It is time for cow’s milk to moo-ove over
Understanding the purchase intention of Dutch consumers towards
plant-based alternatives to milk
Thesis for obtaining the Master of Science in Marketing
By:
University of Groningen
Faculty of Economics and Business Administration
It is time for cow’s milk to moo-ove over
Understanding the purchase intention of Dutch consumers towards
plant-based alternatives to milk
Master Thesis
Author: Tessa Renes
Visschersstraat 41, 8325BS Vollenhove (The Netherlands) t.renes.1@student.rug.nl
Student Number: 3032310
Course of studies: Master Thesis Marketing (EBM867B20)
Year: 2018
Supervisor: Dr. Wander Jager
Management summary
The changing climate of the planet is considered to be a major health issue that impacts the food security. One of the main contributors to this change is the dairy industry, which is responsible for up to 26% of all CO2 emissions. And therefore it is emitting a higher percentage of greenhouse gas than
transportation. But humans can reduce their environmental impact on the planet by changing their diets, such as a diet that includes more plant-based products. One way to increase a person’s intake of plant-based products is to start consuming plant-based milk instead of cow’s milk. This is better for the environment because the production of plant-based milk has less environmental impact than the production of cow’s milk. Producing one liter of cow’s milk results in 1467 grams of CO2, compared
to 397 grams of CO2 for one liter of soy milk and 396 grams of CO2 for one liter of almond milk. Yet,
the market of plant-based alternatives can still considered being a niche market, as plant-based milk accounted for 3% of the total dairy market in Europe in 2015. However, people are increasingly aware that changing their diet can have positive effects on the environment, although environmental concerns do not seem to be the main focus of people when changing to a plant-based diet.
Therefore, this paper aims at studying why people do or do not buy plant-based milk. The research question we want to answer is: what are the driving elements and barriers influencing consumers’ purchase intention of plant-based milk? We answer this question with the determinants of the Theory of Planned Behaviour: the effects of the attitudes, the social norms and the perceived behavioural control on the intention to purchase plant-based milk. The results build on a survey with a sample of 168 respondents, in which the participants were asked to indicate their purchase frequency of dairy products. In the second part of the questionnaire the respondents were assigned to one of two treatment conditions. Both conditions showed an advertisement displaying a descriptive norm for almond milk. The last part of the survey asked the respondents to the determinants of the TPB.
positive influence on the purchase intention and is the second most influential determinant. The perceived availability and whether plant-based milk has a reasonable price do not influence the purchase intention. Yet, the importance of the price of plant-based milk during the purchase decision however does positively influence the purchase intention. Finally, we looked at the different types of descriptive appeals and its effect on the purchase intention. There is no differential effect between a health and environmental appeal.
Our study has implications for the manufacturers of plant-based milk and the retailers that sell them. As our results show is the habitual component critical for the purchase intention of plant-based milk. Since the majority of our sample does not look at alternatives and this has a negative effect on the intention to purchase plant-based milk. Hence, managers should focus on people establishing a new habit with regard to buying plant-based milk. The first step to do this is by letting people taste plant-based milk in order to get a first-hand experience. As our results illustrate that most people indicate that they do not know the taste of plant-based milk, while we did find the taste to have the largest impact on the intention to purchase plant-based milk. The second step for managers to establish a habit is by frequency marketing, because the more frequently a behaviour is performed by a person the more likely it becomes a habit.
This paper contributes to the academic field of plant-based alternatives to animal-based products. We contribute to the literature by showing that the habitual component is critical for a person’s purchase intention. This implies that first the old habits must be broken in order for people to change to plant-based diet. Yet, our results show that the purchase frequency of cow’s milk has no effect on the purchase intention of plant-based milk and it has no effect on whether people look at alternatives. So, other factors affect why people do not look for alternatives, for example the lack of time or effort to look for alternatives, the lack of awareness about alternatives or people are not interested in alternatives.
Preface
This study is written as part of the Master Marketing at the University of Groningen from February to June 2018. The basis for the topic of this thesis originates from my interest in the organic and ‘clean’ products, as during my visit to the supermarket I frequently look for the best environmental option. Since I do think that the easiest, and perhaps most impactful, way for people to lower their impact on the environment is by changing their dietary behaviour. For example by lowering the consumption of meat. That is why for the subject of my thesis I was considering studying organic alternatives for dairy products, after which my supervisor Dr. Wander Jager suggested to study plant-based milk. I hope that my research can make a contribution to the understanding of people’s behaviour regarding plant-based alternatives. As I do hope that people will consume less animal-based products in the near future.
Several people have contributed to this thesis. First of all, I would like to thank my supervisor Dr. Wander Jager for his suggestions, guidance and constructive comments throughout the entire process. Secondly, I thank my family for their continuous support and ideas. Furthermore, I am grateful to everyone that took the time to fill in my questionnaire and without whom I could not finish this thesis. And finally I want to thank you for reading my Master thesis. And I hope you will find this topic and the following results just as interesting as I did. Enjoy reading my thesis.
Tessa Renes
Table of contents
List of abbreviations VII
List of figures and tables VIII
1 Introduction 1
2 Theoretical framework 3
2.1 Habitual consumption 3
2.2 Theory of planned behaviour 5
2.3 Conceptual model 6 2.4 Advertising 12 3 Research methodology 13 3.1 Experiment 13 3.2 Measurement 15 3.3 Data collection 16 4 Data analysis 16 4.1 Descriptive statistics 17
4.2 Factor and reliability analyses 22
4.3 Testing the control variables 23
4.4 Testing the hypotheses 24
4.5 Additional analyses 28
5 Conclusions and recommendation 28
5.1 Discussion 28
5.2 Managerial and academic implications 31
5.3 Limitations and future research 32
5.4 Conclusions 33
References 35
List of abbreviations
CO2 Carbon dioxide
KMO Kaiser-Meyer Olkin measure of sampling adequacy
TPB Theory of Planned Behaviour
U.S. United States
List of figures and tables
Figure Title Page
Figure 1 Model of Theory of Planned Behaviour 5
Figure 2 Conceptual model 12
Figure 3 Purchase frequency of milk 18
Figure 4 Purchase frequency of the different types of milk as indicated by the respondents
19
Figure 5 Taste of the different types of milk as indicated by the respondents 19 Figure 6 Healthiness of the different types of milk as indicated by the respondents 20 Figure 7 Environmental impact of the different types of milk as indicated by the
respondents
21
Figure 8 Conceptual model with the standardized coefficient betas 27 Figure 9 Advertisement with descriptive environmental appeal 43
Figure 10 Advertisement with descriptive health appeal 43
Table Title Page
Table 1 Descriptive statistics 17
Table 2 Amount of euros the each type of milk costs as indicated by the respondents 21
Table 3 Varimax rotated component matrix 22
Table 4 Correlation matrix – attitude 51
Table 5 Correlation matrix - subjective norm 51
Table 6 Correlation matrix - perceived behavioural control 52
1 Introduction
The climate change of the planet is considered to be a significant public health issue that impacts people’s food security (Joyce, Dixon, Comfort, & Hallett, 2012). A major contributor to the climate change is the livestock industry and especially the dairy farms. Globally, the milk production, processing and transport are responsible for 26% of the total CO2 emissions (FOA, 2010). Of which
18,7% originates solely from the production and the processing of the milk (Thoma et al., 2013). The livestock industry is emitting a higher percentage of greenhouse gas than transportation (Steinfeld et al., 2006). In addition to this is the livestock industry a contributor to a limited freshwater availability and the pollution of it, it limits the biodiversity (Steinfeld et al., 2006) and causes deforestation, as plains of the rainforests are cut down yearly for soy plantations to feed cattle (National Geographic, 2016). Thereby indirectly contributing to the extinction of other animal species (Harrison, 2018). But humans can reduce their environmental impact on the planet by changing their diets. It is recommended that people increase their consumption of plant-based products and lower the intake of animal-based products (Joyce et al., 2012; Baumann, 2013). This is because the environmental impact of crop production is less than that of livestock (Davis, Sonesson, Baumgartner, & Nemecek, 2010), as a plant-based diet saves water supplies and crops that otherwise would have been necessary to feed the livestock (Harrison, 2018). Hence, people with a completely plant-based diet have a smaller carbon footprint than those that consume animal-based products or even vegetarians (Nijdam et al., 2012; Baumann, 2013). Therefore what a person chooses to eat can be just or even more important as his choice of transportation (Eshel & Martin, 2006).
