University of Groningen
The effects of message framing and nature imagery on
environmentally friendly purchase intentions
Master’s Thesis Marketing Management
Author: Nicole Biesma
Address: Nieuweweg 40-11, 9711 TH Groningen Email: N.Biesma@student.rug.nl
Phone number: +31646477076 Student number: S2735415
Department: Faculty of Economics and Business Master: Marketing Management
Supervisor: Dr. J.I.M. de Groot Second supervisor: Dr. M. Keizer
Abstract
There is an increasing concern with environmental issues, which shifts consumers’ lifestyle into a more environmentally responsible lifestyle. People care more about reusing products, recycling, and buying more sustainable products; hence, they want to engage in more environmentally friendly consumption. To increase the purchase of environmentally friendly products, the present study aimed to examine the effect of using message framing (gain-framing versus loss-framing) and imagery (nature versus neutral) in advertising sustainable products. A 2 by 2 between-subject experimental design (N=323) showed that using a gain-framed message in an advertisement was not more effective than a loss-framed message to encourage buying intentions of a sustainable washing detergent (Hypothesis 1). Also, an advertisement including a background of nature imagery was not more effective to change one’s purchase intention than a neutral background (Hypothesis 2). Finally, although the purchase intention of a gain-framed message combined with a nature image was higher than a gain-framed message combined with a neutral background, and the purchase intention of a loss-framed message combined with a neutral background was higher than a loss-framed message with a nature background, there were no significant differences between the four conditions on the purchase intention of the advertised product (i.e., no congruency effect; Hypothesis 3). These results were further validated by controlling for potential confounding variables (gender and education level). Furthermore, this study examined the effect of message framing and nature imagery on alternative outcome variables that have been used in prior research to investigate the effectiveness of advertisements, that is, attitude towards the advertisement, and attitude towards the brand. The main conclusions remained largely the same when using these alternative measures. An exception was that using nature imagery as a background in the advertisement resulted in significantly stronger attitudes towards the advertisement and the brand that using a neutral background. The findings and implications of the present results are discussed and integrated in line with three important streams in the field of green marketing and communication, that is, message framing and prospect theory, theories about the restorative effect of using nature imagery, and, the congruency effect.
Table of contents
1. Introduction ... 5
2. Literature background ... 8
2.1 Purchase intention of environmentally friendly products ... 8
2.2 Using message framing to promote purchase intentions of environmentally friendly products ... 8
2.3 Using nature imagery in advertisements to promote the purchase of environmentally friendly products ... 12
2.4 Message framing in combination with nature imagery to promote environmentally friendly purchase intentions: Congruency effect ... 15
2.5 Present study ... 18 3. Methodology ... 19 3.1 Participants ... 19 3.3 Pre-test ... 20 3.4 Materials ... 22 3.4.1 Product ... 22
3.4.2 Manipulation of independent variables ... 22
3.5 Procedure ... 23
3.6 Measures ... 24
3.6.1 Dependent variable: Purchase intention ... 24
3.6.2 Manipulation check: Message framing and imagery ... 25
3.6.3 Alternative depending variables: Attitude towards the brand and attitude towards the advertisement ... 25
3.6.4 Confounding variables: Gender and education level ... 26
3.6.5 Socio-demographics ... 27 3.7 Plan of analysis ... 27 3.7.1 Factor analysis ... 27 3.7.2 Reliability analysis ... 28 3.7.3 Manipulation check ... 28 4. Results ... 30 4.1 Manipulation check ... 30
4.2 Reliability and validity ... 31
4.3 Hypothesis testing ... 33
4.4 Overview of the hypotheses: supported or not supported ... 35
4.5 The influence of message framing and imagery on purchase intentions: Further validation of findings ... 35
4.5.1 The influence of message framing and imagery on attitude towards the advertisement and brand. ... 36
5. Discussion and conclusion ... 39
5.1 Conclusion ... 39
5.2 General discussion ... 39
5.3 Contributions ... 42
5.4 Limitations and future research ... 42
References ... 45
Appendices ... 52
Appendix 1. Pre-test images ... 52
Appendix 2. Advertisements ... 57
Appendix 3. Survey in Dutch ... 60
Appendix 4. Constructs ... 72
1. Introduction
There is a growing international concern about climate change and environmental sustainability (Nidumolu, Prahalad, & Rangaswami, 2009). The increasing awareness on the environmental issues shifts consumers’ lifestyle into a more environmentally responsible one (Cherian & Jacob, 2012). People care more about reusing products, recycling, and buying more sustainable products. Hence, they want to engage in more environmentally friendly consumption. Environmental consumption entails “those consumption activities that have a less negative or more positive effect on the natural environment than substitutable activities’’ (Pieters, 1991, p.59). One example of environmental consumption is the purchase of environmentally friendly products.
The stronger emphasis on environmentally friendly consumption among consumers means for companies that they have to integrate environmental responsibility into their business strategy and activities (Nidumolu et al., 2009). Hence, more firms are adopting green advertising strategies and are putting great emphasis on the value of green marketing strategies (Nidumolu et al., 2009). Green marketing strategies are strategies to promote products by using environmental claims about the products or about the firms that manufacture and sell these products (Prakash, 2002). Even though many firms are adopting these strategies, there is little research about how effective green marketing strategies are (Chang, Zhang, & Xie, 2015).
It is not easy to persuade consumers with green marketing strategies to purchase more environmentally responsible products, because the benefits of purchasing these kinds of products are often not directly for the consumer him or herself, but also for other people and the planet (Kronrod, Grinstein, & Wathieu, 2012). The most powerful tool for green marketing to persuade consumers to act more environmentally responsible is the use of green communication. Green communication is transmitted to the consumers through environmental claims in advertisements (Hu, 2012). So, in other words, green communication is the message used in green advertisements to promote environmentally friendly consumption.
prospect theory, a message can be either framed as a gain or as a loss (Tversky & Kahneman, 1981). In terms of environmentally friendly consumption, the message either emphasizes on what the environment will gain when the consumer buys the environmentally friendly product, or the message emphasizes what the environment will lose when the consumer purchases a not environmentally friendly version of the product. For example: ‘If you buy this product made out of recycled materials, you contribute positively to climate change’ versus ‘if you buy a product not made out of recycled materials, you will contribute negatively to climate change.’ There is little research about the effect of message framing in green advertisements and the effectiveness of such messages to promote environmental consumption (Cheng et al., 2011), while it is used a lot in practice. Therefore, examining the impact of the two different types of message framing will provide essential insights for advertisers and researchers.
