Corporate social responsibility : getting the right message across to consumers : an experiment about the mediating role of message framing and the moderating role of sustainability orientation in the fashion industry

(1)

- 1 -

MASTER’S THESIS

CORPORATE SOCIAL RESPONSIBILITY: GETTING THE RIGHT MESSAGE ACROSS TO CONSUMERS

An experiment about the mediating role of message framing and the moderating role of sustainability orientation in the fashion industry

LIESBETH BOS 10113592

SUPERVISOR LARS MORATIS

“The price of greatness is responsibility.”

(2)

- 2 -

STATEMENT OF ORIGINALITY

This document is written by Liesbeth Bos 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 Economic and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

- 3 -

ABSTRACT

AIM This thesis provides insights in to the effect of message ambiguity, the mediating role of

message positivity (message framing) and moderating role of sustainability orientation, measured in brand attitude towards fictional Fashion Company X.

BACKGROUND Companies can choose to frame their messages in ambiguous, vague and unclear

ways or concrete, informational and clear ways. Second, these messages can be positively framed, what you gain by buying the product, or negatively framed, what you lose by not buying the product.

METHOD This study is based on a 2 (ambiguous versus concrete message) x 2 (positive versus

negative message framing) x sustainability orientation experiment. A total of 196 respondents filled in an online survey provided on Qualtrics. These respondents were divided in four different conditions and were asked after reading the relevant CSR message to evaluate fictive Fashion Company X in terms of brand attitude.

FINDINGS The results show a main effect of message positivity on brand attitude. First, the more

positive a message is framed the higher brand attitude will be. Second, a non-significant effect of message ambiguity on brand attitude was found. When a message is ambiguous this results in lower brand attitude compared to a concrete CSR message. Third, no interaction effect was found between message ambiguity and message positivity. In conclusion, no interaction was found between message positivity and sustainability orientation.

MANAGERIAL IMPLICATION This study shows that companies should rather use positive than

negative message framing, both by itself as well as combined with concrete and ambiguous messages and for sustainable- versus unsustainable-oriented consumers. This research also found out that message framing is not similar to involvement. Also, international companies should take different communication styles per country or culture into account when framing a message.

Keywords: Corporate social responsibility; message ambiguity; message framing; sustainability orientation; sustainable fashion

(4)

- 4 -

2.2 CSR communication 15 2.2.1 Message positivity 17 2.2.2 Message ambiguity 20 2.3 Sustainability orientation 23 2.4 Conceptual model 24 3. Methodology 25 3.1 General methods 25 3.2 Pre-tests 26 3.2.1 Message pre-test 26 3.2.2 Survey pre-test 29 3.3 Experiment 30 3.4 Variables 30 3.4.1 Message positivity 30 3.4.2 Message ambiguity 31 3.4.3 Sustainability orientation 31

3.4.4 Attitude towards the brand 32

3.4.5 Purchase intention 32 3.4.6 Manipulation check 33 3.5 Sample 33 4. Results 34 4.1 Statistical analysis 34 4.1.1 Reliability test 35

(5)

- 5 - 4.1.3 Hypotheses testing 36 4.1.2.1 Hypothesis 1 36 4.1.2.2 Hypothesis 2 38 4.1.2.3 Hypothesis 3 39 4.1.2.4 Hypothesis 4 40

5. Discussion and conclusion 43

5.1 Summary of results 43

5.2 Hypotheses 44

5.2.1 Hypothesis 1: Effect of message positivity 44

5.2.2 Hypothesis 2: Effect of message ambiguity 44

5.2.3 Hypothesis 3: Interaction effect 46

5.2.4 Hypothesis 4: Interaction effect 48

5.2.5 Adjusted conceptual model 49

5.3 Practical implications 50

5.4 Limitations 51

5.5 Avenues for further research 52

5.6 Conclusion 53

6. Acknowledgement 54

7. References 55

(6)

- 6 -

1. INTRODUCTION

1.1 The True Cost

Meet Shima Ahkter, a 23-year-old mother and garment worker in one of the many factories in Dhaka, Bangladesh. The story of Shima Ahkter is one of the many stories featured in The True Cost, a documentary about the toxic production of our western textiles by producer Michael Ross and director Andrew Morgan (2015). She is one of the 40 million garment workers worldwide. Almost 4 million of these workers are located in Dhaka, the capital of Bangladesh. This city counts over 5000 factories that produce clothing primarily for the Western markets. Shima was just 12 years old when she came to Dhaka to start working. While working in the factory, Shima is also taking care of her three-year-old daughter. The workplace is not a suitable environment for children: these so-called sweatshops are very hot and everyone is exposed to chemicals dangerous for humans. The walls in the factory are shattered with cracks; working conditions are outright abysmal. The factory Shima works in is not that far from Rana Plaza, a garment-producing complex that took 1138 lives when the disapproved building collapsed. Garment workers were buried alive making this the biggest disaster in the history of fast fashion. Why are garment workers in Bangladesh paying the price for our cheap clothing?

Let’s go back to the commodity behind this enormous industry: imagine yourself to be a cotton farmer in India. To keep up with the growing demand for cheap clothing, cotton plants are genetically modified to keep up with the huge amounts of pesticides that the plants are treated with. Poor farmers in India are forced into buying these genetically modified seeds from companies like Monsanto. According to The True Cost (2015), these seeds cost 17,000% more than their usual seeds and their use has a devastating effect not only on the soil but also on people living in these areas. Doctor Pritpal Singh, Director of the Faridkot Centre in Punjab and physician (Ross & Morgan, 2015), has been researching the effects of the use of pesticides in India’s biggest cotton producing region, Punjab. His research shows a dramatic

(7)

- 7 -

rise in the number of birth defects and in the occurrences of cancer and mental illness in this region in specific. He has found villages in the Punjab region where a minimum of 60 to 70 children are mentally disabled. Furthermore, in the last 16 years over 250,000 suicides of Indian cotton farmers were recorded, which comes down to one farmer every 30 minutes, making this the biggest suicide wave in history (Ross & Morgan, 2015).

Fashion has a devastating effect on our world. Every step of the clothing life cycle generates pollution (Claudio, 2007; Ro & Kim, 2009). Globalisation was supposed to be a win‒win situation: the Western world would get cheaper goods and people in developing countries would get jobs and a better life (Ross & Morgan, 2015). These people would get a chance to get away from poverty and poor living standards. “So how come this industry, generating over 3 billion dollars in profits, is not able to support its millions of workers and is instead taking so many innocent lives (Ross & Morgan, 2015)?” Now, globalisation is

responsible for producing clothing at low prices, which results in “fast fashion” where

clothing as a product becomes disposable (Claudio, 2007). The solution for companies lies in a more sustainable method of producing. This is something that consumers also find very important (Niinimäki, 2010), and therefore companies often like to communicate their

corporate social activities to consumers. According to The True Cost, there are three activities in the fashion industry that can be made more sustainable: working conditions in factories in developing countries, the production of cotton, and the reduction of textile waste (Ross & Morgan, 2015). The fashion industry developed numerous trademarks and logos pointing to these sustainable developments. However it is unclear to consumers what all these trademarks actually stand for.