People are increasingly aware that changing their diet can have positive effects on the environment, although environmental concerns do not seem to be the main focus of people when changing to a plant-based diet (Joyce et al., 2008; Wyker & Davison, 2010). This is because people believe the main way to help the environment is by changing their choice of transportation (Joyce et al., 2008). The reason why people change to a plant-based diet seems to be the expected health benefits (Lea, Crawford & Worsley, 2006), as Joyce et al. (2008) found that people would consume less meat because of these health benefits. One barrier for adopting a plant-based diet is the lack of information, as a study by Lea, Crawford and Worsley (2006) found that people needed more information on nutrition and on preparing plant-based dishes. Other reasons for sticking to a diet with animal-based products are convenience, ease of preparation, lack of personal interest and that of family members agreeing to change to a plant-based diet (Lea, Crawford & Worsley, 2006; 2005).
producing one liter of cow’s milk results in 1467 grams of CO2, compared to 397 grams of CO2 for
one liter of soy milk and 396 grams of CO2 for one liter of almond milk (Unnasch, 2018).
The Dutch population is changing their dietary behaviour as 55% indicated they consumed less meat (Van Hest, 2017). This dietary change can also be seen in the sales of milk, which lowered with 8% in 2017 (Kassa, 2017). Sugary beverages, like sodas, and plant-based alternatives are taking over cow’s milk place (Miller, 2014; Luckerson, 2014). The growing demand for plant-based dairy in Europe has found its origin in America (Wilschut, 2015), where the sales of non-dairy milk increased by 61% from 2012-2017 (Chew, 2018). But despite the growing sales in America, the share of plant-based dairy products in Europe is limited, accounting for only 3% of the dairy market in 2015 (Wilschut, 2015). Therefore, plant-based alternatives can still be considered a niche market, holding a small percentage of the total market share (Mousel & Tang, 2016). Yet, the market of plant-based milk is predicted to grow with a Compound Annual Growth Rate of 15.5% and is expected to reach 19.5 billion USD in 2020 (Marketsandmarkets, 2015).
Five categories of plant-based alternatives to milk can be distinguished: legume based (soy, peanut), nut based (almond, coconut, walnut), seed based (sesame, hemp), cereal based (rice, oat), and pseudo-cereal based (quinoa) (Sethi, Tyagi, & Anurag, 2016). But not all plant-based milks are beneficial for the environment, as a lot of water is needed for the production of soy and almonds, which increases the environmental impact (Judkins, 2017; Kassa, 2017). Other varieties as oat are a better alternative when it comes to water usages for crop production (Kassa, 2017). Soy milk was long time the favourite non-dairy milk for people, but has now been surpassed by almond milk, accounting for 64% of the market, while soy has 13% market share and coconut 12% (Shoup, 2018). The segment of almond milk is growing the fastest due to reasons that people can ‘understand’ the product and the taste is considered good and neutral for adding other flavors (Green, 2017a). Still people are diversifying their non-dairy purchases and are looking for new varieties such as quinoa and rice (Shoup, 2018), which are expected to be the first segments to do well with the customer (Chew, 2018).
The consumption of plant-based milk is viewed by people as part of a lifestyle choice rather than only for those who are lactose intolerant or allergic to milk protein (HPS, n.d.). As people nowadays purchase drinks for their specific functionalities, such as enhancing one’s energy, instead of drinks just being thirst-quenchers (Sethi, Tyagi, & Anurag, 2016). People purchase plant-based milk as a contribution towards a personal goal, such as consuming less animal products and thereby not contributing to animal mistreatment or lowering their impact on the environment (McCarthy et al., 2017), or because of allergies (lactose intolerance and milk proteins allergy), or concerns for one’s health (Watson, 2018; Cassetty, 2018). Barriers people face for purchasing plant-based products are the taste, the price (Watson, 2018) and the availability of the products (Walton, 2017).
purchase intention of plant-based milk? We do this by using the determinants of the Theory of Planned Behaviour. We choose to study the dairy market, because dairy products are part of people’s daily consumption pattern in the Netherlands (ZuivelNL, 2015). Since milk is consumed daily, there is a lot of potential to reduce the impact of the Dutch population on the environment by changing to plant-based milk instead. But in order to achieve this change in the Dutch consumption pattern, one must understand the determinants of the dietary behaviour and how to change it (Joyce et al., 2012). Currently there exists a small amount of research on the necessary conditions for changing the consumption pattern of people towards a plant-based diet and the processes involved in this change (Joyce et al., 2012). Therefore, we want to research what motivates people to purchase plant-based milk, such as the key drivers and drawbacks. As emphasizing these drivers can help to enhance the sales of plant-based milk.
Next to the drivers and drawbacks of purchasing plant-based milk we analyse whether the messages stressed in the advertisement can help to enhance the intention to purchase plant-based milk. As campaign messages directed towards the determinants of the behaviour can influence people’s dietary choices (Joyce et al., 2012). Therefore, we study campaign appeals directed towards the positive health and environmental benefits of consuming plant-based milk. Concluding, our research will provide an understanding of the intention to purchase plant-based milk and it will lead to new ideas on how to market this product to the consumer. The rest of this paper continues as follows. Chapter 2 presents the theoretical background, the conceptual model and the hypotheses. Chapter 3 discusses the methodology and chapter 4 reports the analyses of the data. The paper concludes with chapter 5, which states the discussion, the managerial implications and suggestions for further research.
