Does green advertising work? : a meta-analytic research of the effects of green ads

Does green advertising work? A meta-analytic research of the effects of green ads Nancy D. van Dongen

University of Amsterdam

10765085

Dr. Marijn Meijers Master Thesis

Research Master Communication Science 08-07-2016

Abstract

This study provides an answer to the question whether and under which conditions green ads are more effective than non-green ads in enhancing positive consumer outcomes (attitude towards the ad, attitude towards the brand, and purchase intention). Green ads are defined as advertisements that advertise or promote environmentally friendly attributes of products and/or services. Using a meta-analytic approach, this study integrates and compares previous findings on green ads and as such provides conclusive results of the effectiveness of green ads. Part I of the meta-analysis shows that green ads have more positive effects on attitude towards the brand, but no effects on attitude towards the ad and purchase intention. Moderator analyses reveal that product

involvement, environmental impact and brand type do not affect the relation of green versus non-green ads on consumer outcomes. Part II of the meta-analysis shows that environmental and personal green ads do not have different effects on attitude towards the ad, attitude towards the brand, and purchase intention. Moderator analyses reveal that country disposition and the product’s goal of fulfillment do not affect the relation of green ads on consumers outcomes. By identifying gaps in previous research, this meta-analysis give guidance for future research.

Introduction

The last 25 years, consumers’ interest in the environment has grown enormously. As a result, consumers begin to realize that their consumption activities contribute to environmental problems (Kumar, 2016; Kangun, Carlson & Grove, 1991). Consumers are becoming more environmentally conscious and are likely to choose one product over another product if they think that they are helping the environment (Kangun et al., 1991; Teisl, Roe & Hicks, 2002). They consider it important that organizations act environmentally responsible (Easterling, Kenworthy & Nemzoff, 1996). On their turn, organizations become aware of the importance of being environmentally friendly (Yoon & Kim, 2016) and as such become more environmentally friendly. For example, by adding green products to their product range (e.g., Coca Cola Life), by making the process more environmentally friendly (e.g., KLM’s use of biofuel), or by

encouraging consumers to act environmentally friendly in general (e.g., recycling of clothing; H&M’s Conscious). In order to promote these green products and processes marketers use green advertising as an important communication tool, which has been established as a sub-discipline of marketing (Chan, Leung & Wong, 2006; Kumar, 2016).

Green ads are defined as advertisements that advertise or promote environmentally friendly attributes of products and/or services (Hartmann & Apaolaza-Ibáñez, 2008). With the growing interest of consumers, research on green advertising has also grown (Banerjee, Gulas & Iyer, 1995; Bickart & Ruth, 2013; Borin, Lindsey-Mullikin & Krishnan, 2013; Chan, 2004; Chen & Lee, 2015; Easterling et al., 1996; Green & Peloza, 2014; Hartmann & Apaolaza-Ibáñez, 2009; Hu, 2012; Jiménez & Yang, 2008; Kong & Zhang, 2014; Matthes, Wonneberger & Schmuck, 2013; Segev, Fernandes & Wang, 2015).

Research has been conducted to find trends and classifications in green advertising (VanDyke & Tedesco, 2016; Kumar, 2016). In addition, research has been conducted to measure

general attitudes toward green advertising and to measure the effectiveness of different message elements of green messages in order to understand green advertising (Yoon & Kim, 2016; Hartmann & Apaolaza-Ibáñez, 2009). However, research of the effects of green advertising is somewhat limited and show inconsistent results.

First, whereas some studies show that green ads are more effective in enhancing more positive consumer outcomes than non-green ads (Schuhwerk & Lefkoff-Hagius, 1995; Kong & Zhang, 2013), other studies show that there is no difference of the effects on consumer outcomes between green and non-green ads (Grimmer & Wooley, 2012; Borin et al., 2013). Secondly, many scholars assume that green ads positively enhance consumer outcomes and started to investigate what type of green ad (environmental versus personal) is more effective in enhancing more positive consumer outcomes (e.g., Borin et al., 2013; Green & Peloza, 2014; Hartmann & Ibáñez, 2008; Kareklas, Carlson & Muehling, 2012; Grimmer & Woolley, 2012). Those results are also not consistent. Whereas some studies show that environmental green ads are more effective in enhancing more positive consumer outcomes than personal green ads (Green & Peloza, 2014; Hartmann & Ibáñez, 2008; Kareklas et al., 2012; Matthes et al., 2013), other studies show that there is no difference of the effects on consumers outcomes between environmental and personal green ads (Borin et al., 2013; Grimmer & Woolley, 2012).

Previous research have failed to provide a conclusive answer whether green ads are more effective than non-green ads and under which conditions green ads are especially effective. This leads to the following research question: Are green ads more effective than non-green ads in enhancing positive consumer outcomes and under which conditions are they especially effective?

No study to date has integrated the effects of green ads. Therefore, a quantitative research that integrates and compares previous findings is essential. The current study fills this gap

the first part of the meta-analysis examines whether green ads are more effective in enhancing consumer outcomes than non-green ads. The second part of the meta-analysis examines what type of green ad (environmental versus personal) is more effective in positively enhancing consumer outcomes. Besides generating empirical generalizations, the meta-analysis tries to explain the inconsistent findings of previous research by applying moderator variables. In order to show the conditions under which green ads are more effective, a quantitative analysis that investigates moderator variables is necessary. Besides revealing the extent to which green ads are effective, the findings will also have practical implications as they give directions to marketers whether and under which conditions green ads are more effective in enhancing consumer outcomes.

The current manuscript is structured as followed. First, a definition is given of green advertising. Second, hypotheses are formulated based on previous research. Literature on green ads in general, as well as literature on the type of green ads (environmental versus personal), and literature on possible moderators that might explain inconsistent findings of previous research will be discussed. Then, the meta-analysis method is explained. After that, analyses are

performed using a random-effects model and moderator variables are examined using sub-group analyses. Finally, results are interpreted and discussed in order to provide an answer to the research question.

Theoretical Background Green versus non-green advertising

Many different definitions are used in the literature to define green ads. The following definition is one of the most cited in the literature for ‘green ads’ coming from Banerjee, Gulas

and Iyar (1995) and is used by many scholar since then (Hartman & Apaolaza-Ibáñez, 2009; Jiménez & Yang, 2008; Alniacik & Yilmaz, 2012):

If the advertisement addresses the relationship between a product or service and the environment, if the advertisement promotes a green lifestyle, but also when an advertisement promotes the environmental initiatives of the company then an

advertisement meets the criteria of being ‘green’ and is therefore considered as green advertising.