Furthermore, the studies that aimed to examine message framing in advertisements showed mixed results (O'Keefe & Jensen, 2007; O'Keefe & Jensen, 2009). Therefore, it is likely that other factors strengthen or weaken the effect of message framing. One of these factors is imagery. Not only written messages are used to influence consumers in advertisements, but visual cues, as in this study referred to as ‘imagery’, are also important (Spack, Board, Crighton, Kostka, & Ivory, 2012). Using imagery in advertisements can increase the effectiveness of a (social) marketing campaign (Perrine & Heather, 2000). Imagery in advertisements is used as a powerful tool for eliciting emotions, turning abstract ideas to real and credible product benefits, and for implanting ideas and suggestions, and to encourage the purchase of products
(Branthwaite, 2002).
The combination of message framing and nature imagery has not been investigated before, but both individual concepts are used to increase the effectiveness of an advertisement. Research has shown that congruence between different advertisement elements increases the effectiveness of the advertisement (Zhang, Bao, & Xiao, 2019). That is why the ‘right’ combination of both concepts could increase the effectiveness of the advertisement even more.
The present study will aim to examine how message framing and imagery can be aligned to increase the purchase intention of environmentally friendly products, hereby aiming to contribute to the field of message framing by examining under which conditions message framing is most effective. Also, this research will make practical contributions by examining what the most effective combinations are that will increase the chance of creating and implementing more effective marketing campaigns that promote the purchase of environmentally friendly products.
This study will answer the following research question:
How do message framing (gain versus loss) and imagery (nature versus neutral) influence the purchase intention of environmentally friendly products?
2. Literature background
2.1 Purchase intention of environmentally friendly products
This study focusses on purchase intention, rather than actual purchase behavior because intentions to perform a specific behavior can be predicted with high accuracy from the attitudes towards the behavior (Ajzen, 1991). Thus, when we discuss the purchase of environmentally friendly products, we mean the intention to purchase environmentally friendly products. Intentions capture the motivational factors that influence certain behavior and indicate how hard people are willing to try to actually perform that behavior (Ajzen, 1991). Therefore, in this study, we study the underlying processes that lead to behavior, such as attitudes and intentions. Purchase intention is a predictor of the possibility that a consumer will buy the product (Dodds, Monroe, & Grewal, 1991). The stronger the intention, the more likely that a person actually performs the behavior (Ajzen, 1991). In this study, purchase intention is referred to as the intention of consumers to purchase environmentally friendly products (Lasuin & Ching, 2014). We regard this type of behavior as a specific type of environmental consumption because environmental consumption is “consumption activities that have a less negative or more positive effect on the natural environment than substitutable activities” (Pieters, 1991, p.59). This study also looks at attitudes, more specifically, attitudes towards the advertisement and towards the brand. Attitudes towards the advertisement and towards the brand affect the purchase intention of environmentally friendly products (Goldsmith, Lafferty & Newell, 2000), because there is a relationship from values to attitudes to purchase intention (Follows & Jobber, 2002). Therefore, it is important to not only examine the effects of message framing and imagery on the purchase intention but also take into account the attitudes that precede this intention.
2.2 Using message framing to promote purchase intentions of environmentally friendly products
Message framing refers to how people respond and encode messages with the same information differently depending on the way the message is framed (Davis, 1995). Message framing is based on the prospect theory (Kahneman & Tversky, 1981). The theory describes how people make decisions that involve a certain level of risk (Kahneman & Tversky, 1981). The outcomes of these decisions can either be framed as a gain or as a loss. For example: ‘if you do not recycle, you contribute to increasing climate change’, ‘if you recycle, you contribute to decreasing climate change.’ Both messages are showing the same content about recycling, but they are framed differently. The first message focusses on the cost (e.g., the risk) of not complying with the behavior, while the second message focusses on the benefits (e.g., the gain) of complying with the behavior.
Message framing is a technique that can be used in a lot of different contexts. Much research about gain-framed versus loss-framed messages has been conducted in the health context (Segev, Fernandes, & Wang, 2015). Examples of these contexts are organ donation (Reinhart, Marshall, Feeley, & Tutzauer, 2007), HIV detection (Apanovitch, McCarthy, & Salovey, 2003) and anti-smoking (Goodall & Appiah, 2008). These studies have mixed findings about which type of message framing is most persuasive (O'Keefe, & Jensen, 2007; O'Keefe, & Jensen, 2009). The effect of gain-framed messages and loss-framed messages in the healthcare context depends largely on the type of behavior (i.e., prevention behavior versus detection behavior). For example, O’Keefe and Jensen (2007) conducted a meta-analysis including 93 studies and found that gain-framed messages are more persuasive than loss-framed messages for screening for diseases because this is a form of prevention behavior. Another meta-analysis, including 53 studies (O’Keefe & Jensen, 2009), found that loss-framed messages aimed at encouraging disease detection behaviors are slightly more persuasive than gain-framed messages. Loss-framed messages are more effective for people who think that disease prevention tests are too risky or unreliable. Thus, gain-framed messages are more effective when targeting prevention behavior (i.e., prevent the development of a disease), and loss-framed messages are more effective when targeting the detection behavior (i.e., detect the presence of a disease). The reason for this is that prevention behaviors have quite certain outcomes (e.g., use sunscreen to prevent our skin from developing skin cancer) and detection behaviors are riskier and have more uncertain outcomes (e.g., the disease might not be detected) (Schneider, Salovey, Pallonen, Mundorf, Smith, & Steward, 2001).
In terms of preventive behavior versus detection behavior, purchasing environmentally friendly products can be regarded as preventive behavior. With purchasing more environmentally friendly products, the consumer is trying to protect the environment (Loroz, 2007). Therefore, it is expected that gain-framed messages are more effective at influencing the purchase intention of environmentally friendly products than loss-framed messages.
conducted in the environmental domain (Baek & Yoon, 2017; Grazzini, Rodrigo, Aiello, & Viglia, 2018; Lord, 1994; Obermiller, 1995; White, MacDonnell & Dahl, 2011). However, most of these studies focused on conservation behaviors like recycling and energy conservation rather than green consumption and found mixed results (Segev et al., 2015). For example, Grazinni et al. (2018) found that loss-framed were more effective at influencing recycling behavior than gain-framed messages. Baek and Yoon (2017) found that a gain-framed message combined with a prime of guilt was more effective at influencing the intention to conserve water and that a loss-framed message combined with a prime of shame was more effective at influencing the intention to save water. Also, Obermiller (1995) found that gain-framed messages were more effective at influencing salient behaviors (e.g., recycling), and loss-framed messages were more effective for less salient behaviors (e.g., conservation of energy).
Furthermore, the purchase of an environmentally friendly product is not only perceived as riskier than purchasing the alternative products, but the potential gains related to this behavior are also different compared to other typical risky behaviors, such as HIV detection (Apanovitch et al., 2003). The difference between typical risky behaviors such as testing for HIV and purchasing environmentally friendly products lies in the concept of social distance (Nan, 2007). When the social distance increases (i.e., when judgments are made for others), the impact of a gain-framed message increases as well (Nan, 2007). The decision to purchase environmentally friendly products do not only have consequences for the person who purchases environmentally friendly products, but also for other people and the environment. The relatively large social distance for the behavioral context of the present study makes it even more likely that a gain-framed message will be more effective than a loss-gain-framed message. Therefore, it is expected that gain-framed messages are more effective at influencing the purchase intention of environmentally friendly products than loss-framed messages.
