• No results found

The good and bad of meat

N/A
N/A
Protected

Academic year: 2021

Share "The good and bad of meat"

Copied!
60
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

1

The good and bad of meat

The relationship between flexitarian’s food motives

and ambivalence towards meat

(2)

2

The good and bad of meat

The relationship between flexitarian’s food motives and ambivalence towards meat

Master Thesis

MSc Marketing Management University of Groningen Faculty of Economics and Business

Department of Marketing January 14, 2019 Nick Dokter Oosterkade 2C2 9711 RS Groningen +31 6 2920 7823 N.dokter@student.rug.nl S2552515

(3)

3

Abstract

(4)

4

Preface

This thesis is part of the final phase of the Master Marketing Management program at the University of Groningen. My interest for the topic ‘flexitarian’ emerged about a half year ago. Until my 23rd year, I ate meat every day. This was not quite normal in my family, since my mother and several nieces in my family followed a vegetarian diet. Becoming more aware of the negative impact of the meat industry and the fact that friends around me limited their meat intake, led me to becoming a flexitarian in April 2018. Since then, I try to limit my meat intake to one or two times a week. Thus, when I could fill in my preferences for the thesis subject, I was glad that I could choose this subject and I am still glad with this choice.

(5)

5

Introduction

The increasing demand for meat is one of the major challenges in the food system (Godfray et al., 2010). This is a trend that has been progressing for some time. In 1961, the average meat consumption was 20 kilograms per capita and it has increased to 43 kilograms per capita in 2014. Moreover, it is projected that the amount of meat per capita will only continue to increase (Ritchie & Roser, 2017). This increase in meat consumption could have disastrous consequences, since the consumption of meat has negative consequences for one’s health, the environment, and for animal welfare. Firstly, the international agencies for research on cancer finds an association between red meat and several forms of cancer (IARC, 2015). Furthermore, processed meat has been associated with diabetes mellitus and a higher incidence of coronary heart disease (Micha, Wallace, & Mozaffarian, 2010). Secondly, meat is identified as the most environmentally harmful form of food consumption (Popp, Lotze-Campen, & Bodirsky, 2010; Austgulen, Skuland, Schjøll, & Alfnes, 2018). This is because the production of meat is associated with large amounts of greenhouse gas emissions. Lastly, meat production has a negative effect on animal welfare, since animals undergo lots of suffering (Singer, 2015). Hence, consumption of meat is a significant problem for both the individual and society, as well as the suffering of animals.

Despite these negative characteristics of meat consumption, people also associate meat with positive characteristics, such as for its familiarity and sensory appeal (Hoek, Luning, Weijzen, Engels, Kok, & de Graaf, 2011). In most Western cultures, eating meat at dinner is a tradition that begins in childhood. People therefore acknowledge that eating meat is familiar to them and that eating meat is an ingrained habit (Mullee et al., 2017). Furthermore, people value the taste and the structure of meat. This is because meat has unique sensory appeals (Kenyon & Barker, 1998).

Individuals may hold ambivalent feelings towards meat, because they associate meat with both negative and positive aspects. The ‘meat paradox’1

describes these ambivalent feelings that people have around the consumption of meat (Loughnan, Haslam, & Bastian, 2010). People

1

(6)

6

dislike the negative aspects associated with the production and consumption of meat, but still eat meat because of its positive characteristics. This ‘meat paradox’ does not apply to every dietary group to the same extent. Vegans or vegetarians do not experience these inconsistent beliefs, since they fully restrict from meat consumption. Meat eaters do experience these inconsistent beliefs (Povey, Wellens, & Connor, 2001), but to a limited extent. However, the beliefs of people who eat meat also differ greatly. For example, flexitarians (people who restrict meat intake) believe that meat consumption has a more negative impact on the environment, one’s health and animal welfare than omnivores (people who do not restrict their meat intake) (Forestell, Spaeth, & Kane, 2012; Mullee et al., 2017). Therefore, it would not be correct to generalize the beliefs of meat eaters.

