indicated that individuals can develop a relationship and feel a sense of connectedness to an organisation (Bhattacharya et al. 1995; Mael and Ashforth 1992; Sen and Bhattacharya 2001). At the heart of these relationships between the consumer and organisation is the sense of commitment and shared values, which lend themselves to ongoing interaction (Gregoire and Fisher 2006). As a consumer ascribing to high levels of consumer-company identification wishes to maintain their relationship with an organisation, as a means of preserving the organisation as a sense of positive identity and self-esteem (Gregoire and Fisher 2008), the existence of negative information may have varying effects. Such that when strongly identified customers’ positive beliefs about an organisation are challenged by negative information, they are likely to try to maintain a positive association with the organisation (Bhattacharya and Sen 2003), by dismissing or downplaying the negative information (Ahluwalia et al. 2000) or judging them less harshly.

Just as identification can insulate an organisation from negative information regarding responsibility, it is possible that it can insulate an organisation from punishment. In order to maintain the relationship with the organisation, the consumer may choose not to punish. Conversely a customer with a high level of identification with a firm may feel more betrayed by the negative behaviours of an organisation, leading them to retaliate to a greater extent (Gregoire and Fisher 2008). Therefore, depending on the situation, consumer-company identification may actually protect the organisation from punishment or result in the consumer punishing to a greater degree. This study argues that the influence of consumer-company identification on the intention to use punishing behaviours can be either positive or negative. Stated more formally:

H1 Consumer-company identification will significantly impact the use of punishing behaviours against the firm.

Identification with victim and punishing behaviours

The degree to which a third-party observer identifies with the victim in a situation is proposed to influence the likelihood of the third party engaging in punishing behaviours. Perceptions that the victim possesses similar attributes to the third party can arouse self-concern, particularly when similar attributes are thought to have played a role in the mistreatment. From the perspective of social identity, the third party may feel a sense of affection for the victim (Brockner and Greenberg 1990) and be motivated to punish on behalf of the victim. From a psychological perspective, the observer of the harm may put themselves into the position of the victim and react as a surrogate victim (Vidmar 2000). Alternatively, Solomon (1990) argues that compassion for victims is a ‘natural justice’ reaction that may have origins in humans’ socio-biological nature. In contrast, if the third party does not feel a sense of identification with the victim, they will not be likely to expend their own resources to punish the firm. Stated more formally:

H2 The higher the identification with the victim, the higher the use of punishing behaviours against the firm.

Perceived injustice and punishing behaviours

It is apparent that many different types of responses to injustice are possible, both behavioural and psychological (O'Reilly and Aquino 2011). However, it still remains unclear which particular behavioural expressions of injustice will manifest and when they will do so. For example, although Tyler et al. (1997) described some of the conditions under which people will respond individually (e.g. by showing personal vengeance) or collectively (e.g. by engaging in riots), it remains difficult to predict exactly what form a response might take (Greenberg 2001).

The perceptual and social nature of injustice makes it difficult to predict exactly how and when people will respond to apparent injustices. Responses to injustice are difficult to predict as many of these responses are undertaken individually and in private. It is possible that an individual will do nothing regarding a perceived injustice. This may be because the costs of responding are believed to be too high, especially in the case of third-party observers. Inaction also may occur because the individual has worked out the injustice by distorting the facts cognitively or by forgiving the source of the injustice for inflicting harm. Assuming that consumers as third-party observers even recognise injustices, their willingness to take action and punish the organisation may be overshadowed by other concerns.

When an individual observes an injustice, they are considered often to feel the intuitive-affective impulse to punish the offender (Carlsmith and Darley 2008), meaning that retribution for injustice is considered by some as a natural instinct. Even if they are aware that punishment will be costly (Fehr and Fischbacher 2004b) and has limited ability to prevent future offences (Tyler 2006), they may still feel the need to punish the offender (Carlsmith et al. 2002). As observers, seeing the offender get their ‘just deserts’ is considered to be psychologically satisfying. It facilitates the psychological need to symbolically reassert the rules of society (Vidmar 2000), standards that are important for defining social bonds, maintaining a just world-view, and reducing subjective uncertainty (Okimoto and Wenzel 2011).

