Consumer boycott behavior: an exploratory analysis of twitter feeds.
The "anticonsumption concept demands action against consumption" for reasons related to consumer values and ethics or as a form of political activism (Yuksel and Mryteza 2009, 2). Consumer boycotts are a form of anticonsumption behavior, where boycotters are market activists who forgo the consumption of certain products and services because of environmental, political, ethical, or social issues (Chatzidakis and Lee 2013; Hoffmann 2011; Yuksel 2013; Yuksel and Mryteza 2009). While anticonsumption in a broad sense is characterized by a negative attitude toward the market and consumption in general, consumer boycotts have a more narrow scope, and are targeted toward specific organizations or entities (Hoffmann 2011; Lai and Aritejo 2010). Formally, consumer boycott is defined as "an attempt by one or more parties to achieve certain objectives by urging individual consumers to refrain from making selected purchases in the marketplace" (Friedman 1985, 97). Consumer boycott is also described as the refusal to conduct market transactions with the boycott target (Garrett 1987).
Boycott movements have been fueled by increasing public attention to corporate social responsibility (Klein, Smith, and John 2004) and are often used as one of the most effective consumer actions against companies that engage in practices deemed unethical or unjustified (Friedman 1999). Consumer boycotts are strongly linked to consumer well-being. In fact, the main tenet of consumer well-being posits that higher levels of consumer well-being are associated with higher levels of satisfaction and life happiness of consumers, and welfare of the society (Sirgy, Lee, and Rahtz 2007). A boycott may occur when boycotters feel that their well-being or a third party's well-being is threatened by egregious behavior (Friedman 1999). By promoting boycotts, consumers express their dissatisfaction with company or country actions or policies (Shaw, Newholm, and Dickinson 2006). This form of anticonsumption behavior is often aimed at affecting public welfare and increasing the "general good" by modifying organizational and government practices or building collective social movements against issues of concern (Yuksel 2013).
Boycott movements negatively affect consumer attitudes and purchase intentions toward the target company and its products (Ettenson and Klein 2005; Klein, Smith, and John 2002). As a result, consumer boycotts can hurt a company's image (Klein et al. 2004) and adversely influence its financial performance. Even though the results from past research about the effect of boycotts on a company's financial performance are mixed (Koku 2012; Koku, Akhigbe, and Springer 1997), several studies have found that boycott movements result in significant stock price drops (Pruitt and Friedman 1986) and influence change in company policies (Davidson, Worrell, and El-Jelly 1995).
The upsurge of consumer empowerment and connectedness can exacerbate the adverse effects of boycott behaviors on company performance. In fact, with the increased use of the Internet and the proliferation of social media, boycott organizers can reach thousands and even millions of consumers effectively and with unprecedented speed (Sen, Giirhan-Canli, and Morwitz 2001). For instance, recently animal rights activists have used social media to organize a "Boycott SeaWorld" movement via Facebook, Tumbler, Twitter, and similar platforms, reaching millions of consumers. Outside the realm of organized boycott movements, individual consumers can also engage in boycott behaviors by sharing their boycott intentions with hundreds of their family members, friends, and followers via personal Facebook pages, Twitter accounts, and online blogs.
Therefore, it is important for anti-consumption researchers and for businesses to understand the motives behind consumer boycotts of specific targets and the causes that drive consumers to engage in boycott behaviors. Prior research on boycott behavior has made great progress especially in identifying consumer boycott motivations (John and Klein 2003; Klein et al. 2004; Kozinets and Handelman 1998) and studying the consequences of boycott movements on firm performance and actions (Ettenson and Klein 2005; Klein et al. 2002; Koku 2012). Despite such research efforts, there are several research gaps that need to be addressed.
Previous research on boycott behavior typically relied on survey and experimental research methods. This might not only limit the in-depth understanding of consumer boycotts, but also more narrowly focus the research context. In fact, there is a tendency in past boycott research to investigate boycott motivations and behaviors related to one specific cause or target company (e.g., Nestle) (Brinkmann 2004; Klein et al. 2002). Thus, there is lack of research on how varying boycott causes and targets influence consumer boycott motivations. The past decade has witnessed increasing calls for studying boycott behavior using qualitative research methods and data collected via the Internet and social media platforms (Braunberger and Buckler 2011; James 2010; Kozinets 1997; Kozinets and Handelman 1998), which allow for a more profound look at consumer boycotts using rich data, from a naturalistic setting, and without context constraints.
Studies of boycott motivations have been mostly grounded in cost-benefit theoretical perspectives (James 2010). Yet, an emerging research area in boycott behavior adopts socio-psychological theoretical views (Farah and Newman 2010; James 2010; Lindenmeier, Schleer, and Price 2012) that emphasize the role of consumer emotions. Boycotts are viewed as a way of emotional expression (Hoffmann and Miiller 2009), where negative consumer emotions, such as outrage, play a key role in increasing boycott participation (Lindenmeier et al. 2012). Despite the importance of consumer emotions, research examining consumer expression of emotions in boycott messages, the emotional intensity of these messages, and their relationship with boycott motivations is scarce and constitutes a valuable research area requiring further attention (James 2010).
The present study aims at addressing these research gaps with two main objectives. First, we explore the motivations, causes, and targets of consumer boycotts using content analysis of Twitter feeds. What drives consumers to participate in boycotts? What are the causes that fuel consumer boycotts? Which entities do consumers target in their boycotts? These are questions we address through exploratory content analysis--without a priori expectations required by survey and experimental research--of actual consumer boycott messages collected via Twitter, a valuable social media source for gathering consumer opinions about a diverse range of topics (Park and Paroubek 2010).
Next, we identify consumer expression of emotions in Twitter boycott messages and employ human sentiment analysis--an application of quantitative language processing models to understand how people intensify their opinions (Carrillo-de-Albomoz and Plaza 2013)--to investigate the intensity of these messages in relation to boycott motivations. The intensity of boycott messages can influence their effectiveness and is worth exploring to allow for a better understanding of consumer engagement in boycott behaviors and the motivations that drive these behaviors, given various boycott causes and targets. Following a review of the research background, we report research methods and findings, and then present conclusions, research implications, limitations, and future research directions.
Consumer Boycott Motivations
The understanding of anticonsumption motives, or reasons against consumption, provides information beyond other fields of consumer behavior that focus on reasons for consumption (Chatzidakis and Lee 2013). Even though boycotting can be seen as a planned and collective anti-consumption action (Friedman 1985; Garrett 1987; Yuksel and Mryteza 2009), boycott behavior is an individual's private response to organizations and their offerings, thus an important dimension of consumer complaint behavior (Singh 1988). Exploring individual motivations is crucial for a better understanding of consumer boycott intentions and decisions. Drawing from the prior literature, consumer boycott motivations can be classified into instrumental vs. noninstrumental (Friedman 1985, 1991, 1999; John and Klein 2003; Klein et al. 2002, 2004; Yuksel 2013), with the possibility of mixed motivations for boycott decisions influenced by both (Ettenson and Klein 2005; Klein et al. 2002).
When individuals are driven to engage in boycotts by instrumental motivations, their goals are explicitly stated or presented concretely (Ettenson and Klein 2005; Klein et al. 2002). The boycott behavior is used as a tactic to influence change in the actions and policies of the target firm (Friedman 1999; John and Klein 2003; Yuksel 2013), e.g., by asking the firm to lower prices or to sign union contracts (Friedman 1999). Instrumentally motivated boycott messages can also be aimed at signaling, to other consumers, the necessity of boycott behaviors or informing them about appropriate conduct (Klein et al. 2004). Consumers with instrumental boycott motivations often have high perceived efficacy (Sen et al. 2001). They believe that boycotts are effective and that their actions can make a difference in outcomes, by influencing change in the target company's practices or in the actions of other consumers (John and Klein 2003).
Instrumental motivations are dominant in boycott decisions because boycotts generally have instrumental goals. Instrumentally motivated boycotts are also more effective in building momentum because of their underlying perceived efficacy and positive message framing related to the belief in making a difference (Sen et al. 2001). However, motivations are complex and consumers can often be driven to engage in boycott behaviors by psychological needs rather than concrete outcome-focused goals (Klein et al. 2002).
