When one bad apple spoils consumers’ judgment of the brand: exposure to an employee’s non-workplace transgression and potential remedies

Abstract

Four experiments investigate consumers’ moral judgments of a firm’s brand reputation when given information about an employee’s non-workplace transgression. To the extent that the employee is perceived to have power in the firm (i.e., control over resources and decisions), the employee’s offensive action damages the firm’s reputation and decreases consumers’ purchase intentions. These effects occur even though the action occurs in the employee’s private life and is unrelated to product quality. The results replicate for three types of products and three types of offenses. The employee’s perceived power in the firm provides the most consistent explanation of customers’ negative responses and is a better predictor than the alternative explanations tested (e.g., perceived status). Results also show that after an offense comes to light, firm reactions that decrease or eliminate the employee’s power in the organization—such as reducing decision-making responsibilities or firing the employee—can help restore the firm’s reputation.

It has long been appreciated that consumers’ judgments about a firm influence their purchase decisions for its products and services. Corporate social responsibility research in particular has addressed the influence of ethical information on consumers’ and other stakeholders’ judgments (e.g., Brown et al. 2006). A company’s immoral and moral behaviors influence consumers’ judgments even when those behaviors have no direct relevance for product quality (Kirmani et al. 2017). Even before the turn of the millennium, most consumers reported that organizational ethics influenced their purchase and word-of-mouth decisions (Ipsos 1999). This “moralization of the marketplace” has been building over the past several decades (Shamir 2008) and seems likely to continue. Hence, firms are wise to be concerned about their moral reputations with consumers.

How vulnerable is a firm to damage to its moral reputation? We know that the ethicality of a firm-level action influences consumers’ judgments and behavior (e.g., Kang et al. 2015; Peloza and Shang 2011; Romani et al. 2013). A firm can be vulnerable to similar moral reputation risk if a small subset of employees make decisions that lead the firm to act unethically, such as in the VW diesel emissions scandal or Equifax’s delay in warning the public about breaches to its database. Moreover, anecdotal information suggests that a firm may be vulnerable to damage to its moral reputation even when an individual employee behaves unethically in his or her private life. Many high-level individual employees who engaged in unethical behavior in their private lives—such as the CEOs of Uber, of Martha Stewart Omnimedia, and of American Apparel—have been fired due to concerns about the impact of their respective firms’ continued association with their behavior. However, the latter examples blur the lines of public and private actions, since those specific CEOs were also founders whose personas became part of the corporate brand and who were in the public eye before any scandals came to light. The unanswered question is whether information about an ordinary employee’s offensive behavior in his or her private life can damage consumers’ moral judgments of the firm.

This is a realistic concern because through social media and other online resources, consumers have access to more information than ever about who works for which company and in what capacity (e.g., president, manager). Whether it is sourced from the media (Gans 2011) or as a result of a public “outing” by motivated internet provocateurs (e.g., Suggs 2016), information about individuals’ private actions can become linked to the firm. Based on reports that many employees have been fired for private, non-workplace behavior (Halper 2015), there is an apparent belief that information about a non-workplace employee offense will cause harm to the company. Yet there is no direct empirical evidence in the marketing literature to support such an effect. Our research addresses this gap.

Our main goal is to identify the conditions in which this damage to the firm’s reputation might occur. We predict that consumers’ beliefs about the employee’s power—defined as influence over the firm’s decisions, control over firm resources, or the ability to reward or punish other employees—will determine the impact of the employee’s transgression on judgments of the firm. In a series of experiments, we present evidence that consumers’ lay-perception of power, which is inferred based on the employee’s position title, predicts whether that employee’s non-workplace transgression damages those consumers’ moral judgments of the firm. Further, we show that power has a different effect on consumers’ moral judgments when the employee behaves virtuously, and a transgressing employee’s past association with the company is insufficient to influence firm moral judgments. In experiments that involve information about real companies and a recent, actual employee offense, we identify specific actions that the firm can take to counteract damaging effects after an offense has come to light.

Consumers’ moral judgment of firms

Our research focuses on consumers’ perceptions of an individual agent or actor who is employed by a firm, and on how those perceptions affect inferences about the firm. Most previous marketing research has examined effects of information about morally relevant actions for three types of agents or actors associated with the firm: (1) the individual firm employee (i.e., an employee who behaves offensively while at work), (2) the firm as a whole (e.g., an organization’s policies and practices of social responsibility), and (3) the public figure associated with the firm (e.g., the paid celebrity endorser who transgresses in his or her private life). Though none has directly addressed the effects of an individual employee’s non-workplace behavior, some aspects of each of these literatures are relevant to our investigation.

Moral judgment based on employee or firm actions

Research on the effects of employees’ behavior suggests that their negative behaviors at work influence consumers’ evaluations of the firm and its services, even if the behavior is merely observed and has no consequence for the consumer (e.g., an employee’s incivility toward another employee; Porath et al. 2010). Similarly, research on corporate image has examined outsiders’ judgments of the organization as a consequence of a company’s moral and immoral actions (e.g., Sen and Bhattacharya 2001; Ellen et al. 2006). Consumers value companies that are perceived to engage in socially responsible activities: both operationally, such as environmental sustainability, and externally, such as charitable contributions (Lacey et al. 2015; Peloza and Shang 2011). Corporate social responsibility (CSR) policies and actions influence consumers’ brand judgments and purchase intentions because the image of the firm as a whole is a component of the overall brand image (Brown and Dacin 1997; Hsieh et al. 2004). These judgments influence the person’s brand purchase behavior (Hsieh et al. 2004); a positive image leads to superior business outcomes (Simões et al. 2005).

Many of the socially responsible activities that contribute to corporate image reflect actions of individual employees undertaken while not at work, whether they are donations of time, money, or one’s own blood to a cause (Pappu and Cornwell 2014). Managers, in particular, can play a key role in conveying the company’s values (Simões et al. 2005). Hence, it is possible that consumers equate managers with firms in their moral judgments. Just as with  ir responsible firm activities, perhaps information about a manager’s unethical behavior outside the workplace can influence consumers’ brand evaluations.

However, the CSR literature suggests that unless the employee’s non-workplace transgression is in some way intended by the company, it will have little influence on consumers’ judgments. Consumers tend to disregard behaviors not intended by the company (Ellen et al. 2006). Hence, consumers seem unlikely to judge companies or their brands harshly for an individual employee’s non-workplace, offensive behavior -- that the firm would not have intended, and that furthermore has no impact on product quality (Folkes and Kamins 1999; Kirmani et al. 2017). Collectively punishing others for the misdeed of an individual member whose behavior the group cannot control is considered unfair (Lickel et al. 2003). In sum, neither the literature investigating effects of individual employees nor the CSR literature provide a clear answer to our question.

Celebrity endorser research

Celebrity endorser research has examined the effects of information about an individual’s non-workplace behavior on brand evaluations. A celebrity endorser is “anyone who enjoys public recognition and who uses this recognition on behalf of a consumer good by appearing with it in an advertisement” (McCracken 1989, p. 310). Mere co-occurrence of the celebrity with the brand can lead the consumer to transfer meanings attached to the celebrity onto the brand.

Evaluative conditioning may account for information about a celebrity enhancing consumers’ brand evaluations when meanings are positive, and detracting from it when negative (Miller and Allen 2012). From the perspective of a network model of associative memory, a scandal involving a celebrity activates memories of other linked concepts, such as the brand (Carrillat et al. 2014). These processes of cognitive association suggest that consumers also would transfer negative inferences from an employee’s transgression to the brand, given the employee’s connection with the firm. Further, the network model suggests that eliminating the learned cognitive association with the brand could ameliorate negative effects.

There is mixed evidence about whether learned associations are eliminated by informing consumers that the endorser has been dismissed. Some research indicates that whether the company retains or dismisses the blameworthy endorser can result in no difference in consumers’ judgment of the company (Louie and Obermiller 2002; Study 1). Yet, in research using stock market returns as indicators, the negative effect of a scandal is ameliorated by terminating the celebrity endorsement contract (Bartz et al. 2013). In other research gauging consumers’ brand attitudes and purchase intentions, terminating the contract of a scandal-ridden celebrity endorser who is seen as a meaningfully good fit (i.e., congruent) with the brand led to more negative consumer reactions than maintaining the relationship (Carrillat et al. 2013).

Perhaps a more critical question is whether the celebrity endorser literature is relevant to understanding how an employee’s transgressions might influence consumers’ evaluations of the firm’s brand. On the one hand, there are similarities between celebrities and regular employees in that decision makers in the firm have chosen to create a link with the person, so the basis for the association is similar. On the other hand, perceptions of employees may differ from celebrities in ways that limit the ability to generalize from the implications of celebrity endorser research. Celebrities have similar properties to brands, so much so that they can be considered “human brands” (Thomson 2006). Being a human brand might lend celebrities to be idealized or put on a pedestal (Rockwell and Giles 2009), which may make immoral behavior less tolerable.