In an organisational context, O'Reilly and Aquino (2011) suggest that third-party attempts to punish perpetrators can be conceptualised as a form of vigilante behaviour if the third parties decide to take the matter of achieving justice into their own hands rather than waiting on formal mechanisms to do so. Similarly, altruistic punishment refers to the theory that humans invest their own resources to redress norm violations without self-interest involved (Lotz et al. 2011).

Building upon the previous discussion, we argue that even as a third-party observer, a consumer will desire to punish an organisation for an injustice against another stakeholder in an attempt to restore justice. In sum when the perceived injustice in a situation is high, the consumer will have a higher willingness or desire to punish the organisation. Stated more formally:

H3 The higher the perception of injustice, the higher the use of punishing behaviours against the firm.

Consumer involvement and punishing behaviours

Research has suggested that consumers will care more about some issues than others (Auger et al. 2010). According to Uusitalo and Oksanen (2004), consumers may find it difficult to take a number of ethical concerns into consideration simultaneously and therefore become selective in the ethical judgements that they oppose. Auger et al. (2010) found that consumers placed different levels of importance on different social attributes. For instance, an individual who values environmental issues does not necessarily value labour issues and vice versa.

Involvement in the social issue is considered to affect a consumer’s punishing behaviours. In psychological terms, involvement represents the degree to which an issue is relevant to an individual (Petty and Cacioppo 1990). Highly involved individuals feel more connected to specific social issues and are more motivated to seek out information about specific issues and even take action with regard to those issues (Golob et al. 2008).

Building upon this logic, we argue that a consumer that feels highly involved in a social issue will seek to punish the organisation for engaging in ethically questionable behaviours resulting in negative outcomes. A consumer that is not involved in an issue is not likely to use their own resources to punish the firm. Stated more formally:

H4 The higher the consumer involvement in the issue, the higher the use of punishing behaviours against the firm.

Consumer-company identification and perceived injustice

Beyond the determinants of punishing behaviours, the relation between consumer-company identification and perceived injustice warrants attention. Some literature suggests that individuals make judgements about fairness and injustice based on the negative impact of the event and whether an individual (or organisation) is accountable and responsible for the injustice (Folger and Cropanzano 1998, 2001). There is a tendency to try to assign blame or responsibility for the outcome. Skarlicki and Kulik (2005) and O'Reilly and Aquino (2011) both propose that third-party observers to potentially unjust situations will try to somehow make sense of a situation and make attributions about whether the outcome or the situation is somehow justified.

A widely used theory to explain third-party justice motives is Lerner’s (1980) belief in a just world. This posits that most people need to believe that they live in a world where people get what they deserve and deserve what they get. Becoming aware of a possible experience of injustice poses a threat to the just world belief and motivates the third party to restore justice either behaviourally or cognitively.

When a consumer with a high level of consumer-company identification becomes aware of a possible injustice situation, there is the possibility that the customer may seek to preserve the relationship with the organisation and their sense of positive identity by downplaying their perception of the company’s responsibility or blame in the situation and therefore downplaying their perception of injustice. In the process of trying to make sense of a potential injustice (O'Reilly and Aquino 2011), they may distort the facts, downplay the severity of harm caused to the victim, or assign blame and responsibility away from the company, therefore reducing the level of perceived injustice in a situation. This discussion leads to the hypothesis that:

H5 The higher the consumer-company identification the lower the perception of injustice.

An overall summary of the conceptual framework and hypothesised relations is provided in Fig. 1.

Fig. 1
figure 1

Conceptual framework and hypotheses

Full size image

Methods

Measures

A combination of previously validated multi- and single-item measures of the constructs is suitably adapted for the study. For a number of constructs, a single attribute that is easily, concretely, and uniformly imagined exists (Rossiter 2002), which validates the use of a single-item measure. In many cases, alternative measures of a construct are evaluated to facilitate a rigorous examination of the posited hypotheses.

In order to measure the key dependent variable of interest (punishment behaviours), we considered the series of measures developed and used by Gregoire and Fisher (2006), which permits respondents to identify/choose different punishment behaviours. Our focus is with the measures of patronage reduction and negative word of mouth. These two reflective measures were demonstrated to have useful reliability and validity properties. We also considered other measures such as the willingness to punish (Creyer and Ross 1997), desire for retaliation and third party complaining (Gregoire and Fisher 2006), and a measure of punishment behavioural intention as articulated by O'Reilly and Aquino (2011). These latter measures were found to be empirically inferior to the measures of patronage reduction and negative word of mouth for punishment behaviours.

