Can Companies Survive a Multi-Brand Crisis? Research on Consumer Scapegoating

: Compared with single-brand crises, multi-brand crises have broader and deeper inﬂuence. However, there is only a limited amount of research regarding this ﬁeld, especially for the phenomenon of consumer scapegoating. This research aims to further explore the consumer scapegoating e ﬀ ect and its psychological mechanism in the context of a multi-brand crisis. Through two experimental studies, this research discusses the inﬂuence of a multi-brand crisis and also the emergence of a scapegoat brand on two brand categories, the crisis brand category and the competing brand category. Furthermore, the current research also explores the mechanism of the consumer scapegoating e ﬀ ect. Results show that when a multi-brand crisis happens, consumer brand trust in the crisis brand category decreases because of assimilation, and consumer brand trust in the competing brand category increases because of contrast e ﬀ ect. Besides, from the perspective of the crisis brand category, the emergence of a scapegoat brand could be treated by consumers as a signal that the crisis is over, especially for a severe crisis. Results also support that cognitive dissonance mediates the process that the emergence of a scapegoat brand leads to an increase in consumer brand trust in the crisis brand category and a decrease of consumer brand trust in the competing brand category. “And Aaron shall lay both his hands on the head of the live goat, and confess over it all the iniquities of the people of Israel, and all their transgressions, all their sins. And he shall put them on the head of the goat and send it away into the wilderness by the hand of a man who is in readiness. The goat shall bear all their iniquities on itself to a remote area, and he shall let the goat go free in the wilderness.”


Introduction
Marketing researchers have recently started to show interest in multi-brand product-harm crises, though there is still a limited amount of research in this area [1][2][3][4]. Multi-brand product-harm crises (hereafter called "multi-brand crises") refer to crises involving multiple brands in an industry. The 2008 "melamine contamination" of dairy products in China is arguably the most influential multi-brand crisis during the past decade and has attracted interest from a number of marketing or crisis management researchers [2][3][4][5][6]. Other cases include the recent discovery of deceptive emissions testing by Volkswagen and other brands across the automobile industry [7] and the accusation against multiple firms in Brazil selling meats with expired due dates and carcinogenic chemical additives [8].
Most research in the previous literature focuses on single-brand crisis [9][10][11][12][13][14][15]. The few existing studies of multi-brand crises have mainly discussed negative spillover of distrust from one group of brands to another group of brands and contrasting effects [4]. Although Gao et al. [2] proposed a consumer scapegoating effect, their research focused on exploring consumers' trust towards the scapegoat brand, not towards related or competing brands in the same industry. Gao et al. [4] explored the influence of multi-brand crises on consumers' trust towards brand categories, but their research assumed the emergence of a scapegoat brand in a multi-brand crisis and did not evaluate and compare consumer responses between the existence of a scapegoat brand and nonexistence of a scapegoat brand. Despite its importance, existing academic research overlooked the mechanism of the scapegoating effect.
Inspired by the research gap, the current research tries to answer the question whether companies involved in a multi-brand crisis can survive and get back consumers' trust, through addressing the following research questions: (1) In the context of multi-brand crises, how do consumers respond to the signal of a scapegoat brand (in comparison with no signals) when rating their trust towards different brand categories? (2) What is the underlying psychological mechanism? More specifically, we aim to evaluate the different effects of scapegoat brands on different brand categories (including crisis brand category and competing brand category), as crises can be contagious [16]. Our study offers a structured understanding of consumers' psychological process towards a scapegoat signal. Insights from this study provide practical implications for brand managers and industry analysts to monitor different stages of multi-brand crises and develop appropriate response strategies.
Unlike previous studies, which mainly applied survey research [2][3][4] and case studies [6], we applied experiments to detect causal effects in the consumer-scapegoating phenomenon. Through two experimental studies, we established the role of a scapegoat brand in changing consumers' brand trust towards the crisis brand category and the competing brand category. We contributed to the psychological mechanism of the scapegoating effect by articulating the mediating role of cognitive dissonance.
In the following sections, we first discuss the theoretical background of multi-brand crises and the scapegoating effect, and propose research hypotheses. Next, we describe our research design and report results of the two studies. Finally, we conclude with theoretical contributions and managerial implications, as well as limitations and future research.