Consumers are sometimes so committed to supporting a celebrity or public figure that they will pursue reasoning strategies that allow them to decouple their judgments of the person’s immoral behavior from their judgments of whatever product or brand associated with the individual (Bhattacharjee et al. 2013). This sort of rationalization is easier when the transgression is in another domain, one that is seen as irrelevant to the product or service. For example, in that research, consumers will support a baseball player who evades taxes more than a baseball player who uses steroids, because the latter is relevant to the individual’s sphere of influence. The source of this effect is a motivation to defend the celebrity or public figure rather than a general motivation to defend all employees, so this research is informative but distinct from the contribution of the research we present here.

Moreover, celebrity conveys status and prestige, which may be damaged more when a celebrity’s reputation is tarnished compared to an employee. To the extent that consumers’ unusually high regard for celebrities accounts for the endorser’s transgression effects on consumers’ brand evaluations, that element would be missing for employees (with the exception of celebrity CEOs who are public figures, such as Bill Gates and Mark Zuckerberg). Another difference between celebrities and employees is that consumers also may perceive more permeable boundaries between the celebrities’ and public figures’ “work” activities and their private lives than they might for a non-celebrity. Public figures lose their privacy in consumers’ eyes (Rockwell and Giles 2009). In contrast, the stronger boundary between a non-celebrity employee’s work and private life may prevent a non-workplace offense from influencing brand evaluations.

In short, differences between consumers’ perceptions of celebrities and non-celebrity employees are such that it is unclear whether the celebrity endorser findings can be generalized to a low-level employee who lacks a prestigious title or even to a high-level employee who has a prestigious title but not the revered characteristics or public renown of a celebrity. Note that the purpose of our research is not to compare effects of information about employees versus celebrities, but instead to understand the impact of employees’ non-workplace transgressions on consumer judgments of their firms as corporate brands.

Hypothesis development

When do non-workplace employee transgressions affect consumers’ moral judgments?

People often generalize from information about one individual group member to form an impression of the group (Pryor et al. 2012). Just as with a celebrity and a brand, mere co-occurrence of an employee with the brand may lead the consumer to transfer meanings attached to the employee’s action onto the brand (Bartz et al. 2013). Whereas an association may be sufficient for the action of any employee to influence brand moral judgment, a powerful employee’s transgression may have even more impact.

We propose that consumers’ perceptions of the transgressing employee’s power predict the extent of influence on brand moral judgments. In research on organizations, power is defined as the extent to which the employee controls important resources within the firm, has the ability to reward or punish others, and influences important decisions by the firm (Magee and Galinsky 2008). Power thus implies control over what the company is and does. The more a person controls the organization’s resources and decisions, the more the person is likely to be seen as responsible for the firm’s behavior and identity. In contrast, people believe that a person who lacks power is more constrained by the situation and has less freedom to express his or her own ideas and to influence others (Galinsky et al. 2008; Magee 2009).

Further, a powerful employee’s transgression may be particularly diagnostic because immoral behavior is less normative and less socially desirable than virtuous behavior. Virtuous behaviors can be motivated by moral goodness, per se, but also by other motives, such as the desire to enhance one’s status, prestige, or respect (Andreoni 1990). That makes an individual employee’s virtuous behavior less diagnostic about the firm’s moral identity. Immoral behavior is more diagnostic and, as with the effect of a powerful person’s responsibility for the firm’s behavior, increases the likelihood of the consumer’s inferring an immoral identity for the firm.

In sum, information about a powerful employee’s transgression should influence judgments of the firm’s morality more than the same information about a less powerful employee’s behavior, but the effect of a virtuous employee’s power should have less impact. Previous research shows that perceptions of a powerful person within a firm often influences external stakeholders’ judgments of the firm, though not that it influences consumers’ purchase intentions (Gomulya et al. 2016). Other research shows that stigmatized leaders can detract from consumers’ valuations of their firms’ products (Avery et al. 2016). However, differential effects of employee power have not been examined.

These potential effects of power are important for managers to understand because prior research on corporate social responsibility indicates that brand moral judgment is likely to influence purchase intentions (e.g., Brown and Dacin 1997). We predict that the effect of perceived employee power in cases of non-workplace transgressions can extend to purchase intentions through its effect on moral judgments of the firm’s brand. Our hypotheses are thus:

• H1: The employee’s perceived power in the firm moderates the effect of the non-workplace action on brand moral judgment, such that a more powerful employee’s action, particularly if it is immoral (i.e., negative valence), has a greater effect than does a less powerful employee’s.

• H2: Brand moral judgment positively influences purchase intention and mediates the effect of a non-workplace employee action on purchase intention.

Potential remedies

We also investigate whether a firm’s response to the offense might mitigate negative consumer judgments. This is important practically as well as theoretically. If, as we have argued, a person’s power over the firm’s resources and decisions accounts for the negative effect of transgressions on brand moral judgments, then reducing that power should help repair the damage. For example, decreasing a powerful employee’s power by terminating employment, or by stripping the employee of his or her responsibilities, should decrease the employee’s influence over others and ability to allocate the firm’s resources toward potentially offensive actions in the future.

After an ethical violation, firms often replace the CEO, which implies an underlying belief that eliminating the association with a powerful person will remedy stakeholders’ negative opinions of the firm (Gomulya et al. 2016). On the other hand, some research suggests that termination is an ineffective remedy. The celebrity endorsement literature suggests that there may be times when the learned association to the brand is not so easily overcome as by merely ousting the transgressor (Carrillat et al. 2013; Louie and Obermiller 2002). In that research, perceived fit between the celebrity and brand made termination less effective as a tactic to prevent damage from a scandal. We therefore test termination and compare it to demotion—specifically achieved by taking away resources and responsibilities, which are indicators of power (Magee and Galinsky 2008)—as another potential remedy for a non-workplace employee transgression. The test also provides further evidence regarding the role of employee power as a mechanism in determining the extent to which a non-workplace transgression affects consumers.

• H3: Firm reaction can reduce the effect of a non-workplace employee transgression, such that decreasing the employee’s perceived power can improve brand moral judgment.

Overview of studies

A series of four experiments test these hypotheses. Study 1 tests H1 and H2 in a controlled experiment. Study 2 provides evidence of the robustness and external validity of the effect of a non-workplace transgression on brand moral judgments. Studies 3 and 4 test H3, investigating the effect of firm reactions and providing further evidence for perceived employee power effects on consumers’ brand moral judgments. The conceptual model in Fig. 1 represents the core constructs and predictions guiding the following research.

Fig. 1

Conceptual model

Our studies are in the common context of a consumer being exposed to a news report about a non-workplace transgression by an employee; within that context we create variation to examine the pattern of effects. Specifically, we vary the employee’s position within the firm (e.g., manager, clerk, assembly-line worker) because this is the kind of information typically conveyed to consumers in a media report. Variation in perceived power between participants in different experimental conditions is achieved indirectly and is externally valid, while also avoiding a potential confound of the manipulation with demand effects that might be spurred by artificially directing attention to the power of the employee’s position. As well, this operationalization allows us to test alternative explanations for the effects observed.

Study 1: Employee power and action valence

Study 1 tested the predicted relationships in a lab experiment designed to minimize extraneous influences. In addition to examining consumers’ reactions to offensive non-workplace behavior, this study investigated reactions to an employee’s virtuous behavior. If power increases the negative effect of offensive behavior, the effects of power are likely to be reversed in regard to virtuous behavior, such that a more powerful employee enhances brand moral judgments, and does so to a greater extent than a less powerful employee. However, we expect stronger effects on brand moral judgments when the individual’s behavior is immoral rather than moral, based on Coovert and Reeder’s (1990) findings and based on research suggesting that negative information is generally weighed more heavily than positive information in judgments (Ahluwalia 2002; Herr et al. 1991). Therefore, we expect negative information to exert an asymmetrically strong influence on brand moral judgments.

As well, we tested two alternative explanations: perceptions of the focal employee’s status and similarity to other employees. Status is the inferred respect for the individual, such as might come from long tenure at the firm, demonstrations of competency, or a prestigious position and title (Magee and Galinsky 2008). In the corporate world, people may gain respect and status if they have power and may gain power from having status and respect. Although status and power co-vary in most organizational contexts, research has shown that power and status can have different effects on an individual’s behavior, as well as on others’ reactions to the person (Magee and Galinsky 2008). Whereas power is bestowed by the firm via various responsibilities and resources, status depends more on others’ bestowing admiration and respect. High status might suggest less likelihood of committing a future transgression and greater likelihood of behaving virtuously than lower status. However, there might also be higher expectations of those we respect, so that the transgressions of a high-status employees seem more severe and their virtuous behavior less good. We therefore include perceived status as a potential alternative explanation and examine its role.