(8)

- 8 - 1.2 Concepts

The hypotheses will be tested in the fashion industry, since the apparel industry is the second largest polluting industry in the world (Sweeny, 2015). The fashion industry is polluting the environment in different ways. According to Fast Company the apparel industry contributes up to 10% of the total global emissions (Bédat & Shank, 2016). To provide more clarity on why the messages for the experiment are formulated like this, this part of the thesis will be dedicated to illustrating sustainability in the fashion industry. The focus on sustainability will be on the two main polluting parts addressed in The True Cost: the cultivation of cotton and the social aspect of the poor working conditions in developing countries.

1.2.1 Cultivation of cotton

Clothing is made of fibers, either natural or synthetic. These fibers are often of natural sources like cotton, wool or satin. From these three natural fibers, cotton, which has a clean and wholesome image in the mind of consumers, is the most often used, in 40% of all garments produced worldwide (Sweeny, 2015). People might think that these commodities are natural and therefore less polluting; however making just a T-shirt and a pair of jeans from cotton takes more than 5,000 gallons of water (Sweeny, 2015). Moreover the crops from which all these natural fibers are produced need to be sprayed with anti-pesticides. An extra problem here is that cotton requires more pesticide use than almost every other crop cultivated worldwide (Clark, 2009). An average of 15% of the worldwide use of pesticides (herbicides and insecticides) is used on cotton crops, whereas cotton only covers a few per cent of the worldwide cultivable soil (Clark, 2009). Big agricultural companies like Monsanto are exhausting soils worldwide. The use of these herbicides, insecticides and chemicals to grow the crops for the fibers are exhausting cultivable soils and giving farmers working with these

(9)

- 9 -

chemicals and their progeny diseases like cancer and born disabilities and malformations (Ross & Morgan, 2015).

1.2.2 Social aspects

Together with these harmful ways of cultivating natural fibers for clothing, the way this clothing is produced is also polluting our world. As can be read in the first section, the apparel industry is one with lots of pressure on low-paid employees in so-called sweatshops. The same low wages that made producing countries like India and Bangladesh so attractive for clothing companies, are now also responsible for forcing people to work extremely long days for an extremely low wage. I visited a few clothing factories myself in Istanbul, Turkey. Although these factories should be ones with proper working conditions, they were not what I expected. The working conditions and the work these people have to do are not human and are not comparable to what I am used to myself.

Another great challenge for the working conditions of factory workers is the rise of fast fashion in the last 15-20 years (Ross & Morgan, 2015). Once all of these clothes have been produced, with lots of harm to the environment, nowadays it is common to wear them only a couple of times and be done with them: fast fashion. You don’t own it; you consume it (Bédat & Shank, 2016). Fast fashion, as the name already states, is produced fast. It is

practically the equivalent of fast food (Claudio, 2007). Workers must work harder and faster and produce more and more in the same time. How can you make a pair of jeans for a price of 30 or 50 cent each in an ethical way (Ross & Morgan, 2015)? This however is what the Western world is asking of these workers. The current market for environmentally friendly apparel, from now on called “eco-fashion” was relatively small in 2012 at 1% of the total market (Yan, Hyllegard & Blaesi, 2012). This market is expected to grow more in the upcoming years. According to Yan et al. (2012) this market is even going to double since

(10)

- 10 -

there is a lot of media attention, greater commitment from fashion companies and growing awareness among consumers over this issue. That is why this research is dedicated to finding out how companies can communicate good intentions better; this way I hope consumers will buy more into eco-fashion.

1.2.3 CSR communication in the fashion industry

The impact of the fashion industry on the environment is well defined and is planet wide. This is why fashion companies are already highly involved in becoming more sustainable and are therefore creating departments concerning corporate social responsibility, hereafter CSR. These CSR departments develop Codes of Conduct in which they take responsibility for the way the garments are produced. However when the “Decent working conditions and fair competition” act was applied in the congress of the United States in 2009, the fast fashion companies all responded the same: we cannot have rules for this and this act would be an impediment of free trade. That’s why all these Codes of Conduct nowadays are voluntary and actually “are worth less than the paper they are written on” (Ross & Morgan, 2015). Now it is time for change. By clearly formulating these Codes of Conduct consumers will know what can be expected of companies and may actually believe that companies are keeping these promises.

1.3 Research gap

Firms are already spending lots of money on CSR activities. The definition of CSR is clearly stated in literature: CSR activities can be defined as ‘a commitment to improve (societal) well-being through discretionary business practices and contributions of corporate resources’ (Kotler & Lee, 2005, p. 3) which ‘occupies a prominent place on the global corporate agenda in today’s socially conscious market environment’ (Du, Bhattacharya & Sen, 2010, p. 8).

(11)

- 11 -

Furthermore, there is awareness for change rising in the fashion industry (Li, Zhao, Shi & Li, 2014); consumers want change. The literature is also clear about how message ambiguity influences consumers’ attitudes (Cotton Incorporated 2008; 2009). Lastly, research also proves the difference between positively and negatively framed messages (Gaeth, Gary, Levin, Cours & Combs. 1990; Levin & Gaeth, 1988; Maheswaran & Meyers-Levy, 1990; Smith, 1996). However, one gap can be identified in the CSR research field: the combination of the kind of message and message framing. The sort of message can be either ambiguous or concrete. Some research has already been conducted on this subject. Vague, abstract or ambiguous messages will make consumers more sceptical (Cotton Incorporated, 2008, 2009; Pickett-Baker & Ozaki, 2008). The opposite, clear formulated informational messages, have a positive effect on the consumer (Yoo & MacInnis, 2005; Yan et al., 2012). Next to that positive message framing tends to have a more positive effect on consumers (Maheswaran & Meyers-Levy, 1990; Tversky & Kahneman, 1981), whereas negative message framing will make consumers look more negatively towards a brand (Gaeth et al., 1990; Levin and Gaeth, 1988). In order to make consumers change their behaviour, companies need to make sure that they develop a positive attitude towards the brand. This research searches for what kind of message will get the consumer to make that change.

1.4 Research questions

Which type of sustainability message (ambiguous vs. concrete) results in positive consumers’ attitude towards a brand?

Is this relationship mediated by message framing (positive vs. negative message) and moderated by sustainability orientation?

The research is an exploratory and quantitative study. The focus of this experiment-based study will be on under what circumstances what kind of message should be used to result in positive attitude towards the brand. This study will focus on what kind of message, positive or

(12)

- 12 -

negative and abstract versus concrete, will lead to the highest outcome of these dependent variables.

This research is conducted in the form of an experiment. Data will be collected using the online survey tool Qualtrics. The type of message respondents will read will be

manipulated (concrete-positive, concrete-negative, positive and ambiguous-negative). After reading the manipulated CSR message, respondents are asked to evaluate fictive fashion company X in dependent variable brand attitude.