Zinkhan and Carlson (1995) conclude that green ads should be described as followed: ‘all advertisements that contain appeals that include ecological, environmental sustainability, or nature-friendly messages that target the needs and desires of environmentally concerned stakeholders’ and Kotler and Armstrong (2009) define green ad as: ‘marketing that meets the present needs of consumers and businesses while also preserving or enhancing the ability of future generations to meet their needs’. The present study combine these definitions but focuses only on green ads that target the needs for consumers, and do not focus on green ads that promote environmental initiatives of companies. This leads to the following definition of green ads: green ads advertise or promote environmentally friendly attributes of products and/or services

(Hartmann & Apaolaza-Ibáñez, 2008). In contrast, green ads are defined as followed: non-green ads do not advertise or promote environmentally attributes of products but promote other non-green attributes products, such as financial benefits (Schuhwerk & Lefkoff-Hagius, 1995). Research of the advertising effects of green ads is not consistent. First, it is not conclusive whether green ads are more effective in enhancing positive consumer outcomes than non-green ads. While some studies (Kong & Zhang, 2013; Chan et al., 2006) show that green ads are more effective in enhancing more positive consumer outcomes (attitude towards the ad, the brand, and

purchase intention) other studies (Tu & Kao, 2013; Grimmer & Woolley, 2012) show that non-green ads are more effective in enhancing more positive consumer outcomes.

Although it is not evident whether green ads work, a lot of scholars assume that it does (Hartmann, Ibáñez & Sainz, 2005; Matthes et al., 2013; Grimmer & Woolley, 2012; Green & Peloza, 2014). Many scholars (Hartmann & Ibáñez, 2009; Kong & Zhang, 2012) think that green ads increase consumers’ satisfaction, as consumers feel a sense of pride, a feeling of well-being that is associated with being altruistic. Whereas this feeling of well-being contribute to positive evaluations of green ads, consumers are also more becoming more skeptical about green ads which lead to negative evaluations of green ads (Matthes et al., 2014; Zinkhan & Carlson, 1995). Some consumers feel that green ads are vague (Hartmann & Ibáñez, 2009) and misleading (Easterling et al., 1996). In sum, findings of the effects of green ads are inconsistent. It is not clear whether green ads work effectively or not. So, the first goal of this meta-analysis is to find out whether, overall, there is a positive effect of green versus non-green ads on consumer

outcomes (attitude towards the ad, the brand, and purchase intention). This leads to the following research question:

Research question 1: Are green ads more effective in positively enhancing consumer outcomes (attitude towards the ad, the brand, and purchase intention) than non-green ads?

Moderating variables affecting the effects of green versus non-green ads on consumer outcomes.

One of the reasons that research show such inconsistent findings can be explained by different factors that influence the effects of green ads. A review of the literature reveals that the effects of green versus non-green ads seem to vary depending upon a number of moderators, such as product involvement (Zaichkowksy, 1985; Kong & Zhang, 2013; Chan et al., 2006; Phelps &

Thorson, 1991), environmental impact of the product (Yam-Tang & Chan, 1998; Kong & Zhang, 2012; Henion, Russel & Clee, 1981), and brand familiarity (Bickart & Ruth, 2013; Mobley, Painter, Untch & Unnava, 1995; Phelps & Thorson, 1991). The following sections discuss these suggested moderator variables and how they may influence the effectiveness of green versus non-green ads.

Effects of product involvement on consumer outcomes.

Product involvement has been recognized as an important variable that could influence the effectiveness of advertisements (Petty, Cacioppo & Schumann, 1983; Kong & Zhang, 2013; Zaichkowksy, 1985). Product involvement refers to a consumer’s attitude and enduring

perceptions of the importance of the product category based on the consumer’s needs, values and interests (Kong & Zhang, 2013). The present study makes a distinction between high involving products and low involving products as followed: high involving products are products that cost relatively more and/or that you do not buy on a daily basis, for example a refrigerator. Products that cost relatively less and/or that you buy on a daily (or weekly) basis are considered as low involving products, for example shampoo (Zaichkowksy, 1985).

According to the Elaboration Likelihood Model, consumers are processing information about products differently for high and low involving products (Petty et al., 1983). Consumers who want to purchase high involving products are more motivated to devote the cognitive efforts to form an evaluation and are processing information through the central route. This means that more elaborate information about the benefits of the product are more effective under high involvement conditions. In contrast, when consumers want to purchase a low involving product, they are less motivated to devote the cognitive efforts to form an evaluation, but instead focus on peripheral cues (e.g. the attractiveness or prestige of the product’s endorser) and are more likely

to process information via the persuasive route. This means that peripheral cues are more effective under low involving conditions (Petty et al., 1983).

In general, green ads contain very accessible peripheral cues like labels stating ‘100% recyclable’, ‘100% environmental’, or ‘green’. Compared to non-green ads, green ads contain less elaborate product information but more readily available information. This readily available information in green ads are considered as peripheral cues (Beltramini & Stafford, 1993;

Atkinson & Rosenthal, 2014; Kong & Zhang, 2013). If consumers want to purchase low involving products, they will make a decision based on these peripheral cues. Green ads advertising a high involvement product would be evaluated more and would therefore be

processed through the central pathway and less influenced by the peripheral cues. This is because high involving products are associated with a greater amount of risk. In this case, consumers’ evaluation is more likely to be influenced by the strength of the message. Therefore, it is

expected that green ads will be more effective when an ad concerns a product of low involvement than when an ad concerns a product of high involvement, as high involving products are less influenced by the peripheral cues that green ads provide. It is assumed that in general, green ads contain green labels. This leads to the following hypothesis:

H1: Product involvement moderates the effectiveness of green ads over non-green ads such that low involving products lead to stronger effects on a) attitude towards the ad, b) attitude towards the brand, and c) purchase intention than high involving products.

Effects of perceived environmental impact on consumer outcomes.

The previous moderator variable classified product type as either low involved or high involved. Products can also be classified based on their impact on the environment. Products can be seen as either less harmful or more harmful for the environment (Yam-Tang & Chan, 1998;

Kong & Zhang, 2012; Henion et al., 1981). Research suggests that consumers choose one product above the other if they think that they contribute positively to the environment (Teisl et al., 2002). The present study make a distinction between less harmful products and more harmful products based on the direct effects of the products on the environment. Products that not directly have an impact on the environment (e.g., paper, glass) are considered as less harmful products. Products that directly have an impact on the environment (e.g., plastic, detergent) are considered as more harmful products (Yam-Tang & Chan, 1998).

Previous research indicates that green ads influence the way consumers feel about less and more harmful products (Kong & Zhang, 2012). Consumers experience feelings of guilt when they buy a product that is perceived as more harmful for the environment. Green ads help by reducing those feelings of guilt that consumers experience as consumers want to make an effort to reduce the feeling of guilt by adopting environmental friendly alternatives for more harmful products (Kong & Zhang, 2012). Because of green ads, consumers feel that they are contributing by protecting the environment which leads to positive emotions and the feeling of altruism, which ensures more positive evaluations of the product and brand (Kong & Zhang, 2012). Those effects of green ads seem to be stronger for products that are perceived as more harmful for the environment. This is because a greater feeling of guilt is reduced (Henion, Russel & Clee, 1981; Kong & Zhang, 2012). Therefore, it is expected that green ads have more positive effects for products that are perceived as more harmful for the environment, than for products that are perceived as less harmful for the environment, leading to the following hypothesis:

H2: Environmental impact moderates the effectiveness of green ads over non-green ads such that advertisement for products that are perceived as more harmful for the environment leads to stronger effects on a) attitude towards the ad, b) attitude towards the brand, and c) purchase intention than for products that are perceived as less harmful for the environment.