Another alternative explanation rests on the perception of similarity between the individual employee and other employees. People are more likely to generalize from information about an individual to the group when there is assumed similarity among its members (Coovert and Reeder 1990). Perceivers may assume some degree of similarity for those who work together (Crawford et al. 2002). However, the extent to which such assumptions are made may vary based on what is known about the employee and other situational factors. Perceivers may generalize to the firm more from an employee who is perceived as similar to other employees than from one perceived as dissimilar. For example, consumers may assume that one teller at a bank is similar to the other employees because many bank employees are tellers. In comparison, a bank officer might be seen as less similar to other employees because that person has been selected for promotion, which could imply the person is exceptional and has different qualities and characteristics from most other employees. This example would imply that a lower level employee could have greater effect on brand moral judgments. However, if a manager is seen as more similar to a “typical” or “ideal” company employee, similarity could also result in the same pattern of effects as we predict for perceived power. Therefore, Study 1 also assesses perceived similarity as a potential alternative explanation.

Method

Participants and design

One hundred and eighty students at the University of Southern California participated for partial course credit (M _age  = 20.6 years, 61% female). Participants were randomly assigned to conditions in a 2 (non-workplace employee action valence: positive, negative) × 3 (employee power: high, low, control) full-factorial between-subjects design.

Procedure

Participants first saw a description of a brand of high quality clothes. The clothing company was referred to by an obviously disguised name, “Company K,” to imply that this might be a real company from which they might have purchased. The description stated that a consumer magazine had rated the company’s clothes as among the top brands of clothing, in terms of style, design, and durability. The company’s customers were described as “college students (aged 18-24),” to indicate to participants that they would find these clothes desirable. As expected, participants indicated on a pre-manipulation measure that they would be willing to purchase the clothes based on this description (M _pretest  = 6.78, SD = 1.52, on a 9-point scale labeled “not at all” to “very” willing to purchase), and there were no preexisting differences between groups before the manipulation was presented (p > .17).

Participants then read a scenario about an employee of the clothing company. The information was described as having been discovered in “reputable news reports.” Similar to previous research (Magee and Galinsky 2008), our studies manipulate perceived power via the person’s position in the organization. For the high level of employee power, the moral and immoral actions were attributed to the head designer of the company’s clothes. An assembly-line worker was used in the low employee power conditions. The actions of a former assembly-line worker were used for the control (no power) employee conditions. In contrast to the high and low power conditions, this group read that the person had not worked for the company at the time of the action described but had worked for the company five years ago for a brief time. The gender of the employee was never referenced: the person was always identified by the job description and no gendered pronouns were used.

To manipulate valence, participants either read that an employee had saved a child from being abused (positive valence) or that the employee had been convicted of child abuse (negative valence). Association with child abuse has been shown to be a stigma for which an individual is judged responsible and leads to avoidance of that individual, even more so than other kinds of stigma (e.g., obesity, depression; Pryor et al. 2004). As well, this private-life offense is similar to that used by Bhattacharjee et al. (2013, Study 1: spousal abuse by a public figure). No details of the offense were given to participants in this study, but they read that the employee was given probation “because it was a first offense,” and (except in the control condition) the person was described as still working for the company.

Measures

Purchase intention and product quality

After reading the description of the employee action (varied by condition), all participants rated again whether they would purchase the company’s clothes on two 9-point semantic differential scales, from “definitely would not” to “definitely would,” and from “not at all likely” to “very likely” (α = .94). They also indicated product quality perceptions on two items: from “bad” to “good” and “low quality” to “high quality.” As intended, no significant effects of non-workplace actions on product quality are observed (M_high-power = 6.96 [SD = 1.78], M_low-power = 7.06 (SD = 1.43), M_control = 7.48 [SD = 1.58]; p’s > .21). Neither the position of the employee nor the valence of the employee’s non-workplace action affects perceptions of product quality (F’s < 2.30, p’s > .13).

Brand moral judgment

Brand moral judgments were measured as the extent to which the company was considered (1) “highly immoral” versus “highly moral,” (2) “not at all harmful” versus “very harmful” (reverse coded), and (3) “very cruel” versus “very compassionate.” Those same scales were then used to measure moral judgments of the employee’s non-workplace action (α’s > .90). Moral judgments of the employee’s action exhibit only a significant main effect of valence (M _moral  = 8.32 [SD = 1.01] vs. M _immoral  = 1.60 [SD = 1.00]; F(1, 178) = 2009.42, p < .01). The main effect of employee position and the interaction of position with valence for employee moral judgments are not significant, as expected.

Power

Participants rated employee power as the extent to which the employee has “power in the company” and is able to “influence other employees” (α = .76). The manipulation of the employee’s position title shows a significant main effect on power perceptions (F(2, 177) = 84.42, p < .001). As expected, the head designer is seen as more powerful (M _high-power  = 6.99 [SD = 1.48]) than the assembly-line worker (M_low-power = 4.46 [SD = 1.57]; p < .001), who in turn is seen as more powerful than the former assembly-line employee (M _control  = 3.04 [SD = 1.94]; p < .001). There is also a main effect of valence on employee power perceptions (F(1, 178) = 3.80, p = .05; M _positive  = 5.43 [SD = 2.04] vs. M _negative  = 4.78 [SD = 2.44]). The interaction is marginally significant [F(2, 174) = 2.89, p < .06]: power perception significantly differs between moral and immoral actions only for the control condition, former assembly-line employee (M _{control -positive}  = 3.79 [SD = 2.13] vs. M _{control-negative}  = 2.39 [SD = 1.53]; p < .01), whereas the other conditions show no effect on power perceptions between moral and immoral conditions as expected (M _{low-power-positive}  = 4.73 [SD = 1.45] vs. M _{low-power-negative}  = 4.20 [SD = 1.66]; p = .17; M _{high-power-positive}  = 6.93 [SD = 1.40] vs. M _{high-power-negative}  = 7.05 [SD = 1.58]; p = .77). This difference in the control condition was unexpected but is not relevant to the interpretation of the negative, immoral action conditions where perceived employee power varies as compared to the negative control. For all the important comparisons, the manipulation worked as intended.

Status

Participants also rated employee status as the extent to which the employee’s position “is likely to be seen as prestigious” (M = 4.39, SD = 2.91). Not surprisingly, the employee position manipulation also has a significant main effect on employee status (F(2, 174) = 94.90, p < .01). As with power, the head designer is seen as higher status in the company (M_high-power = 7.01 [SD = 2.14]) than the assembly-line worker (M_low-power = 3.03 [SD = 1.99]; p < .001), while the current and former assembly-line employees were seen as similar in status (M_control = 2.32 [SD = 1.92]; p = .17). Although power and status are strongly correlated (r = .75, p < .01), differences in these employee positions seem better captured by perceptions of power than of status. Still, we include status in the analyses to further examine its effects as an alternative explanation based on the constructs’ close theoretical association.

Similarity

Measures of the perceived similarity between the employee and other company employees show a marginally significant main effect of the employee position manipulation (F(2, 174) = 2.37, p = .10]. Significant differences in similarity perceptions were found only between the control (former employee) and high-power conditions (M_control = 3.24 [SD = 1.95], M_high-power = 4.14 [SD = 1.99]; p = .01). The low-power position (assembly-line worker) (M_low-power = 3.84 [SD = 1.86]) and high-power position (head designer) were equally similar to other employees (p = .33). Though it does not distinguish our employee positions, similarity to other employees is correlated with power (r = .40, p < .01), status (r = .33, p < .01), and brand moral judgments (r = .14, p = .03).

Results

The results support H1: consumers’ brand moral judgment is influenced by an employee’s nonworkplace behavior. Information about a powerful employee’s transgression damages judgments of the brand more than the same information about a less powerful employee. These moral judgments of the brand also mediate the influence of the employee’s non-workplace action on purchase intentions, as predicted by H2.