This research will be relevant for CSR communication literature since it researches the possible interaction effect of message ambiguity and message framing. It will be an extension of current literature in which interaction effects will be researched. Next to that, the outcomes of this research can be used to transfer the correct message to the consumer in the right way.

The primary goal of this thesis is to develop a theory about what framed messages can be used in a given situation to achieve a set outcome. The experimental use of different kinds of messages and their outcomes will contribute to CSR communication. A theoretical

framework will be developed in which it becomes clear how these variables are connected. Managers, for whom it is really important to understand how consumers interpret different message framing, can use this theoretical framework. Big fashion companies that are not sure how to communicate their CSR activities in a way that consumers do not get sceptical can learn from this thesis. These companies can apply the outcomes of this research in their marketing communication plan and will know if they have to use abstract or concrete and positive versus negative CSR messages.

This thesis starts with an extensive literature review. In this literature review, first CSR literature will be reviewed and the link to the fashion industry will be made. The focus will be on sustainability versus unsustainability in the fashion industry, clearly explaining all terminology. From the literature review, the gap in the literature will be identified and two

(13)

- 13 -

research questions to fill this gap are proposed. This chapter will be followed by a description of the research and data collection methods that were used. The data is also analysed in this chapter. After a description of methodology used, the fourth chapter will include the results following from the data. These results will be followed by a discussion in chapter five. This discussion will be followed by a concluding word about the research question.

(14)

- 14 -

2. LITERATURE REVIEW

This literature review will discuss different terms used in CSR with a specific focus on CSR communication in the fashion industry. First, the generic aspects of CSR will be discussed, followed by CSR communication and different communication styles. A company can send a positive or negative CSR message and can communicate in an ambiguous or concrete way. Furthermore, the literature review will look into sustainability versus unsustainability. Finally, this will point to a gap in the literature, which can be answered with the research question.

2.1 Corporate social responsibility

In the literature there still is confusion regarding a clear and unbiased definition of CSR (Dahlsrud, 2008). In this part of the literature review different definitions will be provided and combined into one clear definition that will be used in this thesis. First, three dimensions can be discerned in CSR: the environmental dimension (as regards the environment), the social dimension (the relationship between business and society), and the economic dimension (the financial aspects) ‒ compressed to Planet, People and Profit. Most definitions are biased because the authors tend to focus more heavily on one of these three dimensions (Dahlsrud, 2008). To get more into depth in this thesis, the following definition will be used: CSR is “a commitment to improve (societal) well-being through discretionary business practices and contributions of corporate resources” (Kotler & Lee, 2005, p. 3), which “occupies a prominent place on the global corporate agenda in today’s socially conscious market

environment” (Du et al., 2010, p. 8). The most comprehensive definition of CSR found in the literature, which includes all three dimensions (with an addition of stakeholder value and voluntary value) (Dahlsrud, 2008), is by the Commission of the European Communities: CSR is “a concept whereby companies integrate social and environmental concern in their business

(15)

- 15 -

operations and in their interaction with their stakeholders on a voluntary basis” (Dahlsrud, 2008, p. 7).

Companies use CSR to make a positive contribution both to themselves and the environment (Skarmeas & Leonidou, 2013). However, since companies also obtain positive outcomes themselves, consumers are often very sceptical about motives behind CSR. To achieve positive brand attitude from consumers, CSR should be communicated the correct way. Are the companies doing it for themselves or to really help the environment and the workers suffering in the factories?

2.2 CSR communication

Communication is the key for CSR to have the desired effect. Companies need to

communicate their CSR activities the right way to get their message across to consumers. Companies can earn substantial benefits from their consumers when engaging in CSR; however, companies have to take consumer scepticism into account. Consumers can

experience CSR activities as being either extrinsic, when it is increasing companies’ benefits such as profit, or intrinsic, in which consumers experience CSR activities as being genuine and for the good of the external environment (Du et al., 2010). According to Forehand and Grier (2003), intrinsic CSR activities have a higher positive effect on consumer behaviour towards the company than its counterpart. Research by Ellen, Webb and Mohr (2006) support this with findings that consumers respond in a positive way to CSR activities driven by value and strategy and in a negative way to CSR driven by stakeholders’ and egoistic motives. According to the same research a high fit between company and cause is essential; when a high fit between company and cause is experienced, the company is perceived as being more expert than in case of a low fit between company and cause. When consumers experience a low fit between company and cause CSR activities might even damage the company (Speed

(16)

- 16 -

& Thompson, 2000). Chaundri and Wang (2007) add that transparency in CSR is important to improve consumer loyalty. The more transparent a company is the higher consumers’ belief in its sincerity. Thus, to reduce consumer scepticism and increase consumer loyalty, companies must engage in intrinsic CSR activities and engage in causes that are experienced to be close to the company.

CSR activities have several communication outcomes. These outcomes can be divided into internal outcomes, like awareness and trust in the company, or external outcomes, like brand attitude, loyalty and intention to purchase. The outcomes mentioned above are

influenced by CSR communication factors and contingency factors (Du et al., 2010). In this thesis the focus will be on how the CSR activities are communicated to the consumers and, moreover, what kind of message will result in the most positive outcome for consumers’ attitude towards the brand.

Whereas the framework developed by Du et al. (2010) focuses more on what external aspects make CSR messages successful, this thesis focuses on the internal factors; which message elements make a CSR campaign successful?

(17)

- 17 - 2.2.1 Message positivity

Companies can communicate their CSR practises with a positive or a negative message. Messages framed in a positive way, which show the gains from using a product, have dissimilar outcomes and should be used in different situations than messages framed in a negative way, which shows the losses one suffers by not using a product (Maheswaran & Meyers-Levy, 1990).

Prospect theory, a theory concerning behaviour of humans when choosing between options involving risk and uncertainty (Kahneman & Tversky, 1979), can be used to explain this phenomenon: when people were asked to either save a fixed number of lives or lose a fixed number of lives, this resulted in people picking the risk averse gain of lives rather than the loss situation of saving lives (Tversky & Kahneman, 1981). So, people choose the safe option of “saving lives” rather than the risky option op “losing lives”. People respond

differently to the way the given information is presented: when there are lives to save people tend to avoid risk, but when lives are given as a loss people tend to take more risk (Rohtman, Salovey, Antone, Keough & Martin, 1993). So if there are lives that can be saved people will chose for a safe, non-risky option, and when there are lives that are “already lost” people will chose more risky options because they think of the loss of lives as a given. To better

understand this theory Levin and Gaeth (1988; Levin, 1987) conducted an experiment in which people had to choose from two options of beef: 75% lean compared to 25% fat. Here, the 75% option is positively framed whereas the 25% fat is negatively framed. The authors found that more people favoured the positive, healthy option. This theory can be used for this research since Puto (1987) states that messages tend to have more influence when described as benefits that are gained instead of benefits lost. It is in our nature to love gain more than loss (Puto, 1987). Prospect theory can also be used to explain contemporary message effects.

(18)

- 18 -

An example of a positive CSR message is Shoe Company TOMS’ One for One campaign in figure 2.