To operationalise consumer-company identification, two alternative measures are considered. Even though a number of related measures such as organisational identification (Mael and Ashforth 1992; Marin et al. 2009) exist, these do not capture ‘unofficial memberships’. To capture the nature of the relationship between the consumer and an organisation, we considered the multi-item scale of Hildebrand et al. (2010), and the single-item measure developed by Sen et al. (2006) was also assessed. The former multi-item measure produced superior results.

Two measures of identification with victim are considered. Skarlicki and Kulik (2005) suggest a focus on ‘similarities of experiences’, while Skarlicki et al. (2012) focus on relationships with the situation. The latter performed best empirically. To capture perceived injustice, two alternative measures were assessed. A multi-item measure based on the conceptualisation of O'Reilly and Aquino (2011) that relates to a third party’s reasoned judgement of injustice was assessed. An alternative single-item measure developed by Mikula et al. (1998) was also examined as argument exists to suggest that perceived injustice is easily and uniformly imagined. The single-item measure performed best empirically. Consumer involvement in social issues is operationalised using the single item developed by Grunig (1979) as refined by Golob et al. (2008). The item reflects the clarity and concreteness of the construct.

Survey instrument

The study utilised scenarios to elicit responses of attitudes and judgements of research participants towards ethically questionable firm behaviours. According to a review by O’Fallon and Butterfield (O'Fallon and Butterfield 2005, p. 403), ‘scenarios remain the most widely used method of assessing constructs in business ethics research’. Alexander and Becker (1978, p. 103) stated that the use of scenarios ‘helps to standardize the social stimulus across respondents and at the same time makes the decision-making situation more real’. Eight scenarios were developed specifically for this study, and they are available from the authors on request.

In developing scenarios, the methodology suggested by Fredrickson (1986) was followed. First, scenario issues and content were sought, then the actual scenarios were developed using appropriate terminology and with sufficient detail, and finally a pilot test involving 32 respondents was conducted to ensure that the scenarios and accompanying questions were understandable and tapped the relevant issues and constructs. Each scenario was written following the same basic format: (1) identification of firm and situation, (2) the issue—identify the type of harm and ethically questionable behaviour, and (3) firm’s response/any actions taken. It is important that the scenarios have a sense of realism (Randall and Gibson 1990), so that consumers can feel a sense of connection. For ethical and legal reasons, the scenarios were based on publicly available information, rather than being hypothetical scenarios about real firms.

In selecting the content for the scenarios, attention was given to the organisation and the type of ethically questionable situation. It was important to select organisations which respondents could feel a sense of identification, in order to assess the impact of consumer-company identification. The chosen organisations include high profile multinational companies. A number of different types of harm were considered in the scenarios. It was important to find situations reflecting physical, economic, and psychological harm (e.g. Stein and Ahmad 2009). The degree of harm was also manipulated by identifying a ‘high’ and ‘low’ scenario for each type of harm. The literature indicates that the type of harm (physical, economic, or psychological) and the magnitude of that harm (high or low) are likely to influence an individual’s ethical decision-making (Collins 1989; Ingram et al. 2005; Stein and Ahmad 2009). For example, a situation involving animal testing is likely to be considered differently than a situation involving possible negligence that led to employee deaths. A schema for the types of harm described in the scenarios is provided in Table 1.

Table 1 Scenarios: types of harm and organisations
Full size table

The nature of the study requires responses from a single respondent, and such issues of common method bias impacting on results may arise (Podsakoff et al. 2003). A number of research design features were implemented to reduce the potential impact of any common method bias. Respondents were advised that only their attitudes and behaviours were being sought, they were assured of their anonymity when answering the questions, and the measures employed have been extensively validated in previous studies. These procedures reduce respondents’ evaluation apprehension and any comprehension ambiguity. Feedback from the pilot test indicated that responses to consumer-company identification items could be impacted by the information contained in the scenario. As a consequence, questions regarding consumer-company identification preceded the scenario content in the final questionnaire. This modification reduces biases emerging from the order of questions.