Effects of brand familiarity on consumer outcomes.

Another variable that seems to influence the effects of green ads on consumer outcomes is the type of brand, that is, a familiar or unfamiliar brand. A familiar brand is defined as a brand that is well-known, whereas, an unfamiliar brand is defined as a brand that is made-up, new, or a brand that is not well-known (Stewart, 1992). A consumer can be familiar with a brand because they tried a product of the brand before or they may have seen ads of the brand before (Phelps & Thorson, 1991). When a consumer is familiar with a brand an attitude about the brand is already formed which leads to different consumer outcomes than when the brand is unfamiliar (Phelps & Thorson, 1991). In general, a familiar brand is evaluated more positive than unfamiliar brands (Hoyer & Brown, 1990). In addition, consumers have more trust in familiar brands because the perceived risk is lower than for unfamiliar brands (Pickett-Baker & Ozaki, 2008). These prior experiences consumers have with familiar brands influences the effects of green ads on consumer outcomes. This is because green ads of familiar brands are accepted more favorably due to brand trust and brand familiarity (Bickart & Ruth, 2013; Mooth, 2009). When an unfamiliar brand advertises green attributes, it is rejected because of unfamiliarity, not because consumers

consciously think about the green attributes that are promoted (Wheeler, Sharp & Nenycz-Thiel, 2013). This leads to different effects of green ads. Due to the lack of prior experiences green ads work less effective for unfamiliar brands (Bickart & Ruth, 2013; Mobley et al., 2004). This is reflected in findings that green ads from familiar brands are evaluated more positively than green ads from unfamiliar brands (Bickart & Ruth, 2013; Mooth, 2009; Mobley et al., 2004; Wheeler et al., 2013). Therefore it is expected that green ads will be more effective for familiar brands than for unfamiliar brands as familiar brands benefit from prior experiences. This leads to the

following hypothesis:

familiar brands lead to stronger effects on a) attitude towards the ad, b) attitude towards the brand, and c) purchase intention than unfamiliar brands.

Environmental versus personal green advertising

Besides research on the effects of green versus non-green ads, literature on green advertising also distinguish between different types of green ads (e.g., self-benefit versus other-benefit, utilitarian benefit versus psychological benefit or abstract versus concrete). However, most studies that examine the effects of different green ads distinguish between environmental green ads versus personal green ads (Yang, Lu, Zhu & Su, 2015; Green & Peloza, 2014; Grimmer & Woolley, 2012). Environmental green ads focus on benefits for society and

environment in general (e.g., reduce gas emissions or pollution reduction). This type of green ad highlights the fact that the main beneficiary of green consumption is some other individual or the society at large (Yang et al., 2015). Personal green ads focus on the benefits of the products that are intended for consumers (e.g., recycled plastic bag or reusable mugs). This type of green ad highlights that the main beneficiary is the consumer (Yang et al., 2015).

Previous literature on environmental versus personal green ads show inconsistent results. Segev, Fernandes and Wang (2015) argue that because consumers are self-indulgent and are primarily interested in feeling good about themselves, personal green ads are more effective in positively enhancing consumer outcomes. They provide evidence for the effects of personal green ads. Hu (2012) advices marketers to adjust their green strategies appropriately in order to be effective. This means that both environmental and personal green ads are effective, but only when they are congruent to the needs of the consumer. Whereas Hu (2012) is aiming for congruency, Borin, Lindsey-Mullikin and Krishnan (2013) advices marketers to use green strategies, irrespective of the type of green ad. They argue that is it not important which green

strategy is used. In sum, findings of the effects of environmental and personal green ads are inconsistent. It is not clear whether environmental or personal green ads work better. Therefore, the second goal of the meta-analysis is to find out which type of green ad is more effective in positively enhancing consumer outcomes (attitudes towards the ad, attitude towards the brand and purchase intention). This leads to the following research question:

Research question 2: What type of green ad (environmental versus personal) is more effective in positively enhancing consumer outcomes (attitude towards the ad, the brand, and purchase intention)?

Moderating variables affecting the effects of environmental versus personal green ads on consumer outcomes.

One of the reasons that research show such inconsistent findings can be explained by different factors that influences the effects of green ads. A review of the literature reveals that the effects of environmental versus personal green ads seem to vary depending upon a number of moderators, such as country disposition (Manrai, Manrai, Lascu & Ryans, 1997; Yoon, Kim & Baek, 2016) and the products’ goal of fulfillment (hedonic vs utilitarian product: Steinhart, Ayalon & Puterman, 2013; Wertenbroch & Dhar, 2000; Voss, Spangenberg & Grohmann, 2003). The following sections discuss the suggested moderator variables and how they may influence the effectiveness of environmental versus personal green ads.

Effects of country disposition on consumer outcomes.

Country disposition has been recognized as one variable that could explain differences in research findings of the effects of environmental versus personal green ads (Manrai et. al. 1997; Yoon et al., 2016). Nowadays, not only consumers in western countries are more concerned

about the environment. Consumers in eastern countries (e.g., China, Taiwan and India) are also becoming more aware and concerned about the environment (Chang, 2013; Xue, 2015). In response to this, companies in eastern countries are also developing more green products and processes (Xue, 2015). However, due to differences in cultures, consumers in western and eastern countries should be approached differently (Han & Shavitt, 1994; Xue, 2015). Western countries are perceived as more individualistic where individual uniqueness and self-determination are valued (Hofstede, 1980; Han & Shavitt, 1994). This means that personal goals and

accomplishment are emphasized. In a more individualistic society, communication should focus on achieving personal excellence (Han & Shavitt, 1994). Personal green ads focus on the benefits of the products which are intended for consumers and highlights that the main beneficiary is the consumer. Eastern countries are perceived as more collectivistic (Hofstede, 1980). This means that individuals identify with and work well in groups in exchange for loyalty and compliance, while personal goals and accomplishment are deemphasized (Han & Shavitt, 1994). In a more collectivistic society, communication should focus on achieving group consensus and harmony (Han & Shavitt, 1994). Environmental green ads focus on benefits for society and environmental in general. It highlights that the main beneficiary of green consumption is some other individual or the society at large (Yang, et al., 2015).

A few scholars (Uskul & Oyserman, 2010; Xue, 2015) believe that green ads should be congruent with the countries’ level of individualism vs collectivism. Previous results indicate that in eastern countries, where the level of collectivism is high, environmental green ads have more positive effects on consumer outcomes than personal green ads (Chan, 2000; Kim & Choi, 2005; Xue, 2015). This indicates that the focus of green ads (environmental versus personal) should be congruent with the culture of the country (collectivistic versus individualistic). Based on these findings and the assumption that consumers should be approached differently in western and

eastern countries, it is expected that consumers in individualistic countries have more positive consumer outcomes after seeing a personal green ad, while consumers in collectivistic countries have more positive consumer outcomes after seeing an environmental green ad. This leads to the following hypothesis:

H4: Country disposition moderates the effectiveness of environmental green ads over personal green ads on consumer outcomes a) attitude towards the ad, b) attitude towards the brand and, c) purchase intention such that environmental green ads will lead to more positive consumer outcomes in an collectivistic country, whereas personal green ads will lead to more positive consumer outcomes in an individualistic country.