Multi-Brand Crisis and Brand Trust
A majority of previous literature on product-harm crisis focuses on single-brand crisis, in which only one brand is involved, showing that a crisis could negatively affect the crisis brand regarding many aspects of consumers' brand evaluation, especially brand trust [17][18][19][20]. Other brands could also be affected. For those that have interests in common with the crisis brand such as its sub-brands and parent brands, because of negative spillover, these brands would be negatively affected even if they are not guilty [20], and the extent of negative spillover is decided by the directional strength of association between themselves and the crisis brand [21]. In contrast, competing brands may benefit from the crisis [22,23].
For multi-brand crisis, the very few studies have shown that a crisis could cause negative spillover of distrust to a group of brands [4], or to the industry [24]. However, when multiple brands are simultaneously involved in a crisis, little could be found in the existing literature about the influence of such a crisis on different brand categories; we tried to make up this research gap in this research.
According to the categorization theory, when the similarity between the new instance and the existing category are high, an assimilation effect would likely occur so that beliefs and effect are more likely to transfer to the new instance [25]; whereas, contrast effects are more likely under conditions of category or domain mismatch or dissimilarity [26,27].
When a multi-brand crisis happens, a group of brands in the industry may receive the most public attention, because the assimilation of crisis information occurs intensively within this group (they have a certain common brand characteristic). Therefore, in this research, this group of brands is defined as a crisis category in a multi-brand crisis, and negative spillover from crisis brands to other brands in the same brand category would be more likely to happen. Comparatively, some other brands that consumers may perceive as alternatives of crisis brands are actually "competitors" to them and consist of a "competing brand category" in a multi-brand crisis. While considering brands in a competing category, contrast is more likely to happen, because it is easy to think of the dissimilarity between the crisis category and the competing category. Therefore, brands in a competing category may benefit from the crisis.

Hypothesis 1a.
When crisis assimilation is contained within a crisis brand category, consumer brand trust in this crisis category decreases.
Hypothesis 1b. When crisis assimilation is contained within a crisis brand category, consumer brand trust in a competing brand category increases.

Scapegoating Effect
At the beginning of this paper, we quote a paragraph from Leviticus of the Hebrew Bible, which is the original text of the "scapegoat." Because that goat carried all the sins of the Israelite, and let the Israelites get relieved, it is called a scapegoat. One of the most famous researches in scapegoating was done by Girard [28]. In his sociopsychological analysis, rivalries among people would cause threats to the existence of a community. When such a threat emerges, people would push communal violence on a single individual that had been chosen as a scapegoat. The elimination of the scapegoat could help the community to restore peace. This process is called "scapegoating" by Girard, who investigated extensively the concept by analyzing multiple historical stories in the literature.
Scapegoating is also widely discussed in other research fields. In psychoanalytic theory, a person may vent his/her negative emotions to other individuals (defined as scapegoats) [29]. Scapegoating frees the individual from negative experiences such as guilt and anxiety [30]. In the area of politics, refugees have been found to be treated as a scapegoat of terror attacks [31]. In the field of business management, research has shown that people may naturally want to blame a scapegoat when facing failures. For example, results of one research study show that, stakeholders of delisted IPO (Initial Public Offering) firms would blame the endorsing venture capitalists and thus bring negative effects on those venture capitalists' reputation. [32] Some scholars propose scapegoating as a "strategy" to solve crisis. In order to obtain a post-crisis reputation recovery, companies could pass the responsibility to a scapegoat [33]. Even though firms involved in scandals choose other response strategies at the beginning, with failures of these strategies, they would finally turn to blaming a scapegoat to survive [34].
To the best of our knowledge, the first and only research investigating consumer scapegoating in multi-brand crises was done by Gao et al. [2]. However, based on attribution theory, they only explored consumers' trust towards the scapegoat brand, not towards the crisis brand category or competing brand category in the same industry. In another study by Gao et al. [4], the authors examined the spillover of distrust from one crisis category to a related brand category using an assimilation perceptive, but their study did not examine whether and how emergence of a scapegoat brand in a multi-brand crisis may save or further derail the crisis brand category in the crisis.
In this research, we try to further explore the mechanism of consumer scapegoating in the context of a multi-brand crisis. Under such context, the punishment of a scapegoat brand could on one hand provide an opportunity for consumers to express their negative emotions, and on the other hand, deter other brands from making similar mistakes in the future. Consumers may treat the emergence of a scapegoat brand as a "signal" showing that the crisis is over and the market is stepping back to normal. Therefore, emergence of a scapegoat brand would weaken the assimilation effect/contrast effect in a multi-brand crisis, which is defined as "consumer scapegoating effect" in this research.

Hypothesis 2a.
When crisis assimilation is contained within a crisis brand category, emergence of a scapegoat brand increases consumer brand trust in a crisis brand category.
Hypothesis 2b. When crisis assimilation is contained within a crisis brand category, emergence of a scapegoat brand decreases consumer brand trust in a competing brand category.
Previous research has shown that, for a single-brand crisis, the severer the crisis, the more negative influence could be brought to consumer evaluations on the brand [35]. Similarly, when a multi-brand crisis becomes severer, consumers face a greater level of uncertainty and engender a greater level of fear, and are therefore more likely to engage in an assimilation process for a crisis brand category and in a contrast process for a competing brand category. Therefore, influence on the two brand categories, that is, the decrease of consumer brand trust in a crisis brand category and the increase of consumer brand trust in a competing brand category, may only show significance for severe multi-brand crises compared with those not severe. Thus, only when a multi-brand crisis is severe, the emergence of a scapegoat brand could significantly improve consumer brand trust in a crisis brand category and remove the "halo" from a competing brand category. We propose that crisis severity would moderate the consumer scapegoating effect, that is, a severe multi-brand crisis goes along with a significant consumer scapegoating effect.