H1: Gain-framed messages increase the purchase intention of an environmentally friendly product more than loss-framed messages.
2.3 Using nature imagery in advertisements to promote the purchase of environmentally friendly products
Not only verbal cues but also visual cues are important to influence consumers (Spack et al., 2012). All types of marketing communications rely on the combination between verbal elements (i.e., messages) and visual elements (i.e., images) to transfer the intended message (Houston, Childers & Heckler, 1987). Therefore, it is important to also examine the influence of images on the purchase intention of environmentally friendly products. Images can be used to capture attention (Mackenzie, 1986), increase the memory of the advertisement (Childers & Houston, 1984), evoke emotions (Seo, Dillard & Shen, 2013), and can lead to an attitude change (Peracchio & Meyers-Levy, 2005).
fractality of natural elements is evident because many shapes in nature consist of smaller copies of themselves in different scales (Joye & Van den Berg, 2011). This phenomenon is also known as self-similarity (Joye & Van den Berg, 2011). Self-similarity makes nature scenes more predictable than urban or neutral scenes and, therefore, easier to process (Joye & Van den Berg, 2011). Nature scenes are easier to process and, therefore, more preferred by most people (Joye & Van den Berg, 2011). Easy processing of a certain stimulus (e.g., advertisement) will elicit a positive affective state that people will link to the stimulus (e.g., advertisement) itself (Winkielman, Schwarz, Fazendeiro, & Reber, 2003). In the case of advertisements, the positive affective state will lead to a positive effect on the attitude towards the brand and the attitude towards the advertisement (Batra & Ray, 1986; Edell & Burke, 1987). Using nature imagery in advertisements will, therefore, lead to a more positive brand and product attitude due to the easy processing, which might indirectly lead to an increased purchase intention of environmentally friendly products (Hartmann, & Apaolaza-Ibáñez, 2009; Hartmann, & Apaolaza-Ibáñez, 2012).
Furthermore, people like natural elements and settings more is because of the restorative responses to nature (Kaplan & Kaplan, 1989; Kaplan, 1995; Ulrich 1981; Ulrich 1983). Natural elements make people feel relaxed and help them to recover from stress by reducing directed-attention fatigue (Kaplan & Kaplan, 1989; Hartig & Staats, 2005). According to Kaplan and Kaplan (1989), contact with nature also restores the attention capacities and improves the performance of memory and attention-related tasks. The reason for this is that the environment is full of fascinating stimuli, which helps to recover the attention capacities (Kaplan & Kaplan, 1989). In the case of advertisements, these restored attention capacities lead to better recall and recognition of the advertisements (Hartmann, Apaolaza, & Alija, 2013). This plays an essential role in the purchase intention because consumers are more likely to purchase a product when they know the brand or the product from the advertisements (Macdonald & Sharp, 2000).
emphasize the environmentally friendly product attributes (Hartmann et al., 2013). Nature imagery could improve the emotional and cognitive processes that affect the effectiveness of an advertisement (Hartmann et al., 2013). According to Hartmann and Ibáñez (2010), nature imagery is often used in advertising because it elicits more positive emotional responses that further enhance attitudes towards the advertisement and the brand. There is a hierarchical relationship from abstract values to attitudes to purchase intentions to actual purchase behavior (Follows & Jobber, 2002). According to Goldsmith, Lafferty, and Newell (2000), attitude towards the advertisement and attitude towards the brand affect purchase intention. Consumers who feel more positive towards the advertisement and towards the brand, usually have a higher purchase intention (Goldsmith et al., 2000). Therefore, using nature imagery in advertisements will likely result in stronger purchase intentions of environmentally friendly products as well.
Moreover, in the case of green advertising specifically, the use of nature imagery could increase the positive emotional response because of the congruency between the green product and the nature image. Research has shown that congruence between different advertisement elements increases the effectiveness of the advertisement (Zhang, Bao, & Xiao, 2019); this is because of the congruency effect. Congruence means that the different elements in an advertisement reflect a common theme (Lochbuehler, Mercincavage, Tang, Tomlin, Cappella, & Strasser, 2018).
Both the product and the advertisement seem to be associated with the environment, which influences the cognitive response of consumers even more (Hartmann et al., 2016). The nature imagery in pro-environmentally advertising can highlight the environmentally friendly product features even more (Hartmann et al., 2016), which may further increase the consumers’ purchase intentions of the product.
recognition. So, nature imagery in advertisements increases the attention towards the advertisement, which leads to an increased message recall and increased recognition levels (Hartmann et al., 2013).
In conclusion, using nature imagery in advertisements will lead to a more positive brand attitudes, and better recall and recognition of the product, and recognition of the brand, which might indirectly lead to an increased purchase intention of environmentally friendly products (Hartmann, & Apaolaza-Ibáñez, 2009; Hartmann, & Apaolaza-Ibáñez, 2012). It is often assumed that these positive attitudes towards the advertisement and the brand and the increased message recall and recognition levels increase the purchase intention of the product as well (Shah, Aziz, Jaffari, Waris, Ejaz, Fatima, & Sherazi, 2012). However, there is little research that confirms this assumption. Also, previous research mostly looked at the differences between the effect of nature imagery and urban imagery in advertisements but did not look at the differences between nature imagery and neutral imagery. In this research, we will test whether the use of nature imagery in advertisements will also result in stronger purchase intentions of the environmentally friendly products than neutral imagery.
Concluding from the previous paragraphs, the following hypothesis is derived:
H2: Advertisements with nature imagery positively influence the purchase intention of environmentally friendly products more than advertisements with neutral images.
2.4 Message framing in combination with nature imagery to promote environmentally friendly purchase intentions: Congruency effect
Gain-framed messages are assumed to be more effective than loss-framed messages to encourage consumers’ intentions to purchase environmentally friendly products. However, a lot of research shows mixed support for this relationship (O’Keefe & Jensen, 2007; O’Keefe & Jensen, 2009), suggesting that there are boundary conditions for the effectiveness of message framing.
in an advertisement are in line with each other, it will lead to a stronger memory of the advertisement (Morelli, 1970). It will also lead to better recall of the information (Emery Romer, Sheerin, Jamieson, & Peters, 2013). Memory and recall of the advertisement have a positive effect on the purchase intention (Thorson, Chi, & Leavitt, 1992). Also, Joireman, Liu, and Kareklas (2018) found that consumers are less skeptic about claims when they are paired with an image that supports the claim, which makes them more likely to purchase products that have aligned advertisements. Hence, gain-framed message should be aligned with nature imagery to be most effective to encourage purchase intentions, because both the message and the nature imagery focus on the positive outcomes of purchasing an environmentally friendly product (e.g., the text of the advertisement states: purchasing this product contributes to decreasing climate change, while the image shows a picture of a beautiful landscape). A loss-framed message would be most effective to encourage purchase intentions when combined with neutral imagery because when combined with nature imagery, both elements would contradict each other (e.g., the text of the advertisement states: not purchasing this product contributes to increasing climate change while showing a picture of a beautiful landscape).