While instrumentally motivated boycotters have clearly stated and practical goals focused on changing the actions or policies of the boycott target, consumers who have noninstrumental motives to engage in boycott actions do so to express their displeasure with the unjustified actions of the boycott target (Ettenson and Klein 2005; Friedman 1999). Noninstrumental boycotts, including "expressive boycotts," are characterized by "vague statements of goals" (Friedman 1991, 153), where the consumers focus on generally protesting the actions of the boycott target, venting their frustrations with these actions (Friedman 1999), and using boycotts as a means of self-expression and self-realization (Kozinets and Handelman 1998; Yuksel 2013). Noninstrumental motivations drive consumers to engage in boycott behaviors based on psychological utility gain or loss. By venting their frustrations, consumers can diminish their negative psychological states and as a result, experience relief (American Psychological Association 2007). Here, it is important to note that venting as a noninstrumental driver of boycott behavior is linked, by definition, to the action of forgoing consumption (Yuksel and Mryteza 2009) or refusing to conduct market transactions with the boycott target (Garrett 1987). It is also therefore directed toward a specific target and differs from other forms of venting, which are not necessarily action-oriented nor aimed at a target. When driven by noninstrumental motivations, consumer decisions to boycott are independent from the desire to have others participate in the boycott (Klein et al. 2004).
John and Klein (2003) conceptualize several noninstrumental motivations that increase the psychological utility of participating in boycott movements. One form is expressive motivations, where individuals take part in boycott behaviors to vent or express displeasure, anger, or outrage toward the acts of the boycott target (Brennan and Lomasky 1993; Friedman 1999). Thus, this type of boycott is usually related to unsuccessful boycott outcomes (John and Klein 2003). Clean hands (Smith 1990), referred to later as self-enhancement motivation (Klein et al. 2004), is another form where individuals increase self-esteem by participating in a boycott out of moral obligation or a mindset of moral superiority. Self-enhancement motivations can also drive consumers to participate in boycotts to avoid the feeling of guilt or discomfort resulting from engaging in marketing transactions with companies deemed unethical (John and Klein 2003). They are based on intrinsic rewards and are often a response to social pressures when individuals want others to perceive them in a positive light (Klein et al. 2004). Noninstrumental boycott motivations can also be exhibited in consumers' desire to punish the target company by boycotting its products/services or by hoping for its demise (Abosag and Farah 2014; Braunberger and Buckler 2011).
Consumer Emotions and Boycott Behavior
An important development in marketing lies in the exploration of emotions and their effects on consumption behaviors (Baggozzi, Gopinath, and Nyer 1999). Affect, especially negative emotions such as anger, is an important part of anti-consumption frameworks (Lai and Aritejo 2010). Prior research suggests that boycott participation is an "emotional expression of a consumer's attitude" (Farah and Newman 2010, 349). However, there is very little research investigating the role of emotions in boycott behaviors and the expression of emotions in boycott messages. In the service recovery literature, emotions have been shown to play a crucial role between consumer perceptions of firm injustice and postpurchase behavioral reactions to these perceptions, especially when they have high intensity (Shoefer and Ennew 2005). The next section tackles the intensity aspects of consumer emotions and boycott messages.
A few studies, exploring the role of emotions in consumer boycott behaviors, show that boycotts are a way for consumers to emotionally express themselves (Hoffmann and Muller 2009), and that emotional expression is a strong predictor of consumer engagement in boycott behaviors (Farah and Newman 2010). In fact, boycotts, particularly those motivated by noninstrumental expressive motives, have a strong emotional aspect (Klein et al. 2002), but this can also be true for other types of motivations. Moreover, emotions can be a critical mediator between the firm's unethical actions or bad reputation and consumer boycott behaviors (Hoffmann and Muller 2009; Lindenmeier et al. 2012). This is especially true for negative emotions such as disgust, which constitute a key clue for a better understanding of boycott motivations and behaviors (Braunberger and Buckler 2011). Despite falling short on offering an in-depth look into consumer emotions and boycott behaviors, these studies emphasize the importance of examining the expression of emotions in boycott messages. The present study addresses this research gap by investigating emotional expressions in boycott messages shared by consumers online using Twitter.
Boycott Message Intensity and Sentiment Analysis
Consumer appraisals of specific events elicit emotional responses of various intensities, which consequently affect their behaviors (Shoefer and Ennew 2005). The intensity of affect experienced by consumers is linked to the likelihood of individuals engaging in more anticonsumption behaviors (Lai and Aritejo 2010). For instance, when actions by a firm are deemed unethical or unjust, consumers might experience intense negative emotions, such as anger. The intensity of such emotions can be a strong predictor of negative behavioral responses (Bougie, Pieters, and Zeelenberg 2003), such as consumers' intentional boycott of the firm's products and services, which is a high-intensity anti-consumption behavior (Lai and Aritejo 2010). Factors such as valence and intensity are important to the understanding of heterogeneous emotional experiences. Researchers in the area of semantic associations have attempted to use language and to develop word and emotional intensity ratings, in order to identify human emotional states (Strauss and Allen 2008).
Sentiment analysis is a research technique that collects and analyzes textual data (Rambocas and Gama 2013). The main purpose of sentiment analysis is to examine language to identify consumer opinions (positive vs. negative) and sentiments to detect message intensity (strong vs. weak) in a given text (Pang and Lee 2008). The proliferation of social media and the increasing access to consumer data available on platforms such as Facebook, Twitter, and online blogs have motivated the use of sentiment analysis, which "offers organizations the ability to monitor various social media sites in real time and act accordingly" (Feldman 2013, 82). Sentiment analysis allows organizations to conduct market intelligence and provides valuable information about consumer perceptions of companies and their offerings, the public mood, and consumer antagonistic feelings toward companies and brands (Carrillo-de-Albornoz and Plaza 2013; Kumar and Sebastian 2012).
There are various sentiment analysis techniques in the field and many commercially available sentiment analysis algorithms based on automated language processing models (Feldman 2013; Kumar and Sebastian 2012). These techniques are usually based on mining opinion words, their semantic orientation, and their intensity. They also allow for the identification of opinion intensifiers used by consumers, such as capitalization, exclamation marks, and profanity, and they take into account emoticons (such as a smiley "[C]" or angry face "X-("), commonly used in online messages as pseudo expressions of emotions (Kumar and Sebastian 2012). Automated sentiment analysis algorithms have been criticized for being limited when it comes to detecting sarcasm and mining through noisy texts that use slang or problematic punctuation (Feldman 2013).
Therefore, instead of using automated sentiment analysis, we use human sentiment analysis performed by two coders, in order to overcome some of these limitations. Human sentiment analysis provides advantages over automated language processing algorithms, by enabling us to: (1) understand consumer sentiment within the context of the message (e.g., love in a sentence might not be aimed at the boycott target but a software analysis would perceive it as a positive opinion word), and (2) interpret slang words and abbreviations and detect sarcasm and profanity, which are difficult to program into dictionary-based algorithms.
There is a call for integrating sentiment analysis with qualitative research techniques in order to provide more accurate insights into consumer opinions and emotions (Rambocas and Gama 2013). The following section provides details about the use of both content analysis and human sentiment analysis approaches in the analysis of consumer boycott tweets.
STUDY APPROACH AND RESEARCH METHOD
In order to better understand the underlying motives, causes, and targets of consumer boycotts, this research uses Twitter feed online data. Twitter, a real-time microblogging service with a massive reach of audiences (more than 645 million users as of 2014), provides a platform with an immense amount of data (Kumar and Sebastian 2012; Savage 2011). The service allows users to post micro blogs limited to 140 characters each, called tweets. These tweets usually have an informal writing style with no censorship for profanity, and can contain slang, emoticons, acronyms, and web links to other online content. The tweets are expressions of consumers in real life and are therefore relevant and salient to the boycotters who share them online (Hoffmann 2011). Textual consumer data available online through social media platforms give researchers low-cost access to data, allow data collection in a naturalistic and unobtrusive setting with a high level of anonymity (Kozinets 1997; Park and Paroubek 2010), and offer many advantages over traditional qualitative data, such as face-to-face interviews (Podoshen 2012). It is therefore well suited for research involving sensitive topics such as boycotts (Kozinets and Handelman 1998; Savage 2011).