2 Theoretical framework
2.1 Habitual consumption
As stated in the introduction do we first need to understand the determinants of the dietary behaviour of the Dutch population regarding milk and how to change it. Currently there exists a small amount of research on the necessary conditions for changing the consumption pattern of people towards a plant-based diet and the processes involved in this change (Joyce et al., 2012). First we evaluate the product milk itself, after which we explore the habitual purchase behaviour of people and on how to break this habit.
greater emphasis is placed on practicality, usefulness and fulfilment of consumers’ basic needs (Kim & Kim, 2016). Whereas hedonic consumption on the other hand is pleasure oriented (Strahilevitz & Myers, 1998) as people seek enjoyment and make decisions based on feelings (Pham, 1998). Therefore, purchasing milk can be seen as a low involvement activity of which the outcome is goal oriented.
Grocery shopping or, as in our study, buying milk is automated by people through the formation of habits (Fennis & Stroebe, 2016). A habit can be defined as a learned sequence of acts that has become an automatic response to a specific cue (Verplanken & Aarts, 1999). In order for people to break a habit a considerable amount of effort is required, as habits cause people to not actively seek for new information and to not take new information into account while performing the certain behaviour (Jager, 2003). Even if people are aware of the negative outcomes of the habit in the long-term, information concerning these negative outcomes does not affect the performance of the habit as longs as the short-term outcomes of this habit are satisfying (Jager, 2003). Hence, the disadvantage of habits is that they are difficult to change, even when people have formed an intention to do so (Fennis & Stroebe, 2016). This is because habits lock people into making the same choice, as changing this choice requires switching costs, such as time and effort (Murray & Häubl, 2007). For example by always buying the same type of milk a person saves time and effort by knowing where to find it. But purchasing a different type of milk would cost more time and effort, as the person now has to search for it (Fennis & Stroebe, 2016).
Since buying milk is a low involvement activity, with people not actively searching for information or evaluating the product characteristics, we expect people that frequently buy milk do not look at alternative types of milk, as this would require more time and effort. So, our first hypothesis is:
H1: The higher the purchase frequency of milk, the less alternatives are being considered
Following hypothesis one we expect people that do not look at alternatives are locked in habitual purchase behaviour and thus have a lower intention to purchase plant-based milk. This could be because people are satisfied with the product they are currently buying and do not take new information into account (Jager, 2003) and therefore are not aware of a possibly better alternative. In addition to this requires making a different choice more cognitive effort than making the same choice (Fennis & Stroebe, 2016). Thus, we expect people that do not look at alternatives while buying milk have a lower purchase intention of plant-based milk. Hence, our second hypothesis is formulated as:
H2: The less people look at alternatives the lower is the purchase intention of plant-based
Subsequently, we expect people that purchase milk more frequently have a lower intention to purchase plant-based milk. An antecedent of a habit is the frequency of the behaviour and in order to form a habit, the certain behaviour must be performed frequently and repeatedly in a stable environment (Chiu, Hsu, Lai & Chang, 2010). The more frequently people perform a behaviour, the more likely it becomes a habit and the stronger a habit is, the higher are the costs to switch to another brand or product (Chiu, Hsu, Lai & Chang, 2010). Accordingly, our third hypothesis is:
H3: The higher the purchase frequency of milk the lower is the purchase intention of
plant-based milk
Jager (2003) indicates three ways to change a habit, namely: 1) removing the stimuli and therefore making the performance of the habit impossible, 2) providing information on the negative outcomes of the habit and on the positive outcomes of the wanted behaviour and 3) emphasize the positive short-term outcomes of the wanted behaviour and thereby increasing the chance of the formation of the new habit. In our study we focus on option 2, to provide information on the positive outcomes of the wanted behaviour, which will be further described in section 2.4 and 3.1. Option 1, removing all the stimuli from the consumer’s environment, is not possible on the short-term as the cow’s milk is not going to be removed from the supermarket aisles. Option 3, that considers the positive short-term outcomes such as the taste of plant-based milk, is studied as a determinant of the Theory of Planned Behaviour and will be discussed in section 2.3.1.
2.2 Theory of Planned Behaviour
In order to research what the driving elements and barriers are for the purchase intention of plant-based milk we use the Theory of Planned Behaviour (TPB) (Ajzen, 1991). This theory is a cognitive model of human behaviour and is based on three determinants of a person’s behavioural intention, as can be seen in Figure 1.
The first determinant attitude stands for the degree to which a person has a favourable or an unfavourable evaluation of performing a behaviour (Ajzen, 1991). The second determinant is the subjective norm, which reflects the perceived social pressure a person feels to perform or not perform a behaviour (Ajzen, 1991). Lastly, the third determinant is the perceived behavioural control, which is the ease or difficulty a person has performing a behaviour (Ajzen, 1991). The more positive the attitude and the subjective norm regarding the behaviour and the greater the perceived behavioural control, the stronger a person’s intention to perform the behaviour should be (Ajzen, 1991). All the three determinants of the TPB are important to consider as persuading consumers of the positive attributes of the product is ineffective when consumers are not going to purchase it, for example when family members do not want them to or when a person cannot afford it (Fennis & Stroebe, 2016). The three determinants of the TPB influence the intention to purchase a product. This purchase intention is an indicator of the effort people are willing to exert and the stronger this intention is, the more likely it is people will actually perform the behaviour (Ajzen, 1991).
The TPB has been used in the environmental literature to analyse the psychological factors that drive consumers’ purchase behaviour towards environmental friendly and plant-based products and has been applied in a variety of contexts. Vermeir and Verbeke (2008) researched the food consumption of organic products in Belgium and found that 50% of the variance in the intention to purchase organic milk was explained by the combination of attitudes, social norms and perceived behavioural control. The attitudes were the main predictor of behavioural intentions. Mousel and Tang (2016) studied consumer behaviour towards plant-based meat and dairy and found the attitude to have a positive effect on the purchase intention, while the subjective norm had an positive effect on the purchase intention through the positive feedback from friends and family. Next to this they found that the purchase intention of plant-based products strongly influences the actual purchase behaviour. Which is similar to the results of Vermeir and Verbeke (2006), who found that the behavioural intention and performing the actual behaviour are correlated, although it is not perfectly aligned.
2.3 Conceptual model
The conceptual model used in this study is anticipated in offering an understanding of the intention to purchase plant-based milk and is based on the aforementioned TPB. Each determinant of the conceptual model will be described next.
2.3.1 Attitude
(Kozup, Creyer & Burton, 2003). The attitude determinant consists of a limited number of salient and accessible beliefs (Ajzen, 2005; Fishbein and Ajzen, 1975). A prior study on plant-based dairy found that environmental consciousness, animal welfare, health-related values and taste influence consumers’ attitudes (Mousel & Tang, 2016). In order to study the determinant attitude we have identified the following variables: environmental consciousness, animal welfare, health consciousness, taste and subjective knowledge. Together these determinants will sum up the attitude consumers have regarding plant-based milk.
