The effects of observing punishment on consumers’ decisions to punish other companies during industry-wide crises

Abstract

Industry-wide crises, characterized by multiple companies within an industry allegedly engaging in similar misconduct, lead to disruptions in the social order. During such crises, one or a few of the involved companies often attract more media attention and receive punishment ahead of the others. Will such punishments take the heat off other involved companies or increase their risk of further punishment? This research shows that the observation of these punishments increases consumers’ intent to punish other involved companies. Observing one involved company get punished signals that the involved companies are indeed at fault, thereby increasing consumers’ certainty in blame attribution. Subsequently, consumers increase their motive to restore social order and their intent to punish other involved companies. Four theoretically and managerially relevant moderators have been identified and tested. Seven studies involving secondary data and experiments with both hypothetical and real behavioral outcomes support the proposed effects.

Introduction

Multiple brands or companies can be publicly alleged to have engaged in similar misconduct, such as product harm, service failure, or brand transgression, which can then escalate into an industry-wide crisis (Chen et al., 2022; Grayson et al., 2008; Han et al., 2024). When such a crisis emerges, one or a few of the involved companies often get reported first, attract more media attention than do the other involved companies, and are used as examples to illustrate the crisis (Zavyalova et al., 2012). Subsequently, these companies are likely to receive heightened scrutiny and negative imposition (i.e., punishment) before the other involved companies (Gao et al., 2012; Zhang et al., 2020). Famous examples include Sanlu in the 2008 Chinese milk scandal (Gao et al., 2012), Mattel in the 2007 toy recalls (NBC, 2007), Volkswagen in the 2014 diesel emission scandal (Carrington, 2015), and JBS in the 2017 Operation Weak Meat scandal (Romero, 2017). How do consumers respond to other involved companies in an industry-wide crisis after seeing one involved company get punished? Are they more or less likely to punish those companies? The answers to these questions have important practical implications for crisis management.

The involved companies may believe that when one company is punished, it can serve as a scapegoat and take the heat off them. However, we argue against this belief and propose that the opposite may happen. According to the altruistic punishment literature, consumers have a tendency to maintain social order (Fehr & Gächter, 2002; Lin et al., 2013). When social order is disrupted (e.g., crises), consumers are motivated to identify and punish the responsible parties to restore social order (Kurzban et al., 2015; Lin et al., 2013). However, consumers’ punitive intent decreases when they lack certainty about whether the alleged parties should be blamed for disruptions in the social order, as punishing innocent parties can cause further disruptions (Inbar et al., 2012; Shaver, 2012; Toribio-Flórez et al., 2023; Wiltermuth & Flynn, 2013).

During industry-wide crises, consumers lack certainty for various reasons when assigning blame for the crises to the allegedly involved companies. Industry-wide crises might have other causes in addition to the alleged companies’ actions, such as misconduct by supply chains or outsourcing partners, gaps in regulatory or legal frameworks, and “unspoken” rules that are implicitly endorsed, perpetuated, and normalized within the industry (Chen et al., 2022; Cleeren et al., 2013; Greve et al., 2010; Han et al., 2024; Shadnam & Lawrence, 2011). Furthermore, media reports, a primary information source for consumers, may offer noisy, incoherent, and incomplete information, hint at multiple potential causes, and fail to provide a definite conclusion regarding the causes of the crises (Clemente & Gabbioneta, 2017). In addition, consumers’ limited knowledge about industry norms, legal standards, and the intricacies of industry systems and supply chains increases their reliance on media information, which reduces their level of certainty in blaming the alleged companies. Consequently, consumers tend to seek external information to validate their blame attribution and gain certainty (Alicke, 2000; Lei et al., 2012; Malle, 2021). During an industry-wide crisis, the punishment imposed on one involved company serves as such validation information: it confirms that companies that have allegedly engaged in the same misconduct should be responsible for the crisis and take the blame. In other words, observing one involved company get punished enhances consumers’ certainty in attributing blame to the allegedly involved companies, subsequently strengthening consumers’ motive to restore social order and driving their intent to punish other involved companies that have not yet been punished.

Based on our theorizing, we identify four theory-driven and managerially relevant boundary conditions. First, when the punished company objects to (vs. accepts) the punishment, the punishment information no longer enhances blame attribution certainty, as consumers may still doubt whether the alleged companies are responsible. Second, when another company in the industry is not explicitly involved in the crisis, consumers will refrain from punishing it because of low certainty regarding whether this company is to blame. Third, when the punisher (i.e., the party who imposes the punishment) is informal (vs. formal, such as the government or official consumer associations), the observed punishment does not provide high validity regarding whether the punished company and other involved companies are at fault. Thus, such punishment does not effectively enhance blame attribution certainty. Finally, when the other involved companies proactively respond to the crisis by remedying the damage, consumers may perceive that the social order is already restored and thus lower their intent to punish these companies. We offer empirical support for the main effect, the underlying process, and boundary conditions across seven studies, including one study based on a secondary dataset and one with real behavioral outcomes.

This research makes several important contributions to the literature. First, while previous studies have focused mostly on single-brand crises (i.e., a crisis involving only one brand or company) (Cleeren et al., 2008; Khamitov et al., 2020; Klein & Dawar, 2004), less attention has been given to consumers’ responses during industry-wide crises (e.g., Chen et al., 2022; Gao et al., 2012; Zhang et al., 2020), especially their reactions to other involved companies after observing one involved company being punished. Existing research on single-brand crises has examined the negative spillover effect of a brand’s scandal to other non-scandalized brands (e.g., Che et al., 2023; Roehm & Tybout, 2006; Votola & Unnava, 2006). Based on the unique characteristics of industry-wide crises, we show a novel effect in which observing one involved company get punished increases consumers’ intent to punish other involved companies, and we provide process explanations and evidence for the underlying mechanism.

Second, prior studies on single-brand crises reveal the important role of blame attribution in consumer responses (e.g., Klein & Dawar, 2004; Lei et al., 2012). We extend the investigation of blame attribution, its subjective certainty, and its validation process in the context of industry-wide crises. Previous studies have shown that contextual information cues may help consumers adjust their blame attribution (Alicke, 2000; Heidenreich et al., 2015; Kranzbühler et al., 2019; Lei et al., 2012), while we reveal a novel and important information cue, i.e., the observed punishment, that can help consumers validate their certainty in blame attribution. Additionally, we demonstrate that consumers assess the characteristics of the observed punishment (e.g., punisher formality and the punished company’s response) to decide whether the punishment information can effectively validate their blame attribution. Hence, we expand the understanding of blame attribution and its validation mechanism.

Finally, prior research has examined primarily the responses of direct victims, particularly in the context of product-harm and service crises, and identifies a revenge motive (i.e., when individuals aim to “get even” with and hurt those who have hurt them) as a key driver of consumer punishment (e.g., Bechwati & Morrin, 2003; Grégoire et al., 2010). Relatively limited research has explored punishment from observers who are not directly impacted by crises (Xie & Bagozzi, 2019; Xu et al., 2021), which is considered “third-party punishment” or “altruistic punishment” (Fehr & Gächter, 2002; Lin et al., 2013; Toribio-Flórez et al., 2023). Our findings suggest that in the context of industry-wide crises, consumer punishment can be driven by a motive to restore social order and can therefore be extended to third-party consumers who are not directly harmed by crises. Thus, we extend the research on third-party punishment by demonstrating another context in which the motive to restore social order can drive third-party consumers to act.

Our findings enrich the crisis management literature (e.g., Bundy et al., 2017) and provide direct strategic implications for companies involved in industry-wide crises. We demonstrate that when consumers lack certainty about whether to blame the companies involved in the crisis, observing one involved company get punished can increase their certainty and thus their intent to punish other involved companies that have not yet been punished. Therefore, after one company is punished, other companies involved in an industry-wide crisis should remain vigilant and consider offering resolutions to the crisis to avoid further consumer sanctions.

Conceptual development

Industry-wide crises, consumer response, and motives

Industry-wide crises are well-publicized negative events involving multiple brands or companies alleged to have engaged in similar misconduct, such as product harm, service failure, or brand transgression (Chen et al., 2022; Khamitov et al., 2020). These crises violate social norms and lead to disruptions in social order that threaten consumer safety and well-being (Gao et al., 2012; Zavyalova et al., 2012; Zhang et al., 2020). Despite their large-scale impact, industry-wide crises have received limited attention in contrast to crises confined to individual brands or companies, known as “single-brand crises” (e.g., Klein & Dawar, 2004; Lei et al., 2012).