Description of the Data Set and Study Sample
In Twitter, data are updated in real time and users are continuously posting tweets about different topics. The hashtag "#" symbol is used by Twitter members to mark the topics of their tweets. In order to retrieve the online data, the Twitter application programming interface (API) was used to crawl Twitter for tweets with the hashtag "#boycott" over the period of one month starting December 12, 2013. Using the hashtag "#boycott" automatically eliminates any tweets that are not related to consumer boycotts, such as tweets containing simple venting. The total number of tweets retrieved by the end of this time period was 14,785.
The data were subsequently cleaned by eliminating non-English tweets (around 65% of all tweets), tweets that are exact duplicates copied by individuals from other Twitter users, and re-tweets, which are also duplicates of existing tweets shared by individuals who give credit to their original author by using the re-tweet (RT) symbol. Although re-tweets can provide very insightful information about the efficacy and reach of different tweets, the data set obtained using Twitter API does not provide the number of times each Twitter message was re-tweeted. Also, given the time frame of the data collection (one month), the original poster of a re-tweeted tweet could fall outside of the time frame and thus be excluded from the retrieved data. The inability to trace the re-tweets to their original poster, or to track the number of re-tweets for each message, constitutes a research limitation discussed later in the study.
However, given that the main objective of this research is to accurately identify opinions, emotions, and sentiment intensity of the actual users who wrote the boycott messages, cleaning re-tweets from the data was a valid approach for creating a data set of original tweets, expressed by consumers in their own words. In fact, a consumer who re-tweets another consumer's boycott message might support the original poster's opinion about the boycott or its target, but would be less likely to share the emotions and sentiment intensity linked to that opinion. However, tweets that are replies to another person's boycott message and which constitute a prominent way consumers engage with tweets were not deleted because they reflect the posters' expressions of their own sentiments.
The clean data set consisted of 4,904 tweets. A sample of 2,000 tweets was then selected randomly for a preliminary analysis, with the intention of expanding or reducing the sample size based on when convergence in themes and data saturation (Strauss and Corbin 1998) is established during the iterative analysis process described in the following section. At this stage, two coders examined the sample to identify spam tweets using guidelines from past research (Pettit 2013). After eliminating the spam--about 29% of the sample including tweets with one repeated word, unusual sales content, undecipherable side conversations between users, or other content that does not fit boycott research--the final sample size was 1,422 tweets.
Data Analysis Procedures
This study uses a mixed-method approach that involves qualitative content analysis of tweets to identify different themes related to boycott motives, causes, and targets, followed by sentiment analysis aimed at examining consumer emotions in boycott messages and quantifying the emotional intensity of these messages. Finally, quantitative data are generated using a tally of frequencies and percentages for different themes in order to run descriptive analyses, analyses of variance, and contingency statistical analyses, for a more generalizable interpretation of our findings (Hoffmann 2011).
Two coders, the authors of this study, performed a content analysis of the boycott messages obtained from Twitter. The main objective of the content analysis was to identify themes related to: (1) whether the motivation of a boycott message was instrumental, noninstrumental, or both (we termed this general motivation), (2) more specific motivations within the instrumental and noninstrumental classification, which are referred to as specific motives, such as making a difference, self-enhancement, or punishment, (3) the causes that triggered consumer engagement in boycott behaviors, such as human rights issues or corporate failures, and (4) the target(s) of the boycott message.
For each of the elements examined in the content analysis, including motivations, causes, and targets of boycott messages, the coders started with a set of categories deductively derived both from previous research theories and findings and a pretest involving the preliminary analysis of 100 tweets. This was followed by an analysis of the entire sample of 1,422 tweets to assign the motivations, causes, and targets of boycott messages to existing categories and inductively come up with additional categories for tweets that do not fit existing ones (Hoffmann 2011). It is important to note that after analyzing around 700 tweets, no additions or changes to the thematic categories for motives, causes, and targets of boycott messages were required. The remaining tweets in the sample fit within the established categories and reflected similar percentage distribution among these categories. This provided support for the generalizability of our findings about the motives, causes, and targets of consumer boycotts that were prominent during the study's timeframe. A further expansion of the sample size was deemed unnecessary.
The data analysis process entailed the coders spending five intensive weeks carefully reading each tweet several times and following an iterative process to identify recurring themes, analyzing the boycott messages for expected and evolving themes, and then coding them accordingly (Kozinets and Handelman 1998). In order to gain the proper background, especially when identifying causes and targets of the boycotts, the authors also immersed themselves in the context of the boycott messages by following message links to online content and examining news stories about the issues or events mentioned in the tweets.
After three iterations, a hierarchical category schedule with main categories and subcategories was developed for each element (Hoffmann 2011). For example, under boycott causes, the iterations revealed a general human rights category as well as subcategories within it, including women's rights, gay rights, second amendment rights, etc. A fourth iteration was then used to recategorize all the tweets in the sample. The final iteration was conducted to check for intercoder agreement. The two coders agreed on all the elements examined by the content analysis for 79% of the tweets. Disagreement on any of the items was then resolved through discussion. After a final coding of the data, a tally of the frequencies and percentages of the themes for general motivations, specific motives, causes, and targets of boycotts was reported.
Often qualitative research is criticized for being subject to research bias (Mays and Pope 1995; Norris 1997). The most practical way to increase the validity of qualitative research results is for researchers to focus on minimizing error and to actively reduce bias (Norris 1997). To minimize bias in the current study, the researchers took several key steps (Merriam 2009). First, Twitter messages were purposefully sampled for tweets with "#boycott" over one whole month and without restrictions to the topic of interest to allow for a greater range of applications. Second, data analysis was conducted over multiple iterations to identify recurring themes that fit the analytical framework (instrumental vs. noninstrumental boycott motives). Third, the "triangulating analysts" method was used to allow investigator triangulation, whereby multiple investigators analyzed the same data independently and compared their findings (Patton 2002). Finally, quantitative analysis served as an additional approach for looking at the data to ensure the validity of qualitative conclusions (Mays and Pope 1995).
During the second step of the data analysis, the goal was to examine and quantify the emotional intensity of boycott messages using sentiment analysis, performed by human coders rather than an automated algorithm. First, consumer emotions, expressed in boycott messages such as anger, dislike, or disgust, were identified. Two coders classified each emotion into one of three intensity levels--high, moderate, and low--by adopting guidelines from past literature (Kumar and Sebastian 2012; Strauss and Allen 2008). Message emoticons, such as a smiley or angry face, were coded similarly, as they are considered pseudo expressions of emotions. Intercoder agreement on emotion intensity levels was 83% and disagreement was resolved through discussion. Table 1 provides examples of emotions and emoticons found in the Twitter sample of boycott messages, and their corresponding intensity classification.
Second, the research looked at elements of boycott messages that could inform its intensity and adapted Kumar and Sebastian's (2012) sentiment analysis technique in order to identify the following message intensifiers: (1) the length of the boycott message, which could reflect the level of consumer engagement in writing it, (2) capital letter words, a pseudo for the human action of screaming online, (3) expressive punctuation such as exclamation marks and question marks unless used appropriately at the end of a question "!!!???", (4) the number of profanity and insult words describing the target of the boycott, and (5) the absence or presence of sarcasm. Most of these intensifiers can be objectively assessed and did not require multiple coders, except for sarcasm. The inclusion of sarcasm as a message intensifier constitutes an important advantage of human sentiment analysis over automated algorithms (Feldman 2013). Because of the lack of past research guidelines on coding sarcasm and because perceptions of sarcasm can be subjective, three coders analyzed each tweet for the presence of sarcasm and coded it as a binary variable (0 for absence of sarcasm; 1 otherwise). The intercoder agreement was 85%, and disagreement was resolved through discussion.
Finally, to quantify the intensity of boycott messages, a numerical value was assigned to each of the elements discussed in this section, and then the sum of all values was computed for each tweet. The number of points assigned to each type of intensifier does not take into consideration their potentially varying weights, due to the lack of established coding systems in past research. However, the main purpose of this research was to consistently code the total intensity for all boycott messages in order to use the data in subsequent comparative analyses aimed at examining the relationships between boycott motivations and the emotional intensity of these messages. The computation formula used for total message intensity is therefore appropriate for this type of analysis. Computation details are provided in Appendix 1.