Effects of hedonic versus utilitarian products on consumer outcomes.

Another well-known variable that has been recognized as an factor that influences the effects of different types of green ads (environmental versus personal) is the products’ goal of fulfillment that can have either a hedonic or utilitarian function (Steinhart et al., 2013; Dhar & Wertenbroch , 2000; Voss et al., 2003). A hedonic product is defined as a product that provides more experiential consumption, fun and pleasure such as clothes, sports cars, or perfume (Steinhart et al., 2013; Wertenbroch & Dhar, 2000). An utilitarian product is defined as a functional tool that helps to complete a practical task such as a computer, toilet paper, and detergent (Steinhart et al., 2013; Dhar & Wertenbroch, 2000). While personal goals and interests are reasons to purchase hedonic products, practical and functional goals are reasons to purchase utilitarian products (Steinhart et al., 2013). These different reasons to purchase either hedonic or utilitarian products, leads to different effects on attitude towards the products and the brand of environmental and personal green ads (Voss et al., 2003; Steinhart et al., 2013). Therefore, scholars (Bjorner, Hanse & Russel, 2004, Steinhart et al., 2013; Kangun et al., 1991) aim for

alignment between the types of green ads (environmental versus personal) and the types of products (hedonic versus utilitarian). Previous research found that environmental green ads enhance consumer outcomes more positively for utilitarian products than personal green ads (Bjorner et al., 2004; Steinhart et al., 2013; Kangun et al., 1991). This is because consumers’ choice for an utilitarian product is motivated by the need to fulfill functional aspects, which is provided by environmental green ads. The opposite is true for hedonic products. Personal green ads enhance consumer outcomes more positively for hedonic products than environmental green ads (Steinhart et al., 2013). This is because the choice for a hedonic product is motivated by self-indulgent desires, which is provided by personal green ads. This indicates that consumers will evaluate the product more favorably when the type of green ad and the type of product are aligned. Therefore, it is expected that environmental green ads will enhance consumer outcomes for utilitarian products whereas that personal green ads will enhance consumer outcomes for hedonic products. This leads to the following hypothesis is employed:

H5: The products’ goal of fulfillment moderates the effectiveness of environmental green ads over personal green ads on consumer outcomes a) attitude towards the ad, b) attitude towards the brand, and c) purchase intention such that environmental green ads will lead to more positive consumer outcomes for utilitarian products, whereas personal green ads will lead to more positive consumer outcomes for hedonic products.

Overview of Studies

Figure 1 shows the conceptual model that is based on previous hypotheses. The

hypotheses will be analyzed through a meta-analysis. Part I examines the effects of green versus non-green ads (RQ1). For this, product involvement (H1), environmental impact (H2) and brand

type (H3) will be examined as moderator variables. Part II examines the effects of environmental versus personal green ads (RQ2). For this, country disposition (H4) and the products’ goal of fulfillment (H5) will be examined as moderator variables.

Figure 1. Conceptual Model

Method Study Retrieval

To identify relevant empirical studies that examine the effects of green ads, a

computerized keyword search using the databases PsychInfo, BSP, CMMC, and ScienceDirect was conducted1, followed by an internet search using Web of Science and Google Scholar. In order to be complete, all the references of the studies that were included in the meta-analysis were searched for further articles. All the studies were carefully examined according to the following inclusion criteria.

In order to be considered, the study had to focus on the effects of green compared to non-green ads, or on the effects of environmental compared to personal non-green ads on the dependent

1

This literature search and the selection of articles (out of more than 10.000 articles) was done as part of a student assistantship. The development of inclusion criteria and the ultimate selection of articles for this thesis was done as part of the Master Thesis.

variables attitude towards the ad, attitude towards the brand, or purchase intention. Furthermore, only studies that were reported in English were included - simply because of the practical difficulties of translation. In addition, the ad promoted either a product or service. Lastly, there were no restrictions with regard to geographical or cultural restrictions, how participants were assigned to different conditions, and no time frame was applicable. This search procedure resulted in 17 studies (the included studies are marked by an asterisk in the reference list). See appendix II for an overview why selected articles were not included in the current meta-analysis.

Next, it was investigated whether each study provided the necessary empirical results (e.g., means with standard deviations, F-values, t-values, Chi-square, or an Beta with standard deviations) in order to compute the effect size. Please note that only data from experimental studies were included as combining study findings derived from different research designs will lead to difficulty by incorporating the findings in one meta-analysis (Lipsey & Wilson, 2011). When the article did not provide sufficient information to calculate effect sizes, the authors of the article were requested for more statistical information. Five studies could not be included due to lack of data for calculating the effect size, so these were dropped, leaving a total of 12 studies (see Appendix II). The included studies were conducted between 1995 and 2014. The effect sizes provided data on 10.714 participants with sample sizes ranging from 35 to 800 participants and were conducted in a variation of western and non-western countries (e.g., United States,

Australia, China and Taiwan). The included studies provided 48 independent effect sizes.

Variables

The first part of the meta-analysis examined the effect of green versus non-green ads. Green ads advertise or promote products or services that are environmentally friendly. Non-green ads do not promote the environmentally attributes of products but promote other non-green

attributes of products (e.g., discount). The second part of the meta-analysis examined the effects of environmental versus personal green ads. An environmental green ad focuses on benefits for society and environment in general, whereas an personal green ad focuses on the benefits of the products which are intended for consumers. The dependent variables in the meta-analysis were attitude towards the attitude towards the ad, attitude towards the brand, and purchase intention. Attitude towards the ad can be thought of as whether the consumers like or dislike the ad (Phelps & Thorson, 1991). Attitude towards the brand is defined as an individual’s internal evaluation of an object, product or brand (Spears & Singh, 2012) and purchase intention is defined as the decision to act that shows an individual’s behavior according to the product (Spears & Singh, 2012).

Coding procedure of the moderators

After the studies were retrieved, two coders coded the information on the moderators independently. Any disagreements were resolved through discussion by the two coders. The first part of the meta-analysis examined the moderator variables product involvement, environmental impact and brand type. Product involvement was coded as either 0 = low involved and 1 = high involved. Products that cost relatively less and/or that consumers buy on a daily (or weekly) basis are products that were coded as less involved (e.g., shampoo, detergent). Products that cost relatively more and/or that you do not buy on a daily (or weekly) basis were coded as high involved (e.g., refrigerator, car). For product involvement there was only discrepancy about vehicle fuel efficiency. It was decided that vehicle fuel efficiency was coded as a low involving product as people buy it regularly, even though it is not particularly cheap. Vehicle fuel

efficiency is a product that you need when it comes to a high involving product (e.g., car), however, the product itself is a low involvement product. Products that are necessary for using

high involving products, such as dishwashing cubes and detergent are therefore considered as low involving products. The intercoderreliability for: product/service involvement was α = 0.842. Environmental impact was coded as either 0 = less harmful for the environment and 1 = more harmful for the environment. Products that are less harmful are defined as products that have an indirect effect on the environment (e.g., papers or glass). Products that are harmful for the environment are defined as products that directly have an impact on the environment (e.g., plastic or pesticide). It was more difficult to find agreement on environmental impact. For environmental impact there was only discrepancy about ecological laundry detergent. It was decided that ecological laundry detergent was coded as a more harmful product. Although

‘ecological laundry detergent’ is a product that is less harmful for the environment in comparison with ‘normal laundry detergent’, consumers evaluate it as being equally harmful. The

intercoderreliability for environmental impact was α = 0.779.