Brand moral judgment

The degree of power implied by the employee’s position influences brand moral judgment (MJ). ANOVA reveals a significant interaction between employee position condition and action valence on brand moral judgment (F(2, 173) = 20.65, p < .01). The moral and immoral actions by the high-power employee (head designer) show the greatest polarizing effect on brand moral judgment (M _MJ-positive  = 7.20 [SD = 1.52] vs. M _MJ-negative  = 3.76 [SD = 1.46], p < .01). The same actions by a low-power employee (assembly-line worker) also reveal significant differences between positive and negative non-workplace actions (M _MJ-positive  = 6.70 [SD = 1.43] vs. M _MJ-negative  = 4.87 [SD = 1.51], p < .01). As expected, the differences in brand moral judgment are not significant for the control employee’s (former assembly-line worker) positive or negative actions (M _MJ-positive  = 5.97 [SD = 1.47] vs. M _MJ-negative  = 6.21 [SD = 1.32], p = .54).

The significant interaction is driven mainly by negative effects in the immoral non-workplace action conditions: specifically, brand moral judgment is more negative for employees with more power (see Fig. 2a). The high-power employee’s immoral action leads to worse brand moral judgment than the low-power employee (M _{MJ-high-power}  = 3.76 [SD = 1.46] vs. M _MJ-low-power  = 4.87 [SD = 1.51], p < .01). The low-power employee’s immoral action also leads to significantly worse brand moral judgment compared to the control condition (M _MJ-control  = 6.21 [SD = 1.32], p < .01).

Fig. 2

Employee power and action valence effects in Study 1

In the positive moral action conditions, the high-power employee’s moral action relates to marginally more favorable brand moral judgment than the low-power employee (M _{MJ-high-power}  = 7.20 [SD = 1.52] vs. M _MJ-low-power  = 6.70 [SD = 1.43], p = .077). Similarly, the high-power employee’s moral action leads to significantly more favorable brand moral judgment than the control condition (M _MJ-control  = 5.96 [SD = 1.47], p < .01). However, a positive, moral action by a low-power employee does not have a significant effect on brand moral judgment compared to the control condition (p = .15).

Purchase intention

Brand moral judgment positively predicts purchase intention (R² = .31; B = .56, t(178) = 8.96, p < .01). ANOVA reveals a significant interaction between employee position and action valence on purchase intention (F(2, 173) = 11.18, p < .01; see Fig. 2b). As with brand moral judgment, the interaction effect is driven by reduced purchase intentions (PI) in the immoral (negative) non-workplace employee action conditions. The high-power employee’s immoral action leads to lower purchase intentions than the low-power employee (M _{PI-high-power}  = 4.39 [SD = 2.07] vs. M _PI-low-power  = 5.71 [SD = 1.81], p < .01). More importantly, the low-power employee’s immoral action leads to significantly lower purchase intentions compared to the control condition employee (former assembly-line worker) (M _PI-control  = 7.36 [SD = 1.28], p < .01). There are no significant differences in purchase intentions between employee positions for the moral (positive) employee non-workplace action conditions (see Fig. 2b).

Moderated-mediation

To test H1 and H2 simultaneously, we used the continuous measure of power perceptions, rather than the manipulation of employee position, because it allows greater differentiation (Irwin and McClelland 2003). Analysis shows a significant moderated-mediation (Hayes 2013, Model 8), including the dichotomous effect of the non-workplace action valence as moderator. Results reveal a significant effect of perceived employee power on brand moral judgment and purchase intention, and the mediating role of brand moral judgment differs based on the non-workplace action valence (B_{moderated-mediation} = −.22; CI = [−.384; −.102]). Specifically, for a negative, immoral action, perceived employee power negatively predicts purchase intention directly and through brand moral judgment (B_{negative-indirect} = −.13; CI = [−.240; −.059]; B_{negative-direct} = −.25; CI = [−.393; −.112]). Conversely, for a positive action, perceived employee power positively predicts purchase intention through brand moral judgment (B_{positive-indirect} = .09; CI = [.020; .187]; B_{positive-direct} = .03; CI = [−.132; .189]).

Status

The effects of employee position and action valence on brand moral judgments and purchase intentions replicate whether or not perceived employee status is included as a covariate. When including perceived status as a control in the moderated-mediation, perceived employee power still significantly negatively predicts purchase intention directly and through brand moral judgment in the immoral action conditions (B_{negative-indirect} = −.14; CI = [−.281; −.053]; B_{negative-direct} = −.18; CI = [−.360; −.005]), and the overall moderated mediation remains significant (B_{moderated-mediation} = −.22; CI = [−.365; −.097]). However, the effect of perceived power becomes insignificant for positive, moral actions (B_{positive-indirect} = .07; CI = [−.022; .205]) when perceived employee status is controlled.

To further investigate, an alternative moderated-mediation analysis (Model 8) shows that employee status perceptions also predict the mediating role of brand moral judgment (B_{moderated-mediation} = −.17; CI = [−.299; −.078]), when action valence is included as a moderator and power perceptions as a control variable. In contrast to the prior result, brand moral judgments mediate between status perceptions and purchase intention only for the moral action (B_{positive-indirect} = .12; CI = [.050; .232]; B_{positive-direct} = .10; CI = [.058; .259]), but not for the immoral non-workplace employee action (B_{negative-indirect} = −.05; CI = [−.146; .016]; B_{negative-direct} = −.17; CI = [−.304; −.034]). In other words, while employee power is clearly responsible for the effect of a non-workplace employee transgression on damage to judgments of brand morality, employee status instead may be responsible for positive effects of non-workplace employee heroism or other virtues on improved judgments of firm morality.

Discussion

The results of Study 1 show support for H1 and H2. Brand moral judgments are influenced more by a powerful employee’s action than by a less powerful employee’s action (H1), and purchase intentions are affected by these brand moral judgments (H2), though only when the employee’s non-workplace act is a transgression (i.e., negative action valence conditions). Overall, the effect is stronger for transgressions than for virtuous behavior. These results are consistent with research showing that information about an immoral individual is weighed more heavily in judgments of social groups than is information about a moral individual (Coovert and Reeder 1990).

Stated simply, the effects of perceived employee power on moral judgments and purchase intentions are clearest for transgressions. In contrast, for a virtuous employee, perceived status influences brand moral judgments, but these do not predict purchase intentions. When power and status are considered together, perceived employee status better accounts for potential positive effects, but perceived employee power predicts the occurrence of negative effects on the firm from a non-workplace employee transgression.

Note also that the perceived similarity of the heroic or transgressing employee cannot explain the effects observed on brand moral judgments. As reported, current employees—whether relatively higher or lower power in the company—are seen as similar to other employees to the same extent, but a former employee is seen as less similar; in contrast, brand moral judgment shows differential effects of a current employee’s position within the company. Moreover, perceived similarity does not predict the mediating role of brand moral judgments on purchase intentions for either moral (B_indirect = .08; CI = [−.008; .193]) or immoral actions (B_indirect = −.10; CI = [−.210; .014]). In other words, whereas there is some evidence that perceived status is important, inferring similarity to other employees in the company is clearly not supported by the data as a viable alternative explanation. It is interesting to note that, if companies are seeking to promote the virtue of their employees, communications to consumers might be more effective if they emphasize the employee’s status, or respect from others in the company. Nevertheless, information about virtue has a weaker effect on purchase intentions compared to transgressions, as found in prior work (e.g., Folkes and Kamins 1999).

Study 1 provides initial support for our core predictions regarding non-workplace actions’ potential effects on brand moral judgments and purchase intentions (H1 and H2). Although the simulation methodology used here offers a high degree of control, it leaves open the question of whether this effect can be replicated for an actual offense committed by an employee of a familiar firm.

Study 2: A non-workplace offense by an employee of a widely-known company

Study 2’s main goal was to replicate the effect of employee power on consumers’ moral judgments of a real firm—one that is widely known, and for which consumer familiarity with the firm varies naturally. We led Study 1 participants to believe they were familiar with the company based on consistent findings in prior research showing minimal familiarity is necessary for any brand meaning effects to occur (e.g., Berens et al. 2005). Nevertheless, it is possible that familiarity with a specific, real brand may have different effects from the vague sense of familiarity we created in Study 1.

We also measured involvement because it might motivate consumers to judge a brand more favorably, according to the moral reasoning literature (Paharia et al. 2013). Consumers may engage in motivated reasoning to diminish the immorality of the employee’s behavior or dismiss the company’s association with it when they are highly involved or connected to the company. Study 2 therefore also tests both familiarity and involvement, not as alternative explanations but rather as potential limits on the generalizability of the effects.

Study 2 departs from the full conceptual model in Fig. 1 by addressing only transgressions. Fig. 3 illustrates the variables as operationalized in the following studies, focused on understanding the negative effects of non-workplace employee transgressions. The existence of a non-workplace employee transgression (negative, immoral action) is held constant, and therefore the non-workplace employee action is exogenous in the following studies.