ANALYSIS AND FINDINGS
Consumer Boycott Motivations
A tally of the frequencies and percentages for general and specific boycott motives showed that, consistent with past research, instrumental boycott motives are dominant and that they can frequently overlap with noninstrumental ones, confirming the complexity of consumer motivations (John and Klein 2003). Of the 1,422 boycott messages, 50% were driven by instrumental motives, 27% by noninstrumental motives, and 23% by both.
The content analysis of boycott messages also led to the identification of eight specific motives, four within each of the instrumental and noninstrumental motivations. A combination of two specific motives was found in 705 tweets (49.6%). Although past research has suggested motivation categories within instrumental and noninstrumental motivations, most researchers used experimental and survey studies, where these motives were predetermined. Also, past research studies focused on examining self-reported intentions to boycott rather than actual boycott behavior. Although this study does not examine the behavior of refraining from purchasing products and services of boycott targets, it contributes to past research by examining actual consumer engagement in sharing boycott messages in a naturalistic online setting.
Within instrumental boycott motivations, four themes emerged for specific motives: call for action, awareness and information sharing, making a difference, and offering alternatives or substitutes. Of the 1,422 boycott messages, 635 consumers (29.9%) clearly expressed a call for action by other consumers such as in the tweet: "Throw your Nintendos out the window #boycott all #japan owned horror #Shut Tajji Down," calling consumers to boycott Japanese manufacturers for animal rights causes. Consumers driven to boycott by instrumental motivations often have the goal of signaling to others the necessity of participating in the boycott (Klein et al. 2004). This was the most common specific motive in our sample.
Among instrumental-specific motives, awareness and information sharing was another common theme (18.2%). This boycott motive is novel and not highlighted in past research because of the unique sharing aspect of social media, which allows consumers to warn others and increase awareness about issues related to ethical violations or unjust organizational conduct. This motive was often seen in combination with a call for action (22.3% of the time).
Consistent with past research (Braunberger and Buckler 2011), many individuals in our sample were motivated to participate in boycotts because they believe that they can make a difference by forcing the boycott target to change or discontinue their behavior. The tweet "@SeaWorld truth would result in them having to close down #boycott #Vote With Your Dollars" is one of 260 tweets (12.2%) illustrating the specific motive of making a difference. Such boycott messages show consumer beliefs in their power to make a change and the success of their boycott efforts; they are also often present in combination with call for action (10.4%). Finally, on rare occasions, in 22 tweets (1.0%), boycott messages were driven by the specific motive of offering alternatives or substitutes for the products and services of the boycott targets.
In our sample of 1,422 boycott messages, noninstrumental boycott motives included venting and expression of anger or displeasure, the desire to punish, threaten, or warn the boycott target, and self-enhancement. These specific motives are consistent with past research findings about noninstrumental motivations (Braunberger and Buckler 2011; Brennan and Lomasky 1993; Friedman 1999; Klein et al. 2004). Venting and expression of displeasure with the actions of the boycott target was a prominent theme with 439 (20.6%) tweets, supporting the proposition that individuals engage in boycotts because they want to express their anger or outrage. In 10.5% of these tweets, venting was expressed in parallel with the instrumental-specific motive of a call for action.
A punitive theme was detected in 214 (10.1%) tweets, whereby consumers wish to punish the target and hope that it will suffer from negative consequences due to their behavior, as in the following tweet: "@Cracker-Barrel #Boycott Nope, looks like IHOP or Bob Evans next time we go out for a nice breakfast. #1 Stand With Phil." The desire to punish the target often came from consumers who believe that they can make a difference (10.1% of the times). Self-enhancement was a specific motive for 165 (7.8%) boycott messages, where consumers engaged in boycott to have clean hands, clear their conscience from guilt, or show moral superiority. Reminiscence, a new specific noninstrumental motive, was the theme of five tweets (0.2%), where boycott messages were used to express consumer longing for an ideal world or for the past state of things. Table 2 provides a list of specific boycott motives, the frequencies and percentages of their occurrence in our sample, boycott message examples for each, and a list of prior research studies validating similar motives.
Causes and Targets of Boycotts
One contribution of this research is its investigation of causes and targets of boycotts without context constraints. In fact, the Twitter data allowed for looking at boycott messages that cover a wide range of issues instead of limiting the research to a specific issue or boycott target. This contributes to a better understanding of boycott causes that consumers are most sensitive to. When examining the causes for consumer boycotts, immersive background research was used in order to identify themes relating to the ultimate cause of the boycott, by asking: "Ultimately, what is this boycott about? Is it a human rights issue? Is it a political issue? Etc."
Originally, 14 themes emerged for boycott causes that were then grouped into five main categories: (1) political causes including issues such as Obama Care, biased political media, and specific government policies; (2) human rights causes, which included freedom of speech and second amendment rights (e.g., gun control), women's rights, gay rights, discrimination and racism, sweatshops and workers' conditions, and human rights violations due to political conflicts (e.g., casualties of the war in Syria); (3) animal rights and environment protection causes (e.g., animal torture and captivity, pollution, and poisonous chemicals); (4) causes related to business strategy decisions or corporate failures (e.g., price fairness, customer service failure, privacy breaches, and data security); and (5) corruption causes that include stealing accusations, lobbying groups, and questionable fund usage by nonprofits, among other issues.
The results from a tally of frequencies and percentages for each category showed that human rights issues are the leading cause of consumer boycotts (34.7%). Business strategy decisions and corporate failures were the cause of 376 (26.4%) of boycott messages, followed closely by political causes (23.2%). Animal rights and environment protection were the cause for 200 (14.1 %) boycott tweets, and corruption causes were the least frequent with 23 (1.5%) boycott messages. Table 3 shows the causes of boycott messages, with frequencies and percentages for each category.
A similar tally of the results for boycott targets revealed five main categories. The most common boycott targets were for-profit providers of goods and services (35.9%). These included manufacturers, retailers, and wholesalers of goods, service providers such as airlines and restaurants, amusement and recreational services, motion pictures, finance, insurance, and real estate services, and telephone, cable, and Internet services. Media outlets such as radio, television channels and shows, print media, and social media were also major targets for consumer boycotts (28.2%). For 18.7% of the tweets, the target of boycott messages was a geographic region (country, state, or city) rather than an organization. Less common targets included people and sports teams (8.6%). These targets, which can be more personal, included commercial sports such as a national baseball team, sports and entertainment celebrities, and public figures. Other boycott targets that did not fit within any of these categories were found in 8.6% of boycott messages. Examples of these targets include nonprofit organizations, public administration, specific policies, and public transportation. Categories of boycott targets are provided in Table 4, with frequencies and percentages of their occurrence in our sample.
The Relationship between Boycott Motivations and Boycott Message Intensity
After computing total intensity scores for all boycott messages, which ranged between 2 and 38 with an average of M = 16.02, a one-way ANOVA was performed to compare mean message intensities among boycott messages that have instrumental vs. noninstrumental vs. both motivations. The results showed a significant difference in means (F = 172.46, p<.001), with instrumentally motivated boycott messages having a significantly lower average emotional intensity (M = 13.24) than boycott messages with noninstrumental motivation (M = 18.71, p < .001) or with both motivations (M = 18.88, p < .001). Figure 1 illustrates these results.
Additionally, because the total intensity score was based on a linear formula that did not account for the relative weights of different types of message intensifiers, we performed multiple one-way ANOVAs looking at each of the elements in our total intensity score. The results consistently showed that noninstrumental motives or complex motives with both instrumental and noninstrumental components lead to boycott messages with higher intensity than instrumentally motivated ones. In fact, consumers express higher intensity emotions and emoticons (F = 25.59, p<.001), write longer tweets (F = 128.24, p < .001), and use more messages intensifiers such as profanity (F = 36.72, p < .001) and exclamation and question marks (F = 12.92, p < .001), when they have noninstrumental boycott motives or a combination of general motives rather than only instrumental motives.
One advantage of human sentiment analysis over automated algorithms is that it allows for the detection of sarcasm in boycott messages. A chi-square test, examining whether the presence or absence of sarcasm is dependent on the general motivation of boycott messages, revealed significant differences in the use of sarcasm between instrumentally and noninstrumentally motivated boycotts ([chi square] = 110.20, p < .001). The results in Table 5 show that while sarcasm is present in only 8.6% of instrumentally motivated messages, it was found in 34.0% of noninstrumentally motivated boycott messages. For messages with both motives, sarcasm was present in 18.3% of the messages in our sample. Sarcasm, as an intensifier, appears to be more common when consumer boycotts are driven by the need to express displeasure, anger, or outrage, by punitive motives, or by self-enhancement motives.