Brands that are well known in general (e.g., Coca Cola, Starbucks) and brands that were specified in the article as familiar were coded as 0 = familiar. Brands that were specified to be made up (fictitious brands) or brands that were not known by consumers were coded as 1 = fictitious/unfamiliar. There was no discrepancy about the moderator variable brand type as each study provided clear information, therefore the intercoderreliability for was α = 1.00. Table 1 (see Appendix I) gives an overview of the studies and moderator variables included in part I.

The second part of the meta-analysis examined the moderator variables country

disposition and the products’ goal of fulfillment. Country disposition was either coded as western or eastern. Countries in Europe, North America, and Oceania were coded as 0 = western

countries and countries in Asia, Africa, and South America were coded as 1 = eastern countries. Country disposition was coded based on information provided in the sample descriptives. If there was no sufficient information provided in the sample descriptives, country disposition was coded

based on where the article was published.

The products’ goal of fulfillment was coded as either 0 = a hedonic product and 1 = an utilitarian product. A hedonic product is defined as a product that provides more experiential consumption, fun and pleasure (e.g., clothes, sports cars, or flowers). An utilitarian product is defined as an important and functional tool that helps to complete a practical task (e.g., computer, toilet paper, and detergent). There was no discrepancy about the moderator variable country disposition as each study provided clear information. The intercoderreliability for each moderator was: country disposition α = 1.00 and the products’ goal of fulfillment: α = 0.842. Table 2 (see Appendix I) gives an overview of the studies and moderator variables included in part II.

Statistical method

The common effect size metric selected for the analyses is the standardized mean

difference (ESd), with a positive ESd value indicating a stronger effect for green ads compared to non-green ads on consumer outcomes for Part I, and with a positive ESd value indicating a stronger effect for personal green ads compared to environmental green ads on consumer outcomes for Part II. For part I, the dependent variables scores of the non-green condition were considered as control group and the green condition as treatment group. For part II, the dependent variables scores of the environmental condition were considered as control group and the

personal condition as treatment group. Effect sizes were calculated with the online effect size calculator of Lipsey and Wilson (2011). Thereafter, an unbiased standardized mean difference has been calculated in order to remove a small positive upward bias that can affect the effect size (Lipsey & Wilson, 2011). Each calculated effect size was treated as statistically independent as each effect size is calculated by unique and independent information (Borenstein, Hedges, Higgins & Rothstein, 2009).

The mean effect sizes were calculated by weighting each individual study effect size by the inverse of its variance, using a random-effects model. A random-effects model was used as there was no reason to assume that there would be one true effect size (Lipsey & Wilson, 2011). The mean effect sizes were significant when the confidence interval did not include zero. In addition, the mean effect sizes were tested for significance by z statistics. A homogeneity test was used to see whether there was variance between studies, that might be due to other factors than sampling error alone. If the homogeneity test was significant, it indicated that there is variance left that can be due to moderator variables. If the test indicated heterogeneity, it was decided whether sub-group analyses were applied in order to test if the suggested moderator variables could explain the inconsistent findings. In order to establish whether there is a different effect of the subgroups of the moderator variables, the magnitude of effects in different subgroups are compared. The moderator variable was not investigated if the variable did not provide a sufficient number of effect sizes for each category. All the analyses were performed in SPSS 22 with

Macros for meta-analyses (Lipsey & Wilson, 2011).

Results

Part I of the meta-analysis examined the effects of green versus non-green ads on consumer outcomes (attitude towards the ad, the brand and purchase intention). First, overall mean effects are reported that examine what type of ad (green versus non-green) is more effective in positively enhancing consumer outcomes. Secondly, results of homogeneity tests are reported that examine whether the differences between effect sizes are caused by other sources than sampling error alone (Lipsey & Wilson, 2011). As a third, results of sub-group analyses are reported that examine under what conditions green ads are more effective in positively enhancing consumer outcomes. Hereafter, results of part II are presented in the same way. Part II of the

meta-analysis examined the effects of environmental versus personal green ads on consumer outcomes (attitude towards the ad, the brand and purchase intention).

Meta- Analysis Part I: Overall mean effects of green versus non-green ads

First of all, with a random-effects model it was examined whether green ads are more effective in positively enhancing attitude towards the ad, attitude towards the brand and purchase intention. Results (see Table 3 in Appendix I) showed that green ads do not have more positive effects on attitude towards the ad than non-green ads. No significant effect was found, d = 0.138, SE = .161, CI = -0.178 to 0.454, p = .392. Though green ads do not have more positive effects on attitude towards the ad than non-green ads, results showed that green ads do have more positive effects on attitude towards the brand than non-green ads, d = 0.397, SE = .998, CI = 0.201 to 0.592, p < 0.001. This concerns an effect with a medium effect size (Cohen, 1977). Similar as attitude towards the ad, results showed that green ads do not have more positive effects on purchase intention than non-green ads. No significant effect was found, d = 0.194, SE = .133, CI = -0.066 to 0.454, p = .143.

The overall mean effects provided an answer to the first research question that questioned whether green ads are more effective in positively enhancing consumer outcomes (attitude towards the ad, the brand and purchase intention) than non-green ads. It is found that green ads have a positive overall effect on attitude towards the brand. The positive overall effect indicates that green ads are more effective than non-green ads in positively enhancing attitude towards the brand. Nonetheless, green ads do not have a positive overall effect on attitude towards the ad and purchase intention. This means that, overall, green ads are not more effective than non-green ads in positively enhancing attitude towards the ad and purchase intention.

Part I: Moderator Analyses

Second, the homogeneity analysis indicated whether the differences between effect sizes are greater than expected from sampling error alone (Lipsey & Wilson, 2011). Thus, high heterogeneity indicated that the variability between studies is greater than the variability within studies which means that there are moderator variables that could possibly influence the overall effect size. Results showed variability across effect sizes for attitude towards the ad (Q(9) = 69.81, p < 0.001), indicating that there is some unexplained variance left. Furthermore, results showed no variability across effect sizes for attitude towards the brand (Q(4) = 8.39, p = .078). This means that the effect on attitude towards the brand is not influenced by moderator variables and that all the studies agree on the magnitude of the effect. Thus, there is no need to test

moderator variables for this relation, as there is no unexplained variance left. Similar to attitude towards the ad, results also showed variability across effect sizes for purchase intention (Q(7) = 17.10, p < 0.05). As the heterogeneity test showed significant results for attitude towards the brand, only sub-group analyses were conducted to examine under what conditions green ads are more effective in positively enhancing attitude towards the ad and purchase intention.