Besides the general benefits of congruency in advertisements, there are also some additional benefits of combining a gain-framed message with nature imagery. Gain-framed messages and nature images are both ‘associative’ claims. Associative claims are used by organizations to develop a positive perception with the consumers by focusing on other things than the advertisers’ own products or processes (Carlson, Grove, Kangun, Polonsky, 1996). Images do not link organizations to environmentally causes, while messages do not link the advertisement to the advertiser him or herself (Carlson et al., 1996). So, combining nature imagery and a gain-framed message will create a link between the advertisement, the company, and the environmentally cause. Also, when processing associative claims, consumers create a more positive perception of the company itself (Carlson et al., 1996).
advertisements are formed. Our attitudes are formed through an interaction of emotional processes and cognitive processes (Hartmann & Apaolaza-Ibáñez, 2009). The emotional responses to nature imagery in interaction with the cognitive processes of understanding the pro-environmental claim will lead to stronger attitude formations. Also, the positive emotional responses to nature imagery will transfer on to brand attitudes (Hartmann & Apaolaza-Ibáñez, 2009). Also, another study of Hartmann, Apaolaza-Ibáñez, & Forcada Sainz (2005) found that to increase the purchase intention of environmentally friendly products, green advertisements should combine images of nature with information on environmentally product attributes. The present study does not only focus on pro-environmental claims (like the study by Hartmann and
Apaolaza-Ibáñez (2009) does) but distinguishes between gain-framed and loss-framed pro-environmental messages. So, based on the study by Hartmann and Apaolaza-Ibáñez (2009), combining a gain-framed message with nature imagery will lead to an increase in purchase intention.
The benefits of combining pro-environmentally messages and nature imagery do not apply to loss-framed messages. A loss-framed message should not be combined with nature imagery because it will lead to incongruency in the advertisement. A loss-framed message emphasizes the negative consequences for the environment when not purchasing an environmentally friendly product, while the nature imagery emphasizes the positive outcomes for the environment by showing a picture of a beautiful landscape. This incongruency is likely to undermine the message processing capacities by priming information that is irrelevant to the message and makes people think less about the message content, which makes the message less effective (Smith & Shaffer, 2000).
Because of the congruency effect, the expectation is that combining gain-framed messages with nature imagery will lead to an increased purchase intention, while on the other hand, combining loss-framed messages with neutral imagery will lead to an increased purchase intention. Hence, the following effect is expected:
2.5 Present study
It is not easy to persuade consumers to purchase more environmentally friendly products, because the benefits of purchasing these kinds of products instead of ‘normal’ products are less directly noticeable (Kronrod et al., 2012). To increase the purchase intention of environmentally friendly products, green advertisements should be designed in the most effective way possible. Both verbal (e.g., text) and visual (e.g., images) elements are essential when influencing consumers to purchase environmentally friendly products (Spack et al., 2012). Hence, this study investigates the effect of message framing and imagery (nature versus neutral images) on the purchase intention of pro-environmentally friendly products. As a case in point, the present study focuses specifically on the purchase of environmentally friendly washing detergent, because it is a low involvement product, and often used in green advertisement campaigns.
Figure 1: Conceptual model
3. Methodology
3.1 Participants
Data was gathered with a questionnaire along 411 participants. Participants were found with the use of the authors’ social media channels, LinkedIn, and email. Participants were also asked to share this questionnaire with their contacts. This is a form of opportunity sampling (Fogelman & Comber, 2002). Opportunity sampling was most appropriate for this study because of the time limitations and the willingness of the participants to participate (Etikan, Musa, & Alkassim, 2016).
First, missing values were removed (n=84) People who were younger than 18 years (n=4) were also removed from the sample because people younger than 18 years are not fully capable of making informed choices (i.e., do probably not decide whether the product under investigation should be bought) (Cauffman & Steinberg, 1995). There were no outliers detected that deviated more than three standard deviations away from the mean, so there were no outliers removed from the sample. As a result, 323 responses were used for analysis. This sample size is big enough according to several rules of thumbs. According to Cohen (1988), for a medium to large effect size, 30 participants per condition is enough to detect differences between conditions, so for this study, the minimal number of participants should be 120. In this research, we will also make use of a factor analysis. According to Tabachnick and Fidell (1996), a sample size of 300 is considered to be good to perform a factor analysis.
The final sample consisted of 74,9 % of women and 25,1 % of men. The participants were between 18 and 75 years old (Mage = 34.43, SD= 14.32). The majority of the participants had advanced educational levels, 63.1 % had an HBO-degree or higher. For household incomes, participants earned relatively a low amount of money, 32.8% earned less than €10.000. To conclude, the sample consisted of relatively young and well-educated people, compared to the Dutch population (StatLine, 2019).
3.2 Research design
were developed, as shown in Table 1: Condition 1 included a gain-framed message and a nature image, Condition 2 a gain-framed message and a neutral image, Condition 3 a loss-framed message and a nature image, and, Condition 4 a loss-framed message and a neutral image. Participants were randomly assigned to one of these conditions. There was no control group in this study because this research focusses on the (dis)similarities between groups and was not aiming at comparing it with a point of reference.
Table 1: Number of participants in each condition (n)
Condition Message frame Image n
1 Gain-framed Nature 79
2 Gain-framed Neutral 81
3 Loss-framed Nature 81
4 Loss-framed Neutral 81
3.3 Pre-test
Pre-testing is testing the survey on a small sample of respondents to identify and eliminate problems (Malhotra, 2010). The respondents need to have the same characteristics as respondents who participate in the real survey (Malhotra, 2010). That is why the pre-test was sent to family members and friends of the author. The people who did the pre-test were excluded from the real test.
The type of nature images used in the pre-test were based on the study of Hartmann and
only 30 degrees’. The gain-framed message was included as well because the advertisements in the pre-test should be as similar as possible to the real test, to prevent that other aspect influenced the outcome of the pre-test. We used the gain-framed message instead of the loss-framed message because of the theory about congruency in advertisements (Zhang et al., 2019).
A nature image combined with a loss-framed message causes incongruency in the advertisement. This incongruency undermines the message processing capacities by priming information that is irrelevant to the message and makes people think less about the message content (Smith & Shaffer, 2000). Due to this incongruency, participants are less likely to form elaborated opinions about the advertisement. The gain-framed message was: ‘Washing at 30 degrees lowers CO2 emissions, reduces energy consumption, and contributes to reducing global warming’ (See Appendix 1). The position of the text, the typeface, and the color of the text were overall the same, to minimize the effect of these variables in the pre-test. Only when the text was not visible enough, the color of the text was changed to black instead of white, or the text was moved to a different position.