Are Consumer Motivations to Boycott Affected by Boycott Causes and Targets?
The results in the previous section provide evidence that individuals who engage in boycotts due to noninstrumental motivations with no concrete instrumental goals express more intense boycott messages. Although this does not necessarily indicate actual consumer experiences of more intense emotions, the results show that these individuals share boycott messages with higher emotional intensity, online via Twitter. However, important questions about consumer boycott behaviors remain unanswered. Are boycotts resulting from certain causes more frequently driven by noninstrumental motives, and consequently lead to more emotionally intense expressions of boycott messages, than those resulting from other causes? What types of boycott targets are possibly subject to more intense emotional expressions of boycott messages?
In an attempt to better understand what causes and boycott targets trigger more noninstrumental consumer motivations than others, and potentially produce more intense boycott messages, contingency analyses were performed to examine the relationships between boycott causes and general consumer motivations ([chi square] = 78.47, p < .001), and boycott targets and general consumer motivations ([chi square] = 87.45, p < .001).
The results in Table 6 show that 41.2% of boycott messages, caused by a business strategy decision or corporate failure, are driven by noninstrumental motivations. This is a higher fraction than that of all other causes, including human rights issues, which comes second with 25.2% of boycott messages having noninstrumental boycott motives. When consumers experience corporate failure such as bad customer service or personal information privacy breaches, e.g., in the case of Target's data hacking crisis, the issues leading to boycott behaviors are more personal. They can consequently cause psychological distress, anger, or outrage expressed in noninstrumentally motivated boycott messages, which have high intensity. Although other causes, such as perceptions of human rights or animal rights violations by the target of the boycott, are a common trigger for consumer boycotts, they appear to have a more distant connection to consumers who might not be experiencing their impact directly. Consumers can therefore engage in these boycotts or share boycott messages online, with more goal-directed thought and a lower need for venting or expression of frustration.
When it comes to boycott targets, although for-profit providers of goods and services constitute the most common target (35.9%) of the tweets in this study's sample, the motivations driving these boycotts are noninstrumental only 28.1% of the time. Consumers appear to engage in more intense expressions of boycott messages when the boycott targets are sports teams or people such as celebrities, where 47.2% of boycott messages are noninstrumental. This is also true for media targets such as TV channels or shows, which triggered noninstrumental motivations to boycott 30.9% of the times. Because people can build fandom for these targets or perceive them as an extension of their selves, such as in the case of sports teams, they might experience more psychological needs and be driven by noninstrumental motivations to boycott these targets. More detailed results for the relationship between boycott targets and boycott motivations are provided in Table 7.
SUMMARY AND CONCLUSIONS
Consumer Boycott Behaviors: A Deeper Look
Consumer boycotts "often reflect a concern for the general good" (Yuksel 2013, 205). They can influence public welfare and have detrimental effects on a company's image and its financial performance. Hence, they provide an important tool for consumer empowerment in their fight against organizational practices deemed unethical or unjust. The objectives of this study are twofold.
First, we offer a deeper understanding of boycott behaviors by using content analysis of boycott messages, shared by consumers on Twitter, to investigate the underlying motivations, causes, and targets of these boycotts. Our findings are consistent with past research on the instrumental and noninstrumental classification of boycott motivations (Friedman 1999, 1991, 1985; Klein, Smith, and John 2004, 2002; Yuksel 2013). The analysis of boycott messages on Twitter also reveals a consistent dominance of instrumental motivations of consumer boycotts compared to noninstrumental and mixed motivations (Sen, Gurhan-Canli, and Morwitz 2001). The results in Table 2 show a clear replication of the specific motives underlying consumer boycotts found in prior research. The current study also shares several boycott motivations with the extant studies conducted online (Braunberger and Buckler, 2011; Hoffmann 2011; Kozinets and Handelman, 1998). However, the prior online boycott studies reveal distinctive motives specific to their respective study contexts (e.g., idiosyncratic motives such as proximity and political consumerism by Hoffmann 2011). The present study also finds distinguishing motives not found in the previous online boycott research, such as awareness and information sharing, which might be unique to Twitter being an effective and efficient online sharing medium. An important contribution of this research is the exploration of prominent boycott causes and targets without constraints to a specific context, such as seal hunting (Braunberger and Buckler, 2011) or factory relocation (Hoffmann 2011). This resulted in finding that human rights issues ranging from freedom of speech to women's rights constitute the leading causes of consumer boycotts and that for-profit providers of products and services are the most common boycott targets.
Second, we examine the emotional intensity of boycott messages, using sentiment analysis, to explore the relationship between consumer boycott motivations and message intensity. We contribute to research on consumer emotions expressed in boycott messages and shed light on how different boycott causes and targets are connected to consumer motivations, and how these motivations influence the emotional intensity of boycott messages. The results from this novel research area indicate that boycott messages driven by noninstrumental motives have higher emotional intensity than instrumentally motivated messages, with longer tweets, higher intensity emotions and emoticons, and more frequent message intensifies, including exclamation and question marks, profanity, and sarcasm. Furthermore, when boycotts are caused by business strategy decisions or corporate failures, their underlying consumer motivations are more frequently noninstrumental than instrumental. The resulting boycott messages have higher emotional intensity than boycott messages caused by prominent boycott triggers such as human rights or animal rights violations. For instance, why would the mass murder of children in Syria be less likely to cause consumer expression of outrage and high sentiment intensity in boycott messages, than Target's data breach of consumer information? The answer might lie in the influence of proximity (Hoffmann, 2011, 2013), including geographic distance between English-speaking tweeters and such political events, or personal distance between these events and consumers' lives. Similarly, boycotts targeting sports teams or people such as celebrities and public figures are more frequently noninstrumentally motivated, and hence have higher emotional intensity than boycotts with other targets.
While these insights are not interpretive given the exploratory nature of our research, they shed light on the interactions between boycott causes and targets and consumer boycott motivations. Do consumers advocate for many boycott causes they deem worthy such as human rights issues, while in reality they are more emotionally vested in boycotts triggered by causes that touch their personal lives directly, even when these causes are related to the firing of an actor, such as Duck Dynasty's Phil Robertson, from a television reality show? Our findings reveal many such questions that require further examination in future research.
Boycott as a Tool for Consumer Empowerment
With the proliferation of social media websites such as Twitter, Facebook, online blogs, and so on, consumers and boycott movement organizers have many platforms with massive reach for their boycott messages. In their fight against big corporations and unethical organizational practices, boycotts via social media outlets may offer consumers a powerful tool that could allow them to influence social change, government policies, and corporate decisions. The ability to have such influence empowers consumers to enhance their own well-being, which is usually diminished by the unfair practices of hegemonic organizations.
The current study finds that certain boycott causes trigger consumer noninstrumental boycott motivations, in order to fulfill psychological needs such as venting their frustration and expressing their anger or outrage. The resulting boycott messages are more emotionally intense but lack clear instrumental goals, which could make them less effective in influencing other consumers or corporate actions (John and Klein, 2003). This is particularly true when the boycott cause is a company strategy or corporate failure, such as a bad customer service experience, affecting the consumer personally. Consumers might need to be more strategic with their boycott messages by limiting the influence of negative emotions, when sharing these messages online.
For-profit providers of products and services are the most common boycott targets. They could be targeted for various causes beyond their business strategy decisions or corporate failures, such as human rights issues or animal cruelty. Corporate social responsibility matters to consumers and should be taken seriously, but some causes can bring out stronger consumer reactions than others. When the cause of consumer boycotts is a business strategy decision or a corporate failure rather than political, social, or other issues, consumers commonly engage in boycotts for noninstrumental motives, which leads to boycott messages with higher emotional intensity. This emphasizes the importance of customer satisfaction and failure recovery strategies. Providing consumers with an outlet to complain within the company and having systems in place for adequately handling these complaints can mitigate the adverse effects of consumer boycotts by decreasing the likelihood of negative word of mouth via social media.
What's Next for Research on Boycott Behaviors?