Hypothesis 3.
When crisis assimilation is contained within a crisis brand category, the consumer scapegoating effect only shows for a severe crisis.

Mechanism of Consumer Scapegoating Effect: Cognitive Dissonance
Cognitive dissonance can be defined as the discomfort that emerges when an individual is aware of two inconsistent cognitions at the same time [36][37][38]. Research has shown that increased discrepancies of explicit-implicit attitudes or beliefs would push individuals to process and consider more subsequent related information. [39]. According to Rydell et al. [40], when people are undergoing gaps between explicit and implicit attitudes, they would experience cognitive dissonance, thus leading to more information processing.
In the context of a multi-brand crisis, consumer brand trust in a crisis brand category decreases, leading to a disparity between previous and current attitudes of consumers. Then, if consumers have to buy a product in the crisis brand category, or even think about purchasing, consumers would experience cognitive dissonance and feel negative emotions caused by cognitive dissonance, because there is discordance between their current and previous attitudes, and between their attitudes and behaviors. Therefore, they would be motivated to process information that could help them deal with cognitive dissonance. The appearance of a scapegoat brand is a piece of information that could be regarded as a signal showing to consumers that the crisis is over, which could reduce cognitive dissonance and thus bring about a change in consumers' attitudes. Without the emergence of a scapegoat brand, though, no evident information can be grasped to reduce consumers' cognitive dissonance levels, which does not change the attitude. Therefore, we propose the following:

Study 1
To test our proposed hypotheses, we applied the method of consumer experiment. The aim of Study 1 was to test the existence of the consumer scapegoating effect in a multi-brand crisis. In the experiment design of Study 1, an industry was divided into two brand categories based on a special brand characteristic, brand nationality. An industry was classified into a domestic brand category and a foreign brand category. However, in some consumer goods industries in China, foreign brands are much stronger than domestic brands in brand reputation and performance. For example, in the shampoo industry, foreign brands dominate the market. According to 2018 C-BPI (China Brand Power Index) results published by Chnbrand, the top 3 shampoo brands in China are all foreign brands (Head & Shoulders, Rejoice, and Pantene). In order to avoid country of origin as a confounding factor, we selected the smartphone industry in this experiment, because domestic brands are comparable with foreign ones. According to the 2017 Research Report of Chinese Smart-Phone Market published by iiMedia Research, domestic brands are even more competitive than foreign brands in smartphone sales, although foreign brands such as Apple still have a strong reputation. A second consideration for choosing the smartphone industry in Study 1 is that we attempted to investigate whether there would be a consumer scapegoating effect for both necessities and non-necessities. Since consumers do not have to make purchases on non-necessities, they may not be influenced much by a multi-brand crisis for these products. Therefore, a consumer scapegoating effect may not emerge for non-necessities. That is why we considered both necessities and non-necessities in Study 1, through manipulating a same product to be a necessity or a non-necessity. The second benefit of choosing the smartphone industry is that smartphones can be described as both a necessity and also a non-necessity.
Specifically in Study 1, we described the domestic brand category of the smartphone industry as where the multi-brand crisis happened and the foreign brand category of the smartphone industry as the competing brand category. One purpose of Study 1 was to test the existence of a consumer scapegoating effect with consideration of crisis severity, and the other purpose was to investigate whether a consumer scapegoating effect would work for both necessities and non-necessities.