The images were shown in random order to the participants. Participants were asked if they perceived the background of the advertisement as pleasant, if they thought that the product fitted with the background, if they thought that the message in the advertisement fitted with the background. All these items were measured on a seven-point Likert scale, ranging from (1) strongly disagree to (7) strongly agree. Lastly, participants were asked how many seconds they needed to read and look at the first advertisement carefully, so this could be used for instructions for the real survey.
3.4 Materials
The advertisements (see Appendix 2) were made with the use of the website pixr.com and with the use of the program photoshop.
3.4.1 Product
The product in the advertisement was a low cost, non-durable product, namely laundry detergent (see Appendix 2). This product was used in this study to minimize the involvement of the consumers so that consumers did not have to think too much if they would actually want to buy this product (Melody & Lefkoff-Hagius, 1995). In other words, the decision to purchase laundry detergent does not require much effort. Also, this product is often used in green campaigns, so this makes the experiment realistic (Melody & Lefkoff-Hagius, 1995). A fictive brand, ‘Eco friend’, was used to prevent pre-existing biases towards the brand. Pre-existing biases influence the attitude formation towards the brand (Mau, Silberer & Constien, 2008). Also, brand familiarity influences the purchase intention (Johnstone & Dodd, 2000).
3.4.2 Manipulation of independent variables
The manipulation of the variable message framing was based on the study of Segev, Fernandes, and Wang (2015). The message in the text of the advertisement either emphasized the positive outcomes of purchasing the suggested product (gain-framed) or emphasized the negative consequences of not purchasing the suggested product (loss-framed). The loss-framed message was: Washing at 60 degrees increases CO2 emissions, increases energy consumption, and
contributes to global warming. The gain-framed message was: Washing at 30 degrees lowers CO2 emissions, reduces energy consumption, and contributes to reducing global warming (see
Appendix 2).
messages was Arial because, according to a study of Brumberger (2003), this typeface is seen as the appropriate typeface for professional text.
The second manipulation variable, nature imagery, also included two levels: nature imagery and neutral imagery. The advertisement had either a nature image or a neutral image on the background (see Appendix 2). The nature image was chosen based on the pre-test. The pre-test found that participants were most positive about the image with the grassy landscape (M=5.38, SD= 1.20), so that was the image that was used in this study. The neutral image was chosen to be a similar color green because this study aimed at finding the differences in using nature imagery versus neutral imagery, and not at finding the difference between a green and a non-green appeal. Therefore, we deemed it essential to use a similar color non-green in the neutral image condition.
Finally, we decided to position the product (i.e., washing detergent) at the left and the text at the right of the advertisement. The reason for this comes from the theory of hemispheric lateralization (Janiszewski, 1988), which argues that it is best to place the text on the right, and the picture on the left because of the distinction of functions in the hemispheres of our brain (Janiszewski, 1988). The final advertisements, representing the four experimental conditions, are presented in Appendix 2.
3.5 Procedure
The study was conducted with the use of an online questionnaire made with Qualtrics. The questions of this survey can be found in Appendix 3. The link for this questionnaire was distributed with the use of the authors’ own social media channels. Because the authors’ social media network mostly consisted of Dutch-speaking people, the questionnaire and the messages were written in Dutch, to ensure that everyone correctly understood the questions.
The order of the questions was based on Krosnick (2018), who describes the ideal order of the items. According to Krosnick (2018), the questions at the beginning of the survey influence the willingness to respond to the survey, because they influence the understanding of the respondents about the subject of the survey. That is why the first few questions should be related to the subject, and demographic questions should be at the end (Krosnick, 2018).
Before the start of the questionnaire, participants were informed about the procedure and the general subject of the study. After this short introduction, participants were randomly assigned to one of the four experimental conditions. Participants were asked to carefully have a look at the advertisement for a minimum of 20 seconds because they have to give their opinion about it during the questionnaire. These 20 seconds were deemed sufficient to read the text and look carefully at the image, as tested in the pre-test. Participants could not go the next questions before these 20 seconds ended.
Next, the dependent variable was measured. Participants were asked to indicate their purchase intention with four questions. This was followed by questions about the attitude towards the brand and towards the advertisement. After these, a question about if people actually purchase laundry detergent was asked. Then two questions were asked to test whether the participants perceived the image and the messages as intended to. In the last questions, participants were asked to fill in their demographic details. The questionnaire ended with a short debrief, and the participants were thanked for their participation.
3.6 Measures
Participants were asked several questions regarding the dependent variable (purchase intention), the independent variables (message framing and imagery, the co-founding variables (attitudes towards the brand and attitudes towards the advertisements), and lastly, some socio-demographics. All measures are included in Appendix 4.
3.6.1 Dependent variable: Purchase intention
ranging from 1= strongly disagree to 7= strongly agree. The following four items, based on Ajzen (1991), were included:
1. I am willing to pay more money for Eco friend than for other laundry detergents.
2. I will take Eco friend as my consideration when purchasing laundry detergents in the future. 3. I am willing to purchase Eco friend in the future.
4. When I see Eco friend in a store, I am willing to try it.
3.6.2 Manipulation check: Message framing and imagery
To ensure that the participants perceived the advertisements correctly, we included a manipulation check for both manipulation variables, that is, message framing and imagery. The manipulation checks were similar to the manipulated checks in the research of Maheswaran and Meyers-Levy (1990). The manipulation checks were measured with a seven-point Likert scale ranging from (1) strongly disagree till (7) strongly agree.
For message framing, the participants were asked to evaluate the extent to which they perceived that the advertisement focused on the positive outcomes of purchasing Eco friend or the negative outcomes of not purchasing Eco friend. The manipulation check for the message framing consisted of two items: ‘the advertisement of Eco friend emphasized the positive outcomes of purchasing Eco friend for the environment’ and ‘the advertisement of Eco friend emphasized the negative outcomes of not purchasing Eco friend for the environment’.
For imagery, participants were asked the extent to which they perceived the natural elements of the image in the background. The manipulation check for imagery consisted of two items as well: ‘the image on the background of the advertisement is an image of nature’ and ‘the image on the background of the advertisement is a neutral image’.
3.6.3 Alternative depending variables: Attitude towards the brand and attitude towards the advertisement
behavioral intentions. There is a hierarchical relationship from attitudes to purchase intentions to actual purchase behavior (Follows & Jobber, 2002). According to Goldsmith, Lafferty, and Newell (2000), attitude towards the advertisement and attitude towards the brand have an effect on purchase intention. Consumers who feel more positive towards the advertisement and towards the brand, usually have a higher purchase intention (Goldsmith et al., 2000). If the purchase intention does not differ for the four conditions, it is still essential to look at the differences in attitudes, because attitudes could also eventually lead to an increased advertisement effectiveness. Therefore, previous research often included attitudes towards the advertisement and product as dependent variables, rather than actual intentions to purchase the product (Goldsmith et al., 2000; Gresham & Shimp, 1985). As the relationship between attitudes towards the advertisement, towards the product and purchase intentions are only implied theoretically in these studies, the present study includes both attitudes towards the advertisement and product as alternative dependent variables. Hereby, we can compare our results with these studies to further validate the present findings.