Fig. 3

Operational model for Study 2

Method

Participants and design

One hundred and sixteen students at the University of Southern California participated for partial course credit (M _age  = 20.2 years, 52% female). The experiment used a single manipulated factor (perceived employee power: high, low) between-subjects design, with random assignment to condition. All participants read an actual article from a Georgia newspaper describing a non-workplace offense by an employee of Bank of America (Suggs 2016). Bank of America has many branches throughout the U.S. and in the local area of the university, so we could assume that most students would have heard of it, though the degree of familiarity is still likely to vary.

Procedure and measures

Participants’ familiarity and involvement with Bank of America were collected prior to the manipulation. Familiarity was measured with two items (Kent and Allen 1994). Participants indicated how much they knew about and how familiar they were with the bank (α = .84). Involvement was also measured with two items (Gruen et al. 2000). Participants rated how much they felt connected with the bank and to what extent the bank mattered to them (α = .85).

Participants then read the newspaper article about a Bank of America employee committing an offense in her private life. To increase confidence in the effect’s robustness, the specific offense was different from that described in study 1, and the offense here is not illegal. Specifically, participants read that this bank employee put up a Facebook post expressing racial hatred. The article describes the uproar the post caused on the internet. The information from the article was modified (a) to manipulate the position held by the employee, which was not described in the original article (though Bank of America was identified explicitly as her employer), and (b) to emphasize the non-workplace context by adding that the post was triggered by the employee’s experience at a local grocery store (see Appendix 1 for stimulus). At the study’s conclusion, participants were debriefed about the modified information.

Two employee positions were used to operationalize power. The higher-power employee was described as a “check processing manager” and thus did not have a unique management position within the firm, as in Study 1, but instead was one of the bank’s many managers. The lower-power employee was described as a “check processing clerk.” Because this news was slightly modified from the real incident involving Bank of America, we collected participants’ familiarity with the incident and removed those indicating they had heard of the incident prior to the study (n = 11).

Each construct in the experiment was measured with single or multiple items rated on 7-point scales. The same three items from Study 1—with one additional item measuring perceptions of the action as “ethical” versus “unethical”—were used to measure brand moral judgments and employee moral judgments (α’s > .92). As in Study 1, ratings of the morality of the employee’s action did not differ between conditions (F(1, 103) = .07, p = .80; M _low-power  = 1.84 [SD = .96] vs. M _high-power  = 1.89 [SD = 1.00]).

Power perceptions were measured with three items. Participants indicated to what extent they perceived the employee as having “influence over other employees,” having “power to affect the bank’s decisions,” as in Study 1, as well as to what extent the employee is like to have “resources to reward and punish other employees” (α > .72). We added the last item because it reflects another element of power as measured and conceptualized in the management literature (e.g., Magee and Galinsky 2008).

In Study 1, we measured employee status by asking participants to indicate how “prestigious” the employee position seemed to be. To make the meaning of status clearer to participants, in this study we instead asked to what extent the employee position was perceived as being “respected and admired” by others in the bank. The three-item power measure and this single-item status measure are positively correlated (r = .67, p < .01).

Manipulation check

The employee position manipulation shows a significant main effect on power perceptions (F(1, 103) = 34.54, p < .01) and status perceptions (F(1, 103) = 14.88, p < .01). Specifically, the manager is seen as both more powerful (M _manager  = 4.29 [SD = 1.53] vs. M _clerk  = 2.55 [SD = 1.58]) and higher status (M _manager  = 4.07 [SD = 1.91] vs. M _clerk  = 2.73 [SD = 1.60]) than the clerk.

Results

Employee position has no significant main effect on brand moral judgment (F(1, 103) = 1.04, p = .31; M_high-power = 3.72 [SD = 1.13] vs. M_low-power = 3.93 [SD = .95]). Because the manipulation of employee position simultaneously changes power and status perceptions, which may exert opposing effects on brand moral judgments as in Study 1, the expected main effect may be cancelled out. To investigate, we turned to the continuous measures of perceived employee power and status.

Supporting H1 and replicating results from Study 1’s transgression conditions, perceived employee power is a significant negative predictor of brand moral judgment (R² = .11; B_power = −.33, t(103) = −3.58, p < .01). In contrast, perceived status has no significant main effect on brand moral judgment (R² = .002; B_status = −.04, t(103) = −.44, p = .66). Including both power and status in the model, the significant negative effects of power perceptions on brand moral judgment still hold (B_power = −.55, t(102) = −4.54, p < .01). However, the variance accounted for by perceived power acts as a suppressor to perceived status, revealed by a significant positive effect of perceived status on brand moral judgment (B_status = .18, t(102) = 2.69, p = .01). That is, the perceived power versus status of the transgressor show opposite effects on moral judgments of a firm, with higher power leading to more negative effects, and higher status leading to less negative effects on brand moral judgment. The positive marginal effect of status is revealed only when controlling for the large and significantly negative effect of employee power (i.e., power acts as a suppressor variable in relation to status).

Power and status

To investigate the distinctive effects of power and status perceptions, we ran a mediation analysis between the manipulation of employee position and brand moral judgments through both power and status (Hayes 2013, Model 4). Perceived power and status together fully mediated the effect of the employee’s position on brand moral judgments. After accounting for both, the effect of employee position on brand moral judgments was insignificant (B_direct = .17; CI = [−.266; .607]) indicating full mediation, with power perceptions as a significant negative predictor (B_{power-indirect} = −.62; CI = [−1.024; −.316]), and status perceptions as a significant positive predictor (B_{status-indirect} = .24; CI = [.064; .576]) of brand moral judgments.

Familiarity and involvement

Familiarity (M = 4.69, SD = 1.53) does not independently influence participants’ brand moral judgments (R² = .02; B_fam = .09, t(103) = 1.30, p = .20). However, its interaction with perceived power is significant (B_interaction = −.09, SE = .04, p = .03, CI = [−.162, −.008]; Hayes 2013, Model 1). The Johnson-Neyman technique was used to identify the range(s) of familiarity for which perceived power significantly influences brand moral judgment. Perceived power negatively impacts brand moral judgment when consumers’ familiarity with Bank of America is greater than 4.03 on the 7-point scale (B_J-N = −.13, SE = .06, p = .05, CI = [−.253, .000]). These results indicate that at least the sense of familiarity, as was created by the scenario in Study 1, may be necessary for the effect to occur.

Involvement (M = 3.54, SD = 1.82) does not independently influence participants’ brand moral judgments (R² = .02; B_involve = .09, t(103) = 1.56, p = .12), nor does it interact with power perceptions to influence moral judgments (B_interaction = −.04, SE = .03, p = .27, CI = [−.100, .028]). Regardless of involvement level, perceived power negatively predicts brand moral judgment. Finding no effect for involvement on brand moral judgment may not be surprising since our participants did not show the very high levels of involvement with our stimulus brand (Bank of America) that seem to be critical in motivated reasoning (Paharia et al. 2013). Involvement was not a significant covariate in any analysis (p > .51).

Discussion

Overall, these results show further support for H1. Consumers’ moral judgments of a firm’s brand are more damaged by a powerful employee’s transgression than a less powerful employee, regardless of the employee’s status. Familiarity with the brand does not protect against information about a transgression but instead makes the firm more vulnerable to damage. In contrast to the damaging effect of employee power, employee status influences moral judgments in the opposite direction, with higher status contributing to more favorable brand morality judgments, once the effects of power are accounted. This result was not anticipated, however, and caution is warranted in interpretation.

Study 3: Employee power and company responses

Studies 3 and 4 turn to testing H3 and investigating reparative actions the firm might take after a powerful employee’s offense comes to light. As suggested by media reports (e.g., Halper 2015), firing the transgressing employee is the most obvious solution. However, not only can that lead to potential challenges, such as reversal, but it is also possible that terminating the relationship with a highly powerful transgressing employee (e.g., the company president) will not remedy the damage that has already been done. To assess this, we varied employee position, as before, and measured the effectiveness of firing the employee as the first test of the effectiveness of firm reactions in moderating consumers’ negative reactions to a non-workplace employee transgression.

To boost the robustness of our findings, this study used a different real company, a new offense taken from recent news, and different employee positions. The high-power employee was in a uniquely high-status position within the company (i.e., the CEO is the individual in the real news report), which may be more similar to the status of a celebrity endorser compared to Study 2’s high-power employee (i.e., a manager). Rather than let familiarity vary naturally as in Study 2, in this case, a relatively new product and company let us control for prior associations. Most participants were not previously familiar with the company, and the product is a beverage that is easy to understand. To create the necessary familiarity, we provided participants with information about the company’s product and gauged their interest prior to their receiving the non-workplace employee transgression information. As well, we assessed perceived power and status prior to giving participants the transgression information to provide a manipulation check of power perceptions before they could be influenced by information about the firm’s reaction.