Environmental consciousness represents the concern for the environment (Tarkiainen & Sundqvist, 2009) and consists of the collectivistic side of motivational values (Schwartz & Bilsky, 1987; 1990). When buying a product consumers do pay attention to its environmental impact (Kassa, 2017), but according to research are environmental concerns not the main focus of consumers when changing to a plant-based diet (Joyce et al., 2008; Wyker & Davison, 2010). Even though environmental concerns matter less when making decisions for consumers these aspects are gaining in importance (Gustin, 2018). We expect that consumers that are more environmental conscious are more likely to purchase plant-based milk. Thereby the fourth hypothesis is formulated as:
H4: The more environmental conscious people are, the stronger is the purchase intention of
plant-based milk
Consumers choose plant-based alternatives to cow’s milk because of concerns for the animal welfare (Gustin, 2018). Cows’ milk is considered to be not animal friendly, as most cows seldom spend time outside and need to give birth annually to be able to produce milk (Wilschut, 2015). We expect that people with a higher concern for the welfare of animals are more likely to purchase plant-based milk. Thus, the fifth hypothesis is formulated as follows:
H5: The higher the concern for animal welfare, the stronger is the purchase intention of
plant-based milk
aware of the potential health-related benefits of a plant-based diet (Lea, Crawford & Worsley, 2006). We predict people that are more health conscious are more likely to purchase plant-based milk than people that are less health conscious. Thereby the sixth hypothesis is formulated as:
H6: The more health conscious people are, the stronger is the purchase intention of
plant-based milk
One barrier for people to purchase plant-based milk is its taste, as a study by Hoek et al. (2011) found the sensory appeal to be a main barrier for consumer to switch to plant-based products. People have the prejudice of soy-based products to be not very tasteful (Wanskin et al., 2005). But the taste of almond milk is found to be neutral and good (Green, 2017a). Technical developments did improve the taste of dairy-free products and helped the products to appeal to a wider audience (Green, 2017b). 40% of U.S. consumers stated flavour is the most important determinant when purchasing alternatives for milk and yogurt (Green, 2017b; Watson, 2018). As for our next hypothesis we expect the better people think the taste of plant-based alternatives is, the more likely they are to purchase these products. And therefore the seventh hypothesis is:
H7: The better the perceived taste of plant-based milk, the stronger is the purchase intention
of plant-based milk
The product knowledge can be described as a person’s understanding and memories associated with the product (Brucks, 1985). A lack of product knowledge can prevent people from purchasing a product, even if they would like to (Fennis & Stroebe, 2016). Which corresponds to the results of Liu, Segev and Villar’s (2017) study, who found that people with more knowledge or information about environmental friendly products are more inclined to purchase these products, as they have the ability to adequately evaluate the products. The current knowledge of people of plant-based alternatives is low, as in a research by De Bakker and Dagevos (2010) people indicated that they lack knowledge and information about the environmental and nutritional benefits of plant-based proteins.
people’s ability differences. This indicates that the subjective knowledge of people is a stronger predictor of the purchase intention than objective knowledge. In line with this theory we expect for measuring the purchase intention it to be more important if people think they have a lot of knowledge about plant-based milk rather than if they have actual knowledge about it. This is demonstrated by House et al. (2004) who showed that high levels of subjective knowledge is significantly and positively linked to the willingness to eat genetically modified food, but did not find a significant relationship for objective knowledge. Aertsens et al.’s (2011) study had similar results, as they found subjective knowledge to be positively related towards the attitude to buy organic food. Therefore, we choose the subjective product knowledge as the focus of our study as the attitude towards performing a specific behaviour is measured. In accordance with the discussed literature we expect that the more knowledgeable a person thinks he is, the more likely he is to buy plant-based alternatives. The eighth hypothesis formulated as:
H8: The higher the subjective knowledge of plant-based milk, the stronger is the purchase
intention of plant-based milk
2.3.2 Subjective norm
The second determinant of the TPB is the subjective norm, which consists of normative beliefs and the motivation to comply (Ajzen, 1991). The normative beliefs are the likelihood that important others approve or disapprove of a certain behaviour, while the motivation to comply is to act in accordance with the important others (Ajzen, 1991). People can experience two types of social influence: normative and informational. Normative social influence is defined by Deutsch and Gerard (1955: 629) as the “influence to conform to the positive expectations of another” and the informational social influence is the “influence to accept information obtained from another as evidence about reality” (Deutsch & Gerard, 1955: 629).
Hence, in accordance with the aforementioned social influence theory of Deutch and Gerard (1955) we expect that if the social expectations are that a person should engage in a certain behaviour, then this person is expected to be more likely to do so. Thus, if purchasing plant-based milk is seen as a social desirable behaviour, people are more likely to buy it. However, we first must know the current social norm regarding purchasing plant-based milk, if purchasing plant-based milk is indeed seen as socially desirable behaviour. In the U.S. consumers’ attitudes towards a plant-based diet are mostly positive, as 26% consider it to be very healthy and 35% somewhat healthy, while 19% of the consumers thought it to not be healthy (Statista, 2016). Thus, we can assume that plant-based milk shares these attitudes. We predict that people who encounter a positive subjective norm regarding purchasing plant-based milk have a higher intention to buy it. Therefore, the ninth hypothesis is formulated as follows:
H9: The more positive the subjective norm towards plant-based milk, the stronger is the
purchase intention of plant-based milk
2.3.3 Perceived behavioural control
The third and last determinant of the TPB is the perceived behavioural control, which is formed by the opportunities and resources that are needed to engage in a certain behaviour (Ajzen, 1991). People form the intention to performing a behaviour if they are confident they can act upon it (Armitage & Conner, 2001). For people the main reason for not purchasing plant-based milk seems to be the access to affordable plant-based alternatives to cow’s milk (Forum for the Future, 2016; Walton, 2017; Watson, 2018). A survey among U.S. consumers shows that the first reason why people hold back from trying new type of ‘healthy’ foods is the price, followed the lack of availability in stores (PRNewswire, 2016). Hence, we expect that the barriers people encounter are price and availability.
The first barrier people encounter is the price of plant-based milk. People use the price of a product as an extrinsic cue to make value judgements (Lee, Bhatt & Suri, 2018), such as the quality of a product. Or to assess the monetary sacrifice people have to make when purchasing the product (Lee, Bhatt & Suri, 2018).The price of food influences what and how much people buy and therefore it is a driver of the food choices they make (Andreyeva, Long & Brownell, 2010).