Pre-Test 1
In this research, we may have measured consumer brand trust in individual brands as most researchers did in previous literature, recognizing each brand as either domestic or foreign according to their nationalities. However, an investigation by the Boston Consulting Group in 2008, in which 1003 consumers in Beijing, Shanghai, and other Chinese cities participated, shows a "chameleon effect" for many foreign brands, that is, many consumers regard them as domestic brands. It demonstrates a big gap between consumer perception of brand nationalities and the genuine brand nationalities. Therefore, even if consumers change their attitudes toward a specific domestic brand, it may not be because their attitudes change toward a domestic brand category, since this brand may not be a domestic one in consumers' perceptions. We conducted Study 1 to check consumers' perception of smartphone brand nationalities. Furthermore, we planned to use a fictive multi-brand crisis context and a made-up scapegoat brand in the experimental design, so the other purpose of Pre-test 1 was to test whether the made-up scapegoat brand is believable and consumers' perceptions of this brand's nationality.
Specifically in Pre-test 1, we listed 19 smartphone brands: 10 foreign brands, 8 domestic brands, and 1 made-up brand (Xunfei). Participants were asked to choose the nationality (domestic or foreign) of each brand based on their first impression. Then, they were asked to answer for how long they have used smartphones and demographic information. Thirty undergraduate students in Peking University participated in Pre-test 1.
Results show that, firstly, all of the 30 participants have used smartphones for more than 1 year, and 80% of them have used them for more than 2 years, so they are very familiar with smartphones. Second, participants had consistently accurate perceptions of brand nationalities for only three foreign brands. For the other brands, there were different perceptions of brand nationalities among participants. Third, 87% of the participants believed that the made-up brand (Xunfei) is a real domestic brand. According to these results, consumers may recognize some domestic brands as foreign brands and treat some foreign brands as domestic brands. Therefore, when measuring consumer brand trust toward a domestic brand category and a foreign brand category, it is not a best choice to ask consumers for their attitudes of specific brands and then make calculations based on the genuine brand nationalities instead of consumers' perceptions.

Main Study
• Design In the main study of Study 1, we applied a 2 (emergence of a scapegoat brand: yes versus no) × 2 (crisis severity: high versus low) × 2 (product type: necessity versus non-necessity) between-subject design. A control group was also added, the condition without a crisis. A nonstudent sample was recruited online, and a total of 375 participants participated in our study, 46% of whom were male. The participants were randomly assigned into 8 experimental groups and 1 control group. They were firstly asked to read one piece of reading material and then answer questions.
For participants in the necessity group, they would read an introduction about smartphones' benefits to people's daily life, showing that a smartphone has currently become a kind of basic and essential product. For participants in the non-necessity group, they read an introduction indicating that a smartphone could be substituted by many other products and can hardly be classified as a necessity. Next, participants in the experimental groups were asked to read one piece of news describing a multi-brand crisis. For high crisis severity groups, the news was about domestic smartphone producers adopting inferior screen materials that could cause sickness to the eyes of users. For low crisis severity groups, the news was also about domestic producers adopting inferior screen materials, but it would only cause dark spots to appear on the screen. Participants in groups with emergence of a scapegoat brand read an introduction of the brand Xunfei following the news, showing that Xunfei has been involved in the crisis and received harsh punishments from both the public and the government, whereas those in groups with no scapegoat brand read an introduction on luminance or screen dark spot. All contents of the news were fictive. According to the results of a pre-test, the news was believable. Participants in the control group read a piece of neutral introduction on smartphones. This introduction was paraphrased based on the definition of smartphone on Wikipedia. The length of all the reading materials was basically the same, in order to exclude the influence of reading time. After reading the materials, each participant was asked to answer some questions designed on the basis of the news contents. Before analyzing the experimental data, participants who gave wrong answers in this part were removed.
Next, participants were asked to answer (a) to what extent they agree that a smartphone is a necessity, (b) the evaluation of crisis severity, and (c) four questions regarding their brand trust [41] respectively in the domestic smartphone brand category and the foreign smartphone brand category. According to the results of Pre-test 1, consumers have different perceptions on brand nationalities, so they were not asked for their attitudes toward specific brands, but rather their attitude toward the two brand categories. Six-point Likert scales were used for all the questions, because compared with American people, the Chinese tend to choose the neutral point more frequently, and setting even-point scales would be a better choice [42][43][44]. Some questions were being reversed. Finally, participants answered some demographic information. After the experiment, participants were informed that the news they read was all fake. •