Data collection

A self-administered online survey was employed. The design of the questionnaire was adapted by following the principles for constructing web surveys as advocated in the literature (Dillman and Bowker 2001). The Online Research Unit (ORU) in Australia was employed to collect the data. The ORU is a leading online market research panel company operating in Australia and New Zealand, which has achieved the QSOAP (Quality Standard for Online Access Panels) ‘Gold Standard’ level of accreditation and also holds ISO 20252 and ISO 26362 certifications.

All ORU panels are recruited primarily offline and by an ‘invitation only’ basis ensuring a representative sample not only measured by demographics but also behavioural, attitudinal, and lifestyle criteria. The sampling frame consists of Australians over the age of 18 who are registered with the ORU, approximately 300,000 respondents. It was deemed that a target sample of 150 responses per scenario would be necessary in order to undertake appropriate analysis. Given the length of the scenarios and anticipated time taken to complete the questionnaire, each respondent was given two scenarios. Removing cases for which 30% (or more) of observations were missing (113 cases, these reflect ‘do not know’ responses) and identifying outliers (10 cases) using Mahalanobis distance statistic (Hair et al. 2010), the total number of observations for analysis is 1165 and ranges from 136 (scenario E) to 157 (scenario G).

To examine any common method data collection bias, we employed Harman’s single-factor test and used both principal components analysis (PCA) and confirmatory factor analysis (CFA) for all the measures to check for the goodness of fit of a single factor. Results imply that a single factor poorly reflects the data indicating the possible absence of any common method bias in collecting the data.

Stratified random sampling was employed, and therefore, the sample is representative of the strata relating age, gender, and location for the ORU sampling frame. The key characteristics of the sample include 53.1% female, 37.0% less than 40 years of age, 59.5% reside in metropolitan areas, and 43.7% having a university qualification. This profile is also indicative of the individual scenarios.

Analysis and results

Initially, the measurement model combining the various constructs is analysed and refined using CFA, and then the complete structural equation model with refined measures is estimated. Given the existence of some missing data, we employ the model-based full information maximum likelihood procedure (FIML) for estimating models with missing observations available in AMOS. In terms of model identification, the SEM is recursive; however, the existence of single-item measures poses some identification issues (Hair et al. 2010). Even though some constructs are envisaged as concretely and uniformly imaged, the existence of some associated measurement error when operationalising the construct using a single item cannot be totally ignored. As a consequence, we fixed the measurement error variance of single items employing 0.8 as a reliability estimate; this is broadly consistent with the estimated reliabilities in previous related studies. As previously indicated, in general, the single-item measures performed better than multi-item measures. A list of the employed items for the preferred measures is provided in the Appendix.

For the collected sample, the measures of patronage reduction and negative word of mouth for punishing behaviours did not empirically discriminate. A very high correlation of 0.853 was estimated. As a consequence, punishment behaviours employ a measure of seven items reflecting both patronage reduction and negative word of mouth. This measurement refinement results in a highly unidimensional measure (all factor loadings greater than 0.8, coefficient alpha = 0.970, and average variance extracted (AVE) = 0.789). Effectively punishment behaviours consist of a series of slightly different patronage reduction and negative word of mouth retaliations which are used in similar degrees. This measure adequately captures the degree to which similar retaliatory behaviours are employed.

The construct correlations, reliabilities, and summary statistics of the measures are presented in Table 2. All (coefficient alpha) reliabilities exceed 0.80 and all AVEs exceed 0.50. Once the unidimensional nature of patronage reduction and negative word of mouth for punishing behaviours is recognised, the refined measures exhibit good discriminant validity properties as for all constructs the average variance extracted value exceeds the respective squared correlations with other constructs (Hair et al. 2010). Overall the preferred measures possess very good reliability and validity properties.

Table 2 Construct correlations and summary statistics
Full size table

For the full sample, the estimated structural model had acceptable fit statistics: χ2 (df = 133) = 414.9, χ2/df = 3.12, CFI = 0.987, and RMSEA = 0.043. Acceptable fit indices, in general, exist for the individual scenarios: the χ2/df ratios vary from 1.49 to 2.26 and average 1.79; CFI’s vary from 0.93 to 0.98 and average 0.96; and the RMSEA’s vary from 0.056 to 0.095 and average 0.073. Scenario G best fits the data and scenario H provided the poorest fit. The chi-square difference tests rejected all parameter restrictions across groups and could not reject the totally unconstrained multiple group model. This result provides support for a detailed examination of the eight scenarios. The structural equation model estimates are provided in Table 3.