When an industry-wide crisis breaks out, it is common for one or a few of the involved companies to be reported first by the media (Gao et al., 2012). As the media tends to extensively cover companies that have already been featured in other media outlets (Graf-Vlachy et al., 2020), these first reported companies attract more media attention than others and become illustrative examples of the crisis (Zavyalova et al., 2012). Consequently, these companies are often punished before actions are taken against other involved companies. However, how consumers respond to the industry-wide crisis after observing these companies being punished remains unclear in the literature.

Punishment is a common individual response to negative events. Following the literature, we define punitive intent as consumers’ intention to punish and cause inconvenience to a company for its norm violations (Grégoire & Fisher, 2008; Grégoire et al., 2010). Increased punitive intent can lead consumers to engage in a range of actions, including filing vindictive complaints (Grégoire & Fisher, 2008; Hibbard et al., 2001), withholding future purchases (Rao et al., 1999), ending their relationship with the company (Grégoire & Fisher, 2008), and actively switching to its competitors (Bechwati & Morrin, 2003).

According to Kurzban et al. (2015), punishment is driven mainly by (1) a revenge motive (i.e., when individuals aim to vent anger and “get even” with those who have hurt them) or (2) a social order restoration motive (i.e., when individuals aim to punish norm violators and maintain social order). Revenge-driven punishment usually occurs when the punisher is a direct victim of the violation (i.e., a second party) (Grégoire et al., 2009, 2018; Joireman et al., 2013; Sembada et al., 2016), while punishment driven by a social order restoration motive can occur when the punisher is either a direct victim or, more commonly, an observer who is not directly affected by the violation (i.e., a third party) (Henrich & Muthukrishna, 2021; Lin et al., 2013). During industry-wide crises, extensive media exposure increases both second- and third-party consumers’ awareness of negative events. Although the number of second-party victims can be substantial during severe industry-wide crises, we focus on third-party consumers, who generally represent a larger proportion of the population and have a larger-scale impact. Thus, during industry-wide crises, the prevailing motive for imposing punishment among these third-party consumers is likely the motive to restore the social order disrupted by the crises (Lin et al., 2013).

The social order restoration motive is generally driven by individuals’ preference for a society with a strong social order that ensures stability and safeguards their interests (e.g., Hechter & Horne, 2003). When social order is disrupted, an effective way to restore it is to punish the party that is responsible for the disruption (Hofmann et al., 2018; Valor et al., 2022). This punishment can serve as a retribution for wrongdoing, reform wrongdoers, and deter future wrongdoing (Carlsmith et al., 2002). Therefore, such punishment is viewed as an “altruistic punishment” or a “moral sanction” (Malle, 2021).

Certainty in blame attribution and the role of punishment observation

Individuals’ punishment decisions driven by social order restoration motives are contingent upon moral judgments of blame—the evaluation of whether the alleged parties are responsible and therefore blameworthy (Darley & Shultz, 1990; Malle, 2021; Monroe & Malle, 2019). Such judgments inform the social order restoration motive to impose fair and justified punishments, as punishing guiltless parties or imposing ungrounded punishments creates further imbalances in the already disrupted social order (Darley & Shultz, 1990; Fehr & Gächter, 2002; Lin et al., 2013; Toribio-Flórez et al., 2023). Based on this reasoning, a prerequisite for consumers’ motive to restore social order by imposing punishments is their certainty in blame attribution: consumers need to be certain that the party they intend to punish is indeed responsible for and should take the blame for disrupting the social order. This proposition is consistent with prior research demonstrating that people have a subjective sense of certainty about their moral judgments (e.g., Jones & Ryan, 1997; Wiltermuth & Flynn, 2013). If other justifications for violations exist or if the violation contexts are ambiguous, individuals’ blame attribution certainty decreases (Inbar et al., 2012; Malle et al., 2014; Shaver, 2012; Wiltermuth & Flynn, 2013), subsequently reducing their motive to restore social order and their intent to punish (Toribio-Flórez et al., 2023).

Existing research on single-brand crises has shown that consumers automatically engage in blame attribution when facing crises to determine whether the alleged party is responsible and should take the blame (e.g., Folkes, 1984, 1988; McGill, 1990; Weiner, 1980). Although consumers usually perceive crises as being company-related (Cowley, 2005; Folkes, 1984; Lei et al., 2012), they are not completely certain that the alleged company should be blamed, especially if the crisis context is ambiguous and other justifications exist (Klein & Dawar, 2004; Lei et al., 2012; Monroe & Malle, 2019).

We posit that during an industry-wide crisis, consumers’ certainty in blame attribution—the involved companies should take the blame—is low. Several reasons may explain low certainty. When multiple companies within the same industry allegedly engage in similar misconduct, the cause of such an industry-wide crisis is complex. The misconduct may be a result of other actors’ actions in the supply chain or outsourcing parties, a loophole in regulatory systems or legal standards, or “unspoken” rules accepted, normalized, and spread in the industry (Chen et al., 2022; Cleeren et al., 2013; Greve et al., 2010; Shadnam & Lawrence, 2011). Furthermore, the media, a primary information source for consumers, is rarely able to offer comprehensive information and usually hints at various possible causes rather than providing definite conclusions about the causes of a crisis (Clemente & Gabbioneta, 2017) (see Appendix A for some real examples). Consumers, with limited knowledge of industry norms, legal standards, industry systems, and supply chains, rely on media information to infer the cause of a crisis and may struggle to confirm the allegations and the actual causes, leading to low certainty in attributing blame to the companies allegedly involved.

When consumers lack certainty in blame attribution, they rely on contextual information to validate their attribution (Alicke, 2000; Lei et al., 2012; Malle, 2021; Malle et al., 2014; Monroe & Malle, 2019). For example, when contextual information suggests that companies are facing external constraints, consumers lower the degree to which they attribute blame to the companies (Gilbert et al., 1988). Similarly, when a specific problem or crisis occurs frequently in an industry, consumers infer that other causes may be at play, which leads to a discounting effect in attributing blame to the companies involved in the crisis (Lei et al., 2012).

We propose that observing one involved company being punished can serve as an information cue for consumers to increase their certainty in attributing blame to the other involved companies. The reason is that if one company is punished, it signals that the punisher has reasons or evidence to confirm that the involved companies are responsible for the crisis and should be blamed. Subsequently, with this increased certainty in blame attribution, consumers will increase their motive to restore social order and punish the other involved companies that have not yet been punished.

Boundary conditions

Thus far, we have established the main effect and the underlying mechanism. Building upon our theorizing, we propose several boundary conditions based on when the observed punishments can or cannot enhance the certainty in blame attribution and when consumers are motivated to punish other companies for the purpose of restoring social order.

The punished company’s response to the punishment

A company being punished can respond to punishment by showing cognitive support (acceptance) or doubt (objection) (Szwajkowski, 1992). Accepting the punishment implies that the company admits that it is guilty (Szwajkowski, 1992). In contrast, objecting to the punishment implies that the company denies its responsibility for the crisis (Greve et al., 2010), raising doubts about whether it should be blamed. Thus, when observing objections to punishment, consumers may no longer use punishment information to increase their certainty in attributing blame to the involved companies, which in turn should not increase their punitive intent.

The explicitness of other companies’ involvement in the crisis

Given our proposition that consumers’ motive for punishment is to restore social order, consumers should aim only to punish other companies that are explicitly involved in the crisis, as they are certain that these companies should be blamed for disrupting the social order. Consumers tend to believe that punishing other companies that are not explicitly involved may further disrupt the social order (Fehr & Gächter, 2002; Lin et al., 2013). Thus, consumers’ punitive intent will not extend to companies with no explicit involvement in the industry-wide crisis because consumers are less certain about attributing blame to them than to companies with explicit involvement.

The formality of the punisher

Punishers, the parties who impose punishment, exhibit varying degrees of formality, signaling different levels of validity concerning whether the involved companies are truly responsible for the crisis. For example, governments are usually viewed as formal punishers because they conduct formal investigations, gather substantial evidence, and adhere to the rule of law to determine whether a company is causally responsible for a crisis (Tyler, 2006; Tyler & Trinkner, 2017). Consumer groups, particularly unorganized consumers, are usually deemed less formal because their punishments often occur in response to incomplete evidence, public sentiment, or relatively ambiguous social norms (Farah & Newman, 2010). However, consumer groups can also differ in formality based on their affiliations with authorities, structures, and endorsements (Doh et al., 2010; Piazza et al., 2022). We posit that punishers with higher formality are more likely to impose punishment based on rigorous investigation, solid evidence, and formal rules (Tyler, 2006). Thus, punishment issued by punishers with higher formality serves as a more valid information cue to signal that the involved companies should take the blame for industry-wide crises, which allows consumers to punish these companies.