Results
Manipulation check. First, participants in low crisis severity groups perceived lower crisis severity than those in high crisis severity groups (mean = 4.72 and 5.01, respectively, p < 0.05). Second, participants in necessity groups agreed more with the statement that a smartphone is a necessity than those in the non-necessity groups (mean = 5.02 and 4.55, respectively, p < 0.01). The manipulations of both crisis type and product type were successful. Dependent variables. The results of the factor analysis show that the four variables of brand-trust scale consist of one factor for both the domestic brand category (Cronbach's α = 0.92) and the foreign brand category (Cronbach's α = 0.83), so the mean of the four variables was taken to represent the level of brand trust. Participants in experimental groups had significantly lower brand trust in the domestic brand category than those in the control group (mean = 3.23 and 3.94, respectively, p < 0.01). Hypothesis 1a is supported. Participants in experimental groups showed higher brand trust in the foreign brand category than those in the control group (mean = 4.05 and 3.75, respectively, p < 0.05). Hypothesis 1b is supported.
Then, we did analyses for the necessity groups and non-necessity groups, respectively. When a smartphone was manipulated as a necessity, we did not find any significant main effects of consumer scapegoating for both the domestic brand category (mean = 3.27 and 3.12, with emergence of a scapegoat brand and with no emergence of a scapegoat brand, respectively, p = 0.262), and the foreign brand category (mean = 3.97 and 4.15, with emergence of a scapegoat brand and with no emergence of a scapegoat brand, respectively, p = 0.072). Hypothesis 2a is not supported and Hypothesis 2b is only marginally supported. As Figure 1 shows, for the domestic brand category, when crisis severity is high, consumer brand trust is higher when there is a scapegoat brand than when there is not (mean = 3.49 and 3.08, respectively, p < 0.05), indicating a significant consumer scapegoating effect; when crisis severity is low, consumer brand trust has no significant difference under the two conditions (mean = 3.05 and 3.15, respectively, p = 0.570).
Sustainability 2020, 12, x FOR PEER REVIEW 7 of 14 the foreign brand category than those in the control group (mean = 4.05 and 3.75, respectively, p < 0.05). Hypothesis 1b is supported. Then, we did analyses for the necessity groups and non-necessity groups, respectively. When a smartphone was manipulated as a necessity, we did not find any significant main effects of consumer scapegoating for both the domestic brand category (mean = 3.27 and 3.12, with emergence of a scapegoat brand and with no emergence of a scapegoat brand, respectively, p = 0.262), and the foreign brand category (mean = 3.97 and 4.15, with emergence of a scapegoat brand and with no emergence of a scapegoat brand, respectively, p = 0.072). Hypothesis 2a is not supported and Hypothesis 2b is only marginally supported. As Figure 1 shows, for the domestic brand category, when crisis severity is high, consumer brand trust is higher when there is a scapegoat brand than when there is not (mean = 3.49 and 3.08, respectively, p < 0.05), indicating a significant consumer scapegoating effect; when crisis severity is low, consumer brand trust has no significant difference under the two conditions (mean = 3.05 and 3.15, respectively, p = 0.570). For the foreign brand category, as shown in Figure 2, when crisis severity is high, consumer brand trust is lower when there is a scapegoat brand than when there is not (mean = 4.08 and 4.41, respectively, p < 0.05), indicating a significant consumer scapegoating effect; when crisis severity is low, consumer brand trust has no significant difference under the two conditions (mean = 3.85 and 3.89, respectively, p = 0.793).   For the foreign brand category, as shown in Figure 2, when crisis severity is high, consumer brand trust is lower when there is a scapegoat brand than when there is not (mean = 4.08 and 4.41, respectively, p < 0.05), indicating a significant consumer scapegoating effect; when crisis severity is low, consumer brand trust has no significant difference under the two conditions (mean = 3.85 and 3.89, respectively, p = 0.793).
Sustainability 2020, 12, x FOR PEER REVIEW 7 of 14 the foreign brand category than those in the control group (mean = 4.05 and 3.75, respectively, p < 0.05). Hypothesis 1b is supported. Then, we did analyses for the necessity groups and non-necessity groups, respectively. When a smartphone was manipulated as a necessity, we did not find any significant main effects of consumer scapegoating for both the domestic brand category (mean = 3.27 and 3.12, with emergence of a scapegoat brand and with no emergence of a scapegoat brand, respectively, p = 0.262), and the foreign brand category (mean = 3.97 and 4.15, with emergence of a scapegoat brand and with no emergence of a scapegoat brand, respectively, p = 0.072). Hypothesis 2a is not supported and Hypothesis 2b is only marginally supported. As Figure 1 shows, for the domestic brand category, when crisis severity is high, consumer brand trust is higher when there is a scapegoat brand than when there is not (mean = 3.49 and 3.08, respectively, p < 0.05), indicating a significant consumer scapegoating effect; when crisis severity is low, consumer brand trust has no significant difference under the two conditions (mean = 3.05 and 3.15, respectively, p = 0.570). For the foreign brand category, as shown in Figure 2, when crisis severity is high, consumer brand trust is lower when there is a scapegoat brand than when there is not (mean = 4.08 and 4.41, respectively, p < 0.05), indicating a significant consumer scapegoating effect; when crisis severity is low, consumer brand trust has no significant difference under the two conditions (mean = 3.85 and 3.89, respectively, p = 0.793).   The above results show that when a multi-brand crisis happens for a necessity, Hypothesis 1a, Hypothesis 1b, and Hypothesis 3 are supported, while Hypothesis 2a and Hypothesis 2b are not supported. On one hand, the crisis brand category would be negatively affected, whereas the competing brand category could gain benefits from the crisis. On the other hand, although no significant main effect of consumer scapegoating has been found, the scapegoat signal does work under conditions with high crisis severity.
When a smartphone was manipulated as a non-necessity, for the domestic brand category, no significant main effect of consumer scapegoating was found (mean = 3.22 and 3.32, with or without a scapegoat brand, respectively, p = 0.474). Besides, consumer brand trust has no significant difference when there is a scapegoat brand compared with when there is not, regardless of whether crisis severity is high (mean = 3.13 and 3.28, respectively, p = 0.445) or low (mean = 3.30 and 3.35, respectively, p = 0.795). For foreign brands, no significant main effect of consumer scapegoating was found either (mean = 4.03 and 4.02, with or without a scapegoat brand, respectively, p = 0.937). Consumer brand trust has no significant difference when there is a scapegoat brand compared with when there is not, regardless of whether crisis severity is high (mean = 4.06 and 4.03, respectively, p = 0.856) or low (mean = 4.01 and 4.02, respectively, p = 0.946). These results show that when a multi-brand crisis happens for a non-necessity, the scapegoat signal does not work.