Figure 2 - TOMS One for one

This is an explicit example of a positive CSR campaign since it emphasizes that buying a pair of TOMS shoes contributes to society by giving: if you buy one pair of TOMS they give one pair away to an unprivileged child. WWF is an example of a brand which uses lots of negative messaging in their campaigns.

(19)

- 19 -

Figure 3 shows the Stop Illegal Wildlife Trade campaign. Pictured are different endangered animal species with main assets coloured in red blood spatters. It shows the consequences, the losses, if you do not donate to the WWF. TOM’s and WWF both have their own reasons for using positive or negative campaigning, for one the gains work better than the losses and vice versa. But when is what messaging the most effective?

Empirical research shows that overall positively framed messages have a more desirable outcome than their negative counterparts (Gaeth et al. 1990; Levin & Gaeth, 1988; Maheswaran & Meyers-Levy, 1990). However, there are situations possible in which a negative message has better outcomes. Smith (1996) found that people who had lower education levels were more strongly influenced by negative messages. The higher educated respondents were more strongly influenced by positively framed messages. This also suggests that message framing is dependent on the kind of product which is advertised for.

Advertisements that are positively framed have a higher impact on purchase decisions than negatively framed advertising for transformational, or attitude shaping products (Laschke, Hassenzahl & Diefenbach, 2011; Smith, 1996).According to Block and Keller (1995)

negative message framing can be more effective when more in-depth processing is necessary. Maheswaran and Meyers-Levy (1990) show evidence that negative message framing is also more effective when information is carefully evaluated. Since these are moderators, a variable which “affects directions and/or strength” (Baron & Kenny, 1986, p. 1174), these will be taken into account in the next section of the literature review. This research also looks for the difference in message framing when looking at brand attitude. Since this is a generic message, where the moderators and mediators, variables responsible for showing certain relations between variables (Baron & Kenny, 1986), are not yet involved, the prospect theory is applied:

(20)

- 20 -

H1: When companies communicate positive CSR messages this will result in higher brand attitude compared to negative messages.

2.2.2 Message ambiguity

Different companies and brands use different message framing. This is because different factors can moderate and mediate these attitudes. CSR activities are often communicated in an ambiguous or vague way. Ambiguous or vague communication means that messages are formulated in a non-specific and imprecise way (Cutting, 2007). Lipman (2008) uses the term “tall” as a vague concept. He states that the term tall will mean something different to

everyone. Perhaps for me a man of 1,80 meters is considered tall, while my father (who is 1,95 meters) would think this same man is rather small. According to previous research, messages that are formulated in a vague way and are therefore not clear create confusion and concern with the consumer and thereby lowers their attitude towards the brand (Cotton Incorporated, 2008; 2009). Although buying brands that have a sustainable image will make people feel more ethical (Niinimäki, 2010), at the same time they have trouble identifying the brands that are truly environmental friendly (Pickett-Baker & Ozaki, 2008). Because CSR is a concept that consumers are already sceptical towards (Du et al., 2010; Kim & Lee, 2009), it is expected that the more ambiguous the CSR message is, the higher the scepticism of the consumer, which will lead to negative brand attitude.

Opposed to ambiguous messages are concrete messages. Concrete messages, or informational messages, communicate objective and clear information to the consumer (Yoo & MacInnis, 2005). Concrete information is to say the same man, who was considered tall by me, is 1,80. A clear number is used now, which will mean the same to everyone. Also named explicit messages, these concrete messaged contain clear structured facts about the

companies’ CSR activities; according to Yan et al. (2012) they will have a more positive effect on the brand loyalty of consumers. This is because the message is better understood and

(21)

- 21 -

there is less confusion among consumers. When CSR messages are really concrete, it is expected that consumers will have lower doubts regarding the company and therefore consumer scepticism will decrease and brand attitude will increase. Cotton Incorporated (2008, 2009) also supports this claim; market researchers argue that claims regarding

environmental-friendly production that are clearer and provide more information will have a more positive response from consumers:

H2: When companies communicate ambiguous CSR messages this will result in low brand attitude compared to concrete messages.

The research by Maheswaran and Meyers-Levy (1990) and Celsi and Olson (1988) illustrates that the outcome of positive versus negative message framing depends on whether people engage in the message in detail or not; positive messages are more effective when the

involvement is low whereas negative message framing is more effective when involvement or processing is high. This is also known as “taking the central route” in the elaboration

likelihood model or elaboration likelihood model (Petty & Cacioppo, 1983). The elaboration likelihood model is a generic framework which explains the effectiveness of persuasive communication (Jones, Sinclair & Courneya, 2003). The elaboration likelihood model proposes that the change of attitude because of a difference in message strength (strong message versus a weak message) depends on the likelihood that this person will elaborate on the message, or think about it more. “Elaborating” means thinking about something really hard, so really putting time and effort into it. The elaboration likelihood model asks the question: “how likely are you to really think hard and deep about a situation?” (Jones et al., 2003). In the elaboration likelihood model there are two main routes: the peripheral route, in which you pay attention to the non-important aspects, or the central route, in which you think hard and long about everything. Zhang and Buda (1999) supplement the theory by adding the need for cognition as an important part of involvement as a moderator of message framing.

(22)

- 22 -

They propose that how someone reacts to a framed message, positive or negative, depends on the individual’s ability and motivation to process a message. When involvement is low, i.e., the motivation to process the advertisement and the message is low, the peripheral route is taken. Need for cognition includes the ability and enjoyment of thinking deep and hard about messages received. People that participated in the research and had a low need for cognition, i.e., who took the peripheral route, were more strongly influenced by the negatively framed message. Block and Keller (1995) support this by proving that negative message framing is indeed more effective when receivers process messages in depth.

An ambiguous message is vague and not clear. Because an ambiguous message is not clear, you cannot really elaborate on it. Here, the consumer takes the peripheral route.

According to Maheswaran and Meyers-levy (1990) and Celsi and Olson (1988) a consumer who takes the peripheral route after experiencing an advertisement, is more strongly

persuaded by a positive message. Therefor it is expected that the combination of an

ambiguous and positive message has a more positive effect on brand attitude compared to an ambiguous and negative message:

H3a: When institutes communicate an ambiguous message, brand attitude will be higher when this message is also positive (compared to negative).

With a concrete message the consumer more often takes the central route, since the message is explicit and clear. With a concrete message, the consumer can really elaborate on the information given. According to Maheswaran and Meyers-Levy (1990), negatively framed messages are more persuasive to consumers when detailed processing is required. Therefore it is expected that, according to the theories stated above (Maheswaran & Meyers-Levy, 1990; Celsi & Olson, 1988), concrete messages will lead to higher brand attitude in combination with a negative framing:

(23)

- 23 -

H3b: When institutes communicate a concrete message (low ambiguity), brand attitude and will be higher when this message is negative compared to positive.