Resolution to the crisis offered by other involved companies

Based on our theory, if consumers’ motive to punish is to restore social order, this motive should decline when they perceive that social order has already been restored. Following the punishment of an involved company, if other involved companies communicate to consumers that they have offered resolutions to the crisis, such as taking proper measures to remedy damage and prevent future crises, consumers will perceive that the social order has been restored and that they no longer need to take punitive actions. We summarize our conceptual framework in Fig. 1.

Fig. 1

Conceptual framework

Overview of the studies

We conduct seven studies to test our propositions. In the first study, we analyze a secondary dataset on consumer complaints against airline companies during the overbooking scandal in 2017. We use this real-world correlational evidence to show that after United Airlines was singled out and punished, the number of complaints against other companies with overbooking practices increased significantly. In Study 2, we provide additional support for the main effect in a real behavior setting. Using a crisis context of widespread research misconduct, we show that observing the punishment of one research team can increase online panelists’ intention to withhold participation in a study by another research team that is allegedly involved in the same crisis. In Studies 3–7, we use five scenario-based experiments across different types of industry-wide crises to test the main effect, the proposed process, and the boundary conditions. In addition, we examine a series of relevant competing explanations. All crises in the scenarios are inspired by real cases, and their descriptions are adapted from existing media articles (see Appendix A). The scenarios used in the online experiments are included in Appendix B. The sample sizes in all studies were determined prior to data collection, and no exclusion criteria were applied in any of the studies. The measurement items are listed in Appendix C.

Study 1: Airline ticket oversales crisis and consumer complaints

In the first study, we aim to use a secondary dataset to provide correlational evidence for our proposed main effect. The ticket oversales practice associated with “involuntary bumps” had been an industry-wide issue among U.S. airlines. In 2016, approximately 475,000 U.S. passengers were denied boarding due to overbooking (Bryan, 2017). A significant negative event occurred on April 9, 2017, when a short video showing a United Airlines passenger being forcibly dragged from his seat on an overbooked flight generated over 6 million views in one day and sparked public outrage. In addition to United Airlines, other major airlines, including Delta Airlines, Southwest Airlines, American Airlines, and SkyWest Airlines, were accused of engaging in the same practice (Rollert, 2017). Soon after the industry-wide crisis broke out, the government opened an investigation into United Airlines on April 11th (Kelly, 2017), the U.S. president publicly criticized the company on April 12th (Wojcik, 2017), and consumers began to boycott the company around the same time (Zorthian, 2017). These punitive actions against United Airlines received extensive media coverage.

To investigate consumers’ reactions to other airlines that were also allegedly involved in the oversales practice associated with “involuntary bumps” after observing the punishment imposed on United Airlines, we focus on consumers’ complaints about these other airlines filed with the U.S. Department of Transportation. Consumer complaints cause inconvenience for norm violators, which matches our definition of punishment (Grégoire & Fisher, 2008; Grégoire et al., 2010). Compared with complaints filed directly to airlines, complaints sent to authorities are considered more relevant altruistic punishment behaviors that aim to punish companies to restore the social order instead of receiving compensation (Grégoire & Fisher, 2006). We predict that after observing United Airlines being punished, consumers’ complaints toward other airline companies involved in the overbooking scandal increase.

Method

We gathered data on consumer complaints about airlines received by the U.S. Department of Transportation (https://www.transportation.gov/airconsumer/air-travel-consumer-report-archive). The database provides monthly consumer complaint information for 12 major U.S. airlines, including United Airlines and all other airlines that had records of involuntary bumps associated with oversales during the United Airlines punishment event.

To analyze the effect of observing the punishment of an involved company on consumers’ punitive behavior toward the other involved companies, we excluded complaints against United Airlines, resulting in a sample of 11 airline companies. To control for the airline size, we standardized the number of complaints based on enplanements (i.e., the number of complaints per 100,000 enplanements) by dividing the number of complaints by the number of enplanements. We used data from the following six months: April and May 2017 (i.e., the window period after United Airlines was punished), March 2017 (i.e., the month prior to the punishment window period), April and May 2016¹ (i.e., the same months in the year prior to the post-punishment window period), and March 2016 to control for the seasonality effect. If our effect holds, we should observe higher complaint numbers in April and May 2017 than in the same months the previous year, as well as in March 2017.

Results and discussion

We conducted a 2 × 3 repeated measures ANOVA on the number of complaints, with year (2016 vs. 2017) and month (March vs. April vs. May) as within-subject independent variables. The results showed that the year had a significant effect on complaints (M₂₀₁₆ = 1.69, SD = 3.62 vs. M₂₀₁₇ = 2.18, SD = 3.91; F(1, 10) = 9.76, p = 0.011, η_p² = 0.49) but not the month (M_March = 1.75, SD = 3.13 vs. M_April = 2.06, SD = 2.67 vs. M_May = 2.00, SD = 3.44; F(2, 20) = 1.52, p = 0.244). The interaction effect was marginally significant (F(2, 20) = 2.93, p = 0.077, η_p² = 0.23). Planned contrasts (Fig. 2) showed that in 2017, the number of complaints increased from March to May. Specifically, the numbers of complaints in both April (M_{April_2017} = 2.45, SD = 1.98 vs. M_{March_2017} = 1.51, SD = 1.61; F(2, 20) = 9.42, p = 0.001, η_p² = 0.49) and May were significantly higher than in March (M_{May_2017} = 2.59, SD = 3.36 vs. M_{March_2017} = 1.51, SD = 1.61; F(2, 20) = 3.13, p = 0.066, η_p² = 0.24). The difference between April and May was not significant (F(2, 20) = 0.08, p = 0.92). Additionally, the number of complaints in both April and May was higher in 2017 than in 2016 (all p’s < 0.05), while the number of March complaints did not differ between 2016 and 2017 (p = 0.26). These results provide correlational evidence of an increase in punitive behaviors toward other involved airlines after the punishment event involving United Airlines.

Fig. 2

The number of complaints against other involved airlines (per 100,000 enplanements)

We also checked the major categories of complaints, including complaints about flight problems and complaints about reservations, ticketing, and boarding, and found that they displayed a pattern similar to that of the total number of complaints (see Web Appendix A). Together, these results suggest a positive correlation between observing one involved company being punished and the number of complaints filed against other allegedly involved companies. This correlational evidence provides preliminary support for our proposed main effect. However, given that the event does not allow us to fully separate the punishment observation and the punitive behaviors chronologically and that the sample size is small, the evidence is limited in fully supporting our proposed main effect. In the following studies, we will use experiments to establish the causality of the effect.

Study 2: Academic misconduct and participants’ withholding participation

This study aims to further test our proposed main effect with another type of real punishment-related behavior. Specifically, we created an industry-wide crisis describing multiple research teams that were questioned about academic misconduct and then allowed panelists to decide to participate in a survey study conducted by one of these research teams. If they chose to participate, they were provided with a separate survey study and received a bonus payment for completing that study. Following prior research (Allard & McFerran, 2022; Lin et al., 2013), participation in the study was used as a proxy for withholding participation or assistance; being unwilling to participate in the study and forgoing the payment indicated that the panelists wanted to punish the research team.

Method

The study employed a one-factor, three-level (observation of the punishment of an involved research team: punishment vs. no punishment vs. control) between-subjects design. Participants completed one survey study unrelated to the current research. At the end of the survey, they were invited to participate in another survey conducted by a different research team. If they chose to participate, they would be compensated for their time. Before deciding to participate, they were presented with information about the research team and were made aware of a current affair involving that team and other teams. Participants across all conditions were shown the following text:

A government-sponsored national research project investigating the lifestyle of disabled consumers is recently questioned for potential academic misconduct. The project is led by multiple research teams across multiple universities, including University X (the name is not disclosed for legal reasons) and this research team at my university that runs the study on individual differences in personal attitudes, opinions, and behaviors. According to a tip from an anonymous source, disabled consumers interviewed by multiple research teams on this project were abused. Some wording in the interview questions was alleged to be discriminatory and insensitive.

Several media outlets have reported this scandal and pointed out that the interview questions were supposed to be reviewed, screened, and checked by the government research council and the university’s institutional review boards.

In the punishment condition, participants read the following text: “After the scandal was reported, the Department of Education has censored the research team at University X by suspending their current research projects and freezing their research funding. The disabled community and the students at University X have started protesting against this research team.” In the no punishment condition, participants read the following text: “After the scandal was reported, the Department of Education has not yet censored the research team at University X. Their current research projects have not been suspended, and their research funding has not been frozen. No protest has been organized by the disabled community or the students at University X.” In the control condition, no information was presented.