Table 3 Standardised path estimates of structural model
Full size table

In general there is varying support for the hypotheses across the full sample and individual scenarios. As expected more hypotheses are supported for the full sample. When using the full sample, all hypotheses are supported except for H4 that relates to the impact of consumer involvement in social issues on punishment behaviours. Across the scenarios, most support exists for H3 and the impact of perceived injustice, being supported in five of the eight scenarios. Least support exists for hypothesis H4 and consumer involvement in the issue not being supported in any scenario. A number of the control variables were also statistically significant for explaining punishing behaviours.

As a robustness check, we examined the possibility of variables moderating the strongest relationship between perceived injustice and punishing behaviours. Using the latent moderated structural equations (LMS) method for latent interactions (Klein and Moosbrugger 2000), we tested for the existence of customer-company identification, identification with the victim, and consumer involvement in the issue, moderating the relation between perceived injustice and punishing behaviours. For the eight scenarios and three potential moderators (24 cases), only in one case (customer-company identification for scenario G) did a statistically significant negative moderator impact on the perceived injustice and punishing behaviours relationship. The average absolute t ratios for the moderators were as follows: customer-company identification (0.95), identification with the victim (1.04), and consumer involvement in the issue (0.82). These results indicate the absence of any important moderating effects for the model.

Discussion

Consumer-company identification and punishment

An examination of the statistical significance and parameter estimates indicated that consumer-company identification has an effect on punishment behaviours in the full sample and three scenarios. Our theoretical discussion of H1 implied that either positive or negative relations may exist. A significant positive relationship was estimated for Nestlé (β = 0.160, p < 0.05). A significant negative relationship was found for the full sample (β = − 0.058, p < 0.10), Apple 1 (β = − 0.191, p < 0.10), and Apple 2 (β = − 0.299, p < 0.01).

The significant negative relationship for the full sample and the Apple scenarios indicates that the more the consumer identifies with the firm, the less willing they would be to punish them. Apple products are used on a daily basis and consumers develop a particularly close relation with their services. Thus it seems intuitive given consumers’ with a high level of identification and commitment with Apple desires to maintain a good relationship with the company (Brown et al. 2005). Therefore, the more a consumer identifies with the firm, the less likely they are to punish the firm.

Gregoire and Fisher (2006) described this as a ‘love is blind’ effect and found in their study that customers with a strong relationship with a company had a very low desire to retaliate compared with customers with a weak relationship. Gregoire and Fisher (2006) also connect this relationship to the perceived controllability of the firm, in that low controllability or low attribution of responsibility means that customers that identify highly with the company are not going to consider punishment as a viable option.

The significant positive relationship found for Nestlé indicates that the more a consumer identifies with the firm, the more willing they are to punish the firm. Unlike Apple products, Nestlé produces foodstuffs which are consumed typically on an infrequent basis. As such the nature of the relationship with the company is less personal. This nature of this relationship implies that consumers’ may feel a sense of ‘betrayal’ or reach a breaking point at which the negative perception of the company makes the consumer want to punish them for acting outside of their expectations, or negatively impacting their sense of self. Gregoire and Fisher (2006) considered a similar relationship which they termed ‘love becomes hate’, although Gregoire and Fisher (2006) did not find support for this relationship in their study. Instead, they found that regardless of the strength of the relationship the consumer has with the firm, if the firm is considered to be in control of the outcomes or otherwise responsible, the consumer will seek to retaliate. In contrast to our study, in Gregoire and Fisher (2006), the customer is the victim. Therefore, the customer has more at stake and is personally impacted, unlike this study in which the customer is a third-party observer.

Identification with victim and punishment

Table 3 indicates that the identification with the victim has an effect on punishment behaviours in the full sample and three scenarios. All identified statistically significant estimates are positive and consistent with the posited hypothesis H2. A significant positive relationship was estimated for full sample (β = 0.111, p < 0.01), Nestlé (β = 0.177, p < 0.05), and Apple 2 (β = 0.272, p < 0.01).