amount of their disposable income on food than people in high-income groups and therefore are more affected by higher prices (Andreyeva, Long & Brownell, 2010; Smed, Jensen & Denver, 2007). The price elasticities of products can vary depending on people’s habits and preferences and the number of available alternatives (WHO, 2015). In the U.S. the price elasticity of almond milk is -3,50 (Dharmasena, Capps & Kosub, 2015), this indicates that almond milk is very price sensitive. Therefore an increase in the price will result in a larger reduction in consumption, but a decrease in the price will result in a large increase in demand. The cross price elasticity of almond milk with soy milk is 0,22, this indicates that soy milk acts as a substitute for almond milk (Dharmasena, Capps & Kosub, 2015). Concluding, we expect people that perceive the price of plant-based milk to be higher have a lower perceived behavioural control and thereby have a lower intention to purchase plant-based milk. Thus, the tenth hypothesis is formulated as follows:
H10: The higher the perceived price of plant-based milk, the weaker is the purchase intention
of plant-based milk
The second barrier is the availability of plant-based milk. The perceived availability is the feeling that a person can easily obtain or consume a product (Sparks & Shepherd, 1992). People have a tendency of purchasing certain products only if these are available on a regular basis (O'Donovan & McCarthy, 2002; Rana & Paul, 2017). In accordance Vermeir and Verbeke (2008) found that people who are convinced environmental friendly products are easily available are more inclined to actually purchase these products. But the availability of plant-based milk in Europe lags behind that of the U.S., as people lack access to a wide range of non-dairy alternatives and seek a greater variety in the plant-based products that currently are available (Forum for the Future, 2016). Hence, we expect people that perceive a limited availability of plant-based milk have a low perceived behavioural control and thereby have a lower intention to purchase it. Thus, the 11th hypothesis is formulated as:
H11: The lower the perceived availability of plant-based milk, the weaker is the purchase
All the hypotheses concerning the TPB are summarized in the conceptual model shown below in
Figure 2.
Figure 2. Conceptual model
2.4 Advertising
As stated by Jager (2003) can a person’s habit be changed by providing information on the positive outcomes of the wanted behaviour. Therefore, we will use advertisements in order to persuade people to change their habit of purchasing cow’s milk and instead opt for plant-based milk. Advertising is defined as a paid form of communication intended to inform and/or persuade the targeted audiences (Fennis & Stroebe, 2016). It has the capability to change people’s behaviours and values, and to influence people to learn and to adopt norms (McQuail, 2010). People can be affected by and respond to advertisements in three ways: 1) affective (emotions and moods), 2) behavioural (intention and actual behaviour) and 3) cognitive (beliefs and thoughts) (Fennis & Stroebe, 2016).
not motivated to do so, they rely more on peripheral processing and thereby do not spend effort on thinking about the presented arguments or to think about counterarguments (Petty & Cacioppo, 1986a). According to the multiple-role assumption can an advertising variable serve as a heuristic cue (peripheral route) but also as an argument (central route), depending on the level of processing of the consumer (Petty & Cacioppo, 1986a).
The advertisements of cow’s milk currently displayed on television and print media focuses mainly on the product characteristics or the health benefits of milk. Although the Dutch brand Campina does use known Dutch athletes to promote their dairy products (Ros, 2015). The brands of plant-based milk brands communicate messages focussed on the health benefits, which are found to impress both dairy and non-dairy milk consumers (Decker, 2018). In order to persuade people to eat more plant-rich diets, brands should highlight the sustainability and nutrition benefits of such a diet (Forum for the Future, 2016), rather than focussing on the negative effects of the diet of animal-based products (Beverland, 2014). Findings of Joyce et al. (2012) show that people respond more readily towards a campaign concerning a plant-based diet that focuses on health-related benefits rather than on environmental benefits. One explanation for this effect can be that health-related values are individualistic whereas environmental values are collectivistic (Schwartz & Bilsky, 1987; 1990). Hence, we think that people are more inclined to purchase plant-based milk when reading about the health benefits instead of the expected environmental improvements. We expect that an advertisement showing the health benefits of plant-based milk is more successful than an advertisement displaying environmental benefits and therefore our 12th hypothesis is:
H12: An advertisement with a health appeal leads towards a higher purchase intention of
plant-based milk, than an advertisement with an environmental appeal
3 Research methodology
3.1 Experiment
& Kallgren, 1990), this learning effect can not only take place from friends and family, but also from advertising, especially when people lack first-hand knowledge (Wanke, 2009). Thus, advertising can also shape social norms (Wanke, 2009). And social norms can affect the behaviour of people even when they are not aware of it (Aarts & Dijksterhuis, 2003; Nolan et al., 2008). Cialdini (2003) suggests that people’s pro-environmental behaviour, such as resource conservation, can be enhanced by insights from social influence theories like social norms and social comparison. Of which social norms are one of the most used social influence approaches to instigate a behavioural change (Abrahamse & Steg, 2013). And as prior research showed, are social norms a powerful inducement for sustainable behaviour (Allcott 2011; Peattie 2010; Thøgersen 1999). Also in the marketing has the usage of norms shown to be effective in changing the behaviour of people, as people engage in more ethical consumption when they find others doing so (Starr, 2009). So did Goldstein, Cialdini and Griskevicius (2008) find in their study that appeals with descriptive norms outperform traditional appeals to behave environmental friendly. As stated previously do people not always thoroughly process an advertisement and therefore rely more on heuristic cues to form attitudes towards products (Fennis & Stroebe, 2016). Hence, for our experiment we expect that people will use the presented descriptive social norm as a heuristic cue.
In order to test H12 we design an experimental study to analyse the effect of the appeal of an advertisement and whether the type of appeal influences the purchase intention for the advertised product. The experiment consists of an advertisement, which displays the product, plant-based milk, with an appeal. The appeal consists of a descriptive norm in combination with a claim. Two treatment conditions are designed to manipulate the descriptive norm the respondents will see in the advertisement, one with an environmental message and one with a health-related message. The environmental message is “30% of the Dutch people drink almond milk because it is better for the environment”, while the health focused message is “30% of the Dutch population drink almond milk because it is better for your health”. The number of 30% of the Dutch population drinking almond milk for a specific reason is a made up number. We solely focus on the differential effect of the descriptive norm itself, rather than focussing on a traditional appeal versus a descriptive appeal, since Goldstein, Cialdini and Griskevicius (2008) found in their study that appeals with descriptive norms outperform traditional appeals. Therefore, we want to study whether there is a differential effect between the specific benefits the descriptive message is directed at.
adjusted a picture posted by the brand Alpro on Facebook to suit the experiment. The brand name Alpro is removed from the advertisement in order to prevent confounding effects of prior brand experience, accessibility and knowledge. For the first treatment advertisement the health related appeal is added to the photo, whereas for the second treatment advertisement the environmental appeal is included. The created advertisements can be seen in Appendix B.
The design of this study is a 2 x 1 between-subjects experimental design. A between-subjects design is chosen, as each respondent will be exposed to only one treatment condition in order to prevent learning effects. The respondents of the questionnaire will be randomly and evenly assigned to the two treatment groups and have an equal chance of being allocated to one of the groups. The remaining parts of the survey, namely the descriptive and multi-scale items will be exactly the same for all respondents.