Then, participants indicated whether they would like to participate in another survey conducted by another involved research team (Yes/No). If they selected “Yes,” they received the link to the other survey and received a bonus payment if they completed that survey. If they selected “No,” they were notified of their completion of the first survey. Finally, all participants were debriefed and made aware that the content of this study was fictitious. The debrief was provided at the end of the first survey for participants who chose “No” and at the end of the second survey for participants who chose “Yes.”

Results and discussion

We recruited 370 U.S.-based Prolific panelists to participate in this study (48.1% female, M_age = 42.06, SD = 14.11).

The chi-square test of indicated participation showed a significant main effect of observing punishment (χ²(2) = 7.82, p = 0.020, Cramer’s V = 0.15). Participants in the punishment condition were less willing to participate in the survey than were those in the no punishment condition (П_punishment = 42.1% vs. П_{no_punishment} = 57.0%, χ²(1, 247) = 5.53, p = 0.019, Cramer’s V = 0.15) and control condition (П_punishment = 42.1% vs. П_control = 57.7%, χ²(1, 249) = 6.11, p = 0.013, Cramer’s V = 0.16). There was no difference between the no punishment and control conditions (χ²(1, 244) = 0.01, p = 0.912).

As participants who received the survey link provided by the other research team would not necessarily click it and complete the survey, we also matched participants’ actual completion of the survey with the condition to which they belonged and created an actual participation measure. The chi-square test of this actual participation also showed a significant main effect of observing punishment (χ²(2) = 8.10, p = 0.017, Cramer’s V = 0.15). Again, participants in the punishment condition were less likely to complete the survey than those in the no punishment condition (П_punishment = 31.0% vs. П_{no_punishment} = 46.3%, χ²(1, 247) = 6.13, p = 0.013, Cramer’s V = 0.16) and the control condition (П_punishment = 31.0% vs. П_control = 46.3%, χ²(1, 249) = 6.22, p = 0.013, Cramer’s V = 0.16). There was no difference between the no punishment and control conditions (χ²(1, 244) < 0.01, p = 0.992).

Thus, both measures confirmed that participants were more likely to punish the other involved research team by not participating in or completing their study only after learning that one involved team had been punished. These results provide support for our main effect with evidence of real behavioral outcomes.

Furthermore, the null effect between the control and no punishment conditions rules out the possibility that the main effect is caused by a decrease in punitive intent when the involved companies are not punished. Instead, the main effect occurs because the observation of an involved company being punished leads to an increase in punitive intent.

Study 3: Meat scandal and punitive intent

In this study, we aim to replicate the main effect in a scenario-based experimental setting and test the proposed primary mediator in the process—certainty in blame attribution—while measuring and controlling for three alternative explanations. The first alternative explanation is that after observing one involved company being punished, consumers may feel a sense of unfairness if other involved companies are not punished. This perception of unfairness subsequently drives them to punish the other involved companies. The second explanation relates to the inference of crisis severity based on the punishment. Learning that an involved company is punished (vs. not punished) may increase consumers’ perceived severity of the crisis, which strengthens their willingness to punish other involved companies. Finally, learning that an involved company is being punished may trigger consumers’ just world belief, which, simply put, refers to people’s general belief that good things should happen to good people and bad things to bad people (Lerner, 1965; Lipkus, 1991). Thus, consumers are more likely to punish the other involved companies if this belief is heightened after seeing one involved company get punished.

Method

We employed a one-factor, two-level (observation of the punishment of an involved company: punishment vs. no punishment) between-subjects design. We used the food industry as our research context and provided all participants with a description of an industry-wide food safety crisis scenario. The descriptions were based on news reports about the 2013 horse meat scandal (Lawrence, 2013; Hutton, 2013), in which undeclared meats were found in beef burgers and other meat products of many companies. To minimize the effect of preexisting brand attitudes, we replaced brand names with alphabetic letters and informed participants that the purpose of doing so was to avoid legal disputes. Participants read that many major food companies, including Company X and Company B, were allegedly involved in this food safety issue. In the condition with punishment observation, participants read that Company X was investigated by the state authority and boycotted by consumer groups. The actions taken by the state authority and consumer groups caused a significant financial loss to the company. In the condition without punishment observation, participants read that Company X had not been investigated by the state authority and that consumer groups had not shown any sign of boycotting the company; moreover, this company had not experienced any financial loss.

Punitive intent was measured by rating the extent to which participants “want to take actions to get Company B in trouble,” “want to take actions to cause inconvenience to Company B,” and “want to make Company B get what it deserved” (α = 0.95; Grégoire et al., 2018). Certainty in blame attribution was measured using a two-item scale adapted from (Klein & Dawar, 2004). The items asked participants to what extent they were certain that the alleged companies (including Companies X and B) were at fault/guilty (r = 0.90). Perceived unfairness was measured using a two-item scale (Komarova Loureiro et al., 2018) that asked participants to rate the degree to which they felt that it was unfair that the two companies (X and B) were treated unequally (r = 0.78). Crisis severity was measured using the two-item scale from Mafael et al. (2022), which asked participants to rate the extent to which they felt that the scandal was severe/serious (r = 0.94). Just world belief was measured using the scale from Lipkus (1991) (α = 0.90). All items were measured on seven-point scales (1 = “not at all” and 7 = “very much”) and are listed in Appendix C. At the end of the survey, we measured participants’ perceived scenario realism and collected their demographic information.

Results

We recruited 201 U.S.-based Prolific panelists (49.3% female; Mage = 39.22, SD = 12.90) to participate in this study and offered a monetary reward. Participants found our experimental scenarios to be realistic, as their perceptions of realism were significantly higher than the midpoint of the scale (4 = “moderate”) based on a one-sample t-test (M = 4.93, SD = 1.59, t(200) = 8.27, p < 0.001, d = 0.58). No significant difference in perceived realism was found between conditions (p > 0.05).

Punitive intent

An independent samples t-test revealed a significant effect of observing punishment on consumers’ punitive intent (t(199) = 3.37, p < 0.001, d = 0.48). As predicted, compared with participants who observed no punishment of an involved company, participants who observed punishment showed a higher level of punitive intent toward the other involved companies (M_punishment = 5.27, SD = 1.40 vs. M_{no punishment} = 4.48, SD = 1.86). Thus, we again find support for the main effect.

Certainty in blame attribution

An independent samples t-test showed that observing punishment had a significant effect on consumers’ attribution certainty (t(199) = 2.24, p = 0.026, d = 0.32). Participants who observed the punishment of an involved company were more certain that the involved companies should be blamed than those who observed no punishment (M_punishment = 5.03, SD = 1.39 vs. M_{no punishment} = 4.54, SD = 1.69).

Perceived unfairness

An independent samples t-test of this variable revealed a significant effect such that participants who observed punishment perceived a greater degree of unfairness than those who observed no punishment (M_punishment = 5.84, SD = 1.25 vs. M_{no punishment} = 5.29, SD = 1.55; t(199) = 2.79, p = 0.006, d = 0.39).

Crisis severity

An independent samples t-test showed that the effect of punishment on crisis severity was not significant (M_punishment = 5.64, SD = 1.34 vs. M_{no punishment} = 5.45, SD = 1.47; t(199) = 0.949, p = 0.344). Thus, crisis severity can be ruled out as an alternative process.

Just world belief

An independent samples t-test showed that the effect of punishment on just world belief was not significant (M_punishment = 3.46, SD = 1.09 vs. M_{no punishment} = 3.34, SD = 1.33; t(199) = 0.72, p = 0.471). We thus ruled out just world belief as an alternative process.

Mediation test

We first performed a mediation test using observed punishment as the independent variable (0 = no punishment; 1 = punishment), punitive intent as the dependent variable, and certainty in blame attribution as the only mediator. The mediation test used the PROCESS macro with 10,000 resamples (Model 4; Hayes, 2018). The indirect effect of blame attribution certainty was significant (B = 0.30, SE = 0.14, 95% CI = [0.036, 0.574]), where the observation of punishment increased blame attribution certainty (B = 0.49, SE = 0.22, p = 0.026), which in turn increased punitive intent (B = 0.60, SE = 0.06, p < 0.001). Thus, we find support for our proposed underlying mechanism.