Low versus high involving products. Hypothesis 1 assumed that low involving products

have stronger effects on consumer outcomes (attitude towards the ad, the brand and purchase intention) than high involving products. Results of the sub-group analysis indicated that the effect size of low involving products, d = 0.141, SE = .208, CI = -0.266 to 0.548, p = .498 was not any different than the effect size of high involving products, d = 0.143, SE = .395, CI = -0.632 to 0.917, p = .718 on attitude towards the ad. An analysis of variance confirmed that there was no significant difference in the average effect sizes of low versus high involvement products or services on attitude towards the ad (Qb(1) < 0.01, p = .997).

product was coded as high involving, so there were not enough effect sizes for performing the sub-group analysis. Based on these results, Hypothesis 1, that assumed that low involving products leads to stronger effects on consumer outcomes (attitude towards the ad, the brand and purchase intention) than high involving products, is rejected. In contrast to what was expected, low involving products do not lead to stronger effects on consumer outcomes than high involving products for green ads.

Less harmful versus more harmful products. Hypothesis 2 assumed that more harmful

products have stronger effects on consumer outcomes (attitude towards the ad, the brand and purchase intention) than less harmful products. Results of the sub-group analysis indicated that the effect size of less harmful products, d = 0.204, SE = .245, CI = -0.276 to 0.683, p = .404, is not any different than the effect size of more harmful products d = 0.088, SE = .213, CI = -0.329 to 0.505, p = .679 on attitude towards the ad. An analysis of variance confirmed that there was no significant difference in the average effect sizes of less versus more harmful products or services on attitude towards the ad (Qb(1) = .128, p = .271).

Similarly, for the relation on purchase intention, results of the sub-group analysis

indicated that the effect size of less harmful products, d = 0.120, SE = .198, CI = -0.267 to 0.507, p = 0.545, is not any different than the effect size of more harmful products d = 0.290, SE = .212, CI = -0.126 to 0.706, p = .172. An analysis of variance confirmed that there was no significant difference in the average effect sizes of less versus more harmful products or services on attitude towards the ad (Qb(1) = .133, p = .557). Based on these results, Hypothesis 2, that assumed that more harmful products leads to stronger effects on consumer outcomes (attitude towards the ad, the brand and purchase intention) than high involving products, is rejected. In contrast to what was expected, more harmful products do not lead to stronger effects on consumer outcomes for green ads.

Familiar versus unfamiliar brands. Hypothesis 3 assumed that familiar brands have

stronger effects on consumer outcomes (attitude towards the ad, the brand and purchase intention) than unfamiliar brands. No sub-group analysis was conducted for the relation on attitude towards the ad as only one product was coded as familiar, so there were not enough effect sizes for performing the sub-group analysis.

Results of the sub-group analysis indicated that the effect size of familiar brands d = 0.137, SE = .219, CI = -0.293 to 0.566, p = .534, is not any different than the effect size of unfamiliar brands d = 0.241, SE = .185, CI = -0.122 to 0.604, p = .193 on purchase intention. In addition, an analysis of variance confirmed that there was no significant difference in the average effect sizes of familiar versus unfamiliar brands on purchase intention (Qb(1) = .133, p = .715). Based on the results, Hypothesis 3, that assumed that familiar brands leads to stronger effects on consumer outcomes (attitude towards the ad, the brand and purchase intention) than unfamiliar brands, is rejected. In contrast to what was expected, familiar brands do not lead to stronger effects on consumer outcomes for green ads.

Meta- Analysis Part II: Overall effects of environmental versus personal green ads The second part of the meta-analysis examined which type of green ad (environmental versus personal) is more effective in positively enhancing attitude towards the ad, attitude towards the brand and purchase intention. Results showed (see Table 4 in Appendix I) that personal green ads do not have more positive effects on attitude towards the ad than personal green ads. No significant effect was found, d = 0.394, SE = .216, CI = -0.029 to 0.817, p = .068. The p-value did gave an marginal trend towards significance (p = .068) and provided an

indication that personal green ads might have more positive effects on attitude towards the ad than environmental green ads.

Results further indicated that personal green ads do not have more positive effects than environmental green ads on attitude towards the brand, d = 0.386, SE = .278, CI = -0.159 to 0.931, p = .165, and purchase intention, d = 0.239, SE = .196, CI = -0.145 to 0.622, p = .222. The overall mean effects provided an answer to the second research question that

questioned what type of green ad (environmental versus personal) is more effective in positively enhancing consumer outcomes (attitude towards the ad, the brand and purchase intention). It is found that personal green ads have, although not significant, a positive overall effect on attitude towards the ad. The marginal trend towards significance provided an indication that personal green ads might have more positive effects in positively enhancing attitude towards the ad than environmental green ads. Results further indicate that personal green ads are not more effective than environmental green ads in positively enhancing attitude towards the brand and purchase intention, or conversely. This means that, overall, not one type is more effective than the other in positively enhancing attitude towards the brand and purchase intention.

Part II: Moderator Analyses

Second, the homogeneity analysis indicated whether the differences between effect sizes are greater than expected from sampling error alone (Lipsey & Wilson, 2011). Results showed variability across effect sizes for attitude towards the ad (Q(7) = 190.02, p < 0.001), attitude towards the brand (Q(7) = 367.73, p < 0.001) and purchase intention (Q(8) = 154.02, p < 0.001). The high heterogeneity indicated that the variability between studies is greater than the variability within studies, indicating that there is some unexplained variance left. Therefore, sub-group analyses are conducted to examine under what conditions green ads (environmental versus personal) are more effective in positively enhancing attitude towards the ad, attitude towards the brand and purchase intention.

Collectivistic versus individualistic countries. Hypothesis 4 assumed that environmental

green ads will lead to more positive consumer outcomes in an collectivistic society and that personal green ads will lead to more positive consumer outcomes in an individualistic society. Results of the sub-group analysis indicated that the effect size of western countries, d = 0.226, SE = .419, CI = -0.596 to 1.047, p = .590, is not any different than the effect size of eastern countries d = 0.453, SE = .245, CI = -0.029 to 0.934, p = .652 on attitude towards the ad. An analysis of variance confirmed that there was no significant difference in the average effect sizes of western versus eastern countries on attitude towards the ad (Qb(1) = .218, p = .640).

Similarly, for attitude towards the brand, results of the sub-group analysis indicated that the effect size of western countries, d = 0.001, SE = .543, CI = -1.062 to 1.065, p = .998, is not any different than the effect size of eastern countries d = 0.514, SE = .313, CI = -0.099 to 1.127, p = .101. An analysis of variance confirmed that there was no significant difference in the average effect sizes of western versus eastern countries on attitude towards the ad (Qb(1) = .670, p = .413).

Lastly, results of the sub-group analysis indicated that the effect size of western countries d = -0.037, SE = .289, CI = -0.602 to 0.529, p = .899, is not any different than the effect size of eastern countries d = 0.430, SE = .237, CI = -0.035 to 0.896, p = .070 on purchase intention. In addition, an analysis of variance showed that there was no significant difference in the average effect sizes of western versus eastern countries on purchase intention (Qb(1) = 1.562, p = .211).