3.2 Measurement
In order to test H1-H3 we will ask the respondents to the frequency of their purchase behaviour, such as the purchase frequency of milk during a week and the purchase frequency of fresh, sterilized and plant-based milk. Next to this we will ask the respondents if they look at alternatives during their shopping behaviour. The scale items used to test H4-H11 in our questionnaire are taken from prior studies, that validated these scales. The items to measure the environmental conscious, the concern for animal welfare, the health conscious and the taste are measured with two-items scales each from the research by Myresten and Setterhall (2015) (α = 0.80). We took two items from Peštek, Agic and Cinjarevic (2018) (CR1 = 0.74) to measure the subjective knowledge, who adapted these from the five
items of Flynn and Goldsmith (1999). The subjective norm is measured with three items by Kumara, Manraib and Manraib (2017) (α = 0.89). The price and availability of the perceived behavioral control component are both measured with a item scale by Vabø and Hansen (2016) (α = 0.77). A two-item scale by Tucker et al. (2012) (α = 0.90) is used to capture the purchase intention. All the scales are measured on a 7-point Likert scale to prevent the number-of-levels effect. Which occurs when an item has a smaller or larger scale than the other items and thereby appears less or more important, biasing the given answers (Currim, Weinberg & Wittink, 1981). The scales are adapted to suit the plant-based milk tested and can be seen in Appendix A, which shows a full overview of all the items used to measure the constructs.
main concerns for based milk differ according to age, with the main reason for purchasing plant-based milk being the concerns for animal welfare for age segments millennials (born in 1980-2000) and generation Z (born in 1995-2010). Therefore, we control for the effects of the demographic variables. The second confounding variable is the purchase frequency of plant-based milk. As people who have previously bought plant-based milk have more experience with the product, which increases the stored knowledge (Flynn & Goldsmith, 1999). Next to this could the mere exposure effect be at work. The mere exposure effect implies that the frequency of exposure increases the liking for an object (Zajonc, 1968). Hence, people that buy plant-based milk more frequent can also like the product more due to frequently being exposed to it. Contrary to people that never or hardly buy it. As a third confounding variable we included whether the respondent is allergic to almonds, since this influences a person’s purchase intention. As a fourth confounding variable we considered to include whether participants are conformist or individualist. Since it could be that people who rank higher on the conformist scale are more sensitive to a descriptive norm, as used in our experiment. Yet, we do not include this variable, since it is not the main focus of this study and marketers cannot segment on these two groups.
3.3 Data collection
A digital questionnaire is developed with the online program Qualtrics in order to study the formulated hypotheses. The questionnaire starts with demographic questions, followed by items that measure the habitual purchase behaviour of milk. Next, the advertisement with the environmental or health message is shown to the respondents. After this the purchase intention is measured, followed by the attitude and the perceived behavioural control items. Finally, the subjective norm is measured. The target group of this research is the Dutch population and therefore we developed the survey in Dutch, to ensure that every participant can truly understand and answer the questions asked. The scale items measuring the attitude, subjective norm, perceived behavioural control and the purchase intention are converted from English to Dutch with back translation (Malhotra, 2010), in which the items are first translated to Dutch after which the items are translated back to English in order to find any translation errors. The full survey in English and in Dutch can be seen in Appendix C. The data is collected through the social media Facebook in a 15-day period lasting from 30-04-2018 until 14-05-2018. Our sampling method is biased by the so-called self-selection bias, which entails that people can chose themselves to participate in the study or not (Malhotra, 2010). A total of 186 responses is collected.
4 Data analysis
are discussed first, followed by the factor analysis, the testing of the control variables, the testing of the hypotheses and the additional analyses.
4.1 Descriptive statistics
The descriptive characteristics of the respondents can be seen in Table 1. The ages of the respondents vary between 19 and 68 years with an average age of 26,48. It can be seen that a large part of our sample consists of females (73,8%) and are aged between the 19 and 29 years (90,5%). The largest part of the respondents (64,3%) has an annual income between the €0 - €10.000. Most participants have finished a higher educational level, as 76% of our sample has finished HBO or a higher level.
Variable n=168 % Gender Female 124 73,8 Male 44 26,2 Age distribution 19-29 152 90,5 30-39 2 1,2 40-49 2 1,2 50-59 4 2,4 60-69 8 4,8
Education Primary education 0 0
High school 19 11,3 VMBO 1 0,6 MAVO 1 0,6 MBO 20 11,9 HBO 43 25,6 WO (bachelor) 54 32,1 WO (master/doctoral) 30 18,3 Annual income €0 - €10.000 108 64,3 €10.000 - €20.000 32 19 €20.000 - €30.000 15 8,9 €30.000 - €40.000 8 4,8 €40.000 - €50.000 1 0,6 €50.000 - €100.000 4 2,4 More than €100.000 0 0
Table 1. Descriptive statistics
HBO or WO bachelor, while 18,4% of the Dutch population did. Far more people in the Dutch population finished an education on VBMO (21%) and MBO (29%) level, than did in our sample (0,6% for VMBO and 11,9% for MBO). The majority of the respondents has an annual income level between €0 and €10.000 (64,3%), while for the Dutch population this is only 16% (CBS, 2018c). The only income level that does look similar to the Dutch population is the percentage of respondents with an annual income of €10.000 - €20.000 (19%), which is 26,3% for the Dutch population (CBS, 2018c). Overall, we can conclude that our sample does not reflect the Dutch population and therefore is not generalizable to represent the Dutch population.
Following the demographic items, the respondents answered questions about their habitual behaviour regarding fresh, sterilized and plant-based milk and what the respondents think of the taste, the healthiness, the environmental impact and the price of the different types of milk. First, we asked the respondents how often they purchase milk during a week. The minimum amount the respondents indicated is 0 times, while the maximum is 5 times. The average is 1,13 times a week. The distribution of the purchase frequency can be seen in Figure 3. When asked if the respondents look at alternatives during purchasing milk 54 (32,1%) indicated that they did, while 114 (67,9%) of the respondents did not look at alternatives. Hence, the majority of our sample does not look at alternatives when shopping for milk.
In the survey the respondents were asked to indicate their purchase frequency of fresh, sterilized and plant-based milk. The answers can be seen in Figure 4, of which the y-axis displays the number of respondents. Most people never purchase sterilized and plant-based milk. But people are divided with their purchase frequency of fresh milk as 32,7% indicate to never buy it, 24,4% sometimes and 23,8% always. Hence, the purchase frequency of sterilized and plant-based milk is low, while fresh milk is purchased more often.
Figure 4. Purchase frequency of the different types of milk as indicated by the respondents
Next, we asked the respondents to the taste of the different types of milk. The answers of the respondents can be seen in Figure 5. People think the taste of fresh milk is good, while for the other types of milk most respondents indicate that they do not know what the taste of the milk is like. This illustrates that most people have not tried the other types of milk besides fresh milk.
Figure 5. Taste of the different types of milk as indicated by the respondents
that fresh milk is reasonable healthy or healthy. While people have differential opinions about sterilized milk, as 23,81% indicated that it is reasonably unhealthy and 29,98% think it is reasonably healthy. All the plant-based milks are considered to be healthy.
Figure 6. Healthiness of the different types of milk as indicated by the respondents
Subsequently, we asked the respondents to the environmental impact of the different milk types.
Figure 7 shows that people do think that fresh and sterilized milk have a lot of impact on the
Figure 7. Environmental impact of the different types of milk as indicated by the respondents
We asked the respondents if they could indicate how many euros one liter of the different types of milk costs. As can be seen in Table 2 respondents think that on average sterilized milk and fresh milk are the least expensive. While coconut milk and cashew milk are thought of to be the most expensive drinks. The difference between the average lowest (fresh milk) and the highest (cashew) is €0,94. In the last column we added the current retail prices of the different types of milk. When comparing the average price of the respondents with the retail prices of the different types of milk, we can see that there is little difference between the two. Only did the respondents think that the price of soy milk is higher than the retail price.