2.3 Sustainability orientation

Sustainability is an “emerging megatrend” (Lubin & Esty, 2010). Since there are different terms for a sustainable way of consuming goods, this section will focus on different views on sustainable (Maignan, 2001), mindful (Sheth, Sethia & Srinivas, 2011) and ethical

consumerism (Carrigan & Attalla, 2001). Since CSR is “a concept whereby companies

integrate social and environmental concern in their business operations and in their interaction with their stakeholders on a voluntary basis” (Dahlsrud, 2008, p. 7), people who are more concerned about social and environmental effects will react differently than people who are not concerned about these aspects. Therefore another independent variable is introduced: sustainability orientation. Sustainability orientation will measure how social and

environmental organized people are (Maignan, 2001). People who are sustainability oriented are influenced in a different way by CSR related messages than people who are not since these messages include a lot of stimuli concerning environmental or societal issues. Another term used for sustainable consumption is “mindful consumption” (Sheth et al., 2011). The authors propose that mindful consumers care for themselves, for the community and for nature. The mindful consumer is opposed against repetitive overconsuming: ethical consumption. So, ethical consumers are more involved with the social responsibilities. As proposed before by Petty and Cacioppo (1986), the elaboration likelihood model states that behaviour is influenced by this involvement. According to Maheswaran and Meyers-Levy (1990) and Celsi and Olson (1988) positive messages are more effective when involvement is low whereas negative message framing is more effective when involvement or processing is high.

(24)

- 24 -

So negative messages are more effective for people with high involvement, in this case people that are more oriented towards sustainability. Positive messages will be more effective for people with low involvement, which in this case will be the less sustainability-oriented consumers:

H4a: More sustainability-oriented consumers have a higher brand evaluation when they read a negative message compared to reading a positive message.

H4b: Less sustainability-oriented consumers have a higher brand evaluation when exposed to a positive CSR message compared to being exposed to a negative message.

2.4 Conceptual model

The literature review describes CSR, CSR communication and sustainability orientation. The hypotheses contain multiple independent variables: message ambiguity, message positivity and sustainability orientation. This study is designed as an experiment. It investigates the relation between and/or moderating effect of several independent variables. The relation between independent, moderating, mediating and dependent variables are presented in the conceptual model below.

(25)

- 25 -

3. METHODOLOGY

This chapter will describe the methodology used in the research. First the general methods will be described including why these specific methods were chosen. Second the survey, pre-test, final experiment and the sample will be discussed.

3.1 General methods

This research is an exploratory study. The approach started with reporting on what has

already been studied and proven on this topic (Stebbins, 2001). Exploratory research is a brief way to get to the real information and will compare different sides of the literature (Stebbins, 2001). This study carried out a quantitative research design to test the hypothesis. The choice for quantitative research over qualitative research is based on the failing of scientific

adequacy of the latter (Sandelowski, 1986). To control certain variables and therefore measure the effect of independent variables, the choice for an experiment was made.

The study hence involves a 2 (ambiguous message versus concrete message) x 2 (positive versus negative message framing) x sustainability orientation in between design experiment. The experiment is an online survey with data collection from Qualtrics. This study was designed to test what type of CSR messages result in the highest attitude towards brand. The prediction was that a concrete message results in higher attitude towards brand than an ambiguous message, a positive message results in higher attitude towards brand than a negative message, attitude towards brand being the way the consumer feels about a brand measured on five different scales (Spears & Singh, 2004). All participants are randomly assigned to one of four groups: they are subjected to either an ambiguous x positive,

ambiguous x negative, concrete x positive or concrete x negative CSR message written about a fictional fashion company X. The fashion company is fictional to ensure that respondents don’t have any bias towards the company. After the respondents are shown the CSR message,

(26)

- 26 -

they answer five questions concerning attitude towards the brand. The goal of the research is to find out what combination of messages works best for this fictive fashion company. After answering the five questions concerning attitude towards the brand, the effect of sustainability orientation had to be measured. Therefore respondents were asked to fill in five questions regarding sustainability orientation. These questions are asked at the end of the survey so as to not prime the respondents with sustainability subjects.

3.3 Pre-tests

Two pre-tests were conducted before the main survey. The first pre-test was to investigate if all CSR messages are understood the way they are supposed to. A second pre-test was used to check if the final survey was clear to and understood correctly by respondents.

3.3.1 Message pre-test

First all respondents read an introductory text about the terminology used in the short questionnaire. Here the researcher explained the terms “ambiguous”, “concrete”, “positive” and “negative” message framing. A total of 41 respondents were all shown one of the CSR messages that were created for this study, after which they had to rate message abstractness on a Likert scale of 5 (1 being very ambiguous and 5 being very concrete) and message framing on a Likert scale of 5 (1 being very negative and 5 being very positive).

Unfortunately the first pre-test did not have the expected outcome. Message 1, meant as ambiguous and positive, had an M = 2.79 (N = 14) and was experienced as being more

concrete than ambiguous. The M = 3.57 showed that the message was experienced as positive. Message 2, ambiguous‒negative, was likewise understood as more concrete (M = 3.44, N = 9). The message framing however was correct here with a mean of 2.00. Both messages 3 and 4 were experienced as being concrete (message 3: M = 3.57, N = 7 and message 4: M = 3.09,

(27)

- 27 -

N = 11). Message 3 was correctly understood as being positive (M = 4.29) whereas message 4 was understood as neither positive nor negative (M = 2.45).

After these results the researcher interviewed two respondents about the outcomes of the pre-test. These two respondents got to see all four texts to compare. The interviewees were then asked how they would change these texts to make them more reliable. To both interviewees it was very clear, after comparing messages, which one was concrete and which one was ambiguous. The example of using the name Thi in the message made it very clear what would be saved or lost when buying a product of Fashion Company X. After talking to the second interviewee the last parts of the concrete messages were rephrased. However, for the experiment to be executed the ambiguous messages needed to be rephrased so that these were not interpreted as concrete. For this the researcher held a group discussion with another two respondents. In this group discussion the researcher asked about their opinion on the messages and together they came up with new messages which could be pre-tested. The final pre-test was tested for N = 31. All 31 respondents were randomly assigned to one of four messages. So these respondents either read a concrete‒positive, concrete‒negative,

ambiguous‒positive or ambiguous‒negative message. After reading the message they were asked: “To what extent do you think this message is ambiguous or concrete (1-5)” and “To what extent do you think this message is negative or positive (1-5)”.

The data for the pre-test was analysed in SPSS. First, all respondents were ordered according to the message group they were in. Both variables, message ambiguity and message positivity were recoded. For message ambiguity the ambiguous message was given the value 1.00 compared to concrete being coded .00. Message positivity was also coded with 1.00 being the positive message and .00 being the negative message. This way four groups were created: 1.00 – .00 (ambiguous – negative), 1.00 – 1.00 (ambiguous – positive), .00 – .00

(28)

- 28 -

(concrete – negative) and .00 – 1.00 (concrete – positive). In table 1 the descriptive statistics are shown.

