RECYLE NOW!
AN EXPERIMENTAL STUDY ON FEAR APPEAL AND
ASSERTIVE LANGUAGE IN GREEN ADVERTISING
Master thesis
MSc in Business Administration, Marketing
Vicky Los
Student number: 10409963 Thesis supervisor: Marlene Vock
University of Amsterdam, Amsterdam Business School 22 June 2018, final version
STATEMENT OF ORIGINALITY
This document is written by student Vicky Los who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
TABLE OF CONTENTS
ABSTRACT ______________________________________________________________________ 1 1. INTRODUCTION _______________________________________________________________ 2 2. LITERATURE REVIEW _________________________________________________________ 5 2.1 Green advertising _____________________________________________________________ 5 2.2 Assertive language and Reactance Theory _________________________________________ 7 2.2 Language Expectancy Theory and fit _____________________________________________ 9 2.3 Fear appeal _________________________________________________________________ 10 2.4 Hypotheses development ______________________________________________________ 12 3. METHOD ____________________________________________________________________ 14 3.1 Experimental stimuli _________________________________________________________ 15 3.2 Pre-test fear appeal and assertive language manipulations ____________________________ 17 3.3 Procedure __________________________________________________________________ 19 3.4 Measurements ______________________________________________________________ 19 4. RESULTS ____________________________________________________________________ 22 4.1 Respondents and data preparation _______________________________________________ 22 4.2 Factor analysis ______________________________________________________________ 22 4.3 Reliability analysis and manipulation check _______________________________________ 23 4.4 Correlation analysis __________________________________________________________ 24 4.5 Normality checks ____________________________________________________________ 25 4.6 Fear now and fear future conditions _____________________________________________ 25 4.7 Hypothesis testing ___________________________________________________________ 27 4.8 Additional findings __________________________________________________________ 32 5. CONCLUSION AND DISCUSSION _______________________________________________ 34 5.1 Theoretical implications ______________________________________________________ 36 5.2 Managerial implications ______________________________________________________ 37 5.3 Limitations and future research directions ________________________________________ 38 REFERENCES __________________________________________________________________ 40 APPENDICES ___________________________________________________________________ 46 Appendix A – Advertisement World Wildlife Fund, 2008 _______________________________ 46 Appendix B – Manipulation materials fear ___________________________________________ 47 Appendix C – Manipulation materials assertiveness ____________________________________ 53 Appendix D – Pre-test results _____________________________________________________ 54 Appendix E – Final survey _______________________________________________________ 56
Appendix F – Demographics ______________________________________________________ 64 Appendix G – Reliability analyses _________________________________________________ 65 Appendix H – Manipulation check _________________________________________________ 69 Appendix I – Differences fear now and fear future conditions ____________________________ 69 Appendix J – Planned contrasts hypothesis 1 and 2 ____________________________________ 71 Appendix K – Output PROCESS model 8 ___________________________________________ 73 Appendix L – Additional findings: perceived issue importance ___________________________ 77 Appendix M – Additional findings: language fit _______________________________________ 79 Appendix N – Additional findings: attitudes and behavioral intentions _____________________ 83
LIST OF FIGURES AND TABLES
Figure 1. Conceptual model _________________________________________________________ 14 Figure 2. Hypothesized relationships __________________________________________________ 15 Figure 3. Plotted effects on attitudes towards recycling ___________________________________ 29 Figure 4. Plotted effects on behavioral intentions to recycle ________________________________ 29
Table 1. Number of respondents in each condition, pre-test ________________________________ 17 Table 2. Items of all measurement scales ______________________________________________ 21 Table 3. Number of respondents in each condition, 2 x 3 design ____________________________ 22 Table 4. Exploratory factor analysis __________________________________________________ 23 Table 5. Means, standard deviations, Pearson’s correlations, and Cronbach’s alpha’s ____________ 26 Table 6. Number of reaspondents in each condition, 2 x 2 design ___________________________ 27 Table 7. Means for attitudes towards recycling for all four conditions ________________________ 28 Table 8. Means for behavioral intentions to recycle for all four conditions ____________________ 28 Table 9. Test of between-subjects effects for attitudes towards recycling______________________ 28 Table 10. Test of between-subject effects for behavioral intentions to recycle __________________ 29 Table 11. Mediation analysis for attitudes towards recycling _______________________________ 34 Table 12. Mediation analysis for behavioral intentions to recycle ___________________________ 34 Table 13. Means for perceived issue importance for all four conditions _______________________ 33 Table 14. Means for language fit for all four conditions ___________________________________ 33
ABSTRACT
Past research has repeatedly shown that assertive language reduces consumer compliance. Still, non-governmental organizations continue to use such language in green advertisements. This study examines a moderator that is frequently used in practice, and that might turn around the common negative reaction to assertive language: fear appeal. Building on reactance theory and language expectancy theory, the hypothesized interaction between assertive language and fear appeal was predicted to be mediated by fit between expected language and actual language use. Specifically, this model was analyzed in the context of plastic pollution and plastic recycling. In an online experiment, all respondents were shown an advertisement that randomly mixed fearful or neutral information with assertive or non-assertive language. Effects of the advertisement on respondents’ attitudes towards recycling and behavioral intentions to recycle were measured. The results indicate that fear appeal does not significantly moderate assertive language, and that there is no mediation of language fit. Instead, a direct effect of fear appeal on attitudes towards recycling is found. Moreover, the results repeatedly show a gap between attitudes and behavior. This study adds new theoretical insights to the still limited research field of assertive language and fear appeals in green advertising. From a practical point of view, non-governmental organizations could use this study to investigate how to effectively design green advertisements.
1. INTRODUCTION
‘Shit disgusting ad !! there were a lot of executions to do that in a different way...’ - ‘Yeah? Like something pretty? That would be beside the point. It IS disgusting what happens to Earth’,
‘Baaahahahahaah!’, ‘This ad is cool’, ‘scary!!’.
(Retrieved from www.adsoftheworld.com/media/print/wwf_fish_0)
Online, people leave mixed reactions to the advertisement from the World Wildife Fund (WWF) (appendix A). A man in a blue shirt is portrayed in front of a dark background. But it does not take much time to see that there is something strange: his head has changed into the head of a fish. The big, slimy eyes are desperately looking up to find some sort of help from above. ‘STOP CLIMATE CHANGE BEFORE IT CHANGES YOU’, states the text that accompanies the uncomfortable image. Hypothesis? If we do not stop climate change, sea levels will rise so high that evolution will turn us humans into fish. ‘Ugh! And, bad science’, is posted below (Retrieved from www.adsoftheworld.com/media/print/wwf fish_0).