Attitude towards the brand and attitude towards the advertisement were both measured by asking participants what they thought about the brand/advertisement by using three seven-point bipolar semantic differential scales, adapted from MacKenzie and Lutz (1989):
What do you think about the brand/advertisement? Good _:_:_:_:_:_:_ Bad
Favorable _:_:_:_:_:_:_ Unfavorable Pleasant _:_:_:_:_:_:_Unpleasant
3.6.4 Confounding variables: Gender and education level
According to some studies, women report greater environmental concern than men (Siwayanan, Bakar, Aziz, Chelliapan, 2015; Xiao & McCright, 2015) and that women often more engage in pro-environmentally friendly behavior than men (Robert, 1996). Furthermore, women pay more attention to labels on detergents because they handle detergents more often than men (Staff, 2014). Therefore, gender was included as a confounding variable in this study.
2006), and with environmentally friendly behaviors (Roberts, 1996). Therefore, education level was included as the second confounding variable in this study.
3.6.5 Socio-demographics
Questions regarding socio-demographics were asked to gain more insights into the sample of this study. As said before, in the confounding variables section, participants were first asked to indicate their gender (1 Male, 2 Female, 3 Other). After that, the age of the participants was asked. Next, participants were asked to indicate the highest level of education they have attained (No school, high school degree, bachelor’s degree, master’s degree, doctorate, other, do not want to answer). Next, participants were asked to indicate their nationality (Dutch, other, do not want to answer) Lastly, participants were asked to indicate their net income in the past year (Below €10000, €10000-24999, €25000-39999, €40000-64999, €65000, and higher, do not want to answer).
3.7 Plan of analysis
The data were analyzed with the use of the program SPSS, version 24. First, a factor analysis was conducted, then a reliability check, then a manipulation check, after that the hypotheses were tested.
3.7.1 Factor analysis
To reduce and summarize the data, a factor analysis was used (Malhotra, 2010). A factor analysis was used to test if all the items regarding one variable are related to each other. A Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy was used to examine if it was appropriate to precede with the factor analysis (Malhotra, 2010). Values between 0.5 and 1 indicate that the factor analysis is appropriate, and values below 0.5 indicate that the factor analysis might not be appropriate (Malhotra, 2010). Also, Bartlett’s test of sphericity was used. The value of this test should be significant (p<.05) in order to perform a factor analysis (Malhotra, 2010).
rotated using the varimax procedure. This made the factors better interpretable (Malhotra, 2010).
3.7.2 Reliability analysis
After the factor analysis, a reliability analysis (Cronbach’s Alpha) was conducted to check if a sum-variable could be computed. The minimum for computed a sum-variable is a Cronbach’s Alpha of .60 (Malhotra, 2010).
3.7.3 Manipulation check
To analyze whether respondents perceived the different kinds of message framing as intended to and whether they perceived the different kinds of background images, two independent sample t-tests were performed to see whether the experimental manipulations were successful. To be able to do an independent sample t-test, several assumptions should be met.
Assumptions for t-tests
Firstly, the dependent variable must be measured at an interval or ratio level. This assumption was met since purchase intention was measured with a 7-point Likert-scale. Even though a Likert-scales is an ordinal scale, it is allowed to treat it like an interval scale (Wu & Leung, 2017).
Secondly, the observations need to be independent of each other. This assumption was also met since participants were only assigned to one of the four groups.
Fifthly, there should not be significant outliers in the data. This assumption was also met because no outliers were detected from the data. Outliers in a large sample will have a limited impact on the results (Daszykowksi, Kaczmarek, Van der Heyden, & Walczak, 2007). Nevertheless, we checked for outliers. There were no outliers detected that deviated more than three standard deviations away from the mean.
To test Hypothesis 1, Hypothesis 2, and Hypothesis 3, a two-way ANOVA was used. To examine the influence of the confounding variables, an ANCOVA was used. In order to conduct an ANOVA, six assumptions should be met. The assumptions for the t-test also need to be met in order to conduct an ANOVA with one addition. For the ANCOVA, two extra assumptions should be met.
Assumptions for ANOVA
The independent variable should consist of two or more categorical and independent groups. This assumption was also met because participants were randomly assigned to one of the four conditions.
Assumptions for ANCOVA
Firstly, there should be independence of the covariate and the treatment effect, so the covariates should not differ between the different conditions. This was tested with the use of an ANOVA, using the conditions as independent variables and the covariates as the dependent variables. The outcomes of this analysis showed not to be significant for gender p=.80 and for education level p=.93, so this assumption was met.
4. Results
4.1 Manipulation check
To check whether the manipulations were successful, two t-tests were conducted. For the message framing condition, the scores of the second item were reversed, and a factor analysis was used to test whether the two items regarding one variable were related to each other. Also, a Kaiser-Meyer-Olkin measure of sampling adequacy was used. After it was determined that a factor analysis was suitable, a principal components analysis and Cronbach’s Alpha was conducted. The Kaiser-Meyer-Olkin tests showed that it was allowed to perform a principal component analysis (KMO=.50). Also, Bartlett’s tests showed to be significant. Therefore, a principal component analysis could be conducted. Cronbach’s Alpha showed ⍺=.86, which was higher than the minimum for computing a sum variable (⍺=.60) (Malhotra, 2010). So, a sum-variable could be computed. Therefore, a new sum-variable to test the extent to which the message was perceived as a gain-framed was computed by summing the items and dividing it by two.
The difference in means and standard deviations implied that the two groups (gain-framed and loss-framed) significantly differed from each other. The gain-framed group respondents perceived that the advertisement was more positively framed (M=5.17, SD=1.00) than the loss-framed group (M=3.85, SD=1.42). The t-test confirmed that these differences were significant, t(290.808)=9.678, p<.001). The manipulation of the gain and loss framing experimental conditions was therefore successful.
The difference in means and standard deviations implied that the two groups (gain-framed and loss-framed) significantly differed from each other. The data shows that in the nature group, respondents perceived the background of the advertisement more as a background with natural elements (M=5.51, SD=1.20) versus the neutral group (M=3.51, SD=1.40). Equal variances are not assumed. The t-test confirmed that these differences between the two groups were significant t(314.529=13.855, p<.001). The manipulation of the nature image and neutral image conditions was therefore successful.