Method

Participants and design

Four hundred and seventeen participants recruited from a Qualtrics panel participated in this study (72% female, M _age  = 52.51, SD = 15.92). The experiment used a 2 (employee power: high, low) × 3 (company response: employee fired, employee not fired, no information) between-subjects design, with random assignment to condition. As in Study 2, all participants read about a real incident, which had been published in the Los Angeles Times (Dave and Masunaga 2016). This newspaper article described a non-workplace offense by an employee of Soylent, a California based company that produces and markets its beverages as “nutritious meal-replacements.”

Procedure and measures

To provide at least minimal product and brand familiarity, the product and its benefits were described along with an attractive photo of the beverage (stimuli available on request). After completing initial pre-transgression ratings of the company (i.e., familiarity) and interest in its products, participants were either informed that Mr. Rob Rhinehart is the CEO (high-power employee) or an assembly-line worker (low-power employee) from Soylent. Based on the title, they were asked to rate the power and status of Rhinehart’s position within Soylent. Then, participants read a newspaper article describing how Rhinehart violated the city’s bylaws by placing an unpermitted shipping container on his land and allowing it to deteriorate (see Appendix 1). As in Study 2, we collected participants’ knowledge of this incident, but none indicated that they had heard of the incident prior to the study.

The other factor varied between conditions was the firm reaction to the offense, which was described at the end of the news article. The critical condition involved firing the employee: “the Soylent board voted to terminate the CEO (assembly-line worker).” The other two conditions described no firm reaction (and so no action to change the employee’s power). The lack of response was either explicit (“Information available after the date of the initial story indicates that Rhinehart is still currently employed by Soylent”) or implicit as in Study 2, where no information was given (“No additional information was available from Soylent after this report’s initial publication”). These latter two conditions were included because these are two potential responses a human resources employee could offer if contacted by the media prior to running a story like this. As well, the explicit lack of response might be seen as an endorsement of the immoral behavior. However, there are no significant differences on any of the studied variables (p’s > .29), so these two conditions are collapsed into a single control condition. At the end of the study, we debriefed participants about the purpose of the study and the modified information in the news article they had read.

Prior to reading the article, participants’ familiarity with Soylent was measured using the same two items from Study 2 (α = .97). Familiarity (M = 1.41, SD = 1.16) has no main effect on brand moral judgments (B_familiarity = −.01, t(416) = −.22, p = .82; R² = .000) and no significant interactive effect with the power manipulation (B_interaction = .08, SE = .12, p = .52, CI = [−.160, .315]). Therefore it was excluded from further analyses.

To ensure equivalency prior to the manipulation, the important constructs were measured prior to reading the newspaper article and were each collected again after participants read about the non-workplace transgression. Specifically, the perceived power and status of the employee, and purchase intention for the Soylent beverage, were all measured pre- and post- exposure to the employee non-workplace transgression (α’s > .86). As in Studies 1 and 2, and using the same items as Study 2 (α = .96), brand moral judgment was measured only after learning of the non-workplace employee transgression, to avoid sensitizing participants to the study’s core construct.

The single-item status measure (i.e., the employee is respected and admired) from Study 2 was split into two items: “the employee is likely to be respected by others in the firm” and “the employee is likely to be admired by other employees.” Combined with the single item from Study 1, “the employee is in a prestigious position in the firm,” these three items were used to measure status pre- and post-transgression (α‘s > .93). Perceived power and status are again positively correlated (r = .87, p < .001). Table 1 contains the means by condition for each variable in the study.

Table 1 Means (SD) by employee position and firm reaction conditions in Study 3

Manipulation check

As expected, purchase intentions did not differ by employee position (F(1, 409) = .93, p = .34), firm reaction (F(2, 409) = 1.34, p = .26), or their interaction (F(2, 409) = 1.66, p = .19), before information about the non-workplace employee transgression was presented. As intended, the employee position title manipulation shows a significant main effect on perceived employee power, measured before the transgression (F(1, 415) = 1037.20, p < .001). The CEO is perceived as significantly more powerful than the assembly-line worker (M _high-power  = 5.72 [SD = .95] vs. M _low-power  = 2.35 [SD = 1.18]). Before the non-workplace transgression was described, the high-power position (CEO) is also perceived as having higher status within the company than the low-power position (assembly-line worker) (F(1, 413) = 1003.798, p < .001; M _high-power  = 6.24 [SD = .78] vs. M _low-power  = 2.89 [SD = 1.31]).

Results

All three hypotheses are supported in this study. The core effects observed in Studies 1 and 2 replicate, further supporting both H1 and H2, despite an obviously less severe transgression. Furthermore, a firm reaction that affects perceived power can improve brand moral judgment after a non-workplace employee transgression, supporting H3. All variables used in the following analysis were assessed after the non-workplace transgression and any additional firm reaction information were provided (i.e., power, status, and purchase intention discussed hereafter were measured post-transgression).

Power

As expected, perceptions of the employee’s power after the firm reacts vary across conditions. ANOVA reveals a significant interaction between employee position and firm reaction on perceived employee power, assessed after the non-workplace transgression and any additional firm reaction information were provided (F(1, 412) = 158.71, p < .01). As expected, firing a high-power employee significantly reduces his perceived power within the company compared to when he is not fired (p < .01). After a high-power employee is fired, perceived power is not significantly different from that of a low-power employee who was also terminated (p = .80). (See Table 1 for all variable means used in planned contrasts.) In comparison, firing a low-power employee does not significantly change perceived power compared to when he is not fired (p = .54).

Status

ANOVA also reveals a significant interaction between employee position and firm reaction on perceived status (F(1, 409) = 74.11, p < .01). Firing high-power and low-power employees significantly reduces status perceptions of each (p’s < .01). However, when post-transgression power perception is statistically controlled in the model, the effect of the manipulations on perceived status is no longer significant (F(1, 407) = .04, p = .83). In contrast, when perceived status is controlled in the model, the interaction effect on power remains significant (F(1, 407) = 71.12, p < .01).

Brand moral judgment

The employee position and firm reaction manipulations interact in predicting brand moral judgments (F(1, 413) = 2.76, p = .097), though the effect is marginally significant. Firing a transgressing high-power employee significantly improves brand moral judgment (p = .02), whereas firing a low-power employee does not change brand moral judgment (p = 1.00).

The effect of employee power is also revealed when the ratings of perceived employee power are used to predict brand moral judgment. Perceived employee power negatively predicts brand moral judgment (R² = .03; B = −.12, t(414) = −3.73, p < .01). Post-transgression status perceptions independently also have a significant negative effect on brand moral judgment (R² = .01; B = −.08, t(411) = −2.19, p = .03). When both perceived power and status are included in the model (R² = .04), perceived power still negatively predicts brand moral judgments (B = −.24, t(409) = −3.64, p < .01) but the effect of perceived status becomes positive (B = .13, t(409) = 2.01, p = .05).

Purchase intention

Brand moral judgments are positively related to purchase intentions (R² = .22; B = .63, t(414) = 10.75, p < .01). In addition, perceived employee power influences purchase intention through brand moral judgments (B_indirect = −.08; CI = [−.130; −.035]; B_direct = .02, CI = [−.066; .098]; Hayes 2013, Model 4), and the effect replicates when status perceptions are included as a covariate (B_indirect = −.15; CI = [−.236; −.062]; B_direct = −.10, CI = [−.257; .062]). Alternatively, brand moral judgment also significantly mediates the relationship between perceived status and purchase intention (B_indirect = −.05; CI = [−.100; −.006]; B_direct = .05, CI = [−.030; .134]). However, when perceived power is included as a covariate, the effect of status on purchase intention through brand moral judgment is reduced to the point of non-significance (B_indirect = .08; CI = [−.006; .176]; B_direct = .14, CI = [−.024; .295]).

Power and status

To further distinguish the roles of power and status, we ran moderated-mediation analyses (Hayes 2013, Model 8) to confirm that our employee position manipulation interacts with firm reaction to influence purchase intention through the dual mediators of post-transgression power and status perceptions. With post- purchase intention as the dependent variable, the moderated-mediation is significant; firm reaction moderates the effect of employee position, and this effect is mediated by post-transgression perceived power (B_indirect = .70; CI = [.087; 1.393]) and status (B_indirect = −.72; CI = [−1.261; −.249]). Specifically, firm reaction significantly affects post- purchase intention through post-transgression perceived power only for the high-power employee (B_indirect = −.73; CI = [−1.450; −.078]). Firm reaction did not significantly affect purchase intentions for the low-power employee (B_indirect = −.03; CI = [−.158; .045]). These results suggest that firing a high-power employee reduces perceptions of his power, improving brand moral judgment and purchase intention.