Quickly scanning this WWF advertisement, two components stand out. First of all, the WWF uses assertive language. Assertiveness is when ‘confident and forceful behaviour’ (Oxford Dictionary of English, 2005) is shown. Such language creates, as Zemack-Rugar, Moore, and Fitzsimons (2017, p. 1) put sharply, ‘the impression that refusal is not an option.’ Studies on the use of assertive language in green advertising are very limited, which is striking since this linguistic tool is used a lot in practice (Baek, Yoon, & Kim, 2015; Kim, Baek, Yoon, Oh, & Choi, 2017; Kronrod, Grinstein, & Wathieu, 2012a). A quick scan of the website of Greenpeace or the WWF clearly illustrates this: ‘Sign the petition for plastic-free supermarkets’, ‘Take action to stop ivory trade’, ‘Join us!’ (Retrieved from www.greenpeace.org.uk; www.wwf.org.uk). Kronrod and colleagues (2012a), being among the few who have researched this phenomenon in green advertising, examined slogans promoting pro-environmental behavior versus slogans promoting consumer goods. The researchers found that of the pro-environmental slogans 57% were assertive, while only 17% of the consumer goods slogans used this kind of language (Kronrod et al., 2012a).
The fact that the pro-environmental industry uses assertive language in advertising so extensively is remarkable to say the least: past research on assertiveness in other domains has created an ‘overwhelming’ (Kronrod et al., 2012a p. 95) amount of evidence that assertiveness works in a counterpersuasive way. Using words such as ‘do’ or ‘must’ that leave no room to refuse a request would threaten peoples’ longing for freedom, and therefore they would not perform the requested behavior (Baek et al., 2015). Instead, non-assertive requests using words such as ‘please’ create more freedom and acknowledge that there might be obstacles that restrain people from complying to the request. This increases the likelihood that the recipient will perform the requested behavior (Kronrod, Grinstein, & Wathieu, 2012b).
Kronrod and colleagues (2012a) find that the negative reaction to an assertive request is reversed when the issue at hand is perceived as important. This interaction between assertiveness and perceived issue importance means that when recipients perceive the assertive request to be fit to the importance of the issue at hand, the probability of compliance to the request is enhanced. It explains why environmental organisations like the WWF often use an assertive tone: they attempt to push the already interested consumers over the edge to take action. To reach consumers who are less concerned about environmental issues, the authors advice environmental agencies to either reduce their assertive tone or elevate the perceived importance of the issue (Kronrod et al., 2012a).
A second component that stands out in the advertisement is that the WWF attempts to persuade recipients to perform a certain behavior – stop climate change – with the use of fear arousal. Fear appeals are persuasive messages that aim to scare recipients into compliance, threatening them with potential negative consequences of noncompliance for their own well-being or that of others (Brooker, 1981; Chen, 2016; Witte, 1992). With public health campaigns having used fear appeal extensively for a long time, for example in anti-smoking advertisements, environmental non-profit organisations increasingly use it to persuade the public (Hartmann, Apaolaza, Barrutia, & Echebarria, 2014; LaTour & Tanner, 2006; Witte & Allen, 2000). Since the publication of the first studies on this persuasive technique in the 1950s, the effectiveness of fear appeals has been researched in many different contexts, mostly in the realm of healthcare. Varying degrees of success on outcomes such as attitude, behavior, behavioral intention or purchase intention have been revealed. Some have concluded fear appeals to be very effective, others have found weak persuasive effects (Brennan & Binney, 2009; Shin, Ki, & Griffin, 2017).
The findings from Kronrod et al. (2012a) on assertive language raise several questions. Could it be that fear can serve the same function as perceived issue importance? When people are scared and concerned about the future of their planet, would they find the assertive tone to be appropriate, maybe even necessary? The current study combines fear appeal and assertive language in one advertisement to explore how both persuasive strategies affect each other. Specifically, this relationship is explored in the context of plastic pollution, with an advertisement from an non-governmental organization (NGO) that aims to stimulate plastic recycling. Using an experiment, this study attempts to answer the following research question:
Does fear appeal in green advertising moderate the effect of assertive language on consumers’ attitudes towards recycling and behavioral intentions to recycle
?
To answer this question, the following sub-questions have been formulated: - What is green advertising?
- How has green advertising developed in academic research? - What is assertive language?
- How do consumers respond to assertive language in advertising, and what theories can explain their responses?
- What is fear appeal?
- How do consumers respond to fear appeals in advertising, and what theories can explain their responses?
Research gap and contributions
The present study addresses several gaps in literature. First, several authors have noticed how academic research on fear appeals is still largely limited to the field of health care (Hartmann et al., 2014; LaTour & Tanner, 2006; Witte & Allen, 2000). Studies on the use of fear appeal in green advertising have been minimal, while it could be a useful tool in that domain (Shin et al., 2017). Therefore simply extending knowledge on fear appeal to this emerging research area is the first theoretical contribution.
Second, the varying success rates of fear appeal effects in research has led to criticism on fear appeal scholars for their inability to predict when fear appeals will or will not work (Shehyar & Hunt, 2005; Witte & Allen, 2000). These prediction difficulties imply that there are many factors that can interact with fear appeals in advertisements to make them effective or not. Some of these interacting variables have been studied in the past, in various research domains. Relatively often, fear appeal has been paired with gain and loss frames (Bilandzic, Kalch, & Soentgen, 2017; Gerend & Maner, 2011; Ruiter, Abraham, & Kok, 2001), but it has also been investigated in interaction with perceived coping efficacy (Hartmann et al., 2014), source of advertisement (for-profit versus non-profit organisations) and consumer involvement (Shin et al., 2017), and together with hope appeal in interaction with global versus local issues (Lee, Chang, & Chen, 2017). However, to date no research in any domain has been done to investigate the possible interaction of fear appeal and assertive language.
Third, we look at studies that focus on assertive requests in advertising. What research on fear appeal and assertiveness in green advertising have in common is that they both are young and developing research fields; assertiveness studies have only begun to emerge in environmental studies since this decennium as well. Assertive language in green advertising has been investigated in interaction with effort investment in completing an environmentally friendly request (Baek et al., 2015) and in interaction with nationality (Americans versus South Koreans) (Kim et al., 2017). Whereas the interaction with perceived issue importance investigated by Kronrod and colleagues (2012a) gave the impetus for this research, perceived issue importance is a fundamentally different construct than fear.
While assertive language and fear appeal both are increasingly studied in environmental research, they have never been studied in the same research. This research involves an experimental manipulation of language expectations that has not been excecuted before, and therefore adds to knowledge on the effect of assertive language (Bullock, Green, & Ha, 2010; Kronrod et al., 2012b). Next to its theoretical contributions, this study is also relevant in practice. NGO’s such as the WWF can
benefit greatly by attaining knowledge on how to use the power of advertising best to stimulate behavior. While creating a shocking, scary advertisement with an assertive request like the WWF did may create buzz and evoke reaction, it does not mean that it is successful in stimulating the behavior that the organization asks for.
The remainder of this thesis is structured as follows. First, important literature on pro-environmental advertising, assertive language, fear appeal, and related constructs is reviewed. Based on this theoretical framework the hypotheses are formulated. What follows is a discussion of the method, explaining the research process and collection and analysis of the data. Third, the hypotheses are tested in an experiment. Subsequently the results and conclusions are discussed, closing off with a discussion of managerial implications and recommendations for future research directions.