4.2 Reliability and validity
To reduce and summarize the data, a factor analysis was used (Malhotra, 2010). A factor analysis was used to test discriminant and convergent validity and reliability. Also, a Kaiser-Meyer-Olkin measure of sampling adequacy was used. First, the KMO measure of sampling adequacy was performed to check whether it was appropriate to combine the individual attitudes (6) and intentions (4) items into factors. The KMO=.85 and Bartlett’s Tests showed to be significant (see Appendix 5), so a factor analysis was suitable. To determine how many factors we should create, we looked at the scree-plot, eigenvalues, and the total variance explained. Eigenvalues should be >1, and the total variance explained should be about 70/80% (Kootstra, 2004). The scree-plot, the eigenvalues, and the total variance explained indicated that it was appropriate to create three factors (see Appendix 6). The factors did not have high cross-loadings on other components. There were no multicollinearity issues since the VIF scores were low since the maximum VIF score was 1.35. To determine the reliability of the theoretical constructs, a Cronbach’s alpha reliability test was conducted per construct. Table 2 shows the results of the factor analysis.
Purchase intention. The four items correlated strongly and positively with one component, indicating that the items measured the same construct, namely purchase intention. Cronbach’s Alpha showed ⍺=.89, which was higher than the minimum for computing a sum variable (⍺=.60) (Malhotra, 2010). So, a sum-variable could be computed. Therefore, a new variable ‘purchase intention’ was computed by summing the four items and dividing the score by four (M=4.39, SD=1.35).
computing a sum variable (⍺=.60) (Malhotra, 2010). So, a sum-variable could be computed. Therefore, a new variable: attitude towards the brand was computed by summing the three items and dividing the score by three (M=4.47, SD=1.37).
Attitudes towards the brand Eco friend. The three items correlated strongly and positively with one component, indicating that the items measured the same construct, namely attitude towards the brand. Cronbach’s Alpha showed ⍺=.91, which was higher than the minimum for computing a sum variable (⍺=.60) (Malhotra, 2010). So, a sum-variable could be computed. Therefore, a new variable: attitude towards the brand was computed by summing the three items and dividing the score by three (M=4.37, SD=1.13).
Table 2: Results principal component analysis using the varimax rotation method.
Items Component 1 Component 2 Component 3
Purchase intention
I will take Eco friend as consideration when buying laundry detergent in the future
0.890 0.095 0.157
I am willing to purchase Eco friend in the future
0.876 0.091 0.225
When I see Eco friend in a store, I am willing to try it
0.825 0.110 0.196
I am willing to pay more for Eco friend than for other kinds of laundry detergents
0.758 0.208 0.124
Attitude towards the advertisement Q10 What do you think about the advertisement
0.106 0.913 0.194
Q13 What do you think about the advertisement
Q12 What do you think about the advertisement
1.155 0.868 0.259
Attitude towards the brand Eco friend Q8 What do you think about the brand Eco friend
0.233 0.212 0.873
Q7 What do you think about the brand Eco friend
0.165 0.260 0.871
Q9 What do you think about the brand Eco friend
0.245 0.244 0.850
4.3 Hypothesis testing
A two-way ANOVA was performed including the message framing conditions and the imagery conditions as independent variables and the purchase intention of environmentally friendly products as the dependent variable (Table 3).
Hypothesis 1 proposed that a gain-framed message more strongly and positively influences the purchase intention more than a loss-framed message. The mean scores of the two different groups indicated that there was a small difference: the gain-framed group (M=4.46, SD=1.35) showed slightly higher purchase intentions than the loss-framed group (M=4.33, SD=1.35). However, the two-way ANOVA revealed that these differences were not significant, F(1, 319) = 0.758, p = .38. Therefore, Hypothesis 1 was rejected.
Hypothesis 2 proposed that a nature image more strongly influences the purchase intention than a neutral image. The mean scores of the two different groups indicated that there was a small difference. The mean score of the nature image group (M= 4.37 SD=1.39) showed slightly lower purchase intentions than the neutral image group (M= 4.42 SD=1.31). The analysis revealed that these differences were not significant, F(1, 319)= 0.102, p=.75. Therefore, Hypothesis 2 was rejected.
imagery. An advertisement with a loss-framed message will be more effective in combination with neutral imagery relative to nature imagery. Figure 2 shows the mean scores per different condition. The mean score of purchase intention of the combination gain-framed message and nature image is higher (M=4.48, SD=1.35) compared to the gain-framed neutral image condition (M=4.44, SD=1.37). The mean score of the purchase intention of the loss-framed neutral condition (M=4.40, SD=1.26) was higher than the loss-framed nature condition (M=4.26, SD=1.43). However, the analysis revealed that there was no interaction effect F(1, 319)= 0.390, p=.53. Therefore, Hypothesis 3 was rejected.
Figure 2: Interaction effect message framing and imagery on the purchase intention
Table 3: Results two-way ANOVA. The influence of message framing and imagery on the purchase intention of environmentally friendly products
Source F (1,319) p-value R²
Corrected model 0.419 .740 0.004
Message framing 0.758 .385
Imagery 0.102 .750
Message framing * Imagery 0.390 .533 0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5 Gain-framed Loss-framed In te ra cio n m es sa ge fra m in g an d im ag ery Frame conditions
Mean scores purchase intention per condition
4.4 Overview of the hypotheses: supported or not supported
Table 4 provides an overview of the hypotheses and shows if the hypotheses were supported or not supported.
Table 4: Overview of the hypotheses
Hypothesis Supported or not supported
H1: Gain-framed messages increase the purchase intention of an environmentally friendly product more than loss-framed messages.
Not supported
H2: Advertisements with nature imagery positively influence the purchase intention of environmentally friendly products more than advertisements with neutral images.
Not supported
H3: An advertisement with a gain-framed message will be more effective in influencing green purchase intentions when combined with nature imagery than with neutral imagery. An advertisement with a loss-framed message will be more effective in combination with neutral imagery relative to nature imagery.
Not supported
4.5 The influence of message framing and imagery on purchase intentions: Further validation of findings
variables in research related to advertisements effectiveness (e.g., Goldsmith et al., 2000;
Gresham & Shimp, 1985), probably because, according to the theory of planned behavior (Ajzen, 1991), attitudes are one of the most important predictors for behavioral intentions. Indeed, a Pearson correlation test was conducted and showed a medium-strong correlation (cf. Cohen, 1988) between purchase intention and attitude towards the advertisement (r=.34, p <.01) and a medium to strong correlation between intention and attitude towards the brand (r = .46, p < .01).
4.5.1 The influence of message framing and imagery on attitude towards the advertisement and brand.
We first tested Hypothesis 1, 2, and 3 with attitude towards the advertisement as the dependent variable (instead of purchase intention). The means of gain and loss framing on attitudes towards the advertisement showed that respondents receiving a gain-framed message (M=4.52, SD=1.40) had a slightly stronger attitude towards the advertisement than respondents receiving a loss-framed message (M=4.42, SD=1.34). However, the two-way ANOVA revealed that these differences were not statistically significant, F(1, 319)= 0.460, p=.49, hereby again rejecting Hypothesis 1 for attitude towards the advertisement.