Since the effect of observing punishment on perceived unfairness was significant, we performed a mediation test that included certainty in blame attribution and perceived unfairness as two parallel mediators. We also controlled for crisis severity and just world belief, given the significant correlations between certainty in blame attribution and perceived unfairness, crisis severity, and just world belief (all p’s < 0.05). The indirect effect of blame attribution certainty remained significant (B = 0.16, SE = 0.09, 95% CI = [0.001, 0.355]), while the indirect effect of perceived unfairness was not significant (95% CI = [− 0.019, 0.174]). Thus, after controlling for these other alternative processes, we find support that the focal process of certainty in blame attribution still mediates the main effect. To further explore possible roles of perceived unfairness in the conceptual model, we ran several additional models. All the model results suggest that certainty in blame attribution is a key process underlying the main effect. We report these additional analyses in Web Appendix B.

Even though just world belief was not significantly affected by punishment observation conditions, it may serve as a trait variable (vs. a state variable) that influences participants’ propensity to punish other companies after observing the punishment of an involved company. Thus, we ran a moderated mediation test (Model 8; Hayes, 2018), adding just world belief as the moderator to the parallel mediation model tested above. We did not find a significant moderated mediation effect of blame attribution certainty (B = 0.03, SE = 0.03, 95% CI = [− 0.029, 0.107]). Thus, just world belief did not moderate the main effect or serve as the underlying process.

Discussion

This study shows that consumers increase their intent to punish other companies after observing one involved company get punished. This effect occurs because consumers become more certain that the involved companies should be blamed for the crisis after observing the punishment imposed on one involved company.

Study 4: Frozen fruit scandal and the punished company’s response to punishment

In this study, in addition to replicating the main effect and testing the certainty in blame attribution as the primary mediator, we test the first boundary condition: the punished company’s response to the punishment. We predict that when consumers observe that the punished company objects to (vs. accepts) the punishment, they will no longer apply the related punishment information to increase their certainty in attributing blame or increase their punitive intent toward the involved companies. Furthermore, we again include a control condition in this study to provide a baseline level of punitive intent.

Method

This study employed a single-factor, three-level (the punished company’s reaction: object to the punishment vs. accept the punishment vs. control) between-subjects design. As Study 3 showed the mediating process by comparing the punishment versus no punishment conditions, this study aimed to replicate the underlying process by comparing the punishment condition to the control condition, which does not explicitly mention the punishment outcome of an involved company. The manipulation of the company’s response to the punishment does not make sense in the control condition, where the punishment is not explicitly mentioned. Therefore, instead of using a two-factor between-subjects design, we used a single-factor, three-level between-subjects design to test the moderating effect. If our prediction holds, punitive intent should be higher in the condition of accepting the punishment than in the condition of objecting to the punishment and in the control condition.

Participants read about a food safety scandal (based on the 2023 frozen fruit contamination scandal; Musto, 2023), where multiple companies’ frozen fruits were found to be contaminated with the hepatitis A virus. Among the involved companies, Company X had been investigated by the state authority and had been issued a big fine. In the object [accept] condition, participants continued to read that Company X had made an official statement to formally appeal the fine [had accepted the penalty and paid the fine]. In the control condition, no information was provided on the investigation or penalty. Afterward, participants rated their punitive intent toward Company B, which was also involved in the scandal (α = 0.92), and they reported their certainty in blame attribution (r = 0.87) using the same items as in our previous studies.

Results

We recruited 300 U.S.-based Prolific panelists (47.7% female; Mage = 41.21, SD = 13.63) to participate in this study and offered a monetary reward. Participants found our experimental scenarios to be realistic, as their perceptions of realism were significantly above the midpoint of the scale (4 = “moderate”; M = 5.21, SD = 1.63, t(299) = 12.80, p < 0.001, d = 0.74). No significant difference in perceptions of realism was found between conditions (all p’s > 0.05).

Punitive intent

A one-way ANOVA revealed a significant effect of the punished company’s response on consumers’ punitive intent (F(2, 297) = 4.60, p = 0.011). Planned contrasts show that compared with participants in the object condition, participants in the accept condition showed a higher level of punitive intent (M_object = 4.88, SD = 1.63 vs. M_accept = 5.40, SD = 1.52; t(297) = 2.35, p = 0.020, d = 0.33). In addition, the accept condition triggered higher levels of punitive intent than did the control condition (M_control = 4.75, SD = 1.67; t(297) = 2.82, p = 0.005, d = 0.40), while the object condition did not trigger significantly higher punitive intent than did the control condition (t(297) = 0.54, p = 0.59). Thus, when the company objects to the punishment, consumers’ punitive intent toward other involved companies decreases to the same level as when no information is provided about the punishment.

Certainty in blame attribution

A one-way ANOVA revealed a significant effect of the company’s response on consumers’ blame attribution certainty (F(2, 297) = 3.23, p = 0.041). As predicted, compared with participants in the object condition, participants in the accept condition showed a higher level of blame attribution certainty (M_accept = 5.34, SD = 1.22 vs. M_object = 5.00, SD = 1.32; t(297) = 1.90, p = 0.058, d = 0.26). In addition, the accept condition triggered higher levels of blame attribution certainty than did the control condition (M_control = 4.89, SD = 1.41; t(297) = 2.40, p = 0.017, d = 0.34), while the object condition did not trigger significantly higher blame attribution certainty than did the control condition (t(297) = 0.55, p = 0.581).

Mediation test

We conducted a mediation analysis with the condition as a multi-categorical variable using the PROCESS macro with 10,000 resamples (Model 4; Hayes, 2018). We dummy coded the conditions and set the accept condition as the baseline. The results revealed that certainty in blame attribution significantly mediated the effect of accept (vs. appeal) (B = 0.23, SE = 0.12, 95% CI = [0.003, 0.482]) such that the accept condition triggered higher certainty in blame attribution (B = 0.35, SE = 0.18, p = 0.058) and subsequently led to greater punitive intent (B = 0.68, SE = 0.06, p < 0.001). Certainty in blame attribution also significantly mediated the effect of accept (vs. control) on punitive intent (B = 0.30, SE = 0.13, 95% CI = [0.060, 0.572]) such that the accept condition showed higher certainty in blame attribution (B = 0.45, SE = 0.19, p = 0.017) and then triggered greater punitive intent (B = 0.68, SE = 0.06, p < 0.001).

Discussion

This study replicates the main effect and the proposed underlying process in another crisis context. Furthermore, we show that if a company appeals their punishment, consumers become less certain that the involved companies should take the blame for the crisis and lower their punitive intent accordingly. The findings have practical implications in that a punished company’s response to the punishment has an impact on the other involved companies.

Study 5: Meat scandal and the explicitness of another company’s involvement in the crisis

This study has two objectives. First, we aim to test the full process in the conceptual framework. That is, the effect of observing punishment on the intention to punish other companies is driven serially by certainty in blame attribution and the social order restoration motive. Second, to explore the characteristics of a company that has not been punished yet, this study tests the explicitness of such company’s involvement in a crisis as a boundary condition.

Method

This study employed a single-factor, three-level between-subjects design with two punishment conditions that manipulate the explicitness of another company’s involvement in a crisis (explicitly involved vs. not explicitly involved) and a no punishment condition. Participants were presented with a scandal scenario similar to the one used in Study 3. In the punishment and explicit involvement condition, participants were informed that both Company X and Company B had been reported to be involved in the crisis, and Company X had been punished. In the punishment and non-explicit involvement condition, participants read that Company X had been reported to be involved in the crisis while Company B had not been mentioned in the news reports to be involved in the crisis; participants were then informed of the punishment imposed on Company X. In the no punishment condition, participants read that both Company X and Company B had been reported to be involved in the crisis and that Company X had neither been fined nor boycotted.

Afterward, the participants rated their punitive intent toward Company B (α = 0.93). Furthermore, participants rated the attribution certainty toward the involved companies (including Company X) (r = 0.93) and attribution certainty toward Company B (r = 0.93) using the same items as in Studies 3 and 4. We measured the social order restoration motive by asking people to what extent they agreed with the statement, “I feel that I need to make sure the social order in the affected industry is fully restored by punishing Company B,” using a 7-point scale where 1 = “strongly disagree” and 7 = “strongly agree.” Finally, we measured perceived unfairness using the same scale as in Study 3 (r = 0.92) and examined its role when the other company was not explicitly involved. We report the analyses involving perceived unfairness in Web Appendix C.

Results

We recruited 302 U.S.-based Prolific panelists to participate in this study (51% female, M_age = 37.69, SD = 13.29). Participants found our experimental scenarios to be realistic, as their perceptions of realism were significantly above the midpoint of the scale (4 = “moderate”; M = 4.69, SD = 1.58, t(301) = 7.56, p < 0.001, d = 0.44). No significant difference in perceptions of realism was found between conditions (all p’s > 0.05).