2. LITERATURE REVIEW
This chapter provides an extensive overview of literature on assertive language and fear appeal. First, the context of this study is evaluated, looking into past research on green marketing communication. Second, assertive language is discussed using two closely related theories: reactance theory and language expectancy theory. Both theories have repeatedly demonstrated their robustness in previous research. The effect of assertive language on the persuasive power of advertisements will be examined, looking into variables that can alter this effect. Third, academic research on the use of emotional, specifically fear, appeals is reviewed.
2.1 Green advertising
In the early- to mid-1990s, green advertising started to flourish. ‘Undeniably’, as Banerjee, Gulas, and Iyer put in 1995, ‘marketers have discovered a trend and are attempting to exploit it’. Market numbers support this statement, showing that from 1986 to 1991, the share of new green product introductions among the total of new product introductions grew from 1.1% to 13.4%. Not surprisingly, advertising followed: between 1989 and 1990 green advertisements grew by 430% (print ads) and 367% (television ads) (Ottman, 1993). Together with this big increase, research on green advertising grew. Academic research on this phenomenon in this period is mostly descriptive, discussing criteria for when an advertisement can be called ‘green’ (Fowler & Close, 2012; Hartmann & Apaolaza-Ibáñez, 2009). Different theories on classifications and dimensions of green and non-green advertisements have been introduced, illustrating that the concept of green advertising is not simple or unambiguous, but instead dynamic and multidimensional (Hartmann & Apaolaza-Ibáñez, 2009). Banerjee and colleagues (1995, p. 22) for example define green advertisements as any advertisement that meets one or more of the following three criteria:
1. Explicitly or implicitly addresses the relationship between a product/service and the bio physical environment.
2. Promotes a green lifestyle with or without high lighting a product/service. 3. Presents a corporate image of environmental responsibility.
After the short outburst of green advertising at the end of last century, it decreased for almost a decade (Hartmann & Apaolaza-Ibáñez, 2009). When the trend was picked up where it left off, academic researchers introduced more simplified definitions of green advertising. Hartmann and Apaolaza-Ibáñez (2009, p. 717) understand green advertising as ‘advertising a product or brand with environmental claims’. The current study will use the more recent definition of Fowler and Close (2012, p. 121), finding a balance between the elaborate and simple interpretations. Green advertising here is defined as ‘any advertising that explicitly or implicitly promotes an awareness of environmental issues and/or suggests behaviors useful in minimizing or correcting these environmental issues. Green advertising may be associated with either commercial for-profit enterprises or not-for-profit initiatives’.
Research on green advertising can be divided into two categories. The first category investigates real-life green advertising practices. Through systematic analysis of content and form, studies in this category seek to uncover trends and develop frameworks or classification systems. Banerjee et al. (1995) analyzed green advertisements and distinguished between shallow, moderate and deep advertisements, with on the extreme left side (shallow) vague environmental terms like ‘earth-friendly’, and on the extreme right side (deep) elaborate descriptions of environmental behaviors or of firms’ new environmental policies. A content study by Fowler and Close (2012) classified advertisements in the issues they address: macro, meso and micro level.
Greenwashing is also investigated in this category. Companies making false advertising claims, messages using puzzling terms that confuse consumers, scandals, and in retrospect exaggerated messages have bruised the credibility of green advertising. Carlson, Grove, and Kangun (1993) examined environmental claims and the extent to which they are misleading or deceptive. These authors classified on one hand product-oriented, process-oriented, image-oriented, and environmental facts advertisements, and on the other hand vague/ambiguous, omission, false/outright lie, and acceptable claims, to create a matrix (Carlson et al., 1993). Studies using this categorization matrix find that image-oriented claims are used most by green advertisers, but that these kinds of claims are also most ambiguous (Leonidou, Leonidou, Palihawadana, & Hultman, 2011). Last, this category contains studies that analyze the physical layout of green advertisements, including colors, images of certain landscapes or kinds of nature, and logos (Segev, Fernandes, & Hong, 2016).
The second research category as identified here contains mostly experimental studies, focusing on the psychological processes of consumer reactions to green advertising. The framing of a message, especially gain and loss framing – either showing the gain that can be obtained when consumers execute the request or showing the loss that can happen when consumers do not execute the request – has been
researched remarkably often (e.g. ; Bilandzic et al., 2017; White, MacDonnell, & Dahl, 2011). Several studies conclude loss framing to be more effective than gain framing in promoting pro-environmental behavior, although Cheng, Woon, and Lynes (2011) warn that the most effective frame might in fact depend on the measures that are used to assess the frames’ influence (Davis, 1995). This category has frequently borrowed theories from psychology to form hypotheses and explain results of consumer reactions to green advertisements, for example Petty and Cacioppo’s (1986) elaboration likelihood model or Skinner’s (1948) theory of operant conditioning (Hartmann, Apaolaza-Ibáñez, & Forcada Sainz, 2005). Also Ajzen and Fishbein’s (1980) theory of reasoned action and its extension, the theory of planned behavior, have been popular in green advertising research (Cheng et al., 2011). Frameworks of psychological theories on emotional appeals are frequently used as well (e.g. Baek & Yoon, 2017; Bilandzic et al. 2017).
The current study can be classified in the second category. It researches the effect of fear appeal and assertive language on two dependent variables: attitudes towards recycling and behavioral intentions to recycle. These dependent variables have been used in the past by Baek et al. (2015), who studied the interaction between assertive language in green advertisements and effort investment in signing a recycling pledge. The possibilities for the set-up of an experiment and the collection of data for the current study make attitudes towards recycling and behavioral intentions to recycle particularly attractive as dependent variables. Opposite to for example purchasing solar panels, recycling plastic is a pro-environmental behavior that does not require excessive spatial or monetary resources: it is something that can be performed by virtually everyone, including the expected respondents in this study. In addition, recycling plastic is a quite direct pro-environmental act. It is very likely to be performed only with ‘green’ intentions, as opposed to other pro-environmental acts such adopting a vegetarian diet. To illustrate how fear appeal and assertiveness could affect these two dependent variables, several theories on green advertising and psychological processes of consumers will be set out in the following section.
2.2 Assertive language and Reactance Theory
Advertisements using assertive language aim to guide a consumer towards performing a certain act by using imperative verbs such as ‘buy’, ‘do’, ‘click’, or ‘visit’ (Dillard & Shen, 2005; Zemack-Rugar et al., 2017). Such language creates, as Zemack-Rugar and colleagues (2017, p. 1) put sharply, ‘the impression that refusal is not an option.’ Therefore, assertive language is defined as language that gives directions to the recipient without providing him or her an option to refuse (Katz, Kronrod, Grinstein, & Nisan, 2018).
Past research has shown that the use of assertive language can be detrimental to the persuasiveness of a message. Quick and Conside (2008) investigated how to encourage college students to participate in exercise programs and found unfavorable responses to assertive messages. Several authors have used politeness theory to explain their findings that gentle, more polite requests from
teachers, friends, or romantic partners are more effective in gaining compliance (Wilson & Kunkel, 2000; Zhang & Sapp, 2013). Researching assertive messages in the context of water conservation campaigns, Katz and colleagues (2018) find that non-assertive language is more likely to positively influence residential water conservation behavior. Most studies that focus on a direct effect of assertive language draw a consistent conclusion: assertive language fails to persuade consumers (e.g. Baek et al., 2015; Dillard & Shen, 2005). Even more, it has shown its ability to elicit a boomerang effect, producing outcomes that go directly against the intent of the persuasive message (Stewart & Martin, 1994).