This study is exploratory in nature and has a number of limitations that should be considered when assessing the results and their generalizability. The main limitations are related to the Twitter sample used for this study; specifically, the absence of identifiable personal information, the lack of message effectiveness data, and the data collection timeframe.
Although Twitter offers a platform for gathering a massive amount of consumer opinion data in a naturalistic setting, the anonymous nature of this data prevents researchers from identifying the demographic characteristics of the sample to explore their relationship to boycott behaviors. It also inhibits them from conducting follow-up studies using face-to-face interviews or surveys in order to have a more profound understanding of the actual boycott behaviors of tweet authors and the link between these behaviors and the motives, causes, targets, and emotional intensity of boycott messages on Twitter. How is the emotional intensity of a boycott tweet related to its author's actual boycott behavior? Future research could further investigate the predictive value of boycott messages and whether they reflect actual boycott behavior by (1) using different social media platforms such as consumerist blogs, which offer better potential for approaching bloggers to conduct follow-up surveys or in-depth interviews; or (2) using different research methods such as narrowly focused case-based studies that monitor the effects of boycott messages and their emotional intensity on company stock prices or media viewership over time.
Due to the Twitter API used for this research, the retrieved data set also lacks message effectiveness metrics, e.g., the number of people who saw or read each boycott message or the number of times a message was re-tweeted. Such metrics would allow for studying the link between consumer boycott motivations, boycott message intensity, and message effectiveness in grabbing attention or influencing consumer action. Future research can further explore the relevance and strength of boycott calls by examining the number of times boycott messages are re-tweeted. This would require the use of different Twitter analytics tools and Twitter crawling techniques. Also, measuring the effectiveness of boycott messages in experimental research by manipulating message framing for various motives, causes, and intensity levels constitutes a rich area for future research. The findings of this study suggest the presence of empirical interactions between boycott causes and targets on one hand, and consumer boycott motivations on the other. Future research should investigate the effects of these interactions on outcome measures for the effectiveness of boycott messages.
Additionally, for this study, Twitter data were collected over a one-month period starting December 12, 2013. Although Twitter provides access to boycott messages about a plethora of topics without context constraints, this timeframe is linked to the news headlines and the events that occurred around that time period, which limits the generalizability of our findings, especially for boycott causes and targets. Future research could use a more extended time frame or longitudinal data collected periodically, to examine whether the results are replicable in order to answer questions such as: Do human rights violations trigger boycott behaviors more frequently than corporate failures and strategy decisions? Are big corporations more likely to be the target of boycott movements than other organizations, countries, public figures, and so on?
In this study, consumer boycotts are explored based on the instrumental vs. noninstrumental classification of boycott motivations (John and Klein 2003; Klein, Smith, and John 2004). While this classification is valid, the emerging distinction of macro vs. micro-boycotts should be noted. Micro-boycotts tend to focus on campaigns against products from a single company (Ettenson and Klein 2005; Hoffmann and Muller 2009), whereas macro-boycotts encompass campaigns against foreign products from countries involved in conflicts (Abou Aish et al. 2013; Farah 2014), companies of specific countries, or government policies in a company's country (Abosag 2010). Macro-boycotting is becoming an increasingly important consumer empowerment tool, impacting the economies of boycotted countries (Abou Aish et al. 2013). In fact, Abosag and Farah (2014) have found empirical support for the negative impact of a macro-boycott resulting from religious animosity, on customer loyalty to and brand image of Danish companies. Future research could use this study's mixed methods approach to explore macro vs. micro-boycott behaviors and boycott messages on social media.
One contribution of this study is the consideration of consumer expressions of emotional intensity in boycott messages. Yet, how consumers express intensity in writing might not be an accurate reflection of the actual emotions they experience as a result of a boycott cause. This is another research limitation that should be addressed in future research, perhaps using physiological and neuro-marketing experiments to examine consumer arousal and emotional reactions resulting from being exposed to different scenarios of ethical violations or unfair practices. Can consumers who experience anger or outrage engage in sharing their thoughts through instrumentally motivated boycott messages that for example call for action, without expressing their frustrations?
Research on consumer boycott behaviors has important implications for consumers and businesses. This study takes an important step toward better understanding consumer boycott motivations--given various boycott causes and targets--and how they affect the emotional intensity of boycott messages shared on social media sites, such as Twitter. Our findings can guide future research, especially linking consumer emotional responses to boycott causes, consumer sharing of boycott messages and their effectiveness, and actual consumer behaviors of boycotting the target's products and services.
APPENDIX 1 Computation of Total Boycott Message Intensity Number of Message Element Coding Points Assigned Emotions and emoticons High intensity 3 Moderate intensity 2 Low intensity 1 Length of Twitter Total number of Total number of words message words excluding web links Capital letter words Percentage of Number of capital capital letter letter words/total words number of words Expressive punctuation Number of Number of ! and ? exclamation and question marks * Profanity and insults Number of profanity Number of profanity and insult words and insult words Sarcasm Presence or absence 0 for no sarcasm 1 for of sarcasm sarcasm Total message intensity Sum of all points * Question marks used appropriately at the end of a question were not counted.
Abosag, Ibrahim. 2010. Dancing with Macro-boycotters: The Case of Aria Foods. Marketing Intelligence & Planning, 28 (3): 356-373.
Abosag, Ibrahim and MayaF. Farah. 2014. The Influence of Religiously Motivated Consumer Boycotts on Brand Image, Loyalty and Product Judgment. European Journal of Marketing, 48 (11/12): 2262-2283.
Abou Aish, Ehab, Sally McKechine, Ibrahim Abosag, and Salah Hassan. 2013. The Mystique of Macro-boycotting Behaviour: A Conceptual Framework. International Journal of Consumer Studies, 37: 165-171.
American Psychological Association. 2007. Dictionary of Psychology. Washington, DC: American Psychological Association.
Baggozzi, Richard P., Mahesh Gopinath, and Prashanth U. Nyer. 1999. The Role of Emotions in Marketing. Journal of the Academy of Marketing Science, 21 (2): 184-206.
Bougie, Roger, Rik Pieters, and Marcel Zeelenberg. 2003. Angry Customers Don't Come Back, They Get Back: The Experience and Behavioral Implications of Anger and Dissatisfaction in Services. Journal of the Academy of Marketing Science, 31 (4): 377-393.
Brinkmann, Johannes. 2004. Looking at Consumer Behavior in a Moral Perspective. Journal of Business Ethics, 51: 129-141.
Braunberger, Karin and Brian Buckler. 2011. What Motivates Consumers to Participate in Boycotts: Lessons from the Ongoing Canadian Seafood Boycott. Journal of Business Research, 64 (1): 96-102.
Brennan, Geoffrey and Loren Lomasky. 1993. Democracy and Decision: The Pure Theory of Electoral Preference. Cambridge: Cambridge University Press.
Carrillo-de-Albomoz, Jorge and Laura Plaza. 2013. An Emotion-Based Model of Negation, Intensifiers, and Modality for Polarity and Intensity Classification. Journal of the American Society for Information Science and Technology, 64 (8): 1618-1633.
Chatzidakis, Andrea and Michael S.W. Lee. 2013. Anti-Consumption as the Study of Reasons Against. Journal of Macromarketing, 33 (3): 190-203.
Davidson, Wallace N., Dan Worrell, and Abuzar El-Jelly. 1995. Influencing Managers to Change Unpopular Corporate Behavior through Boycotts and Divestitures. Business & Society, 34 (2): 171-196.
Ettenson, Richard and Jill Gabrielle Klein. 2005. The Fallout from French Nuclear Testing in the South Pacific: A Longitudinal Study of Consumer Boycotts. International Marketing Review, 22 (2): 199-224.
Farah, Maya F. 2014. An Expectancy-Value Approach to the Study of Beliefs Underlying Consumer Boycott Intention. International Journal of Business and Management, 9 (10): 101-115.
Farah, Maya F. and Andrew J. Newman. 2010. Exploring Consumer Boycott Intelligence Using a Socio-Cognitive Approach. Journal of Business Research, 63 (4): 347-355.
Friedman, Monroe. 1985. Consumer Boycotts in the United States, 1970-1980: Contemporary Events in Historical Perspective. Journal of Consumer Affairs, 19 (1): 96-117.