Interestingly, Nestlé, and Apple 2 relate to psychological harm (child labour and poor working conditions) and include individuals as the victims. The more that a consumer can relate to the victim through perceived similarities, or the higher the feeling of affection or liking of a victim (Brockner and Greenberg 1990), the more motivated they will be to punish on behalf of the victim. It is likely that consumers can more easily relate to employees as victims and therefore feel more compelled to punish.

It appears that in the situations where the victims are a firm (such as reputational damage), the consumer is less likely to relate to them, and therefore, their desire to punish the firm is not influenced by their relationship or identification with the victim.

Perceived injustice and punishment

The positive influence of perceived injustice on punishing behaviours was the most supported (H3) hypothesis in this study. It was found that perceived injustice had a significant positive effect on punishing behaviours in the full data set (β = 0.483, p < 0.01) and also in five of the eight scenarios. A strong positive relationship was estimated for the following scenarios: L’Oreal (β = 0.633, p < 0.01), British Petroleum 1 (β = 0.452, p < 0.01), Pacific Brands (β = 0.709, p < 0.01), British Petroleum 2 (β = 0.368, p < 0.01), and Nestlé (β = 0.717, p < 0.01).

The finding of this hypothesis provides support to the literature that in situations where the consumer as a third party has judged a situation as unjust, they are more willing to punish the organisation. That is, the consumers show a willingness to take the matter of achieving justice into their own hands, in a loose form of vigilante behaviour (O’Reilly and Aquino, 2011), or to symbolically reassert the rules of society (Vidmar 2000). Even as a third party, a consumer can seek or desire to restore justice through punishing an offender.

Perceived injustice was not found to have a significant relationship with punishing behaviours in the Apple1 (β = 0.118, p > 0.10), Qantas (β = 0.096, p > 0.10), and Apple 2 (β = 0.185, p > 0.10) scenarios. These three scenarios are characterised by having two firms involved in the situation. Also, two of the scenarios, Apple1 and Qantas, are characterised as containing economic harm through reputational damage.

Consumer involvement in social issue and punishment

The results of this study indicated that consumer involvement in social issue had no significant positive relationship on punishing behaviours for the full sample and for any of the individual scenarios. Therefore, hypothesis H4 has no support.

According to Golob et al. (2008), highly involved individuals feel more connected to specific social issues and are more motivated to allocate resources to processing information about those issues and even take action with regard for those issues.

The social issues represented in the non-significant scenarios include workplace fatalities or injuries, moving domestic production offshore, reputational damage, child labour, labour conditions, and environmental damage. It is somewhat surprising that for these scenarios, the level of involvement in the social issue does not impact punishing behaviours. This may be due in part to the perceived seriousness of the situations, whereby the desire to punish is not related to whether or not a consumer has a specific interest in the ethical issue but rather to the absolute severity of the harm. This finding also suggests that a consumer’s involvement in the issue by itself is insufficient to activate retaliatory behaviour and that the perceived injustice and identification with the victim may are more important drivers of punishing behaviours.

Consumer-company identification and perceived injustice

There appears to be some support for the impact of consumer-company identification being mediated through perceived injustice and then onto punishing behaviours. For H5 it was found that consumer-company identification had a significant negative effect on perceived injustice in the full data set (β = − 0.150, p < 0.01) and also in three of the eight scenarios. A negative relationship was estimated for the following scenarios: British Petroleum 1 (β = − 0.432, p < 0.01), Pacific Brands (β = − 0.166, p < 0.1), and Nestlé (β = − 0.228, p < 0.05). Interestingly a positive significant relationship was estimated for Qantas (β = 0.220, p < 0.05). Again the Qantas finding may be unique and be the result of the scenario involving two firms and characterised as containing economic harm through reputational damage.

Given the model’s structure, then the impact of customer-company identification on punishing behaviours occurs both directly (H1) and indirectly through perceived injustice (H3 through H5). Table 3 reports on these combined total effects in parentheses. In the full sample and six of the eight scenarios, the total effect is negative. The largest standardised impacts for customer-company identification occur for Apple 2 (− 0.323), Apple 1 (− 0.187), and L’Oreal (− 0.141). The only important positive total effect occurs for the unique Qantas case (0.134).

Control variables

A number of control variables were included in the study to assess their potential impact on punishing behaviours. This was done as previous studies have shown mixed results regarding the influence of gender, age, location, and level of education in ethical decision-making. Similarly, this study also produced mixed influences of these control variables on punishing behaviours, with only age and education having an important impact.