Reactance theory offers insights into this mechanism. Jack Brehm, the founder of the theory, defines psychological reactance as ‘the motivational state that is hypothesized to occur when a freedom is eliminated or threatened with elimination’ (Brehm & Brehm, 1981, p. 37). Psychological reactance occurs when an individual is pressured to comply with a request. As pressure is put on a choice, for example in the situation of parents forcing their children to eat vegetables, human beings tend to act against this pressure as a way to preserve their own freedom and control (Brehm & Brehm, 1981; Brehm, 1966). In the process of persuasion, reactance mediates the effects of a threat to freedom on results such as attitudes, beliefs, or behaviors (Dillard & Shen, 2005). Perceived freedom threat is therefore seen as the process that precedes reactance (Brehm & Brehm 1981). Several studies support this notion, having investigated it across different contexts (Quick, Scott, & Ledbetter, 2011).
Reactance often results in an increase in the attractiveness of the restricted behavior, and a decrease in the evaluation of the source of restriction (Fitzsimons & Lehmann, 2004). This process can function in multiple ways: whereas forcing something on an individual leads to opposition, making something difficult to attain leads to increased desirability (Brehm & Brehm, 1981). Reverse psychology, for example, relies on the mechanism of reactance. According to reactance theory, the most effective way to let children eat vegetables is not pressuring them to finish their plate; it is enjoying the vegetables as much as a parent can and forbidding the children to eat it (Brehm & Brehm, 1981). Reactance theory has been popular in advertising research: it has been studied in context of negative political advertising (Meirick & Nisbett, 2008), as well as extensively in health care advertising such as alcohol prevention messages (Bensley & Wu, 1991), anti-smoking campaigns targeted at adolescents (Grandpre, Alvaro, Burgoon, Miller, & Hall, 2003), and organ donation advertisements in gain- and loss-message frames (Reinhart, Marshall, Feeley, & Tutzauer, 2007). Different forms of advertising have been studied as well, with for example Edwards, Li, and Lee (2002) investigating psychological reactance to online pop-up ads.
Given what we have learned from reactance theory and studies on forceful phrasings, assertive language alone does not work as a persuasive technique. Past research has found this backlash effect repeatedly, with an ‘overwhelming’ (Kronrod et al., 2012a p. 95) amount of evidence.
2.2 Language Expectancy Theory and fit
With past research being able to state so clearly that assertive language has negative persuasive effects, one would expect advertisers to take this information into account when designing their campaigns. However, regardless of the theoretical findings, assertive requests by firms and institutions continue to exist in practice. A recent advertisement analysis by Zemack-Rugar et al. (2017) in America’s top ten print magazines has shown that 72% of the advertisements contained assertive language, with on average two assertive statements per advertisement. In the following paragraphs, this paradox will be explained by covering literature that has identified situations in which assertive language might lead to higher compliance.
As researchers have searched for reasons why assertive advertisements still prevail, several moderators that might lower or even reverse the negative effect of assertiveness have been investigated. Zemack-Rugar and colleagues (2017) have shown that consumers that are committed to the advertising brand feel a higher pressure to comply to an assertive request, and therefore show a higher level of reactance. Uncommitted consumers on the other hand display a much lower level of reactance (Zemack-Rugar et al., 2017). Product category has also been researched as a moderator; assertive language shows to be more effective for marketing hedonic products than utilitarian products, because hedonic products create a positive mood (Kronrod et al., 2012b). Interactions with assertive language have also been studied in a health care context: Buller, Borland, & Burgoon (1998) researched sun safety messages and found that assertive (high-intensity) language is more effective in motivating sun-safety behavior, but only when the arguments were formatted in a deductive style. For an inductively formatted message, non-assertive (low-intensity) language showed to be more persuasive (Buller et al., 1998).
Such interactions with assertive language have also been studied in green advertising, specifically recycling requests. Academic research has studied the moderation of nationality, finding that Americans are less receptive to assertive recycling messages than to non-assertive recycling messages, but that South Koreans do not show this reactance response. This finding is substantiated with the notion that people with an individualistic cultural background are more easily threatened in their freedom (Kim et al., 2017). Kronrod and colleagues (2012a) manipulate perceived issue importance and find this to interact with assertive language in several environmental contexts such as recycling plastic containers or economizing water. Baek et al. (2015) study recycling intentions and find an interaction between message assertiveness and effort investment. In the high-effort investment condition, the assertive message shows to be more effective than the non-assertive message (Baek et al., 2015).
A common thread that runs through many of these studies on interactions with assertive language is the notion of fit. In order for the assertive language not to provoke reactance, it needs to fit with the advertised product, the mood or cultural background of the consumer, the perceived issue importance, or the effort investment. A theory that looks into this link between assertiveness and fit more is Language Expectancy Theory (LET), which is mentioned repeatedly in research on advertising, assertiveness, and compliance. LET started flourishing thirty years ago, when Michael Burgoon and
Gerald Miller (1978) felt a renewed interest in persuasive communication and from there lay the foundations of the theory. Their LET serves as a framework that can explain the potential power of persuasive language with source, message and receiver variables, linking language in a message to persuasive outcomes (Dillard & Pfau, 2002). This theory regards language to be a system governed by rules. Throughout their lives, human beings develop ‘rules’, or expectations and preferences, for language strategies that others may use in persuasive attempts in given situations. LET views these rules as rooted in culture, making them a function of cultural and sociological norms (Dillard & Pfau, 2002).
When real-life assertive attempts do not meet linguistic expectations, these expectations are violated. This can be an either positive or negative violation. Language positively violates expectations when a communicator uses language that is more preferred than expected, or when an initially negatively evaluated communicator surprisingly uses language that is perceived to be appropriate by the receiver (Cameron, 2008). When positive violation happens, the persuasive effectiveness is enhanced and changes in for example beliefs or attitudes will occur in favor of the persuasive message. On the other hand, negative violations of language expectancy seem to strengthen reactance. When language negatively violates expectations – when the communicator uses language that the receiver perceives as inappropriate – the persuasive message will simply not change beliefs or attitudes or these will even change opposite to what the message advocates (Dillard & Pfau, 2012; Buller et al., 1998).
Past research has repeatedly concluded that there are certain contexts in which assertive language is more effective in gaining compliance than non-assertive language: it depends on the perceived fit between language expectations and actual language use. The current study explores whether fear appeal is one of the variables that could influence this perception of fit.