• Discussion
The results of Study 1 generally show the existence of a consumer scapegoating effect for necessities in a severe multi-brand crisis. Under such a condition, the emergence of a scapegoat brand could help consumer brand trust in the crisis brand category to recover, and take away the advantage gained by the competing brand category through the crisis, relieving other brands in the crisis brand category from pressures. The consumer scapegoating effect would disappear as the multi-brand crisis becomes less severe. For non-necessities, however, we did not find a significant consumer scapegoating effect, whether the crisis was severe or less severe. Therefore, if something really bad happens, from the perspective of the crisis brand category, emergence of a scapegoat brand could be beneficial, whereas the competing brand category may benefit from a multi-brand crisis. It could be a great opportunity for brands in the competing brand category to establish brand reputation and explore markets.

Study 2
Although the results of Study 1 provide basic supports to a consumer scapegoating effect, one potential drawback of Study 1 is that we divided brands into two categories based on brand nationalities in the experiment design. In this research, we conceptualized the consumer scapegoating effect as the difference between with or without emergence of a scapegoat brand, and we tested it by comparing the two conditions within a same brand category, so brand nationality should not be a confounding factor. However, adopting the design of Study 1, we could not completely rule out the possible influence caused by brand nationality, since it may influence consumers' attitudes towards a crisis brand category and a competing brand category. Therefore, we conducted Study 2 to exclude the alternative explanation of brand nationality. The other target of Study 2 was to test the mediating role of cognitive dissonance.
• Design A 2 (scapegoat brand: yes versus no) × 2 (crisis severity: high versus low) between-subject design was used. We applied a similar design as in Study 1, the only difference being that the multi-brand crisis happened in a foreign brand category, so the domestic brand category became the competing brand category. A non-student sample was recruited online, a total of 183 participants participated in our study, 48% male, and they were randomly assigned into the 4 groups. Participants were firstly asked to read a piece of news regarding a multi-brand crisis. The news for each group was almost the same as that used in Study 1, with only two differences. For one thing, the introduction to manipulate product type was omitted, and for the other, the multi-brand crisis was described to happen in a foreign brand category.
After reading the news, each participant answered some questions designed on the basis of the news contents, and those who gave wrong answers were removed from the dataset. Then, participants were asked to evaluate the crisis severity and answer about their brand trust in the foreign smartphone brand category and the domestic smartphone brand category. Six-point Likert scales were used for all the questions.
Next, we asked the participants to imagine that they were in urgent need of buying a smartphone because their own were lost. Participants needed to choose the words that best described the extent to how they felt when they were trying to choose one smartphone in the marketplace among the five answers: not at all, a little bit, somewhat, very much, and extremely. Four items of negative emotions were listed: discomfort, distressed, irritable, and upset, chosen from items used in the Positive and Negative Affect Schedule-Expanded Form (PANAS-X) [45]. These items were used to measure participants' cognitive dissonance levels. Finally, participants were asked about demographic information and informed that all reading materials were made up. •