N Mean SD

Message Ambiguity .00 concrete 14 4.00 .277

1.00 ambiguous 17 1.88 .342

Message Positivity .00 negative 15 1.53 .516

1.00 positive 16 4.25 .775

Table 1 - descriptive statistics

Table 1 already shows that the messages are interpreted as intended. The ambiguous message is close to 1 = very ambiguous, whereas the concrete message is evaluated more towards 5 = very concrete. Next to that the negative message is close to 1 = very negative and the positive message is evaluated close to 5 = very positive. To check the significance of the messages a one-way ANOVA was executed. First, Levene’s statistic of Homogeneity of Variances was not significant (p=.292), so the ANOVA could be executed. Table 2 shows the outcome of the ANOVA. The ANOVA was used to compare the means of the abstract and concrete and positive and negative groups. The results will determine if the messages differ significantly and can be used for the final research.

SS df MS F Sig.

Message Ambiguity Between Groups 32.429 1 34.429 21.817 .000* Within Groups 45.765 29 1.578

Total 80.194 30

* p <.05

Table 2 – one-way ANOVA

From table 2 we can conclude that there is a significant difference between the ambiguous (1.00) and concrete (.00) messages. Second, another one-way ANOVA was performed to

(29)

- 29 -

investigate if there is a significant difference between positive and negative message framing. For this ANOVA Levene’s statistic for Homogeneity of Variances was not significant

(p=.629) so the ANOVA could be performed. The results are shown in table 3.

SS df MS F Sig.

Message Positivity Between Groups 57.138 1 57.138 130.130 .000* Within Groups 12.733 29 .439

Total 69.871 30

* p < .05

Table 3 - one-way ANOVA

Table 3 shows that the positive and negative message also differ significant (p=.000).After these two ANOVA analyses, conclusion is that the messages created for the research stand up to the standards and can be used in the final research. These messages are shown in appendix 1.

3.2.2 Survey pre-test

The second pre-test was conducted among ten respondents to test whether respondents were able to understand the questions asked in the online experiment. Ten respondents were asked to answer the questions asked in the survey while the researcher was in the room. The

respondents helped the researcher with perfecting the spelling and formulating the questions the right way. The survey was clear to the respondents the first time. The researcher did decide to change the final page with a clear statement (bold and in red letters) that respondents had to click to the final page to finalize their response.

(30)

- 30 - 3.3 Experiment

After the pre-test finalized the messages used for the experiment, this section will describe the method of the experiment. The experiment was an online survey on Qualtrics.

The survey started with a welcome page which introduced the research. Here respondents were asked to tick the box if they “understand the above text and agree to participate with the research”. The second page shows one out of four CSR messages (as described in appendix 1) randomly. So, here respondents are assigned to one of four

conditions. The third page shows the manipulation check. Here respondents are asked to name the fashion company used to check whether they read the message in detail. Respondents could pick between Fashion Company X, Y and Z in random order. If respondents did not give Fashion Company X as answer, data was not used. Next, brand attitude was measured. Respondents were asked to evaluate their feelings about Fashion Company X. The fifth page of the survey was assigned to measure purchase intention. Respondents were asked to fill in if they considered buying a product from Fashion Company X. After purchase intention,

sustainability orientation was measured. Here respondents were asked to describe their feelings about five statements (Maignan, 2001). Finally, additional demographic information such as gender, age and country of origin was asked.

3.4 Variables

The conceptual model and online survey makes use of different variables. These variables will be discussed in this section.

3.4.1 Message positivity

Independent variable message positivity has two levels: the message is either positive or negative. To shape the messages for the pre-test, literature is used. Messages framed in a

(31)

- 31 -

positive way show the gains of using a product, whereas messages framed in a negative way show the losses one suffers by not using a product (Maheswaren & Meyers-Levy, 1990). Two options, either saving children from factories, or leaving kids to work in factories, represent these frames in the fictional CSR messages. The direct effect of message framing will be tested as well as the mediating effect of message framing on message abstractness.

3.4.2 Message ambiguity

The second independent variable message ambiguity has two values: the message is either ambiguous (vague) or concrete. If the message is ambiguous or concrete is tested in the first pre-test. Important for composing the message texts is literature about these concepts. Ambiguous, vague or abstract messages are formulated in a vague way, are not clear and create confusion with consumers (Cotton Incorporated, 2008; 2009). This was represented in the message text by not being clear about what is exactly gained or lost when buying or not buying a product of fashion company X. According to Yoo and MacInnis (2005) concrete messages communicate objective and clear information to the consumer. This was represented in the message text by using a clear description of what exactly the consumer will gain or lose by buying a product by fashion company X. The difference in ambiguity is represented in the message text by the difference between rescuing unnamed children versus recuing one specific child named Thi, aged 9.

3.4.3 Sustainability orientation

The third independent variable in the conceptual model is sustainability orientation. This study tests if sustainability orientation moderates the effect of message framing. To measure this variable the scale of Maignan (2001) was used. Maignan (2001) uses five items on a seven-point Likert scale. This scale ranged from absolutely disagree (1) to absolutely agree

(32)

- 32 -

(7). An example of a statement used by Maignan (2001) is “I avoid buying products from companies that have engaged in immoral actions”.

3.4.4 Attitude towards the brand

The dependent variable of the conceptual model is attitude towards the brand. Attitude towards the brand is a common way to measure what people think about a brand and how different independent variables influence this. This dependent variable is measured according to Spears and Singh’s (2004) framework. Spears and Singh (2004, p. 62) established a way to measure attitude towards brand by asking one question: “Please describe your overall feelings about the brand described in the ad you just read”. Respondents are then asked to evaluate the brand on five seven-point Likert scales: unappealing – appealing, bad – good, unpleasant – pleasant, unfavourable – favourable, unlikeable – likable.

3.3.5 Purchase intention

To obtain additional results, another dependent measure is used in the research. Purchase intention can be measured according Spears and Sing’s (2004) framework and was therefore easily added to the model and the research. Purchase intention is also measured on a seven-point Likert scale with five items. Here the question was rephrased: “Please describe your feelings about Fashion Company X, described in the CSR message you just read. Would you buy a product from Fashion Company X?” Respondents were asked to give their purchase intention on five aspects: never – definitely, definitely not intend to buy – definitely intend to buy, very low purchase interest – very high purchase interest, definitely not buy it – definitely buy it and probably buy it – probably not buy it. The last question was counter-indicative, to catch the acquiescence bias, which are people who answer strictly on the positive or the

(33)

- 33 -

negative side. This scale was not used in subsequent research since results were the same as brand attitude.

3.3.6 Manipulation check

The experiment also contained a manipulation check, where after reading the CSR message the respondents were asked a question to check whether they had read the text carefully. Respondents who did not give a correct answer to the question what Fashion Company was mentioned in the text (choose from randomized ordered answers X, Y or Z) were excluded from the research.

3.5 Sample

In total 224 respondents filled in the survey. However after cleaning the responses based on incomplete data the sample amounted to 196 complete respondents. The smallest condition was group 2, concrete–positive, which consisted of 45 respondents. This meets the minimum requirement of 40 respondents1. Use was made of the own network of the researcher and used Amazon Mechanical Turk because of time pressure. Here 100 respondents were bought for €16,40, which made the sample more diverse in sense of age and country of origin. The data now includes other nationalities like Indians and Americans, instead of merely Dutch

participants2.