Minimum Maximum Average Retail price2
Fresh milk (whole, semi-skimmed, skimmed) €0,40 €3,00 €1,24 €1,13
Sterilized milk (UTH) (whole, semi-skimmed, skimmed) €0,40 €4,00 €1,19 €1,32
Soy milk €0,80 €4,00 €1,83 €1,39
Almond milk €0,90 €4,00 €1,98 €1,99
Coconut milk €0,90 €4,00 €2,05 €2,05
Rice milk €0,50 €4,00 €1,87 €1,86
Cashew milk €0,80 €5,00 €2,18 €2,09
Table 2. Amount of euros the each type of milk costs as indicated by the respondents and the retail price
2For the retail price of fresh and sterilized milk we used the semi-skimmed milk of the brand Campina, while for the
4.2 Factor and reliability analyses
The components of our conceptual model, the attitude, the subjective norm, the perceived behavioural control and the purchase intention, are each measured with multi-scale items. In order to reduce and combine these items we will perform factor analyses. A factor analysis with all the determinants of the conceptual model showed that the items measuring the attitude determinant can be grouped into two components, while the items measuring the subjective norm are one component. Lastly, the perceived behavioural control determinant can be divided into two components as well.
The attitude is measured with ten multi-scale items. The correlation matrix, which can be seen in Appendix D, shows that all variables are significantly correlated, apart from the two knowledge items and the two environmental items. Not all items are strongly correlated with each other, as some variables are below a correlation of 0,3. Bartlett’s test of sphericity is significant with p=0,000, and therefore the items are correlated with each other (Malhotra, 2010). The Kaiser-Meyer-Olkin (KMO) measure is 0,815, higher than the initiated threshold of 0,5 (Malhotra, 2010), this indicates that a factor analysis is appropriate. The communalities show that only one taste item (A_T2) is not above
the recommended threshold of 0,4 (Malhotra, 2010). According to the factor analysis have two components an eigenvalue above the recommended level of 1 (Malhotra, 2010), component 1 has an eigenvalue of 4,519 and explains 45,186% of the variance, and component 2 has an eigenvalue of 1,813 and explains 18,127%. Combined both components explain 63,313% of the variance. The scree plot indicates using two separate components for the attitude determinant as well. The varimax rotated component matrix (Table 3) shows that the environmental, animal friendly and health items load high on component 1, whereas the knowledge items and one taste item (A_T1) loads high on component 2.
The other taste item (A_T2) has a low cross loading on both component 1 and component 2.
Item Component 1 Component 2
I think that plant-based milk is tastes good (A_T1) 0,240 0,709
The taste of plant-based milk is important to me (A_T2) 0,346 0,378
I feel like I know a lot about plant-based milk (A_K1) 0,067 0,930
Compared to most other people, I know more about plant-based milk (A_K2) 0,090 0,913
I think that the consumption of plant-based milk is environmentally friendly (A_E1) 0,659 0,009
I think it is important that plant-based milk is environmentally friendly (A_E2) 0,831 0,181
I think that the consumption of plant-based milk is animal friendly (A_A1) 0,819 0,221
I think it is important that plant-based milk is animal friendly (A_A2) 0,847 0,142
I think that the consumption of plant-based milk is healthy (A_H1) 0,754 0,199
I think it is important that plant-based milk is healthy (A_H2) 0,735 0,186
Table 3. Varimax rotated component matrix
We decided to include A_T2 to component 2 rather than component 1, as the other taste item
A_T1 loads high on component 2. We redid the factor analysis with component 2, of which the rotated
component matrix shows that the A_T2 has a loading of 0,494, which is just below the recommended
Cronbach’s alpha of 0,881 for component 1 and a Cronbach’s alpha of 0,783 for component 2. A new variable is created for component 1, which is called AttitudeConscious and a new variable is created for component 2, which is called AttitudeKnowledge.
The subjective norm is measured with three multi-scale items that all strongly and positively correlate, see Appendix D. The KMO is 0,75, which is beyond the initiated threshold of 0,5 (Malhotra, 2010). The Bartlett’s test of sphericity is significant with p=0.000. Hence, we can continue with the factor analysis. The factor analysis indicates that one component is appropriate, as firstly the commonalities of each item are all above the threshold of 0,4, secondly component 1 has an eigenvalue of 2,631 and explains 87,707% of the variance and thirdly the scree plots also indicates 1 component. The Cronbach’s alpha is 0,928, which is higher than the recommended score of 0,60 (Malhotra, 2010). This implies that the scale is reliable to use. Hence, we combined the items into one variable.
The perceived behavioural control is measured with four items of which one item, P_P2, does
not correlate with the other items, see Appendix D. The KMO score of 0,650 is above the recommended 0,5 (Malhotra, 2010), and Bartlett's test of sphericity with p=0,000 is significant. All the communalities are above the threshold of 0,4, apart from the item P_P2 (0,069). The eigenvalue of
component 1 is 1,804 and explains 45,092% of the variance, while component 2 has an eigenvalue of 1 and explains 24,994% of the variance. Both components are above the recommended eigenvalue of 1 or greater (Malhotra, 2010) and together explain 70,087% of the variance. The varimax rotated component matrix shows that only item P_P2 has a loading below the recommended value of 0,5
(Malhotra, 2010). The Cronbach's alpha is 0,542, which is less than the recommended score of 0,6 (Malhotra, 2010). However after deleting item P_P2 the alpha increases to 0,656. Therefore, we
created a new variable, named Perceived Behavioural Control, with the three correlating items and the item P_P2 is kept apart as a second component.
Lastly, the purchase intention is measured with two multi-scale items that strongly correlate, as can be seen in the correlation matrix in Appendix D. The KMO score is 0,50, which is exactly the recommended threshold (Malhotra, 2010). Bartlett's test of sphericity is significant with p=0,000. The factor analysis shows that component 1 has an eigenvalue of 1,831 and explains 91,531% of the variance. The Cronbach’s alpha is 0,907. Hence, we combined the two items into one variable.
4.3 Testing the control variables
influence the purchase intention, β=-0,036, t=13,788, p=0,007. With a R2 of 0,0043 does the
regression explain 4,3% of the variation in the model, which is smaller than the suggested value of 0,60 (Malhotra, 2010). Hence, the age has a small but significantly negative influence on the purchase intention. A One-way ANOVA is performed to test whether the purchase intention differs per level of education. This One-Way ANOVA is significant, F(6,161)= 4,148, p=0,001. Thus, the education level of people does influence their purchase intention. The average purchase intention is the highest for the WO (bachelor) and WO (master/doctoral) having both an average of 4,7. The lowest two means are of the VMBO (M=2) and the MBO (M=2,95). Although this does not imply that the educational level has a negative influence on the purchase intention, as the average purchase intention of the educational level high school (M=3,8) is the same as the average purchase intention of the HBO (M=3,8). The effect of the income level on the purchase intention was tested through a One-way ANOVA. The One-Way ANOVA is significant, F(5,162)= 3,442, p=0,006. The income level of the respondents negatively influences the purchase intention, as the average purchase intention is higher for the lower income levels (M=4,4 for €0 - €10.000 and M=4,2 for €10.000 - €20.000) and gradually lowers for the higher income levels (M=2 for €40.000 - €50.000 and M=2,1 for €50.000 - €100.000). However, this negative effect of the annual income on the purchase intention is probably due to the fact that 83,3% of the participant has an annual income level between the €0- €10.000 and the €10.000 - €20.000.