The means of nature and neutral image on attitudes towards the advertisement showed that respondents receiving a nature image (M=4.65, SD=1.39) had a stronger attitude towards the advertisement than respondents receiving a neutral image (M=4.29, SD=1.34). The two-way ANOVA revealed that these differences were statistically significant, F(1, 319)= 5.440, p < .05. Hence, Hypothesis 2 was supported for attitude towards the advertisement.
significant interaction, F(1, 319)= 0.074, p=.78, hereby rejecting Hypothesis 3 for attitude towards the advertisement.
Next, we tested Hypothesis 1, 2, and 3 with the attitude towards the brand as the dependent variable. The means of gain and loss framing on attitudes towards the brand showed that respondents receiving a gain-framed message (M=4.35 SD=1.13) had a slightly weaker attitude towards the brand than respondents receiving a loss-framed message (M=4.40, SD=1.14). However, the two-way ANOVA revealed that these differences were not statistically significant, F(1, 319)= 0.113, p=.73 hereby again rejecting Hypothesis 1 for attitude towards the brand.
The means of nature and neutral image on attitudes towards the brand showed that respondents receiving a nature image (M=4.50 SD=1.12) had a slightly stronger attitude towards the brand than respondents receiving a neutral image (M=4.25, SD=1.13). The two-way ANOVA revealed that these differences were statistically significant F(1, 319)= 3.756, p=.05 hereby validating Hypothesis 2 for attitude towards the brand.
The mean score of the combination of message framing and imagery on attitudes towards the brand showed that respondents receiving a gain-framed message and nature image (M=4.49, SD=1.10) had stronger attitudes towards the brand compared to the gain-framed neutral image condition (M=4.20, SD=1.14). However, the same direction was present in the loss-framed condition, where respondents receiving the loss-framed nature condition (M=4.49, SD=1.14) had stronger attitudes towards the brand compared to the loss-framed neutral condition (M=4.29, SD=1.13), implying that there was only a main effect of imagery but no interaction effect. Indeed, the two-way ANOVA revealed that these differences were not statistically significant F(1, 319)= 0.123, p=.72 hereby again rejecting Hypothesis 3 for attitude towards the brand.
4.5.2 Gender and education level as confounding variables
5. Discussion and conclusion
5.1 Conclusion
The present study examined the effect of message framing (gain-framing versus loss-framing) and imagery (nature imagery versus neutral imagery) on the purchase intention of environmentally friendly products. Based on the literature about imagery (Kaplan & Kaplan, 1989; Kaplan, 1995; Ulrich 1981; Ulrich 1983) and the prospect theory (Kahneman & Tversky, 1981), the expectation was that gain-framed messages would be more effective than loss-framed messages and that nature images would be more effective than neutral images. The expectation was, based on the theory about congruency in advertisements (Emery et al., 2013;
Joireman et al., 2018; Morelli, 1970) that the combination of a gain-framed message with nature imagery would be more effective than a gain-framed message with neutral imagery. Also, a framed message combined with neutral imagery would be more effective than a loss-framed message with nature imagery. The findings did not support the three hypotheses. Additional analyses revealed that these conclusions were robust, even when using alternative dependent variables (attitudes towards the advertisement and brand) and controlling for gender and educational level. One exception was the effect of the use of nature (versus neutral) imagery and attitudes towards the advertisement and the brand. Respondents receiving a nature image showed more positive attitudes towards the ad and the brand than those who received a neutral image. Next, we will discuss these results more in dept.
5.2 General discussion
Based on the prospect theory (Kahneman & Tversky, 1981) and previous literature about message framing (Chang, 2011; Loroz, 2007; Nan, 2007; Pickett-Baker & Ozaki, 2008; Segev et al., 2015) the expectation was that gain-framed messages would be more effective at influencing the purchase intention than loss-framed messages. Purchasing environmentally friendly products can be seen as preventive behavior (Loroz, 2007), riskier (Chang, 2011; Pickett-Baker & Ozaki, 2008), and the social distance increases (Nan, 2007) when purchasing environmentally friendly products.
mechanism (i.e., risk) that influences the way people respond to certain messages, might be less relevant. The purchase of environmentally friendly products might not really be regarded as riskier than other types of purchase behavior, because risks are regarded as more relevant when they directly affect the person who makes the decision, rather than more social distant gains. The consumer does not perceive the risks, or the gains directly, and therefore be less influenced by loss-framed/gain-framed messages.
Also, the health literature about message framing suggests that gain-framed messages are more effective for preventive behaviors, but pro-environmentally friendly purchase intention might not be regarded as ‘real’ prevention behavior. The preventive function of pro-environmentally purchases is directly less directly visible than in the healthcare context. For example, you cannot directly see the results for the environment by purchasing one environmentally friendly bottle of laundry detergent, but you directly see the results of using sunscreen to prevent the skin from burning.
To conclude, the results of our study did not find that gain-framed messages are more effective at increasing the purchase intention of environmentally friendly products. A lot of studies explored the effects of message framing in the environmentally behavior context (Cheng et al., 2011), and other contexts (O'Keefe, & Jensen, 2007; O'Keefe, & Jensen, 2009), and found mixed results. So, the results of this study are in line with the discussion about the influence of message-framing. It might not be effective to use message framing in the context of pro-environmentally purchase intentions, as this type of behavior might not be really regarded as risky or preventive behavior.
Our results are interesting, as the mean score of purchase intention of the neutral imagery conditions is higher than the mean score of purchase intention of the nature imagery conditions. One explanation for this is that the neutral image was also a green colored image. The color green could unconsciously have been linked to nature by the participants (Studente, Seppala, & Sadowska, 2016), causing that participants in the neutral imagery condition perceived the same benefits as the participants in the nature imagery condition. So, the color green might already be enough to experience the benefits of nature. However, the results of our study on the alternative dependent variables ‘attitudes towards the advertisement’ and ‘towards the brand’ were in line with the study by Hartmann and Ibáñez (2010), as findings implicated that nature imagery elicits more positive emotional responses that lead to enhanced attitudes towards the advertisement and the brand. Nevertheless, these attitudes did not lead to an increased purchase intention.
Based on the theory about congruency in advertisements (Emery et al., 2013; Joirema et al., 2018; Morelli, 1970; Smith & Shaffer, 2000), the expectation was that the congruency of a gain-framed message and nature imagery and a loss-framed message and neutral imagery would lead to an increased purchase intention of environmentally friendly products. Congruency in advertisements will lead to a stronger memory of the advertisement (Morelli, 1970), better recall of the information (Emery et al., 2013) and less skepticism (Joireman et al., 2018), which eventually leads to an increased purchase intention (Thorson et al., 1992). This study did find that the mean scores of the purchase intention were the higher for the gain-framed nature condition than for the gain-framed neutral condition and for the loss-framed neutral condition than for the loss-framed nature condition, but these results were not significant. So, our study invalidates that congruency in a green advertisement leads to an increased purchase intention.