2.3 Fear appeal
Emotions drive human being’s motivations and actions (Schwartz & Loewenstein, 2017). Everyday, human choices on certain products, brands, and services but also on behaviors, opinions and attitudes are made, and even though they might seem rational, emotions always interfere. For that reason, it is no surprise that in marketing communications, emotional persuasion has been ever-present (Taute, McQuitty, & Sautter, 2013). Negative emotions such as sadness, anger, or fear have been researched in donation advertising (Bagozzi & Moore, 1994; Liang, Chen, & Lei, 2016). Guilt and shame, two more refined emotions that motivate people to avoid wrongdoing, have also been a popular topic in persuasive advertising (Baek & Yoon, 2017; O’Keefe, 2002). In the context of green advertising these emotions have been studied in advertisements on green consumption, water conservation and recycling (Amatulli, De Angelis, Peluso, Soscia, & Guido, 2017; Baek & Yoon, 2017). A positive emotion that has been researched in the light of green advertising multiple times is hope (Bilandzic et al. 2017; Lee et al., 2017). That emotions are substantially researched in green marketing communications is not a coincidence: several researchers emphasize the potential of emotions in gaining awareness and a sense of urgency of future climate change outcomes (e.g. Bilandzic et al., 2017; Moser, 2010; Roeser, 2012).
An emotional appeal that has not been left behind in this stream of research is fear appeal. With the first scientific study on fear appeals being published in 1953 (Janis & Feshbach), this method has quickly grown into the well-established tool in persuasive communications that it is today (Dillard & Anderson, 2004). Fear appeals are defined as persuasive messages designed to scare people into doing what the message recommends by showing the severe consequences of noncompliance (Morales, Wu, Fitzsimons, 2012; Witte, 1992). Most fear appeal persuasion research done in the past uses one or more of the following theories as a theoretical foundation: drive theory and the drive-reduction model (Hovland, Janis, & Kelley, 1953; Janis & Feshbach, 1953), protection motivation theory (Rogers, 1983), and the parallel process and extended parallel process model (Chen, 2016; Tanner, Hunt, & Eppright, 1991; Witte, 1992). All indicate that fear appeal messages cause feelings of fear or threat that, when paired with a message recommendation that can reduce these feelings, become persuasive (Keller, 1999).
Advertisements using fear appeals typically provide recipients with two kinds of information. The first kind is information that evokes fear by showing something that poses a threat to the recipient, for example confronting smokers with pictures of diseased lungs or with a text such as ‘SMOKING KILLS’. Second, a way to prevent this fearful outcome is presented, possibly nicotine patches that can be used to quit smoking (Ruiter et al., 2001). This way, by creating discomfort, a firm producing nicotine patches persuades people to perform behavior (buy patches) that can make the discomfort go away (Brennan & Binney, 2009). Fear appeal theory therefore links to Burrhus Skinner’s theory of operant conditioning: by using the psychology of negative reinforcement, a fear appeal aims to make a receiver perform behavior to avoid an aversive and fearful stimulus (Rossiter & Thronton, 2004).
Since the 1950s, the effectiveness of fear appeals has been researched in many different contexts, especially in the realm of health care. Fear appeals have been used to research how to address a wide range of public health topics such as flossing teeth for dental hygiene, using sunscreen to prevent skin cancer, not drinking and driving, condom use for AIDS prevention, and breast self-examination (Morales et al., 2012; Witte & Allen, 2000). While the field of fear appeal research was expanding, in 2006, authors LaTour and Tanner (2006) stated that this field was too limited to a health care context. They called for a broader view on fear appeals to understand this complex construct (LaTour & Tanner, 2006). When applying this remark to the context of green advertising, Hartmann et al. (2014) explain that fear appeals in green advertisements are different from those in health care advertisements. With fear appeal-advertisements in health care, the recipient can take entire responsibility for eliminating the discomfort and fearful consequences that he or she is confronted with. However, as Hartmann and colleagues (2014) explain, the threat of climate change is perceived in a different way. Recipients of fear appeal-advertisements on the dangers of climate change are aware that they cannot defeat this fear on their own; if they perform the behavior that the advertisements request, they will only marginally contribute to the elimination of the threat. The authors conclude that this results in the recipient having a low perceived efficacy, which in turn acts as a brake on the intention to perform the requested behavior
(Hartmann et al., 2014). For this reason, one must be careful to generalize findings of fear appeal effectiveness in health care advertising to green advertising.
Starting mostly in the new millennium, LaTour and Tanner’s (2006) quest for a wider view on fear appeals was answered by green advertising studies. Academics started to study how fear appeals could be used to stimulate pro-environmental behaviors. Chen (2015) researched differences between high and low fear appeal, finding that the relationship between level of fear and persuasiveness of an environmental advertisement is U-shaped. Fear appeal has shown its ability to promote several pro-environmental behaviors, such as purchasing green electricity and supporting tax on carbon dioxide emissions (Hartmann et al., 2014), willingness to sacrifice money (i.e. paying more for eco-friendly products) and conveniences (Bilandzich et al., 2017), and purchase intention of eco-friendly LED light bulbs (Shin et al., 2017). However, it is difficult to draw general conclusions from these studies. Although all authors conclude fear appeals to be effective in persuasion (e.g. Hartmann et al., 2014; Hartmann, Apaolaza, D’Souza, Barrutia, & Echebarria, 2016), some conclude this with having studied fear appeal in interaction with other variables such as global or local framing of environmental issues (Lee et al., 2017).
In other contexts, this difficulty has been noted as well. Having developed more than sixty years into research on this persuasive tool, academic research has not been able to reach consensus on the effectiveness of fear appeals: some studies have concluded fear appeals to be very effective, others have found weak persuasive effects (e.g. Brennan & Binney, 2009; Morales et al., 2012; Shin et al., 2017).
2.4 Hypotheses development
Consensus on fear appeal effects might be difficult to achieve because there are many factors that can influence the effects of this emotional appeal. Accordingly, academic research has studied fear appeal in combination with several variables. In both a health care and a green context, fear appeal has been investigated repeatedly with gain and loss frames (Bilandzic et al., 2017; Gerend & Maner, 2011; Ruiter et al., 2001). Health behavior studies have concluded that the persuasive impact of fear appeal is enhanced by the adoption of a loss-frame (Gerend & Maner, 2011; Ruiter, Kok, Verplanken, & Eersel, 2003). Bilandzic and colleagues (2017) studied how fear appeal can affect willingness to sacrifice money and conveniences for climate change. The authors find that pairing fear with a gain-negative frame (‘if we reduce emissions, the global temperature will not rise’) or a loss frame (if we do not reduce emissions, the global temperature will rise’) increases perceived threat and willingness to sacrifice.
In green advertising, more influences on fear appeal’s effects have been examined. Hartmann and colleagues (2014) have investigated how coping efficacy information could influence fear appeal to affect pro-environmental behaviors. Providing efficacy information after fear appeals shows to have a twofold effect: it improves persuasive effects, but it also lowers fear responses (Hartmann et al., 2014). An interaction between fear appeal and environmental involvement of consumers has also been researched, but Shin et al. (2017) conclude that this does not moderate the effect of fear appeal on
consumers’ attitudes towards and purchase intention of a green product. Lee et al. (2017) investigated fear and hope in interaction with geographical distance of a fearful environmental issue. They find that when an environmental issue is framed as global, fear appeal enhances pro-environmental behavioral intentions; when an issue is framed as local, hope appeal is more effective (Lee et al., 2017).