Next to the demographic variables we test whether the indicated purchase frequency of plant-based milk has any influence on the purchase intention. The regression is significant, R2=0,200,
F(1,166)=41,547, p=0,000. Therefore, the purchase frequency of plant-based milk does positively influence the purchase intention, β=0,638, t=14,928, p=0,000. Finally, the last control variable is if the respondents are allergic to almond milk, as this type of plant-based milk is shown in the experimental advertisements. The independent samples t-test is not significant, t(166)=-1,369, p=0,173. This indicates that the average purchase intention of people that are allergic (n=2, M=2,5, SD=2,12) is not different from the average purchase intention of people that are not allergic (n=166, M=4,187, SD=1,73).
4.4 Testing the hypotheses
Following the testing of the control variables we will test the hypotheses. We start with analysing the hypotheses concerning the habits (H1-H3), following with the hypotheses of the TPB (H4-H11), the hypothesis of the advertisement (H12) and ending with additional analyses.
4.4.1 Habits
alternatives while buying milk. The model is statistically significant X2(4)=20,459, p=0,000. The
model has a Nagelkerke R2 of 0,406 and therefore explains 40,6% of the variance in looking at
alternatives. It correctly classified 78% of the cases. Increasing the purchase frequency of all types of milk (fresh, sterilized and plant-based) is associated with a decreasing likelihood of looking at alternatives. Although only the purchase frequency of plant-based milk has a significant effect (β=-1,282, p=0,000), whereas fresh (β=-0,077, p=0,577) and sterilized milk (β=-0,144, p=0,372) do not. Hence, the purchase frequency of milk (fresh and sterilized) does not influence whether people look at alternatives or not and thereby H1 is not supported. However, the purchase frequency of plant-based milk does have a significant negative effect on whether people look at alternatives.
Next, we analyse the second hypotheses, which states the less people look at alternatives the lower is the purchase intention of plant-based milk. Therefore, we use an independent samples t-test to look if the average purchase intention of plant-based milk is different for people that do look at alternatives and that of those who do not. The test is significant, t(166)= 5,158, p=0,000. Thus, the average purchase intention for people that do look at alternatives (n=54, M=5,10, SD=1,405) is higher than the average purchase intention of people that do not look at alternatives (n=114, M=3,72, SD= 1,7). And thereby H2 is supported, as people that do not look at alternatives while buying milk have a lower intention to purchase plant-based milk.
For our third hypothesis we look if the higher the purchase frequency of milk the lower is the purchase intention of plant-based milk. Therefore, we analyse if the purchase frequency of fresh, sterilized and plant-based milk has any influence on the purchase intention of plant-based milk with a regression analysis. The regression analysis is first performed with each variable independently and after that we put all the different variables in the regression simultaneously. Independently the purchase frequency of fresh milk does negatively influence the purchase intention (R2=0,028,
F(1,166)=4,848, β=-0,183, t=17,888, p=0,029), the purchase frequency of sterilized milk does not influence the purchase intention (R2=0,013, F(1,166)=2,232, β=-0,166, t=19,075, p=0,137) and the
purchase intention of plant-based milk does positively influence the purchase intention (R2=0,200,
F(1,166)=41,547, β =0,638, t=14,928, p=0,000). However, in the multiple regression (R2=0,209,
F(3,164)=14,406, p=0,000) only the purchase frequency of plant-based milk seems to have a significant effect on the purchase intention (β=0,611, p=0,000), while fresh milk (β=-0,041, p=0,611) and sterilized milk (β=-0,130, p=0,202) do not. Therefore, the purchase frequency of plant-based milk has a strong positive effect on the purchase intention of plant-based milk, while the purchase frequency of fresh and sterilized milk have no effect on the purchase intention. This implies that H3 is not supported.
4.4.2 TPB
frequency of plant-based milk and being allergic to almond milk. In order to test for multicollinearity we looked at the VIF-scores of the determinants of the TPB. All scores are between the 1 and 1,5 and therefore below the value 4, this indicates that there is little multicollinearity (Malhotra, 2010). And therefore we can continue with the multiple regression.
We first looked at the hypotheses concerning the attitude determinants. We predict that a person’s environmental consciousness (H4), concern for animal welfare (H5) and health consciousness (H6) positively affects the intention to purchase plant-based milk. The attitude variable AttitudeConscious is significant with β=0,159, p=0,008 and thereby positively influencing the purchase intention. Hence, H4, H5 and H6 are supported.
Our hypotheses 7 and 8 predict that the better the perceived taste (H7) and the higher the subjective knowledge (H8) is, the stronger is the intention to purchase plant-based milk. The AttitudeKnowledgeable is significant with β=0,390, p=0,000 and has a positive influence on the intention to purchase plant-based milk. We can conclude that H7 and H8 are supported. For hypothesis 9 we predicted that the more positive the subjective norm is towards plant-based milk, the stronger is the intention to purchase plant-based milk. The subjective norm is significant with β=0,231, p=0,000, so the subjective norm positively influences the purchase intention. Thereby H9 is supported. For H10 we infer that the higher the perceived price of plant-based milk is the weaker is the intention to purchase plant-based milk. For H11 we predict that the lower the perceived availability of plant-based milk is the weaker is the intention to purchase plant-based milk. Both hypotheses are measured with the perceived behavioural control determinant. The perceived behavioural control is measured with one component combining three items and one component with one item on its own. The perceived behavioural control component is not significant with B=0,062,
p=0,249, and therefore does the perceived behavioural control have no significant influence on the
purchase intention. Accordingly, H11 is not supported. This is the only determinant of the TPB that is not significant. The single perceived price item however is significant with β=0,199 and p=0,000 and thereby has a positive effect on the purchase intention. Hence, H10 is not supported as it states that we expected the price to have a negative influence on the purchase intention, while the item P_P2 has a
positive effect.
When looking at the betas we can see that the AttitudeKnowledgeable variable of the attitude determinant has the largest effect on the purchase intention, followed by the subjective norm and the perceived price item (P_P2). Of the control variables only the annual income (β=-0,125, p=0,024) has
multiple regression can be seen in Figure 8, which displays the standardized coefficient betas. We use the standardized coefficient betas as our control variables differ in scale size.
Figure 8. Conceptual model with the standardized coefficient betas * significant at p=0,001
4.4.3 Advertisement