Age has a significant negative influence on punishing behaviours for Qantas (β = − 0.212, p < 0.05) and Apple 2 (β = − 0.159, p < 0.10). This implies that older people are less likely to punish in scenarios characterised as the ‘high’ economic damage through reputational harm (Qantas) and ‘high’ psychological damage through labour conditions (Apple 2).

Education has a significant inverse relationship with punishing behaviours for the full sample (β = − 0.055, p < 0.05) and Apple 2 (β = − 0.181, p < 0.05). That is, the higher educated the individual the less likely they are to punish in the full sample and in the ‘high’ psychological damage through labour conditions scenario (Apple 2).

Conclusion

The purpose of this study was to develop and test a model explaining the punishment behaviours of consumers towards organisations that have engaged in third-party observed ethically questionable situations. Most organisational justice research has focussed on the justice evaluations of individuals directly involved in justice situations. Researchers have begun to acknowledge the role of individuals as observers or third parties in others’ justice situations (e.g. Lind et al. 1998). As suggested by Miles and Naumann (2003), this study has controlled for the identity of the third party that being consumers and contributes to the literature by demonstrating that consumers as third parties can and will make judgements regarding justice situations and subsequently act to restore justice between the parties. Even when they are not the direct victim of the outcome, this study provides support to the idea that a consumer will punish a firm using various means, for actions they judge as an injustice.

A comprehensive model of the determinants of punishing behaviours has been examined. The empirical findings suggest that the majority of the posited hypotheses are supported to varying degrees. The use of a variety of ‘real-life’ scenarios based on high profile multinational companies uncovers a series of rich and interesting results. Overall, the research model developed and tested here has significant value in our understanding of consumer punishing behaviours as third-party observers.

The research has highlighted that marketers and managers need to understand what information consumers are receiving about a company and its practices. Consumers seem willing to share negative information, and in some cases, a consumer will reduce spending if they feel a situation is unethical and unjust and was the responsibility of the firm. Reduced spending is typically found to be the most likely form of punishment used by a consumer, which is also potentially the most difficult for the firm to identify and quantify. Marketers should also be aware of the likely reactions of strongly identified customers, realising that a strong relationship or sense of identification will not always protect a firm from punishment.

The research also indicates that the type of ethical situation and type of perceived injustice can also influence whether a consumer as a third-party observer is willing to act to restore justice on behalf of the victim. That is, the type of injustice and the severity of the harm can also be important in the consumer’s decision-making. This has important implications for organisations. For instance, the results indicated that perceived injustice has a significant effect on punishing behaviours in some scenarios but not others. When the situation involved physical harm or death, financial loss or severe economic damage, or psychological harm such as through the use of child labour, the consumer was more willing to punish the organisation. This indicates that the desire to punish is not necessarily related to whether or not a consumer has a specific interest in a particular ethical issue but rather to the absolute severity of the harm that was perceived in the situation. Consumers will not always act to restore justice on behalf of a victim through punishing a firm, but for some types of injustice and when the severity of harm is perceived to be higher, they will.

The findings of this study were interpreted with the following limitations in mind, which suggest a series of avenues for future research. This study used a scenario-based survey method. Hence, captured in this study are behavioural intentions as opposed to actual behaviours in the market place. Some research suggests that vignettes are particularly good at investigating how people think they should react in a given situation, but not necessarily how they will respond (Carlson 1996; Kiyonari and Barclay 2008). Further, even though the employed scenarios do have some features of a controlled experiment design by controlling for the type and severity of harm in realistic situations, this may be insufficient to accurately estimate causal links unlike a fully controlled experimental design.

The model developed in this study is based on a cross-sectional sample that does not offer the insight into the dynamics for third-party judgements and responses as a longitudinal design. At best, cross-sectional designs present a ‘snapshot’ of understanding, as they do not allow an understanding of the relationships between the constructs over time. A direction of future research could be the application of the model in a longitudinal study.

Using ‘real-life’ scenarios based on publicly available information regarding ethically questionable situations or strategies of large multinational corporations provided rich information. The results found in this study should be generalisable to other situations. However, future studies could include scenarios of other multinational companies.