--. 1991. Consumer Boycotts: A Conceptual Framework and Research Agenda. Journal of Social Issues, 47(1): 149-168.
--. 1999. Consumer Boycotts: Effecting Change through the Marketplace and the Media. New York: Routledge.
Feldman, Ronen. 2013. Techniques and Applications for Sentiment Analysis. Communications of the ACM, 56 (4): 82-89.
Garrett, Dennis E. 1987. The Effectiveness of Marketing Policy Boycotts: Environmental Opposition to Marketing. Journal of Marketing, 51 (April): 48-57.
Glazer, Amihai, Vesa Kanniainen, and Panu Poutvaara. 2010. Firm's Ethics, Consumer Boycotts, and Signaling. European Journal of Political Economy, 26: 340-350.
Hoffmann, Stefan. 2011. Anti-Consumption as a Means to Save Jobs. European Journal of Marketing, 45(11/12): 1702-1714.
--. 2013. Are Boycott Motives Rationalizations? Journal of Consumer Behavior, 12 (3): 214-222.
Hoffmann, Stefan and Stefan Muller. 2009. Consumer Boycotts Due to Factory Relocation. Journal of Business Research, 62 (2): 239-247.
James, Victoria. 2010. A Socio-Cutural Approach to Exploring Consumer Boycott Intelligence: A Commentary Essay. Journal of Business Research, 63 (4): 363-365.
John, Andrew and Jill Klein. 2003. The Boycott Puzzle: Consumer Motivations for Purchase Sacrifice. Management Science, 49 (9): 1196-1209.
Klein, Jill G., Richard Ettenson, and Marlene D. Morris. 1998. The Animosity Model of Foreign Product Purchase: An Empirical Test in the People's Republic of China. Journal of Marketing, 62(1): 89-100.
Klein, Jill G., Craig Smith, and Andrew John. 2002. Exploring Motivations for Participation in a Consumer Boycott. Advances in Consumer Research, 29 (1): 363-369.
--. 2004. Why We Boycott: Consumer Motivations for Boycott Participation. Journal of Marketing, 68 (July): 92-109.
Koku, Paul Sergius. 2012. On the Effectiveness of Consumer Boycotts Organized through the Internet: The Market Model. Journal of Services Marketing, 26 (1): 20-26.
Koku, Paul Sergius, Aigbe Akhigbe, and Thomas M. Springer. 1997. The Financial Impact of Boycotts and Threats of Boycott. Journal of Business Research, 40 (1): 15-20.
Kozinets, Robert V. 1997. T Want To Believe': A Netnography of The X-Philes' Subculture of Consumption. Advances in Consumer Research, 24 (1): 470-475.
Kozinets, Robert V. and Jay Handelman. 1998. Ensouling Consumption: A Netnographic Exploration of the Meaning of Boycotting Behavior. Advances in Consumer Research, 25 (1): 475-480.
Kumar, Akshi and Teeja Mary Sebastian. 2012. Sentiment Analysis on Twitter. International Journal of Computer Science, 9 (4): 372-378.
Lai, Meng-Kuan and Bayu A. Aritejo. 2010. Anti-Consumption and Consumer Resistance: A Conceptual Overview. Paper presented at the ICAR/NCARE Symposium, Marseilles, June 25-26.
Lindenmeier, Jorg, Christoph Schleer, and Denise Price. 2012. Consumer Outrage: Emotional Reactions to Unethical Corporate Behavior. Journal of Business Research, 65 (9): 1364-1373.
Mays, Nicholas and Catherine Pope. 1995. Rigour and Qualitative Research. British Medical Journal, 311 (July): 109-112.
Merriam, Sharan B. 2009. Qualitative Research: A Guide to Design and Implementation. San Francisco, CA: Jossey-Bass.
Morwitz, Vicky and Sankar Sen. 2004. The Effects of Pro-Boycott and Anti-Boycott Communications on the Individual Boycott Decision. Advances in Consumer Research, 31: 331.
Norris, Nigel. 1997. Error, Bias and Validity in Qualitative Research. Educational Action Research, 5 (1): 172-176.
Pang, Bo and Lillian Lee. 2008. Opinion Mining and Sentiment Analysis. Foundation and Trends in Information Retrieval, 2 (1-2): 1-135.
Park, Alexander and Patrick Paroubek. 2010. Twitter as a Corpus for Sentiment Analysis and Opinion Mining. Proceedings of the Seventh Conference on International Language Resources and Evaluation: 1320-1326. Valletta, Malta: European Language Resources Association.
Patton, Michael Quinn. 2002. Qualitative Research and Evaluation Methods. 3rd edition. Thousand Oaks, CA: Sage.
Pettit. Annie. 2013. Identifying the Real Differences of Opinion in Social Media Sentiment. International Journal of Market Research, 55 (6): 757-767.
Podoshen, Jeffrey S. 2012. Dark Tourism Motivations: Simulation, Emotional Contagion, and Topography Comparison. Tourism Management, 35 (April): 263-271.
Pruitt, Stephen W. and Monroe Friedman. 1986. Determining the Effectiveness of Consumer Boycotts: A Stock Price Analysis of Their Impact on Corporate Targets. Journal of Consumer Policy, 9 (4): 375-388.
Rambocas, Meena and Jono Gama. 2013. Marketing Research: The Role of Sentiment Analysis. Working Paper 489, School of Economics and Management, University of Porto.
Savage, Neil. 2011. Twitter as Medium and Message. Communications of the ACM, 54 (3): 18-20.
Sen. Sankar, Zeynep Giirhan-Canli, and Vicki Morwitz. 2001. Withholding Consumption: A Social Dilemma Perspective on Consumer Boycotts. Journal of Consumer Research, 28 (December): 399-417.
Shaw, Deidre, Terry Newholm, and Roger Dickinson. 2006. Consumption as Voting: An Exploration of Consumer Empowerment. European Journal of Marketing, 40 (9/10): 1049-1067.
Shoefer, Klaus and Christine Ennew. 2005. The Impact of Perceived Justice on Consumers' Emotional Responses to Service Complaint Experiences. Journal of Services Marketing, 19 (5): 261-270.
Singh, Jagdip. 1988. Consumer Complaint Intentions and Behavior: Definitional and Taxonomical Issues. Journal of Marketing, 52 (January): 93-107.
Sirgy, Joseph M., Dong-Jin Lee, and Don Rahtz. 2007. Research on Consumer Well-Being (CWB): Overview of the Field and Introduction to the Special Issue. Journal of Macromarketing, 27 (4): 341-349.
Smith, N. Craig. 1990. Morality and the Market: Consumer Pressure for Corporate Accountability. London: Routledge.
Strauss, Gregory P. and Daniel N. Allen. 2008. Emotional Intensity and Categorization Ratings for Emotional and Nonemotional Words. Cognition and Emotion, 22 (1): 114-133.
Strauss, Anselm and Juliet Corbin. 1998. Basics of Qualitative Research: Techniques and Procedures for Developing Grounded Theory. Thousand Oaks, CA: Sage.
Yuksel, Ulku. 2013. Non-participation in Anti-Consumption: Consumer Reluctance to Boycott. Journal of Macromarketing, 33 (3): 204-216.
Yuksel, Ulku and Victoria Mryteza. 2009. An Evaluation of Strategic Responses to Consumer Boycotts. Journal of Business Research, 62 (2): 248-259.
Suzanne C. Makarem (firstname.lastname@example.org) is an Assistant Professor of Marketing and Haeran Jae (email@example.com) is an Associate Professor of Marketing, both at Virginia Commonwealth University.