Given what the academic field has learned from LET, this study expects fear appeal and assertive language to significantly influence each other, with fear appeal reversing the negative effects of assertive language. A study by Kronrod and colleagues (2012a) is especially important in forming this idea. Investigating the interaction between perceived issue importance and assertive language use in green advertisements, the key idea of these authors is that perceived issue importance affects linguistic expectations. In their research, they manipulate perceived issue importance in several studies by showing an environmental clip, varying the geographical domain where an issue is or is not important, and using the Google AdWords advertising system. They measure compliance intentions to assertive requests, asking people to reduce air pollution, economize water or soap, or sign a petition by clicking on an online link (Kronrod et al., 2012a).
In all their studies, through lab as well as field studies, the authors find significant interaction effects between assertive language and perceived issue importance. When an issue is seen as very important and urgent, assertive language is more likely to be expected because it is in line with the urgency. The opposite is true when the issue is not considered to be important: then an assertive request is likely to elicit reactance because of the ‘excessive forcefulness’ (Kronrod et al., 2012a, p. 96) of the request. When no high issue importance is perceived, a non-assertive request leads to higher compliance because it fits better with the expectations of the consumer (Kronrod et al., 2012a).
While fear appeal is a distinct construct from perceived issue importance, it might have the same moderating effect on assertive language. Emotions have proven their ability to add urgency or importance to environmental problems: specifically, fear appeals add this sense of urgency by showing the discomforting consequences of noncompliance (Moreales et al., 2012; Roeser, 2012; Witte, 1992). In turn, the feelings of urgency aroused by fear appeal influence consumers’ language expectations: because consumers see the need for immediate action, they expect an assertive request. As LET has demonstrated in the past, when an actual assertive request follows, no negative violation of expectations takes place. Instead, there is a high perceived fit between language expectations and actual language use. This averts the common negative reactance response to assertive language, leading to higher compliance with the request (Dillard & Pfau, 2012). However, when an assertive message is used without being preceded by fear appeal, such forceful language violates expectations. In that case, the common reactance response is evoked: consumers feel threatened in their freedom, and act counter the request (Baek et al., 2015; Brehm & Brehm, 1981; Dillard & Shen, 2005). Following this reasoning, reactance theory, and LET, assertive language is expected to interact with fear appeal to affect attitudes towards recycling and behavioral intentions to recycle. This puts forth the following hypotheses:
H1: When fear appeal is present, an assertive language text will lead to (a) higher attitudes towards recycling and (b) higher behavioral intentions to recycle than a non-assertive language text.
H2: When fear appeal is not present, a non-assertive language text will lead to (a) higher attitudes toward recycling and (b) higher behavioral intentions to recycle than an assertive language text.
As explained above, the main mechanism underlying this interaction hypothesis is expected to be the ‘fit’ of expectations and reality, an idea originating from LET. This theory states that the fit or discrepancy between the expected language use and actual language influences persuasive outcomes (Dillard & Pfau, 2002). It is expected that when consumers are fearful and worrisome about plastic pollution and the future of planet earth, they will find assertive language more appropriate, increasing the fit between expected and actual language use and therefore positively influencing persuasiveness. On the other hand, when consumers are not worried about plastic pollution, they will find the assertive language too strong, violating expectations and creating misfit, and therefore negatively influencing persuasiveness. In line with these ideas from LET, it is hypothesized that:
H3: Perceived fit between expected language and actual language use mediates the interaction-effect of assertive language and fear appeal on attitudes towards recycling and behavioral intentions to recycle.
Figure 2. Hypothesized relationships
3. METHOD
3.1 Experimental stimuli
The design used in this study is a 2 (assertive language: assertive vs non-assertive) x 3 (fear appeal: fear now vs fear future vs fear control) between-subjects experimental design. The two independent variables are manipulated with an advertisement that consists of two parts. In the first part, fear appeal is manipulated with a series of 11 images on either plastic pollution (fear appeal) or plastic production and characteristics of plastic materials (control). Each image is accompanied by a line of text. The second part of the advertisement manipulates assertiveness with a piece of text that either demands (assertive) or requests (non-assertive) respondents to recycle plastic. At the start of the experiment, respondents are informed that the source of the advertisement is a non-governmental organization. This source was chosen because the combination of fear appeal and assertive language is mostly used in advertisements from such organizations. The manipulations of the two variables can be found in appendix B and C.
Fear appeal
Fear appeals can be framed in terms of showing the benefits of performing the requested behavior (gain frame) or showing the costs of neglecting the requested behavior (loss frame). The fear appeal manipulation of this study, in the fear conditions, adopts a loss frame, showing the negative consequences of not engaging in programs to recycle plastic: polluted oceans and rivers, dying animals, disrupted eco-systems, and poisoned food that causes damage to the human body. The 11 images that illustrate these losses have been collected via various online news articles.
The loss frame was chosen for theoretical as well as practical reasons. From a theoretical point of view, past academic research has shown loss framing to be more effective than gain framing in promoting health and environmental behaviors, with or without combining it with fear appeals (Cheng et al., 2011; Davis, 1995; Gerend & Maner, 2011; Ruiter et al., 2003). From a practical standpoint, non-governmental organizations in the environmental sector select the loss frame most frequently when
using fear appeal in advertisements. Therefore, adopting a loss frame makes the findings of this study more relevant for current advertising practices.
In academic research, it is not uncommon to manipulate fear with more than two levels. An example is Hartmann and colleagues (2014), who include control, moderate, and strong threat appeals in their study, finding no difference between the moderate and strong conditions. The current study takes another angle to create three groups within fear appeal. The incentive for this angle is a study by Cheng and colleagues (2011). These authors note that an issue with the use of fear appeals in promoting sustainable behavior is the temporal distance of the addressed negative consequences of unsustainable behavior. Because the environmental losses that could follow from failing to perform the requested sustainable behavior seem so far away, the current generation often does not feel real fear (Cheng et al., 2011).
A theory that is especially interesting in this light is construal level theory, which distinguishes between abstract and concrete mental levels. When an event or object feels far away from someone, then this person will think about it in an abstract way. This is thinking at a high level of construal. The process also works the other way around: when an event or object feels psychologically close, the brain will think about it on a more concrete level. This is called low-level construal (Trope & Liberman, 2010; White et al., 2011). The psychological distance in construal level theory can take several dimensions: temporal distance, spatial distance, social distance, and hypotheticality (Bar-Anan et al., 2006). In the past, research on green advertising has drawn on construal level theory to create experiments and explain findings. White and colleagues (2011) investigated how advertisements can influence recycling behavior with construal level theory and find that gain and loss frames interact with a low-level construal (‘how will I recycle’) and a high-level construal (‘why should I recycle’). Chang et al. (2015) investigated this same interaction with a study on the ad attitude, brand attitude, and purchase intention of eco-friendly dishwashing liquid and hybrid cars. Both studies conclude a match between a loss frame and low-level construal, and between a gain frame and high-level construal (Chang et al., 2015; White et al., 2011).