Punitive intent

A one-way ANOVA on punitive intent showed a significant effect of the manipulation condition (F(2, 299) = 36.78, p < 0.001; see Table 1 for the summary of means). Planned contrasts showed that participants had greater punitive intent in the punishment and explicit involvement condition than in the no punishment condition (M_involved = 5.29, SD = 1.57 vs. M_{no_punish} = 4.42, SD = 1.75, t(299) = 3.60, p < 0.001, d = 0.51), while punitive intent was greater in the no punishment condition than in the punishment and non-explicit involvement condition (M_{no_punish} = 4.42, SD = 1.75 vs. M_{not_involved} = 3.22, SD = 1.83, t(299) = 4.92, p < 0.001, d = 0.70). Moreover, comparing the two conditions that varied Company B’s involvement revealed that participants had a lower punitive intent when Company B was not explicitly involved (vs. explicitly involved) (t(299) = -8.55, p < 0.001, d = 1.20). Thus, we found that when the other company in the industry was not explicitly involved in the crisis, consumers lowered their intention to punish this company, even after observing the punishment.

Table 1 Summary of results (Study 5)

Certainty in blame attribution to the involved companies

A one-way ANOVA on blame attribution certainty toward the involved companies showed a significant effect of the manipulation condition (F(2, 299) = 12.07, p < 0.001). Consistent with the findings of our previous studies, planned contrasts showed that observing punishment increased the certainty in attributing blame to the involved companies (including Company X) compared to observing no punishment, regardless of Company B’s involvement (M_involved = 5.67, SD = 1.28 vs. M_{no_punish} = 4.78, SD = 1.51, t(299) = 4.62, p < 0.001, d = 0.65; M_{not_involved} = 5.51, SD = 1.32 vs. M_{no_punish} = 4.78, SD = 1.51, t(299) = 3.77, p < 0.001, d = 0.53). Moreover, certainty in attributing blame to the involved companies (including Company X) did not differ between the two conditions varying Company B’s involvement (t(299) = 0.84, p = 0.403).

Certainty in blame attribution to Company B

A one-way ANOVA on the blame attribution certainty toward Company B showed a significant effect of the manipulation condition (F(2, 299) = 32.35, p < 0.001). Planned contrasts showed a pattern similar to that of punitive intent. The participants felt more certain about Company B’s blame attribution in the punishment and explicit involvement condition than in the no punishment condition (M_involved = 5.12, SD = 1.33 vs. M_{no_punish} = 4.46, SD = 1.48, t(299) = 3.11, p = 0.002, d = 0.44), while participants’ certainty in blame attribution was greater in the no punishment condition than in the punishment and non-explicit involvement condition (M_{no_punish} = 4.46, SD = 1.48 vs. M_{not_involved} = 3.41, SD = 1.74, t(299) = 4.85, p < 0.001, d = 0.67). Moreover, comparing the two conditions that varied Company B’s involvement showed that participants had lower certainty in attributing blame to Company B when it was not explicitly involved (vs. explicitly involved) (t(299) = 7.99, p < 0.001, d = 1.12). Thus, certainty in blame attribution to Company B was a function of both the observed punishment and Company B’s involvement in the crisis.

Social order restoration motive

A one-way ANOVA on the social order restoration motive showed a significant effect of the manipulation condition (F(2, 299) = 16.24, p < 0.001). Planned contrasts showed a pattern similar to that of punitive intent and certainty in attributing blame to Company B. Planned contrasts showed that participants had a greater motive to restore social order in the punishment and explicit involvement condition than in the no punishment condition (M_involved = 4.42, SD = 1.76 vs. M_{no_punish} = 3.76, SD = 1.80, t(299) = 2.63, p = 0.009, d = 0.37), while their motive was greater in the no punishment condition than in the punishment and non-explicit involvement condition (M_{no_punish} = 3.76, SD = 1.80 vs. M_{not_involved} = 2.99, SD = 1.80, t(299) = 3.05, p = 0.003, d = 0.43). Moreover, comparing the two conditions that varied Company B’s involvement showed that participants had a lower level of social order restoration motive in the non-explicit involvement condition (vs. explicit involvement) (t(299) = 5.70, p < 0.001, d = 0.80).

Serial mediation tests

To provide evidence for the social order restoration motive, we ran serial mediation tests based on 10,000 bootstraps (Model 6; Hayes, 2018). We used the experimental condition as a multi-categorical independent variable (punishment and explicit involvement condition as the baseline). We selected blame attribution certainty toward Company B as the primary mediator, the social order restoration motive as the secondary mediator, and the punitive intent toward Company B as the dependent variable. In the first contrast between the punishment and explicit involvement condition and the no punishment condition, we found a significant serial mediation effect (B = -0.16, SE = 0.06, 95% CI = [-0.29, -0.06]). This result suggests that after consumers are certain that the involved companies, such as Company B, are at fault, the motive to restore social order drives them to punish these companies. When replacing “blame attribution certainty toward Company B” with “blame attribution certainty toward the involved companies (including Company X)”, we observed similar significant serial mediation in this contrast (B = -0.20, SE = 0.06, 95% CI = [-0.34, -0.09]). Note that in this contrast, consumers’ blame attribution certainty toward Company B was equivalent to their blame attribution certainty toward the involved companies (including Company X) (r_{no_punish} = 0.77, p < 0.001; r_involved = 0.63, p < 0.001). Thus, this serial mediation also replicates the effects reported in our previous studies.

In the second contrast between the punishment and explicit involvement condition and the punishment and non-explicit involvement condition, we also observed a significant serial mediation effect (B = -0.43, SE = 0.09, 95% CI = [-0.63, -0.26]). This result indicates that consumers lower their motive to restore social order and punitive intent when other companies, such as Company B, are not explicitly involved in the crisis. In this contrast, when Company B was not explicitly involved, consumers’ blame attribution certainty toward Company B was not equivalent to blame attribution certainty toward the involved companies (including Company X) (r_{not_involved} = 0.05, p = 0.593). Therefore, blame attribution certainty toward Company B would serve as a better predictor of punitive intent toward Company B than would blame attribution certainty toward the involved companies (including Company X). To maintain analytical consistency with our previous studies, we performed an additional analysis that included both blame attribution certainty toward the involved companies (including Company X) and blame attribution certainty toward Company B as the parallel primary mediators. We found consistent results (see Web Appendix D).

Discussion

In this study, we show that observing an involved company being punished increases consumers’ intention to punish other companies that are explicitly involved in the crisis. We provide process evidence that this increase in punitive intent is driven by enhanced certainty in attributing blame to the involved companies. However, when a company is not explicitly involved, observing punishment does not increase the certainty of assigning blame to this company.

To offer direct evidence of the motive to restore social order, we show that this motive serves as the secondary mediator after blame attribution certainty. These results suggest that, after consumers become certain that the involved companies are at fault, the motive to take action to restore social order drives consumers to punish the other involved companies.

Study 6: Hospitality hygiene scandal and punisher formality

In this study, we aim to test whether punisher formality can serve as a boundary condition. Specifically, this study focuses on a non-government punisher—a consumer group—and varies its formality. We predict that observing a punishment imposed by a formal consumer group (vs. by an informal consumer group) will increase consumers’ intention to punish other involved companies.

Method

This study employed a single-factor, three-level (punishment imposed by a formal punisher vs. punishment imposed by an informal punisher vs. no punishment) between-subjects design. Participants were presented with a description of a scandal scenario in which many hotels were reported to have hygiene issues. The descriptions of the scandal were based on a real event (ABC, 2018; Zhang, 2019; Zielinski & Clay, 2013): released hidden camera footage shows that more than a dozen hotels used dirty towels to clean guests’ cups and showers. In the formal punisher condition, participants read that a large, registered consumer rights association had started a boycott of Company X. In the informal punisher condition, participants read that verified online consumers (i.e., not online trolls or bots) on social media platforms had started a boycott of Company X. In the no punishment condition, participants read that Company X had been neither fined nor boycotted. Afterward, participants rated their punitive intent toward Company B (α = 0.93) using the same items as in our previous studies.

Results and discussion

We recruited 303 U.S.-based Prolific panelists to participate in this study (48.8% female, M_age = 37.29, SD = 12.94). Participants found our experimental scenarios to be realistic, as their perceptions of realism were significantly above the midpoint of the scale (4 = “moderate”; M = 5.23, SD = 1.44, t(302) = 14.88, p < 0.001, d = 0.86). No significant difference in perceptions of realism was found between conditions (all p’s > 0.05).