TABLE 1 Intensity of Emotions and Emoticons Expressed in Boycott Messages Example Emotions from Example Emoticons from Intensity Boycott Messages Boycott Messages High intensity Love, hate, anger, Angry face X-(, flip outrage, rage, pissed off, heart <3 off, sick, contempt, make you cry Horror, terror, disgrace, egregious Moderate Suck, dislike, can't Slap in face, pouting intensity stand him, disgust, face, smiley face eek, disturbing, [happy], thumbs down, shocked, insulted, upset face [fraun], offended shocked face, praise Shame, unhappy, sad, hands, face with look upset, proud, fear of triumph, broken Cruel, awful, nasty, heart </3 heartless Low intensity Disappoint, annoyed, Displeased face, shaking disrespect, doomed, my head bad, uncaring, intolerant, insensitive Unpleasant, tired, glad, hope TABLE 2 Specific Motives of Boycott Messages (a) Specific Boycott Motive Prior Research Replications Instrumental motivations Call for Braunberger and Buckler (2011) action Friedman (1991) Hoffmann (2011) Awareness/ Novel finding information sharing Make a Braunberger and Buckler (2011) difference Ettenson and Klein (2005) Friedman (1991, 1999) Hoffmann (2011, 2013) John and Klein (2003) Klein, Smith, and John (2004) Sen, Giirhan-Canli, and Morwitz (2001) Offer Braunberger and Buckler (2011) alternatives/ Hoffmann (2011) substitutes Morwitz and Sen (2004) Noninstrumental motivations Venting/ Braunberger and Buckler (2011) expression of Ettenson and Klein (2005) displeasure, Friedman (1991, 1999) anger, or Hoffmann and Muller (2009) outrage Hoffmann (2011) John and Klein (2003) Lindenmeier, Schleer, and Price (2012) Punish/threat/ Abosag and Farah (2014) warn the Braunberger and Buckler (2011) boycott target Ettenson and Klein (2005) Hoffmann and Muller (2009) Hoffmann (2011) John and Klein (2003) Klein, Ettenson, and Morris (1998) Self- Braunberger and Buckler (2011) enhancement Brinkmann (2004) Glazer, Kanniainen, and Poutvaara (2010) Hoffmann (2013) John and Klein (2003) Klein, Smith, and John (2002) Kozinets and Handelman (1998) Smith (1990) Reminisce Novel finding (ideal world or past) Specific Example Boycott Messages Boycott Motive (Sample Tweets) Frequency (%) Instrumental motivations Call for 1. #BOYCOTT #Israeli academic 635 (29.85) action institutions! Join us in PROTESTING their actions! 2. @Nintendo Throw your Nintendos out the window #boycott all #japan owned horror Awareness/ 1. African American & latino 387 (18.19) information neighborhoods are seldom sharing promoted by the San Francisco Travel Association #boycott 2. I see Phase 8 are still making Angora products available-- #boycott! Make a 1. @SeaWorld Truth would result in 260 (12.22) difference them having to close down. #Boycott #Vote With Your Dollars 2. @AETV #boycott A&E, Bring Back Phil! Offer 1. @Sochi2014 @US01ympic #LGBT for 22 (1.03) alternatives/ those who don't want 2 totally substitutes #boycott the #Olympics, #instead boycott the #medal #ceremony 2. #Boycott @blogtalkradio and everyone go to @spreaker Noninstrumental motivations Venting/ 1. @rugbyworldcup Checking out ticket 439 (20.64) expression of prices. R U having a laugh??? displeasure, Absolute rip off--and held at anger, or soccer grounds. Shocker #boycott outrage 2. @hm useless delivery times, STILL NO PARCEL, again! #When Will I Learn #boycott ffhm #Awful Customer Service Punish/threat/ 1. @Target officially on #boycott 214 (10.06) warn the clean up your fucking act boycott target 2. @CrackerBarrel #Boycott Nope, looks like IHOP or Bob Evans next time we go out for a nice breakfast. #I Stand With Phil Self- 1. I refuse to turn @NBCNews on in my 165 (7.76) enhancement house on Sundays, not worth my time to watch lies and BS. #boycott Sunday 2. @SeaWorld #Boycott!! No animal should be used for entertainment purposes #Free Them #Greedy Bastards Reminisce I can't believe I now need to pay for 5 (0.24) (ideal world the full library ... it used to be or past) included in the original price. What a joke!! #boycott (a) The total frequency for specific motives is 2,127, which is higher than the sample size of n= 1,422 tweets because many boycott messages (705) had more than one specific motive. TABLE 3 Causes of Boycott Messages Frequency Cause of Boycott Message n = 1,422 % Human rights issues 493 34.67 Examples: * Women's rights issues (e.g., TV show with offensive content such as Rush Limbaugh) * Freedom of speech (e.g., A&E firing Phil Robertson for stating his anti-gay opinions) * Human rights in political conflicts (e.g., Casualties of war in Syria) 376 26.44 Business strategy decisions and corporate failure Examples: * Target data breach * Bad customer service (e.g., US Airways, H&M, USPS) * Unfair pricing (e.g., Starbucks, Rugby World Cup ticket prices) Political issues 330 23.20 Examples: * Obama Care * Biased media on issues of left- vs. right- wing politics (e.g., MSNBC, Fox News) * Academic boycott of Israel for mistreatment of Palestinians Animal rights and environment protection issues 200 14.06 Examples: * Animal captivity in Sea World * Use of Angora rabbit wool by companies (e.g., Gap) * Boycott of China for causing pollution and "killing the planet" Corruption-related issues 23 1.53 Examples: * Boycott of WePay, for "stealing" fundraising money * Boycott of Koch Brothers for corrupt politics * Boycott Boston Children's hospital for forcibly removing children from their parents' houses TABLE 4 Targets of Boycott Messages Frequency Main Target Category n = 1,422 % For-profit providers of goods and services: 511 35.90 Manufacturing/retailAvholesale General services: airlines, restaurants, automotive ... Amusement, recreational, and art services Motion pictures and theater services Finance, insurance, and real estate services Telephone, cable, and internet services Multiple businesses Media: 401 28.20 Radio and television broadcasting Print media: newspapers and magazines Social media and search engines (ex: Facebook, Google) Geographic region: country, state, city 266 18.70 People and sports teams: 122 8.60 Commercial sports and sports teams Sports and entertainment celebrities Public figures Other: 122 8.60 Nonprofit organization Public administration, government, policy General products category with no specific target Flealth, education, and social services Utilities and public transportation Other TABLE 5 General Motivation of Boycott Messages and the Use of Sarcasm: Chi-Square Test (a) Absence of Sarcasm Presence of Sarcasm (Frequency and (Frequency and General Motivation Percentage) Percentage) Instrumental motivation 649 61 91.40% 8.60% Noninstrumental motivation 250 129 66.00% 34.00% Both motivations 272 61 81.70% 18.30% Total 1171 251 82.30% 17.70% (a) [chi square] = 110.20, p < .001; N = 1,422. TABLE 6 Boycott Causes and General Motivation of Boycott Messages: Chi-Square Test (a) Frequencies and Percentages Instrumental Noninstrumental Both Boycott Cause Motivation Motivation Motivations Human rights issues 246 124 123 49.90% 25.20% 24.90% Business strategy 136 155 85 decisions and 36.20% 41.20% 22.60% corporate failure Political issues 194 72 64 58.80% 21.80% 19.40% Animal rights and 121 25 54 environment 60.50% 12.50% 27.00% protection issues Corruption-related 13 3 7 issues 56.50% 13.00% 30.40% Total 710 379 333 49.90% 26.70% 23.40% (a) [chi square] = 78.47, p < .001; N = 1,422. TABLE 7 Boycott Targets and General Motivation of Boycott Messages: Chi-Square Test (a) Frequencies and Percentages Instrumental Noninstrumental Both Boycott Target Motivation Motivation Motivations For-profit providers of 243 143 123 goods and services 47.70% 28.10% 24.20% Media 168 124 109 41.90% 30.90% 27.20% Geographic region 174 32 60 65.40% 12.00% 22.60% People and sports teams 42 58 23 34.10% 47.20% 18.70% Other 83 22 18 67.50% 17.90% 14.60% Total 710 379 333 49.90% 26.70% 23.40% (a) [chi square] = 87.45, p < .001; N = 1,422. FIGURE 1 Boycott Message Intensity by General Boycott Motivation: One-Way ANOVA; F = 172.46, p < .001 Instrumental 13.24 Non-Instrumental 18.71 Both Motives 18.88 Note: Table made from line graph.
|Printer friendly Cite/link Email Feedback|
|Author:||Makarem, Suzanne C.; Jae, Haeran|
|Publication:||Journal of Consumer Affairs|
|Date:||Mar 22, 2016|
|Previous Article:||Brand iconicity vs. anti-consumption well-being concerns: the Nutella palm oil conflict.|
|Next Article:||Rhetorical analysis of resistance to environmentalism as enactment of morality play between social and ecological well-being.|