The largest parts of the respondents were from the Netherlands (N = 83) and India (N = 56). Other countries of origin included in the research were the United States, Italy and the United Kingdom. The average age of the respondents was 31.2. The gender of the

respondents was 56% being male and 44% female.

1_{Rule of thumb by Joris Demmers} 2_{Elaboration in discussion (Chapter 5)}

(34)

- 34 - 4. RESULTS

In order to calculate the results, the data had to be recoded. One question, concerning

purchase intention, was counter-indicative so this was recoded into 1=7, 2=6, 3=5, 4=4, 5=3, 6=2 and 7=1. This is to catch the acquiescence bias, or simply said people who answer strictly on the positive or negative side of the questionnaire and will result in higher validity

(Schriesheim & Hill, 1981). Next three new variables were created to measure Brand Attitude, Purchase Intention and Sustainability Orientation. To do so, the mean of the five questions per variable were calculated. As said before, first data was organized so that all respondents that read the same message were aligned. Afterwards two extra columns with variables were added. First all respondents were divided in groups: 1, 2, 3 and 4. Here group 1 consists of respondents reading an ambiguous – positive message, group 2 an ambiguous – negative message, group 3 a concrete – positive message and group 4 a concrete - negative message. After dividing the respondents in groups, two new variables were created: Group Ambiguity and Group Positivity. Both columns were provided with a 1 or a 0. This resulted in each group having their unique code (0-0, 0-1, 1-0, 1-1, see table 4), which can later be used for the Factorial ANOVA.

Positive (1) Negative (0)

Ambiguous (1) 1.00-1.00 1.00-.00

Concrete (0) .00-1.00 .00-.00

Table 4 – coding messages 4.1 Statistical analysis

This section is dedicated to an analysis of all data exported from Qualtrics to SPSS. To find general relations between the variables, first a correlation matrix with five variables is analysed. Second, to find possible mediating and moderating effects, several ANOVAs will be performed to compare means. This section will be closed by testing all hypotheses from the literature review.

(35)

- 35 - 4.1.1 Reliability test

To check the scales used in the research, derived from literature, a reliability test was conducted. Here, three scales were measured: brand attitude, sustainability orientation and purchase intention. First, the brand attitude scale had high reliability, with Cronbach’s Alpha = .948. The corrected item-total correlation indicates that all the items have a good correlation with the total score of the scale; these are all above .30. Next to that, deleting one of five items would not substantially affect reliability. Second, the scale of sustainability orientation also had high relatability with a Cronbach’s Alpha of .771. All corrected item correlations are above .30, which indicates that all five items have a good correlation with the total score of the scale. If one of the items was deleted, the Cronbach’s Alpha would be below .771, so deleting an item would not affect reliability of the scale. Last, the purchase intention scale was also tested. This scale was also reliable: Cronbach’s Alpha = .816. All corrected item correlations are above .30, which indicates that also this scale has good correlation with the total. This was the scale were the counter-indicative question was used. The results show that removing this question would improve the scale by more than .10 (from .816 to. .941). So removing the last question would improve the scale. However, purchase intention was not used as dependent variable in the end, since it had the same outcome as brand attitude and was therefore double.

4.1.2 Pearson correlation matrix

In the correlation matrix, all variables interesting for research are compared. The second dependent variable, Purchase Intention, and the independent variable Sustainability

Orientation were also added. Table 5 shows the correlation between these variables; it also includes the descriptive statistics mean (M) and standard deviation (SD).

(36)

- 36 -

M SD 1 2 3 4 5

1. Ambiguity group .5051 0.50125 1

2. Positive group .5204 0.50068 0.051 1

3. Attitude towards Brand 4.401 1.45489 -0.075 .277** 1

4. Purchase Intention 4.3908 1.13613 -0.053 .196** .762** 1

5. Sustainability Orientation 5.4551 1.06189 0.067 0.04 .230** .247** 1

**Correlation is significant at the .01 level (2-tailed)

Table 5 – Pearson correlation matrix

From the correlation matrix it can be concluded that there is a significant effect between being in a positive or negative group and brand attitude. Which of these variables is responsible for this effect will come out of later analyses. Second, there is some significant effect between being in a positive or negative group and purchase intention. The effect of message positivity on brand attitude is bigger (.277>.196) compared to the effect on purchase intention. Third, a high correlation (.762>,0.53) can be seen between the two dependent variables brand attitude and purchase intention. Since measurement of these two variables is based on the same

literature of Spears and Singh (2004) they more or less measure the same things. To conclude, a significant correlation is found between sustainability orientation and the two dependent variables. This means that the higher sustainability orientation, the higher the purchase intention. The same holds true for sustainability orientation and attitude towards brand, since these dependent variables measure the same thing.

4.1.3 Hypotheses testing

The following section will explain the analysis done to test the hypotheses. 4.1.2.2 Hypothesis 1

H1: When companies communicate positive CSR messages this will result in higher brand attitude compared to negative messages.

3_{Rule of thumb by Sofija Pajic}

(37)

- 37 -

To test hypothesis 1 a Factorial ANOVA was run in SPSS. Here we can find the main effects of message ambiguity and message positivity and the possible interaction effect of message ambiguity and message positivity measured on brand attitude. First, Levene’s test of

Homogeneity of Variances was conducted. The results were not significant (p = .07) for brand attitude, however for purchase intention Levene’s test for homogeneity was significant (p = .000). Table 6 shows that there is a significant difference in means of brand attitude whether respondents read a positive or negative message (p = .000).

SS DF MS F N2 Sig. Message ambiguity 3.479 1 3.479 1.78 .009 .184 Message positivity 32.403 1 32.403 16.577 .079 .000** M ambiguity * M positivity 2.654 1 2.654 1.358 .007 .245 Error 375.295 192 1.955 Total 4744.28 196

**Significant at the p<.05 level

Table 6 – factorial ANOVA

However, which message is evaluated higher is not shown in either of these tables. Since the analysis consist of two levels of the variable a means plot can be made to see the differences in means. This plot is presented in figure 5.

(38)

- 38 -

Figure 5 shows a clear difference how the two groups evaluated Fashion Company X. The means plot shows that the attitude towards Fashion Company X is lower for group .00 (who read a negative message) compared to respondents from group 1.00 (who read a positive message). This indicates that respondents who were presented with a positive message evaluated Fashion Company X higher (measured in brand attitude and purchase intention) compared to respondents presented with a negative message and that this effect is significant. There was a statistically significant effect of message positivity on brand attitude, F(1, 192) = 16.577, p < .05, η²=.079.

4.1.2.3 Hypothesis 2

H2: When companies communicate ambiguous CSR activities this will result in low brand attitude compared to concrete messages.

To test hypothesis 2 we used the same Factorial ANOVA used to test hypothesis 1. First, table 7 concludes the difference in mean between respondent who read a concrete (.00) versus an ambiguous (1.00) message.