Discussion

Study 3, like Studies 1 and 2, finds that information about a powerful employee’s non-workplace transgression damages brand moral judgments (H1), and these judgments have consequences for purchase intentions (H2). Study 3 also provides initial support for H3. Reducing the employee’s perceived power can be an effective remedy, provided the employee had sufficient power before the transgression that a reduction is possible and impactful.

Study 4: Comparing remedies that modify power

Next, we provide a more nuanced test of H3 by investigating potential firm reactions that vary in the extent to which the employee’s power is reduced. The new condition describes the employee being stripped of the power to influence company decisions but keeping her prestigious management position title. This study used the same real-life situation as in Study 2 and compared the effect of firing the offending employee to the effectiveness of demotion by reducing the employee’s power.

Method

One hundred and ten students at the University of Southern California participated in this study for partial course credit (M _age  = 19.82 years, 67% female). We used the same procedure and stimulus as in Study 2, with the exception that the employee was described as a Vice President at Bank of America, and participants were told this position held “major responsibilities in the check processing division.” The purpose was to create a relatively high-power perception of the employee’s position. Prior to the manipulation, participants’ familiarity with the bank was measured (M_familiarity = 4.91, SD = 1.34; α = .75). Familiarity did not predict brand moral judgments (R² = .01; B = .09, t(108) = 1.00, p = .32) or interact with the remedy manipulation to influence brand moral judgments (B = .06, t(106) = .53, p = .60).

After reading the same article describing the incident as in Study 2, participants were randomly assigned to one of three conditions. In one condition, the bank did not take any action and the employee was still employed in the same position (no response), while in the second condition, the bank was described as having taken an immediate action to fire the employee (fired). In the third condition, the bank took away the employee’s managerial responsibilities but she still kept her title of vice president (reduce power). With this manipulation, we expected to reduce the power of the employee while maintaining her status, as power is more associated with managerial responsibilities and resources, but status is more likely to result from a respected and admired position title (Magee and Galinsky 2008).

After the manipulation, moral judgment of the bank and the employee were collected using the same items from Studies 2 and 3 (α’s > .90). Firm reaction had no significant influence on moral judgment of the employee (F(2. 107) = 1.96, p = .15). We also asked participants to indicate whether the bank should be held responsible for the actions of the employee (anchored with the labels of “not at all responsible/definitely responsible,” “not at all at fault/definitely at fault,” and “should not be blamed at all/should definitely be blamed”; α = .93), but found no differences between conditions (F(2, 107) = 1.49, p = .23). Finally, we measured power and status perceptions of the employee’s position using the same items as in Study 3 (α’s > .87).

Results

Power

The firm’s reaction influences perceived employee power (F (2, 107) = 9.76, p < .01). The employee is perceived as most powerful when the company does not respond to the transgression (M _no-response  = 4.53, SD = 1.65), is perceived as marginally less powerful when responsibilities are reduced (M_reduce-power = 3.86, SD = 1.68; p = .09), and, when fired, is perceived as significantly less powerful than the reduced power condition (M _fired  = 2.82, SD = 1.69; p = .01).

Status

Firm reaction also affects status perceptions (F(2, 107) = 21.35, p < .01), but the pattern is different, as we expected. As with the power ratings, status perceptions were lowest when the employee was fired by the bank (M _fired  = 2.18, SD = 1.79). However, as intended, reducing the employee’s power (i.e., removing the employee’s responsibilities but having her keep the position title) did not change status perceptions over no response (M _{no response}  = 4.60 [SD = 1.84] vs. M _reduce-power  = 4.40 [SD = 1.67], p = .64). As in previous studies, perceived power and status are positively correlated (r = .81, p < .01).

Brand moral judgment

Supporting H3, the results show that decreasing employee power improves brand moral judgment. Specifically, terminating employment significantly improved brand moral judgments over no response (F(2, 107) = 25.02, p < .01; M _{no response}  = 2.76 [SD = .95] vs. M _fired  = 4.52 [SD = 1.24], p < .01). Further, reducing the employee’s power by taking away her managerial responsibilities also improved brand moral judgment over no response (M _{no response}  = 2.76 [SD = .95] vs. M _{reduce power}  = 3.63 [SD = .97], p < .01), but not to the same extent as termination. Reducing power resulted in significantly lower brand moral judgment than firing the transgressing employee (M _{reduce power}  = 3.63 [SD = .97] vs. M _fired  = 4.52 [SD = 1.24], p < .01).

Perceived power is a marginally significant negative predictor of brand moral judgment (R² = .03; B = −.16, t(108) = −1.73, p = .087). Status perceptions also negatively predict brand moral judgment (R² = .03; B = −.16, t(108) = −1.94, p = .055). However, when including both power and status perceptions in the model (R² = .03), neither was a significant predictor of brand moral judgment (B_power = −.15, t(107) = −.91, p = .36; B_status = −.04, t(107) = −.27, p = .79).

The weaker findings are likely due to the insufficient power of a small sample size or to an effect of the firm reaction without the anchor of pre-measure. In Study 3, the pre-measure of power may have anchored the ratings such that its influence remained evident even after the firm reaction. Here, participants in the two focal conditions were asked about the employee’s power only after the firm’s reaction, which was explicitly designed to change power perceptions.

Overall the results of Studies 3 and 4 show promise for remedies firms can consider when a non-workplace employee transgression occurs. When it is not feasible to terminate the employee, publicly reducing the employee’s power may improve consumers’ moral judgment of the company.

General discussion

Across a series of four studies describing different immoral behaviors (a criminal behavior, an illegal but not criminal behavior, and a legal but offensive behavior), different employee positions (CEO, president, head designer, manager, clerk, assembly-line worker) and different types of companies (an apparel manufacturer, a well-known bank, a nutritional drink company), we find that information about the non-workplace, immoral actions of a powerful employee consistently influence brand moral judgments (H1). We also find that those brand moral judgments influence purchase intentions (H2), and that firm reactions that reduce the employee’s power can mitigate the damaging effects of a powerful employee’s offense (H3).

Our research contributes to a greater understanding of consumers’ responses to negative events associated with brands. Although many negative events are direct consequences of a firm’s actions (e.g., product harm crises), others are unintended but occur in the workplace (e.g., employee incivility), and still others are not under a firm’s control and seem outside of the firm’s purview (e.g., celebrity endorser transgressions). Finding that even a non-celebrity employee’s transgression that occurs outside of the workplace can harm the brand expands the horizons of this literature.

Previous research indicates that consumers react differently to the immoral behavior of a current employee who is an admired public figure (Bhattacharjee et al. 2013). That research deals with admired public figures that a consumer is strongly motivated to support. Although consumers condemn the transgression, when the transgression is irrelevant to the reason the person is admired (e.g., performance of some skill, sport, or art form), they can decouple that judgment from assessments of the admired employee’s performance. This can be seen in sports fans reactions to scandals by talented players: condemnation of the act does not preclude enjoyment of the game (Young 2018). Our findings are different from that research in that we investigate employees who are not admired public figures and focus on the effect on judgments of the firm rather than on judgments of the individual transgressor. Hence, our research contributes particularly to an understanding of consumers’ moral judgments of the firm in response to negative events.

Implications for the CSR and celebrity endorser literatures

Prior research suggests that consumers generalize from employees’ workplace behaviors to their impressions of the firm as a whole. Our research is novel in showing that inferences extend beyond the workplace to an employee’s behavior in private life. The marketing literatures that investigate similar issues investigate the consumers’ perceptions of corporate socially responsible actions that are often undertaken out of the workplace and the effects of offenses by a celebrity endorser.

CSR studies generally report virtuous actions tied to the firm without taking into account that consumers might attribute certain practices to subsets of employees. Recall that Study 1 shows that employee status was a better predictor of virtuous behavior on brand moral judgment than power. This suggests that if companies are seeking to promote the virtue of their employees, communications to consumers might be more effective if they emphasize the employee’s status, or respect from others in the company. Consumers may assume that decisions about virtuous actions on behalf of the brand (e.g., allocating a large part of a firm’s profit to a charity) are made by a high-status employee, but that lower status employees are more likely to contribute their time to helping others. Further, consumers may infer different motives for the virtuous actions of high status employees as compared to those of low status employees. For example, perhaps inferences about the employee’s extrinsic rewards from CSR are different (e.g., donations as a tax benefit vs. donating time to get time off from dull work). Future research might consider whether inferences about individual employees and their perceived power and status can account for some of the inconsistencies in the CSR literature (Peloza and Shang 2011).