Based on the results, Hypothesis 4, that assumed that environmental green ads will lead to more positive effects on consumer outcomes (attitude towards the ad, the brand and purchase intention) in an collectivistic society and that personal green ads will lead to more positive effects on consumer outcomes (attitude towards the ad, the brand and purchase intention) in an

not lead to more positive consumer outcomes in collectivistic societies and personal green ads will not lead to more positive consumer outcomes in individualistic societies.

Utilitarian versus hedonic products. Hypothesis 5 assumed that environmental green ads

will lead to more positive consumer outcomes for utilitarian products and that personal green ads will lead to more positive consumer outcomes for hedonic products. Results of the sub-group analysis indicated that the effect size of hedonic products, d = 0.053, SE = .253, CI = -0.444 to 0.549, p = .835, is different than the effect size of utilitarian products d = 0.715, SE = .240, CI = 0.244 to 1.185, p = .003 on attitude towards the ad. Although the effect size of utilitarian products on attitude towards the ad is significant, an analysis of variance showed that the difference in the average effect sizes of hedonic versus utilitarian products on attitude towards the ad, was not significant (Qb(1) = 3.60 p = .058). However, the p-value did indicate an marginal trend toward significance (p = .058) and gave directional evidence that the products’ goal of fulfillment (hedonic versus utilitarian products) have different effects on attitude towards the ad. This means that when the advertisement concerns an utilitarian product, an personal green ad will be more effective in positively enhancing attitude towards the ad.

Results of the sub-group analysis further indicated that the effect size of hedonic products, d = 0.054, SE = .344, CI = -0.620 to 0.782, p = .875, is different than the effect size of utilitarian products d = 0.715, SE = .342, CI = 0.046 to 2.095, p = .036 on attitude towards the brand. Although the effect size of utilitarian products on attitude towards the brand is significant, an analysis of variance confirmed that there was no significant difference in the average effect sizes of hedonic versus utilitarian products on attitude towards the brand (Qb(1) = 1.859, p = .173).

Similar to attitude towards the brand, results of the sub-group analysis indicated that the effect size of hedonic products, d = 0.180, SE = 0.357, CI = -0.519 to .880, p = .613, is not any different than the effect size of utilitarian products d = 0.266, SE = .260, CI = -0.244 to 0.775, p =

.307 on purchase intention. An analysis of variance confirmed that there was no significant difference in the average effect sizes of hedonic versus utilitarian products on purchase intention (Qb(1) = .037 p = .847).

Based on the results, Hypothesis 5, that assumed that environmental green ads will lead to more positive effects on consumer outcomes (attitude towards the ad, the brand and purchase intention) for utilitarian products and that personal green ads will lead to more positive effects on consumer outcomes (attitude towards the ad, the brand and purchase intention) for hedonic products, is rejected. In contrast to what was expected, environmental green ads will not lead to more positive consumer outcomes for utilitarian products and personal green ads will not lead to more positive consumer outcomes for hedonic products.

Conclusion and Discussion

The current study offered a quantitative research that integrated and compared previous findings and provides conclusive results of the effects of green ads. The findings of the meta-analysis provided an answer to the question: are green ads more effective than non-green ads in enhancing positive consumer outcomes (attitude towards the ad, the brand and purchase

intention) and under which conditions are they especially effective?

Part I of the meta-analysis provided an answer to the question: are green ads more effective in positively enhancing consumer outcomes than non-green ads? The findings partially support the notion that green ads have more positive effects than non-green ads. The medium effect of green ads on attitude towards the brand shows that consumers evaluate brands more positively after seeing green ads than after seeing non-green ads. This indicates that consumers evaluate brands more positively that engage in green strategies. In spite of the positive effects of

green ads on brand attitudes, the study did not find positive effects of green ads on attitude towards the ad and purchase intention.

Additionally, Part I did not find evidence to support H1, H2 and/or H3: The moderator variables product involvement, environmental impact and brand type did not affect the relation of green ads on attitude towards the ad, attitude towards the brand and purchase intention

differently, in contrast to what many scholars argued (Kong & Zhang, 2013; Chan et al., 2006; Yam-Tang & Chan, 1998; Kong & Zhang, 2012; Henion, Russel & Clee, 1981; Bickart & Ruth, 2013).

Part II of the meta-analysis provided an answer to the question: what type of green ad (environmental versus personal) is more effective in positively enhancing consumer outcomes? The findings of part II of the meta-analysis provided an indication that personal green ads might have more positive effects on attitude towards the ad than environmental green ads. Nonetheless, it cannot be assumed that environmental and personal green ads have different effects on attitude towards the ad. Additionally, environmental and personal green ads do not have different effects in positively enhancing attitude towards the brand and purchase intention. This means that it does not matter which type of green strategy is used in enhancing more positive consumer outcomes. With this, part II support the notion that the type of green strategy chosen is not critically important (Borin et al., 2013).

Furthermore, Part II did not find evidence that support H4 and H5: The moderator variables country disposition and the products’ goal of fulfillment did not affect the relation of environmental versus personal green ads on attitude towards the ad, attitudes towards the brand and purchase differently, in contrast to what many scholars argued (Manrai, Manrai, Lascu & Ryans, 1997; Yoon, Kim & Baek, 2016; Steinhart, Ayalon & Puterman, 2013; Voss,

Theoretical implications

The findings of the present meta-analysis provide theoretical implications for research on green advertising. First and foremost, the meta-analysis provides evidence about the effectiveness of green versus non-green ads. The overall findings of green versus non-green ads provide new insights. Herewith, it contributes to the discussion whether green ads are overall more effective in positively enhancing consumer outcomes than non-green ads and fill up the research gap,

showing that green ads are, overall, more effective in enhancing attitude towards the brand than non-green ads. With this, it supports research on green advertising that claim that green ads are more effective than non-green ads (Hartmann, Ibáñez & Sainz, 2005; Matthes, Wonneberger & Schmuck, 2013; Grimmer & Woolley, 2012; Green & Peloza, 2014). By way of contrast, green ads and non-green ads are equally effective in enhancing attitude towards the ad and purchase intention.

Findings further imply that whether products are considered as low or high involving or perceived as less or more harmful for the environment does not affect consumer outcomes differently, despite previous research results which show they do (Kong & Zhang, 2013; Chan et al., 2006; Yam-Tang & Chan, 1998; Kong & Zhang, 2012). An explanation for the different findings may be that previous research examined the effects of green ads under certain circumstances (e.g., low environmentally involved versus high environmentally involved), whereas the current study examined the overall effects. Herewith, it demonstrates the direction of all the effect sizes together (Lipsey & Wilson, 2011).

Secondly, the findings of the current meta-analysis further contribute to research on environmental versus personal green ads, showing that both green ads (environmental and personal green ads) are equally effective in enhancing attitude towards the brand and purchase intention, despite results of previous research (Laroche et al., 2001; Hartmannn & Ibáñez, 2005;

Pickett-Baker & Ozaki, 2008; Segev et al., 2015; Hu, 2012)Although findings of the overall effects of personal green ads over environmental green ads are not significant (p = .068) it does provide directional evidence that personal ads might be more effective in enhancing attitude towards the ad than environmental green ads. The attractiveness of an ad plays a major role in how consumers evaluate an ad (Belch & Belch, 2015). This can be an explanation why there are different effects of personal and environmental green ads. For this, more experimental research is necessary that can confirm the findings found in this meta-analysis.