Eager to explore construal level theory in green advertising more, the current study includes temporal distance as a research factor. The fear appeal manipulation is therefore divided into three levels: fear now, fear future, and fear control. Drawing on construal level theory, it is examined how the temporal distance between the recipients and the object (frightening plastic pollution consequences) shape their response (attitudes, behavioral intentions) to this object (Bruehlman-Senecal & Ayduk, 2015). No hypotheses on potential differences in fear arousal or subsequent effects have been formulated because these differences are not necessarily expected. The three levels in the manipulation serve more as an exploration of construal level theory in the hypothesized interaction.
Whereas Hartmann et al. (2014) manipulate moderate and strong threat appeal with a series of nine images, they do not show anything to the control condition. In their discussion, the authors identify this as a limitation to their study. They acknowledge that ‘a more balanced experimental design would
have exposed participants to the same number of slides and approximately the same amount of information within any of the six factor treatments’ (Hartmann et al., 2014, p. 757). For that reason, the current study deliberately provides the fear control condition with the same number of images as the two other fear conditions. The images in the control condition are also about plastic, but provide neutral facts about plastic production, material properties of plastics and uses of plastics.
Assertive language
Assertiveness is manipulated with two short pieces of text, either demanding or requesting consumers to recycle. The assertiveness condition contains phrases such as ‘must’, ‘should’, and ‘have to’; the non-assertive condition contains more polite phrases such as ‘could’ and ‘may’. The phrasings that are specific to the condition appear in bold. Except for these textual manipulations in assertiveness, all other aspects of the advertisement such as layout and size remain constant. The format and text of this part of the advertisement is extracted from the assertiveness manipulation of Baek et al. (2015), which focuses on recycling. Small alterations were made to text and lay-out.
3.2 Pre-test fear appeal and assertive language manipulations
To examine whether the manipulations are effective, a pre-test study was conducted. The sample for this study consisted out of 31 respondents, collected via a snowball method. The pre-test survey was distributed via Qualtrics and randomly assigned respondents to one of six conditions: fear now/assertiveness, fear now/non-assertiveness, fear future/assertiveness, fear future/non-assertiveness, fear control/assertiveness, or fear control/non-assertiveness. The number of respondents in each condition can be found in table 1. It must be noted that the sample sizes in this pre-test are very small to conclude significance.
During the pre-test, respondents were exposed to the 11 images, immediately followed by the manipulation check for fear. Next, they were asked to read the piece of text, followed by the manipulation check for assertiveness. A one-way ANOVA and independent samples t-test were used to identify whether the materials are effective in manipulating fear and assertiveness and whether the measurement scales for both manipulation checks are consistent (appendix D).
Table 1. Number of respondents in each condition, pre-test
Fear now Fear future Fear control Total Assertiveness 5 4 5 14 Non-assertiveness 6 5 6 17
Fear appeal
The manipulation check of fear appeal is measured with a five item, seven-point Likert scale in which respondents indicated to what extent they experienced five emotions related to fear: ‘afraid’, ‘scared’, ‘fearful’, ‘concerned’, and ‘worried’. This manipulation check has been used before by Hartmann et al. (2014), who found this scale to have a Cronbach’s alpha of 0.95. Three filler emotions were added to this scale: ‘curious’, ‘calm’, and ‘enthusiastic’.
Analyses indicate the measurement scale of fear to have a good scale reliability (α = .78) that would not improve with deletion of any of the items. The output of a one-way ANOVA test shows that the fear now condition evokes the highest level of fear (M = 5.29, SD = .84), followed by the fear future condition (M = 5.11, SD = .10) and the fear control condition (M = 4.09, SD = .10). There is a statistically significant effect of fear appeal conditions on fear, F(2, 28) = 5.08, p <.05. Tukey post-hoc tests reveal that the perceived level of fear is significantly higher in the fear now condition compared to the fear control condition (p <.05). The difference in evoked level of fear between the fear future and the fear control condition is marginally significant (p = .06). Last, there is no significant difference between the fear now and fear future condition (p = .91). It can be concluded that only the fear now condition and the fear control condition arouse significantly different levels of fear in respondents. Because the fear future condition still shows marginally significant higher scores than the fear control condition, the manipulations for fear are accepted and not changed for the main experiment.
Assertiveness
The manipulation check of assertive language is measured with a four item, seven-point Likert scale. Respondents were asked to indicate the extent to which they agreed with the statement ‘The phrasing of the text of this advertisement is…’, followed by several adjectives. Three items of the assertiveness scale have been used before by Baek et al. (2015): ‘forceful’, ‘intense’, and ‘powerful’. Baek et al. (2015) found these three items to have a Cronbach’s alpha of 0.94. In the current study, a fourth item was added: ‘assertive’. This item was used as a manipulation check by Kim et al. (2017). Last, three filler adjectives were added in the experiment: ‘irritating’, ‘detailed’, and ‘original’. An overview of the scale items for assertiveness and fear can be found in table 2.
The combined four-item assertiveness scale in the current study shows to be reliable (α = .86). The output of the independent samples t-test indicates that there is a significant difference between the scores for the assertive language condition (M = 4.88, SD = 1.35) and non-assertive language condition (M = 3.41, SD = 1.31), t (28) = 3.03, p <.01. These results suggest that respondents in the assertive language condition evaluated the advertisement as significantly more assertive than respondents in the non-assertive language condition.
Based on the pre-tests, the manipulation materials for fear and assertiveness were accepted for the main study. However, based on insightful feedback from the respondents, slight changes in the questions were made. There appeared to be some confusion on if the manipulation check questions
appealed to the 11 images, to the text, or to both parts of the advertisement. For that reason, this was clarified throughout all questions of the final survey. Also, in the pre-test, each series of images ended with a blank page to give respondents the opportunity to digest what had seen. In practice, respondents regarded this blank page to be a picture that failed to load. For that reason, the blank page was deleted.
3.3 Procedure
For the main study, respondents were collected via a snowball method and via Amazon Mechanical Turk (MTurk). Amazon MTurk is a global web service that gives access to online, on-demand workforce. For this study it served as an efficient and effective way to collect data in a short period of time. Due to restrictions in time, only collecting data via a snowball method could jeopardize the planning of this study. Also, no discrepancies between responses from the snowball sample or Amazon MTurk sample were expected: this will be explained further in the results section.
The survey of this study was written in the English language, as to make sure that the foreign respondents provided by Amazon MTurk would see the exact same survey as the Dutch respondents that were collected via the snowball method. While collecting data via friends and family, attention was paid to this English format, requesting friends and family to send the survey only to people that they thought would comprehend the English language fluently enough to understand the entire survey. The survey was distributed via online survey software Qualtrics.