Punitive intent

A one-way ANOVA revealed a significant effect of punisher formality on consumers’ punitive intent (F(2, 300) = 7.93, p = 0.050). Planned contrasts showed that participants in the formal punisher condition had a higher level of punitive intent than did those in the informal punisher condition (M_{formal punisher} = 4.59, SD = 1.52 vs. M_{informal punisher} = 4.10, SD = 1.63; t(300) = 2.15, p = 0.032, d = 0.30) and in the no punishment condition (M_{no_punishment} = 4.11, SD = 1.70; t(300) = 2.11, p = 0.036, d = 0.30), while the informal punisher condition did not trigger significantly higher punitive intent than did the no punishment condition (t(300) = 0.05, p = 0.961). Thus, when punishment was imposed by informal punishers, it did not increase consumers’ intent to punish other companies after observing the punishment. Our proposed main effect occurred only when the punishment was imposed by a formal punisher.

Study 7: Data management scandal and other involved companies’ resolutions to the crisis

In this study, we aim to replicate the main effect in a different crisis context and test the involved companies’ resolution to the crisis as another boundary condition for the main effect. Based on our theorizing, the motive for consumers to punish the other involved companies is to restore social order. If consumers perceive that social order has already been restored such that the other involved companies have offered resolutions, such as taking corrective actions to address the disruptions in the social order, they will no longer feel the need to take their own punitive actions.

Method

This study employed a single-factor, three-level (social order restoration after punishment vs. punishment vs. no punishment) between-subjects design. We used the financial services industry as our research context and an experimental procedure similar to that used in our previous studies. The description of the crisis was based on the 2022 cyber breaches in Australian financial service firms (Morse, 2023; Redrup, 2022). It informed participants of an industry-wide crisis in which many financial services companies, including Company X and Company B, were alleged to have data security loopholes in their systems, which allowed their customers’ data to be accessed by other parties and put customers at risk of identity theft and fraud. In the punishment condition, participants were informed about the punishment of Company X. In the social order restoration after punishment condition, participants read that following the punishment of Company X, other allegedly involved companies announced that they had hired experts to resolve the data security loopholes, heavily invested in new data management systems and infrastructures, and compensated customers who had been adversely impacted. In the no punishment condition, participants read that Company X had been neither fined nor boycotted.

Punitive intent (α = 0.92) was measured using the same items as in our previous studies. We also measured the social order restoration motive using the same item as in Study 6 to serve as a manipulation check. All items were measured on a 7-point scale (1 = “not at all” and 7 = “very much”). At the end of the survey, we measured participants’ perceived scenario realism and collected their demographic information.

Results

We recruited 305 U.S.-based Prolific panelists (47.2% female; M_age = 37.78, SD = 12.64) to participate in this study. Participants found our experimental scenarios to be realistic, and a one-sample t-test showed that their perceptions of realism were significantly higher than the midpoint of the scale (4 = “moderate”; M = 5.39, SD = 1.33, t(304) = 18.24, p < 0.001, d = 1.04). The differences in perceived realism between conditions were not significant (all p’s > 0.05).

Manipulation check

A one-way ANOVA on the social order restoration motive revealed a significant effect of the manipulation condition (F(2, 302) = 7.49, p < 0.001). Further planned contrasts showed that consistent with our theorizing, participants showed a greater social order restoration motive when they observed punishment only than when they observed no punishment (M_punishment = 4.83, SD = 1.76 vs. M_{no_punishment} = 4.16, SD = 1.77, t(302) = 2.69, p = 0.007, d = 0.38). Moreover, participants had a lower social restoration motive when they observed that other involved companies had taken social order restoration actions than when participants observed punishment only (M_{restoration_after_punishment} = 3.91, SD = 1.79 vs. M_punishment = 4.83, SD = 1.76, t(302) = 3.74, p < 0.001, d = 0.52). There was no significant difference in the motive between the condition with social order restoration after punishment and the condition with no punishment (t(302) = 0.97, p = 0.334). Thus, the manipulation of the social order restoration motive was successful.

Punitive intent

A one-way ANOVA on punitive intent showed a significant effect of the manipulation condition (F(2, 302) = 31.72, p < 0.001). Planned contrasts showed that, consistent with our previous studies, participants showed greater punitive intent in the punishment condition than in the no punishment condition (M_punishment = 5.12, SD = 1.46 vs. M_{no_punishment} = 4.35, SD = 1.68, t(302) = 3.33, p < 0.001, d = 0.47). Supporting our prediction, participants who learned that the involved companies had taken actions to restore the social order showed lower punitive intent than did participants who observed punishment only (M_{restoration_after_punishment} = 4.05, SD = 1.77, vs. M_punishment = 5.12, SD = 1.46, t(302) = 4.71, p < 0.001, d = 0.65). There was no significant difference between the condition with social order restoration after punishment and the condition with no punishment (t(302) = 1.29, p = 0.200).

Discussion

This study replicates our proposed main effect in a new crisis context of data management and privacy breaches. We show that observing one company being punished increases consumers’ intention to punish other involved companies. Moreover, we demonstrate that if other involved companies proactively take resolution actions to remedy damage, consumers will perceive that the social order has been restored and that they no longer need to take any punitive actions. Thus, this study provides additional support that consumers’ motive to punish companies is to restore social order. More importantly, the findings offer an ethical crisis resolution strategy to companies involved in an industry-wide crisis.

General discussion

Across various industries and different types of industry-wide crises, after observing one involved company being punished, consumers increase their intent to punish other involved companies (Studies 3–7) and exhibit actual punitive behaviors such as complaints (Study 1) and withholding participation (Study 2). The reason for this effect is that observing such punishment grants consumers certainty in attributing blame to the involved companies, which elevates their motive to restore social order and drives them to punish other involved companies that have not yet been punished. We provide direct process evidence for the proposed main effect (Studies 3–5) and rule out competing explanations such as crisis severity and just world belief (Study 3).

Our research further identifies boundary conditions in the main effect based on factors that can influence certainty in blame attribution and the social order restoration motive. Specifically, when a company objects to a punishment (Study 4), punishment information no longer enhances consumers’ certainty in blame attribution, which in turn decreases their punitive intent. When other companies are not explicitly involved in the crisis (Study 5), consumers’ certainty in attributing blame to these companies becomes low, which reduces their motive to restore social order and their intent to punish these companies. When the punishment is imposed by a party with low formality (Study 6), the observed punishment cannot validate blame attribution effectively, and consumers lower their intention to punish. Furthermore, when the other involved companies have offered resolutions to restore social order (Study 7), consumers’ social order restoration motive declines, subsequently lowering consumers’ punitive intent.

Theoretical contributions

Our research makes three theoretical contributions. First, although extant research has investigated consumer responses to single-brand crises (e.g., Cleeren et al., 2008; Khamitov et al., 2020; Klein & Dawar, 2004), research on consumers’ responses to industry-wide crises remains relatively limited (Chen et al., 2022; Gao et al., 2012; Zavyalova et al., 2012). More importantly, even though it is common for one company or a few companies to be exposed and punished prior to other involved companies during industry-wide crises (Zavyalova et al., 2012; Zhang et al., 2020), how consumers respond to other involved companies after observing such punishment has not been examined. We address this research gap and show that observing the punishment of an involved company can increase consumers’ intention to punish other companies implicated in similar issues. Thus, our findings enrich the literature on industry-wide crises by documenting a novel effect and providing process evidence for its underlying mechanism that is unique to this context.

Second, extant research on single-brand crises has applied blame attribution theory to investigate how consumers make sense of crises and determine their responses (e.g., Klein & Dawar, 2004; Lei et al., 2012). Building on this stream of literature, we explore blame attribution, its subjective certainty, and its validation process in a new context of industry-wide crises. Previous research on product-harm and service crises (mostly single-brand crises) has shown that base rate information (i.e., how common a particular type of crisis is) (Lei et al., 2012), branded outsourcing information (Kranzbühler et al., 2019), and information about customer co-creation (Heidenreich et al., 2015) can serve as signals to help consumers adjust their blame attribution. We identify a novel signaling effect of the punishment of an involved company in the context of complex industry-wide crises, and we demonstrate how it affects consumers’ certainty in assigning blame and the subsequent punishment. We further show that consumers evaluate the characteristics of the observed punishment to decide whether this punishment information can be applied to validate their blame attribution. When the punishment is imposed by a party with lower degrees of formality or appealed by the punishment recipient, consumers no longer use it to enhance their blame attribution certainty.