Findings, furthermore, give interesting implications about the moderator variable product’s goal of fulfillment (hedonic versus utilitarian products). The current study does give directional evidence that it might be the case that personal green ads are more effective in enhancing attitudes towards the ad for utilitarian products. These findings are not in line with findings that show that utilitarian products have stronger effects for environmental green ads instead of personal green ads (Steinhart et al., 2013). The different results may be due to whether it was a product or service that was advertised. Steinhart, Ayalon and Puterman (2013) based their research on products, whereas in this meta-analysis services were also included. Therefore, more experimental research is necessary in order to examine whether the product’s goal of fulfillment indeed affect the relation of environmental versus personal green ads on consumer outcomes differently.

The results fill up the research gaps that was caused by the inconsistent findings of previous results. As the results of the main effects are based on a meta-analytic approach that compare findings of experimental studies, broad generalizations can be made. Herewith, an decisive answer to the question whether green ads are more effective and under which conditions can be given.

Practical implications

Besides providing theoretical implications, the findings are of great importance for practitioners. Hence, practical implications are provided.

First, practical implications for marketers are discussed whether and how they should use green ads or non-green ads. Findings are of high relevance for organizations who are integrating environmentally responsible initiatives to their program in order to positively enhance their image and consumer outcomes. Organizations that become aware of the importance of being environmentally friendly and that respond to this by, for example, engaging in environmentally friendly processes (e.g., KLM’s use of biofuel) or adding green products to their product range (e.g., Coca Cola Life), experience benefits of these innovations, as it enhances the overall brand attitude. This implies that consumers evaluate brands that engage in green strategies more positively than brands that do not engage in green strategies. Furthermore, the results indicate that marketers do not have to take into account whether a product is a low or high involving product, or a less or more harmful product for the environment, as the findings showed that product involvement (low versus high) and environmental impact (less versus more) do not affect the consumers outcomes differently. The same applies to whether a brand is familiar or

unfamiliar. It makes no difference whether a familiar brand or an unfamiliar brand advertises or promotes environmentally friendly attributes of products and/or services. This means that brands that are new or less known, also can integrate environmentally responsible initiatives to their program. As new or less known brands equally affect consumer outcomes than familiar brands, it implies that the prior experiences consumers have with familiar brands are less important when it comes to green ads. This support the notion that consumers are likely to choose one product over another if they think that they are helping the environment, even if the brand is unfamiliar

Secondly, practical implications are discussed how marketers can use green ads (environmental and personal) beneficial and effective. The current meta-analysis reveals that marketers do not have to choose between environmental (emphasizing the benefits for the environment) or personal (emphasizing the benefits for the self) green ads. As it turns out, marketers can use both types of green ads (environmental and personal) equally effective. This implies that, overall, environmental and personal green ads have the same effect in enhancing consumer outcomes. These findings are especially useful for marketers who want to introduce a new brand to the market and do not have (yet) chosen a strategy. However, marketers should note that the findings give direction of the effect sizes. It especially gives a clear overview of how, in general, green advertising work.

Findings further give practical implications for international organizations.Organizations notice that not only consumers in western countries become more aware and concerned about the environment (Chang, 2013; Xue, 2015). Similarly, consumers’ interest in the environment in eastern countries has also grown enormously. The findings of the current meta-analysis imply that when marketers want to advertise internationally, they do not have to take into account whether the country is more collectivistic or individualistic, as previous research argued (Manrai, Manrai, Lascu & Ryans, 1997; Yoon, Kim & Baek, 2016). This means that marketers can use the same ad in in a collectivistic country as in an individualistic country.

Whereas it does not matter what type of green ad (environmental or personal) is used in what country (collectivistic or individualistic), findings do show that the product’s goal of fulfillment (that can have either a hedonic or utilitarian function) may matter. Findings gave directional evidence that when the advertisement concerns an utilitarian product (e.g., computer, toilet paper, or detergent), an personal green ad will be more effective in positively enhancing attitude towards the ad. These findings indicate that marketers should take into account the

reasons why consumers purchase a product, and adjust their green strategy to this. So, when the ad advertises for example toilet paper, a slogan like ‘You can help: 100% recyclable’ would work better than ‘You can help the environment: 100% recyclable’, in enhancing attitude towards the ad. Please note that this difference in effect only applies for attitude towards the ad.

Limitations and future research

Although a quantitative research that integrates and compares findings is essential in order to provide empirical generalizations, conducting a meta-analysis is accompanied by some general limitations. For a start, the findings are limited to information that is available in the studies (Lipsey & Wilson, 2011). A number of studies could not be included due to the lack of statistical information (see Appendix II), which resulted in less effect sizes. This is partly because the current meta-analysis focused explicitly on green ads and excluded all the studies that did not meet the requirements of a green ad, which is considered a more narrowed down approach. A narrowed down approach is useful as it gives marketers more detailed and precise advice about the effectiveness of green ads. A few scholars pointed out that green advertising is the most important communication tool of green marketing (Chan, Leung & Wong, 2006; Kumar, 2016). This illustrates the importance of a detailed advice about how green ads should be used properly in order to positively enhance consumer outcomes. This detailed advice gives directions about how consumers evaluate green ads, and thus how marketers should approach consumer

effectively. As a detailed and precise advice is a strength of this narrowed down approach, it also has a downside. Marketers should take into account that this advice is based only on green ads and not on green messages in general (e.g., information messages about the corporate social responsibility claims; Chen & Lee, 2015) As green advertising is part of green marketing, it should be valuable to also investigate how consumers evaluate green messages that promotes, for

example, social responsibility initiatives of an organization. In that way, green messages are examined in a more broad way, which can give different insights or strengthen the findings of the present meta-analysis.

A second common limitation of the current meta-analysis is that not all studies provide enough information to test for further moderator variables. Only studies that provide enough information could be examined. Meta-analysis is limited to variables that each included study measured, or to variables that each study manipulated. That is why the variable environmental involvement could not be examined in the current meta-analysis. This is unfortunate, as research show that consumers respond differently to green ads based on the consumers’ level of

environmental involvement (Grimmer & Woolley, 2012; Schuhwerk and Hagius, 1995; D’Souza, 2005). As research on green messages is limited, more experimental research is needed in order to investigate additional moderator variables that could explain the differences between studies. In this way, research can provide a clearer picture of the effectiveness of green ads.

Conclusion

The current meta-analysis provided conclusive results of the effects of green ads (green versus non-green ads and environmental versus personal green ads), and revealed the effects of some important moderator variables. The most important findings are that marketers can use green ads in order to enhance brand attitudes instead of using non-green ads, as consumers value brands more positively that advertise or promote environmentally friendly attributes of products and/or services, than brands that are not. However, this does not mean that consumers evaluate green ads also positively, as findings show that green ads are not more effective in enhancing attitude towards the ad than non-green ads. The same applies for purchase intention. Although green ads are more effective in enhancing attitude towards the brand, it does not necessarily mean