Results
Manipulation check. Participants in low crisis severity groups perceived lower crisis severity than those in high crisis severity groups (mean = 4.51 and 5.04, respectively, p < 0.01). The manipulation of crisis type was successful.
Dependent variables. The four variables of brand-trust scale consist of one factor for both the foreign brand category (Cronbach's α = 0.86) and the domestic brand category (Cronbach's α = 0.85), so the mean of the four variables was taken to represent the level of brand trust. Results show significant main effects of consumer scapegoating for both the foreign brand category (mean = 3.34 and 2.98, with and without emergence of a scapegoat brand, respectively, p < 0.05) and the domestic brand category (mean = 3.52 and 3.79, with and without emergence of a scapegoat brand, respectively, p < 0.05). Hypothesis 2a and Hypothesis 2b are supported.
As shown in Figure 3, for the foreign brand category (the crisis brand category), when crisis severity is high, consumer brand trust is higher when there is a scapegoat brand than when there is not (mean = 3.41 and 2.85, respectively, p < 0.01), indicating a significant consumer scapegoating effect; when crisis severity is low, consumer brand trust has no significant difference under the two conditions (mean = 3.26 and 3.08, respectively, p = 0.286).
Sustainability 2020, 12, x FOR PEER REVIEW 9 of 14 foreign smartphone brand category and the domestic smartphone brand category. Six-point Likert scales were used for all the questions. Next, we asked the participants to imagine that they were in urgent need of buying a smartphone because their own were lost. Participants needed to choose the words that best described the extent to how they felt when they were trying to choose one smartphone in the marketplace among the five answers: not at all, a little bit, somewhat, very much, and extremely. Four items of negative emotions were listed: discomfort, distressed, irritable, and upset, chosen from items used in the Positive and Negative Affect Schedule-Expanded Form (PANAS-X) [45]. These items were used to measure participants' cognitive dissonance levels. Finally, participants were asked about demographic information and informed that all reading materials were made up. •

Results
Manipulation check. Participants in low crisis severity groups perceived lower crisis severity than those in high crisis severity groups (mean = 4.51 and 5.04, respectively, p < 0.01). The manipulation of crisis type was successful.
Dependent variables. The four variables of brand-trust scale consist of one factor for both the foreign brand category (Cronbach's α = 0.86) and the domestic brand category (Cronbach's α = 0.85), so the mean of the four variables was taken to represent the level of brand trust. Results show significant main effects of consumer scapegoating for both the foreign brand category (mean = 3.34 and 2.98, with and without emergence of a scapegoat brand, respectively, p < 0.05) and the domestic brand category (mean = 3.52 and 3.79, with and without emergence of a scapegoat brand, respectively, p < 0.05). Hypothesis 2a and Hypothesis 2b are supported.
As shown in Figure 3, for the foreign brand category (the crisis brand category), when crisis severity is high, consumer brand trust is higher when there is a scapegoat brand than when there is not (mean = 3.41 and 2.85, respectively, p < 0.01), indicating a significant consumer scapegoating effect; when crisis severity is low, consumer brand trust has no significant difference under the two conditions (mean = 3.26 and 3.08, respectively, p = 0.286). For the domestic brand category (the competing brand category), as Figure 4 shows, when crisis severity is high, consumer brand trust is lower when there is a scapegoat brand than when there is not (mean = 3.49 and 3.97, respectively, p < 0.05), indicating a significant consumer scapegoating effect; when crisis severity is low, consumers' brand trust has no significant difference under the two conditions (mean = 3.56 and 3.67, respectively, p = 0.489). Results support Hypothesis 3. For the domestic brand category (the competing brand category), as Figure 4 shows, when crisis severity is high, consumer brand trust is lower when there is a scapegoat brand than when there is not (mean = 3.49 and 3.97, respectively, p < 0.05), indicating a significant consumer scapegoating effect; when crisis severity is low, consumers' brand trust has no significant difference under the two conditions (mean = 3.56 and 3.67, respectively, p = 0.489). Results support Hypothesis 3. Next, we ran moderated mediation analyses proposed by Hayes (model 8; 5000 bootstrap samples) [46] to test our hypothesis that a consumer scapegoating effect, moderated by crisis severity, would be mediated by cognitive dissonance in a multi-brand crisis. The four items of negative emotions consist of one factor (Cronbach's α = 0.86), so we used their mean as the score of cognitive dissonance level. As Table 1 shows, the results were consistent with our prediction; for the foreign brand category (the crisis brand category), a   Next, we ran moderated mediation analyses proposed by Hayes (model 8; 5000 bootstrap samples) [46] to test our hypothesis that a consumer scapegoating effect, moderated by crisis severity, would be mediated by cognitive dissonance in a multi-brand crisis. The four items of negative emotions consist of one factor (Cronbach's α = 0.86), so we used their mean as the score of cognitive dissonance level. As Table 1 shows, the results were consistent with our prediction; for the foreign brand category

Conclusions
Extending the scope of previous literature on brand crisis, most of which only considered single-brand crises, this research has explored the consumer scapegoating effect in the context of a multi-brand crisis through two studies. Two brand categories were considered, a crisis brand category and a competing brand category. Consistent with the categorization theory [25][26][27], results of this research show that when a multi-brand crisis happens, consumer brand trust in a crisis brand category would decrease because of assimilation, while that in a competing brand category would, on the contrary, increase because of a contrast effect. Although the scapegoating effect has been proposed and verified in various research areas [29][30][31][32][33][34], there is only one piece of research exploring consumer scapegoating, and it only considered the scapegoat brand itself [2]. The current research, in comparison, shows that the emergence of a scapegoat brand would bring changes to the post-crisis "destiny" of both a crisis brand category and a competing brand category. Generally speaking, our results show a consumer scapegoating effect in a severe multi-brand crisis. From the perspective of the crisis brand category, consumers grasp the scapegoat signal, leading the crisis brand category to recover, whereas for the competing brand category, the emergence of a scapegoat brand in the crisis brand category may take away the benefits it has gained from the crisis. We also discussed the psychological mechanism of a consumer scapegoating effect, suggesting negative emotions caused by cognitive dissonance as the mediator of this process.