Finally, prior studies on consumer punishment have focused mostly on direct victims and their revenge motives (e.g., Bechwati & Morrin, 2003; Grégoire et al., 2010), while relatively less attention has been given to third-party consumers who are not directly impacted by crises (Lin et al., 2013; Xie & Bagozzi, 2019; Xu et al., 2021). Building on the literature on “third-party punishment” or “altruistic punishment” (Fehr & Gächter, 2002; Lin et al., 2013; Toribio-Flórez et al., 2023), we show that third-party punishment can also occur in the context of industry-wide crises and is driven by a motive to restore social order. In addition to social order restoration motives, recent research shows that third-party punishment may be influenced by negative emotions such as moral outrage (Fehr & Gächter, 2002; Pedersen et al., 2018; Xie & Bagozzi, 2019). Based on this perspective, consumers may experience less outrage after observing the punishment of an involved company, and thus, they may be less likely to retaliate against other involved companies. In a separate study (n = 156), we found that observing one involved company get punished can lead to consumers’ discharged outrage (p < 0.001), but this discharged outrage had no effect on punitive intent toward the other involved companies (p = 0.696). This result is somewhat consistent with the findings of Pedersen et al. (2018) that third-party consumers also experience outrage when observing strangers being harmed by violations, but this type of emotional response may not be strong enough to override the social order restoration motive to influence consumers’ punitive intent. Thus, our findings deepen the understanding of consumer third-party punishment from an altruistic punishment perspective (Fehr & Gächter, 2002; Lin et al., 2013).

Managerial implications

Our findings have important practical implications for managing industry-wide crises. An outbreak of a crisis is similar to a storm characterized by public outrage and an urgent need for scrutiny. When a company is put under the spotlight and punished during an industry-wide crisis, other involved companies may believe that company serves as a scapegoat and will take the heat off them, sheltering them from blame and punishment.

However, our research shows the perils of adopting this belief. Specifically, we show that observing the punishment of an involved company can increase consumers’ intention to punish other companies facing similar issues. The downstream consequences of punitive intent include negative word-of-mouth, complaints (Grégoire et al., 2018), and refusal to buy (Pigors & Rockenbach, 2016). Thus, it is wise for other involved companies to initiate crisis management procedures, especially if they are directly responsible for the crisis. As shown in our research, consumers’ punitive intent declines when they see that the social order in the affected industry is largely restored. Therefore, offering resolutions to the crisis, addressing the damage, and providing compensation may be appropriate courses of action for the involved companies (Nikkhah & Grover, 2022).

Not all observed punishments will trigger consumers’ punitive reactions. When a company is punished, other companies facing similar issues should quickly evaluate the punisher and the punished company’s response. By analyzing the characteristics of the punishment, companies can develop effective crisis management plans and avoid potential backlash from consumers. First, companies need to evaluate the formality of the punisher: if the punishment is imposed by an entity that has high formality, other companies facing similar issues will be in greater danger. Therefore, these companies need to prepare for public communications in advance, engage crisis management experts to stem the rise of potential backlash from consumers, and demonstrate their efforts to restore the social order in the affected industry. Second, companies need to pay attention to whether the recipient of the punishment has filed an appeal. Given the interconnected and complex nature of the global supply chain, industry-wide crises can be attributed to various parties (e.g., suppliers and retailers) and even regulatory systems. This offers opportunities for companies not directly responsible for the crisis to appeal their punishment. Such appeals can raise doubts about blame attribution and reduce the possibility of consumers punishing other involved companies. Therefore, industry associations should deploy legal experts to assist companies that are not directly responsible for the crisis.

However, simply objecting to the punishment is not a viable strategy for companies truly responsible for the crisis, as it would trigger further reviews, investigations, and exposures. Considering the large-scale impact of industry-wide crises, scrutiny from stakeholders and authorities would intensify along with media coverage, ensuring the eventual revelation of the truth. In such circumstances, consumers may await further evidence or seek alternative sources of information. Once they have sufficient certainty in blame attribution, they may impose harsh punishment on the truly responsible companies.

Limitations and future research

We have identified several remaining questions that offer directions for future studies. First, we have not tested our effect in other crisis contexts, such as crises with less ambiguity, crises caused by parties external to the affected industry, crises that occurred in other countries, and crises of personal relevance to consumers (e.g., consumers are close to the victims). Based on our theorizing, when consumers become certain in their blame attribution, their motive to restore social order should apply to any parties that are responsible for crises (see consumers protest against governments for their lack of food regulation; Lee, 2015). Within specific crisis contexts, consumers’ punishment intent may be affected by other factors, such as ethnocentrism (i.e., consumers’ biases in favor of domestic over foreign brands and products) (Gürhan-Canli & Maheswaran, 2000; Siamagka & Balabanis, 2015) and revenge on behalf of in-group members (Zhang et al., 2022). Future research can further explore these crisis contexts.

Second, other characteristics of the observed punishment may influence consumers’ punitive intent. Future research could investigate the impacts of punishment severity. It is plausible that harsher punishments are associated with stronger evidence and justification, influencing consumers’ certainty in blame attribution. Additionally, future studies could test what type of punishment (e.g., warnings, fines, criminal prosecution) or threshold of punishment is deemed adequate for restoring social order, thereby reducing the need for further consumer punishment. It is also necessary to consider contexts where consumers lack opportunities to impose their own punishment after observing a punishment event.

Third, we have not fully explored consumers’ moral characteristics and their relationships with brands. Previous studies have shown consumers’ empathetic traits (Karampournioti et al., 2018), other-regarding virtues and prosocial values (Mar García-de los Salmones et al., 2021; Grappi et al., 2013; Russell et al., 2016), identity, and collective self-concepts (Xie & Bagozzi, 2019) all affect their evaluations of and responses to wrongdoers. Additionally, if consumers have a strong relationship with or a positive prior perception of a company (Mafael et al., 2022), they may discount the information they receive regarding its involvement in a crisis (Ingram et al., 2005), justify its involvement as inevitable due to uncontrollable factors (Eckhardt et al., 2010), and discount the credibility of sources that accuse the company of wrongdoing (Yuksel, 2013). We invite future research to examine these factors.

We examine perceived unfairness in Studies 3 and 5 and show that in the presence of this alternative process, the mediating effect of blame attribution certainty still holds. Furthermore, we ran additional tests (see Web Appendices B and C). The results show that, across all the models, certainty in blame attribution always has a significant mediating effect, providing additional support for our proposed mechanism. However, as perceived unfairness also has significant mediating effects in some models and seems to serve as an alternative motivation to drive punitive intent, it cannot be ruled out as an alternative process. Perceived unfairness may play different roles in this context, and we invite future research to explore them.

Finally, future research should further compare our effect with two established effects in the crisis management literature, the scapegoating effect and the spillover effect. The scapegoating effect occurs when a member of a group is singled out and punished, thereby sheltering the remainder of the group from blame and punishment (Douglas, 1995). Scapegoating strategies are studied mainly in organizational research. When an organization faces a performance crisis, letting a CEO or manager shoulder the blame can provide closure or signal a resolution of the crisis, enhancing the organization’s survival prospects (Boeker, 1992; Bonazzi, 1983; Cameron et al., 1987; Cannella & Lubatkin, 1993; Gangloff et al., 2016). Thus, the scapegoating effect may suggest that observing one involved company being punished will spare the other involved companies from being punished. In the context of industry-wide crises, two papers applied the term “scapegoating” (Gao et al., 2012; Zhang et al., 2020); however, neither examined consumers’ responses to other involved brands. Our findings seem to diverge from the prediction of the scapegoating effect. We speculate that this discrepancy could be driven by the extent to which consumers feel that the social order has been restored (or a full closure or resolution is offered). In industry-wide crises where the damage could be severe and multiple companies are accused, punishing one company may not be viewed as offering full closure or a resolution. However, we find that when other involved companies offer resolutions to the crisis, consumers’ punitive intent declines (Study 7). In other words, the resolutions offered by other involved companies may also create closure for consumers, which is aligned with the prediction based on the scapegoating effect. We invite future research to empirically delineate these two effects.

The negative spillover effect is documented in studies about single-brand crises, and it shows that one brand’s scandal can negatively impact non-scandalized brands in the same industry via “guilt by association” (Che et al., 2023; Roehm & Tybout, 2006; Votola & Unnava, 2006). Thus, a negative spillover effect may predict and explain how punishing one involved company increases consumers’ punitive intent toward other involved companies. Our context differs from prior research in the context of a single-brand crisis, as in our research, consumers’ received information concerns punishment (rather than only the crisis), and the other companies are implicated in the crisis (rather than not being implicated). However, we cannot fully rule out that “guilt by association” may also play a role. We invite future research to empirically separate the negative spillover effect from our proposed effect.