Given that most celebrity endorsers are not thought of as having power within the firm, our findings for status may offer more insight for that literature. The celebrity endorser literature emphasizes liking for the celebrity and perceived expertise as mechanisms accounting for benefits to the brand (Miller and Allen 2012), as well as positive learned associations between the celebrity and the brand. Additionally, celebrity endorsers generally have status and prestige, which should benefit judgments of the firm when the celebrity acts in a virtuous way. Effects of status are more complicated for transgressions (Carrillat et al. 2013), which may account for some of the mixed findings in prior research. Perhaps a celebrity’s perceived power over consumers affects reactions to his or her transgression. For example, parents may express more alarm over a transgressing teen idol with ardent fans than a celebrity that elicits less fervor. Those parents may then also judge a more powerful celebrity’s brand sponsor more harshly for its support following a scandal.

Potential research directions

Our research raises a number of issues that might be pursued in future research. One important question in regard to transgressions is identifying when status can have an opposite effect from power on brand moral judgment. The negative influence of power may relate at least partly to a learned aversion to associating with immoral others who are powerful. The potential harm that a powerful and immoral person might loom large as a reason to avoid that person. The potential benefits that might accrue from attaching oneself to a powerful person who turns out to be truly virtuous would be comparatively minor. In contrast, there may be a learned positive association with status. Attaching oneself to a high-status entity might allow the consumer to share in the warm glow of that status more than associating with a low-status entity both when the entity is moral and immoral. Future research might use perceived status to clarify the effects of celebrity endorser transgressions on brand judgments and purchase intentions, as well as to further investigate different forms of corporate social responsibility (Peloza and Shang 2011).

We found that the focal employee’s similarity to other employees could not explain the observed variation in the influence of different employees on brand moral judgments. Yet, we investigated only perceptions of for-profit organizations. The importance of similarity rather than power or status in the firm may be more typical of organizations perceived as relatively democratic organizations, such as is characteristic of nonprofits, rather than those perceived as hierarchical organizations with stratified leadership.

The role of the familiarity with the firm’s brand to customers in moderating the effect found here could also be further explored. Whereas familiarity was a significant moderator in Study 2, its results did not hold in further studies. Further research might discover a more complex role than expected for familiarity and involvement in moral judgment, just as we found here for perceived employee status.

Additionally, our research manipulated power by describing the employee’s position within the firm. Other cues to power might be investigated, as well as other cues to employee status. Though status here was found to vary based on employee’s position titles (just as was perceived power), other signals of respect could serve as well, such as other employees speaking about the person or mention of awards or promotions the person has received. These might serve to increase the perceived status of an employee with a title that might otherwise be perceived as lower status. Future research could examine whether the extent of respect afforded in various ways to a virtuous employee signals the company’s moral values positively to consumers.

Our research suggests some remedies to the damage to the firm’s reputation from an employee’s non-workplace offense, based on reducing the amount of power customers perceive the employee to have. Termination, as well as reducing power (even while retaining a high-status title), appears effective. Future research may pursue this finding to determine whether this is effective because these actions represent a punishment for the employee, satisfying a desire for justice or retribution, or because it signals the firm’s values to customers. Other kinds of remedies that might provide more diagnostic evidence about the brand’s morality compared to information about a single employee might also be investigated.

Managerial implications

Our research provides practical implications for marketers. First, firms need to be aware of their brand’s vulnerability to a “one bad apple” effect. Although people enact multiple roles so that a person’s work identity can be distinct and different from their other identities, such as parent or political party member (Roccas and Brewer 2002), consumers’ judgments here suggest that such boundaries do not limit inferences across an employee’s different roles. Firms may assume erroneously that they can rely on these workplace versus home-life boundaries to protect the brand when non-workplace transgressions occur. Additionally, firms may assume that extensive familiarity with their brand can insulate them from this kind of information. However, our findings suggest greater susceptibility for consumers familiar with the brand, and further suggest no buffering effect of involvement (Study 2). These results suggest that customer engagement and outreach are not sufficient to prevent damage from information about non-workplace employee transgressions.

However, the findings hold good tidings as well: mere association with a transgression is not sufficient to cause the effect. If a non-workplace transgression occurs for a low-power employee, the best way to protect a brand’s reputation among consumers seems to be avoiding a publicized response. Publicly terminating a low-power employee has little impact on consumers’ brand moral judgments and may just reinforce a negative association with the brand. While negative effects can be significant even for a low-power employee transgression, firm reactions as severe as termination do not significantly improve consumer judgments. If a high-power employee commits a non-workplace transgression, however, firm reactions that reduce perceived power do significantly improve consumer judgments implying that firms should publicize this information and do it promptly.

Although our work focuses on the effects of transgressions, we showed that information about positive employee behaviors also influences judgments of the firm. In light of the reputational rewards to the firm, firms might proactively hire qualified employees who engage in prosocial activities on their own time. Helping employees do the right thing and refrain from doing the wrong thing in their private lives may seem intrusive, but firms can adopt policies (e.g., giving time off for prosocial activities, or helping employees deal with stresses in their life that may cause offensive behaviors) that facilitate desired behaviors without interfering with employees’ choices.

Should such policies focus on powerful employees in particular? On the one hand, more powerful employees’ non-workplace actions have more impact on consumers’ moral judgments, so the firm may wish to pay particular attention to them. Further, leaders who have a lot of power are more likely to immorally abuse their power for personal gain than leaders who have less power (Bendahan et al. 2015). On the other hand, less powerful employees may be more likely to engage in immoral behaviors because some research indicates they are more likely to justify behaviors that are ethically suspect (Chen 2014). Furthermore, none of the remedies explored here improved the (smaller but still significantly) negative effect of a low-power employee’s non-workplace offense, which suggests that policies attempting prevention would be beneficial for all employees regardless of rank or actual power. When organizations encourage discussing work-related ethical issues and enforce an ethical code, this enhances the workplace ethical climate (Schwepker Jr and Hartline 2005) and may also promote more ethical non-workplace behavior.

Our research has focused on the direct effects on consumers, but information about another employee’s non-workplace transgression also may damage employee morale. The public association of the firm with immoral behavior by one employee may feel to other employees as if their own reputations have been tarnished. This demoralization could lead to actual quality decrements in the firm’s products, which is potentially catastrophic. Future research should investigate whether these intra-organizational effects occur and what might be done to counteract negative effects on other employees.

Finally, firms should be cautious in giving power or sending cues that suggest that any person has power within the company. A recent trend is for celebrity endorsers to have positions that suggest power within the firm, such as “creative director” or “brand ambassador” (Parekh and Zmuda 2013). Firms who grant celebrity endorsers such titles may be more vulnerable to damage from a celebrity’s bad behavior than they would be otherwise. If you bring the apple into your barrel, so to speak, it could spoil consumers’ perceptions of your brand.

Appendix 1: Sample Stimuli from Studies 2 and 3

Study 2 (Added to original article).

Racist Facebook Post by Bank of America Employee Causes Uproar.

By Ernie Suggs - The Atlanta Journal-Constitution.

Posted: 3:41 p.m. Thursday, June 2, 2016.

Bank of America employee Christine McMullen Lindgren gained considerable attention for her expletive-laced Facebook posting that trashed black people.

Another day, another debacle of social media. Thursday’s example was Christine McMullen Lindgren, an Atlanta-based Vice President at Bank of America.

Lindgren, who described herself as a “personal banking manager” at Bank of America in her profile, posted recently that she hated Facebook because, “you f—-ing (n-words) and yes if (you) can call each other that well I can too f—-ing (n-word) go back to Africa get over your pity party you created this hatred and your own kind that brought your great great grandparents over here and sold them.”

It is unclear who Lindgren is specifically targeting in her rant. However, comments on the post implied that the event that triggered the post occurred at Lindgren’s local grocery store and indicate that the post was made when she returned to her home after that shopping trip.

Study 3 (Italicized text is modified from original article).

Soylent CEO Charged over Defiance of Permits.

By Paresh Dave - The LA Times.

Posted: 4:35 p.m. Dec 10, 2016.

Soylent’s Chief Executive Officer (CEO), Mr. Rob Rhinehart, has gained considerable attention for the way he treated his neighbors after his unpermitted construction of a personal living space.

Rhinehart put a giant, metal shipping container on his personal property several months ago and planned to convert it into a place to live. But over time, said a neighbor, “he did nothing and just left it there to deteriorate. It became a vandalized graffiti eyesore and a magnet for unwelcome visitors, both human and animal. He seemed not to care about these problems created for the people and pets who live in the neighborhood.”

Although the city’s building and safety inspectors continue to express their concerns, Mr. Rhinehart has not responded. His neighbors are worried. He continues to violate codes put in place by the city to protect both people in the surrounding neighborhood and the health and safety of the property for those who may enter it.