After respondents clicked on the survey link, they were presented with a short introduction to the topic of the questions (‘your reaction to an advertisement’), the average amount of time spent on the survey, their anonymity, and their option to answer questions via e-mail. On the next page, the respondents were informed that they would get to see an advertisement by a non-governmental organization, and that the advertisement consists out of two parts: part one being 11 images, each accompanied by one line of text, and part two being a short piece of text. Then the respondents were randomly assigned to one of six conditions. Right after the manipulation, the two dependent variables were measured. This was followed by questions about a mediator, subsequently measuring control variables and the manipulation checks for fear appeal and assertive language. The manipulation checks referred explicitly to part one (the 11 images) or part two (the text) of the advertisement. What followed was an attention check question about the topic of the text in the second part of the advertisement, creating the ability to identify respondents that indifferently filled in the questions. The last stage of the survey started with a question about the extent to which a respondent is currently recycling plastic and closed off with questions about demographics (i.e. age, nationality, gender, employment status).
3.4 Measurements
After seeing the 11 images and the short text, respondents were asked to fill in items that measured the different constructs in the conceptual model. The measures used for the variables are obtained from
previous academic research, some containing small adaptations to better match the context of this study. Some items in the measurement scales were reversed coded, and in most scales the order of the items was randomized in Qualtrics. All items of the scales can be found table 2; the final survey is included in appendix E.
Dependent variables
Attitudes towards recycling was measured with seven item, seven-point scale semantic differential. The items here finish the sentence ‘Recycling plastic is…’ (e.g. bad VS good, foolish VS wise, unfavorable VS favorable – all items are in table 2). This scale is derived from Baek et al. (2015), who found it to have a Cronbach’s alpha of 0.96. The other dependent variable, behavioral intentions to recycle, is measured with a four item, seven-point Likert scale. These four items have also been used by Baek et al. (2015) in the past, with Cronbach’s alpha in their study being 0.90.
Mediator
Language fit is measured by a four item, seven-point Likert scale. Respondents were asked to what extent they found the phrasing of the text ‘typical’, ‘expected’, and ‘standard.’ This scale is derived from Kronrod et al. (2012b), who have demonstrated this scale’s reliability with a Cronbach’s alpha of 0.82.
Control variables
The first control variable is skepticism towards environmental claims by non-governmental organizations. This construct is measured using a four item, seven-point Likert scale. The scale used in the current study is an altered version of scale for skepticism towards environmental claims marketer’s communications from Mohr, Eroǧlu, and Ellen (1998), researching skepticism towards claims on package labels and advertising. Cronbach’s alpha of the scale of Mohr et al. (1998) is 0.87.
The second control variable is situational factors. It might be that some of the respondents are very willing to recycle, but that restrictions in time or space keep them from actually having behavioral intentions to recycle. These inconveniences on the respondents’ side were controlled for by including them as a control variable. Situational factors is measured with a three item, seven-point Likert scale, derived from Bezzina and Dimech (2011). A slight alteration to the original scale was made: the word ‘waste separation’ was replaced with ‘recycling’ in each item. Bezzina and Dimech (2011) found their scale to have a Cronbach’s alpha of 0.82.
Third, perceived issue importance is measured with a four item, seven-point Likert scale. This scale for perceived issue importance has been used before by Kronrod et al. (2012a). Their scale has a Cronbach’s alpha of 0.88.
Table 2. Items of all measurement scales
Attitudes towards recycling (dependent variable, 7 items, 1 = strongly disagree, 7 = strongly agree) Bad VS good Foolish VS wise Negative VS positive Unfavorable VS favorable Unnecessary VS necessary Harmful VS beneficial Undesirable VS desirable
Behavioral intentions to recycle (dependent variable, 4 items, 1 = strongly disagree, 7 = strongly agree) I intend to recycle more in the next few weeks
I will consider participating in a recycling program in the future
I would recommend participating in a recycling program to my friends or relatives I am likely to make an effort to recycle to protect the environment
Language fit (mediator, 3 items, 1 = strongly disagree, 7 = strongly agree) The phrasing of the message text of this advertisement is typical
The phrasing of the message text of this advertisement is expected The phrasing of the message text of this advertisement is standard
Skepticism (control variable, 4 items, 1 = strongly disagree, 7 = strongly agree)
Most environmental claims made by non-governmental organizations in advertising are true (reversed) Because environmental claims are exaggerated, consumers would be better off if such information would not be given
Most environmental claims in advertising are intended to mislead rather than inform consumers I do not believe most environmental claims made in advertisements
Situational factors (control variable, 3 items, 1 = strongly disagree, 7 = strongly agree) Recycling plastic takes too much time
Recycling plastic takes too much space Recycling plastic is too complicated
Perceived issue importance (control variable, 4 items, 1 = strongly disagree, 7 = strongly agree) It is important for me to help reduce plastic pollution
I try to help reduce plastic pollution
I think a lot about ways to help reduce plastic pollution
Helping reduce plastic pollution is not at the top of my priorities list (reversed) Fear (manipulation check, 5 items, 1 = strongly disagree, 7 = strongly agree) Afraid Scared Fearful Concerned Worried Curious (filler) Calm (filler) Enthusiastic (filler)
Assertiveness (manipulation check, 4 items, 1 = strongly disagree, 7 = strongly agree) Forceful Intense Powerful Assertive Irritating (filler) Detailed (filler) Original (filler)
4. RESULTS
4.1 Respondents and data preparation
In total, 247 fully filled-in surveys were collected via a snowball method and Amazon MTurk. Although it was requested in the survey title that only US respondents filled in the survey on Amazon MTurk, 35 surveys were completed by respondents from India. These responses were excluded from the analyses for two reasons. First, there is a big difference in the advancement of plastic recycling processes between Western countries and India. The United States and Europe are quite equally advancing in plastic recycling processes and policies; in India, it has only recently – but greatly – started to boom. Second, the plastic pollution problem in India is a lot more visible than in Western countries (Krishnamoorti, 2018). These two differences between India and Western countries could influence the dependent variables in ways that are not controlled for in this study.
The next round of data cleaning eliminated respondents on the basis of an attention check question that was included at the end of the survey. There were 29 respondents that answered this question wrong: they were excluded from the analyses. Next, outliers that had a Z-score of > 3 or < -3 were eliminated, resulting in the elimination of another 8 respondents. The final number of respondents assigned to each condition can be found in table 3. While the goal was to collect 180 respondents, through the data preparation, 175 respondents were left. However, this study still contains the minimum of 20 – 30 respondents in each condition, which is enough to draw reliable conclusions (Sekaran & Bougie, 2016).
Of the 175 respondents left, 75 are respondents that were collected from Amazon MTurk. In total, 56.6% is Dutch and 38.9% American. The respondents ranged in age from 17 to 70 (M = 30.84,
SD = 12.20). Regarding gender, the data is quite equally divided: 43.4% is male, 54.9% is female. Most
respondents are full-time employees (44.6%) or students (31.4%), and on a 7-point Likert scale, they score quite high on current average recycling behavior (M = 4.77, SD = 1.70). Detailed demographics can be found in appendix F.
Table 3. Number of respondents in each condition, 2 x 3 design
Fear now Fear future Fear control Total Assertiveness 24 31 27 82 Non-assertiveness 38 28 27 93
Total 62 59 54 175
4.2 Factor analysis
Because some of the scales included counter-indicative items, these items had to be recoded before conducting a factor and reliability analysis. After doing so, possible similarities between the two dependent variables (attitudes towards recycling and behavioral intentions to recycle) and the mediator