Table 7 – descriptive statistics

ambGroup posGroup M SD N .00 (concrete) .00 (negative) 4.5224 1.52743 49 1.00 (positive) 5.1042 1.40394 48 total 4.8103 1.48895 97 1.00 (ambiguous) .00 (negative) 4.0222 1.57839 45 1.00 (positive) 5.0704 1.07328 54 total 4.5939 1.42010 99 Total .00 4.2830 1.56395 94 1.00 5.0863 1.23369 102 total 4.7010 1.45489 196

(39)

- 39 -

It is true that respondents who read the concrete message (M = 4.8103) evaluated Fashion Company X higher compared to respondents who read the ambiguous message (M = 4.5939). However table 6 shows that this difference is not significant (p=.184). The same can be concluded from table 5: there was no statistically significant effect of message ambiguity on brand attitude, F(1, 192) = 1.78, p = .184, η²=.009.

H3a: When companies communicate an ambiguous message, brand attitude and purchase intention will be higher when this message is also positive (compared to negative).

H3b: When companies communicate a concrete message (low ambiguity), brand attitude and purchase intention will be higher when this message is negative compared to positivity.

To test hypotheses 3a and 3b we will analyse the descriptive statistics from the Factorial ANOVA, in which the interaction effects between the different levels of message ambiguity and message positivity are presented, and the descriptive statistics are presented in table 8.

ambGroup posGroup M SD N .00 (concrete) .00 (negative) 4.5224 1.52743 49 1.00 (positive) 5.1042 1.40394 48 total 4.8103 1.48895 97 1.00 (ambiguous) .00 (negative) 4.0222 1.57839 45 1.00 (positive) 5.0704 1.07328 54 total 4.5939 1.42010 99 Total .00 4.2830 1.56395 94 1.00 5.0863 1.23369 102 total 4.7010 1.45489 196

(40)

- 40 -

First, for hypothesis 3a we see that if a message is ambiguous (1.00) brand attitude will be higher when respondents read a positive message (M = 5.0704) compared to reading a negative message (M = 4.0222). However, in table 7 we find that this effect is not significant (p = .245). Hypothesis 3a is rejected. Second, hypothesis 3b can be rejected by looking at table 8 itself. Here we find that when a message is concrete (1.00) brand attitude is higher when the message is positive (1.00) (M = 5.1042) compared to when the message is negative (M = 4.5224). So hypothesis 3a is correct yet not significant, whereas hypothesis 3b is neither correct nor significant. There was a non-significant interaction effect of message ambiguity on message positivity F(1,192) = 2.654, p = .245, η2 = .007.

H4a: Sustainability-oriented consumers have a higher brand evaluation when they read a negative message compared to reading a positive message.

H4b: Less sustainability-oriented consumers have a higher brand evaluation when exposed to a positive CSR message compared to reading a negative message.

Sustainability orientation was measured with five factors on a seven-point Likert scale. First, the mean was extracted as an average derived from these five questions. Second, the total sample was divided in two groups: oriented consumers and less sustainable-oriented consumers. To make this distinction between these two groups the researcher looked at the median of the seven point Likert scale. Since respondents could evaluate their

sustainability the lowest at one and the highest at seven, four is the median in the Likert scale. So, respondents replying under a score of four were acknowledged as being “less sustainable” whereas respondents replying above a score of four were seen as “more sustainable”. This makes the variable nominal and dichotomous, and ready for analysis. To analyse hypotheses

(41)

- 41 -

4a and 4b, a Factorial ANOVA was run in SPSS. First, the Levene’s test of Homogeneity of Variances was not significant (p = .05), so the ANOVA was valid. First, the descriptive statistics from the ANOVA will be analysed. These results are shown in table 9.

susGroup posGroup M SD N

.00 (less sustainable) .00 (negative) 3.5500 1.51846 8 1.00 (positive) 4.8857 1.24288 7

total 4.1733 1.51349 15

1,00 (more sustainable) .00 (negative) 4.3807 1.57809 83 1.00 (positive) 5.1370 1.24209 92

total 4.7783 1.45736 175

Total .00 4.3077 1.58249 91

1.00 5.1192 1.23748 99

total 4.7305 1.46689 190

Looking at hypothesis 4a and table 9 conclusion is that when a person is more sustainable oriented (1.00) brand attitude is higher when this person reads a positive message (M = 5.1370) compared to reading a negative message (M = 4.3807). Therefore the conclusion can be drawn that hypothesis 4a is not correct. Hypothesis 4b states that less sustainable

consumers (.00) have a higher brand attitude when exposed to a positive message compared to a negative message. Table 9 shows the mean in brand attitude for both positive and

negative messages. We conclude that less sustainable consumers (.00) evaluate Fashion Brand X higher when reading a positive message (M = 4.8857) than when reading a negative

message (M = 3.5500). However, from the descriptive statistics we cannot conclude whether this effect is significant.

(42)

- 42 -

Table 10 – factorial ANOVA

Table 10 presents the outcomes of the Factorial ANOVA when looking at the interaction between sustainability orientation and message positivity. There was a non-significant

interaction effect of sustainability orientation on message positivity F(1,186) = .581, p = .447, η2 = .003. So hypothesis 4b is rejected. SS DF MS F N2 Sig. susGroup 4.206 1 4.026 2.024 .011 0.157 posGroup 15.050 1 15.050 7.566 .039 0.007 susGroup*posGroup 1.155 1 1.155 .581 .003 0.447 Error 370.012 186 1.989 Total 4658.480 190

(43)

- 43 -

5. DISCUSSION AND CONCLUSION

This chapter will first summarize the results from the analysis. Second, findings will be discussed for each hypothesis. This will be followed by the practical implication of this research. The chapter will be concluded with limitations and topics for further research.

5.1 Summary of results

The purpose of this study was to research the effects of message ambiguity, message

positivity and sustainability orientation on brand attitude for a fictive fashion company X. The research questions “Which type of sustainable message (ambiguous vs. concrete) influences consumers’ attitude towards a brand? Is this relationship mediated by message framing (positive vs. negative message) and moderated by sustainability orientation?” were examined for 196 respondents. By manipulating CSR messages into four different groups, four

hypotheses were tested. Across all studies one significant effect was found; that of message positivity. First, there was no significant effect found of message ambiguity on brand attitude. Respondents’ brand attitude was higher when reading a concrete message compared to

reading an ambiguous message. This effect was predicted. However, this effect was not significant. Second, respondents experienced higher brand attitude when the message was framed in a positive way compared to a negative way. This was also a predicted effect. Consumers have the highest brand attitude when reading a positive message. Third, there was no interaction effect between message ambiguity and message positivity. One hypothesis was predicted falsely. It was predicted that ambiguous messages would result in higher brand attitude combined with a positive message whereas concrete messages had a higher outcome combined with a negative message. Here the overall effect of message positivity is clear again; both concrete and ambiguous messages result in higher brand attitude when combined with a positively framed message. Lastly, there is no interaction between sustainability