Theoretical Contributions
Theoretically speaking, our research contributes to the existing literature on multi-brand crisis and extends the research boundary of a scapegoating effect. In previous literature, scholars mainly discussed the influence of a single-brand crisis. As more multi-brand crises emerged, scholars have started to pay attention to such crises. A few scholars have discussed how individual brands or the industry would be influenced when multiple brands are involved in a crisis, and thus proposed the concept of consumer scapegoating. However, no research has deeply discussed how the crisis brand category and the competing brand category may be influenced by multi-brand crises or consumer scapegoating. In this research, we extended discussions from individual brands to the crisis brand category and the competing brand category. Results of this research have shown that, although the two brand categories belong to a same industry, a multi-brand crisis would have different influence on them regarding brand evaluations.
Second, we explored a consumer scapegoating effect and its psychological mechanism in multi-brand crises. Although scapegoating is widely discussed in other academic fields [29], to our best knowledge, there are only two research studies discussing consumer scapegoating in the field of marketing [2,4]. However, they overlooked the mechanism of a consumer scapegoating effect. In the current research, we explored the influence of a scapegoat signal on both the crisis brand category and the competing brand category and applied experimental methodology to test the underlying psychological process of a consumer scapegoating effect, which could broaden the research scope of scapegoating.

Managerial Implications
By offering a deep and structured understanding of a consumer scapegoating effect, this research not only contributes to existing theories, but also has managerial implications for both marketers and industrial regulators. For individual brands, our results show two important suggestions. First, the best solution to a crisis is prevention. We have shown that it is better for a brand not to be the scapegoat in a crisis. However, it is quite hard for individual brands to control which brand would be the scapegoat, yet they can control things done by themselves. A perfect situation is one in which individual brands should do things right and do right things. Making no mistakes means no punishments. Although they may be affected by others' mistakes, and they could not require all the other brands to do right things, they could still recover once a scapegoat brand is punished. Therefore, rather than racking their brains to figure out how to avoid being a scapegoat brand or how to regain consumers' trust, it would be better for individual brands to keep their conscience when doing business and try to create a win-win situation for both consumers and themselves.
Second, this research shows that, though brands in a same industry are producing similar products, they may be influenced differently by a multi-brand crisis. For those in the crisis brand category, they would be negatively affected because of consumer assimilation, whereas for those in the competing brand category, they could even benefit from such a crisis. Therefore, if an individual brand has been involved in the crisis, the brand could endeavor to get rid of its affiliation with the crisis brand category and make itself look more like a brand that belongs to the competing brand category. This could be achieved through weakening of some brand characteristic with which consumers treat it as a brand in the crisis brand category, since brand categories are classified based on some certain brand characteristic. If this is difficult to achieve, the brand could have another choice, by creating a new sub-brand which can be easily classified by consumers as one member of the competing brand category. When a multi-brand crisis happens, such a sub-brand could have immunity against the consumer assimilation. If a brand has multiple sub-brands with quite different brand characteristics, it would be easier to get through a multi-brand crisis.
In addition, this research also has significant implications for industrial regulators. When a multi-brand crisis happens, especially when it is very severe, we suggest that industrial regulators take actions as quickly as possible. The sooner regulators punish brands that have made mistakes, the sooner consumers would get back into the market, since the results of this research show that consumers need a signal (such as emergence of a scapegoat brand) to get over the crisis.

Limitations and Future Research
Though we deeply explored the consumer scapegoating effect in the multi-brand-crisis context, making contributions both theoretically and practically, there are still two limitations in this research that could be discussed in future studies. First, scapegoating actually emerged from an interactive dynamic process of multiple players, such as the media, firms, consumers, and regulators. In this research, we did not explore the emerging process of the scapegoat brand, which could be further explored in the future. Second, in this research we analyzed the consumer scapegoating effect from the perspective of individual consumers, discussing how individual consumers would be influenced by the emergence of a scapegoat brand. However, both formation and impacts of scapegoating involve group behaviors. Deeper discussions from the perspective of groups in the future may make further contributions to research on the consumer scapegoating effect and help us better understand things happening before and after the scapegoat brand appears.