Flexitarians should experience the greatest amount of ambivalent feelings of the previously mentioned groups, since eating meat remains a dilemma for them (Rothgerber, 2014). People who identify themselves as vegetarians, vegans, and omnivorous experience this dilemma to a much lesser degree, since they have a clear dietary pattern with regards to meat. Flexitarians need to make decisions about whether to eat meat or not on a daily basis and often have conscious reasons for eating less meat. Three commonly cited reasons for why flexitarians adopt a more vegetarian diet are for one’s health, the environment, and animal welfare. Although flexitarians have these conscious reasons, they still like to consume meat (Mullee et al., 2017). Hence, this research predicts that flexitarians experience ambivalent feelings with regard to the consumption of meat and thus experience the ‘meat paradox’.

Research aim

(7)

7

Table 1 - Literature overview Study Theoretical focus Dietary groups Main findings

Cliceri et al., (2018)

Comparison of psychological traits, beliefs and responsiveness on implicit attitudes towards plant-based diets.

Vegetarians, flexitarians, and omnivores.

- Vegetarians and flexitarians have more positive attitudes towards meat-free dishes than omnivores.

- Food consciousness is important in determining food habits.

De Backer and

Hudders, (2014)

Motives for meat reductions. Vegetarians, semi-vegetarians2, and light semi-vegetarians.

- Most differences exist between the vegetarian group and the semi-vegetarian groups.

- Based on different motives (taste preferences, health motive, ecological concern, & animal concern) different clusters emerge for all dietary groups.

De Backer and

Hudders, (2015)

Relationship between morality and diet choice by investigating attitudes and donation

behaviors.

Vegetarians, flexitarians, and meat eaters.

- Vegetarians score the highest on animal concern, then flexitarians, and meat eaters the least.

- Flexitarians score as high as meat eaters on donation behaviors.

Forestell et al., (2012)

Comparison of dietary habits and lifestyle behaviors.

Vegetarians, pesco-vegetarians, semi-vegetarians, flexitarians, and omnivores.

- Semi vegetarians and flexitarians are more cognitively constrained to eat animals’ flesh than omnivores.

- Flexitarians and semi-vegetarians were less motivated by animal welfare concern than vegetarians.

2 The terms ‘flexitarian’ and ‘semi-vegetarian’ are used interchangeably in the literature. For reading the literature table overview, you could treat the terms

(8)

8 Hoek et al.,

(2004)

Comparison of food-related lifestyle and health attitudes.

Vegetarians, semi-vegetarians, and omnivores.

- Vegetarians stress most importance towards health and food-related lifestyle instruments, omnivores the least importance. - Semi-vegetarians are mostly in-between vegetarians and omnivores.

Mullee et al., (2017)

Comparison of attitudes and beliefs about meat.

Vegetarians, semi-vegetarians, and omnivores.

- Vegetarians have the strongest belief that meat is bad for the environment and one’s health, followed by semi-vegetarians and then omnivorous.

- Health is an important motive for considering a more vegetarian-based diet.

Rothgerber, (2014)

Comparison of attitudes towards meat and animals.

Vegetarians and semi-vegetarians.

- Vegetarians dislike meat more than semi-vegetarians.

- No difference between human-animal similarity between the groups.

This thesis, Dokter, (2018)

Level of ambivalence towards meat, resulting from different food motives.

(9)

9

As opposed to comparing the attitudes of flexitarians alongside other dietary groups, this research will take the novel approach of focusing on the (ambivalence) attitudes of just flexitarians. More specifically, this will be achieved by linking the individual’s food motives to their ambivalence towards meat. Flexitarians have various motives to eat meat or to restrict their meat intake, ranging from motives that are subjective (i.e. taste preference) to motives whereby the negative impact can be measured objectively (i.e. environment) (De Backer & Hudders, 2014; Mullee et al., 2017). This thesis will focus on three often mentioned food motives to limit meat intake by flexitarians, namely: health-, environmental- and animal welfare motives. Moreover, this thesis is about two often mentioned food motives to eat meat by flexitarians: sensory appeal- and familiarity motives.

Consequently, the aim of this thesis is to investigate whether different food motives of flexitarians are associated with different levels of ambivalence towards meat. In order to achieve this, the following research questions will be addressed: how do flexitarians’ different motives

(health, environmental, animal welfare, sensory appeal, & familiarity) influence the ambivalence towards meat?

The effect of the different motives on various eating frequencies, and the mediating effect of ambivalence towards meat in the previously mentioned relationship, will be investigated exploratively.

Contributions

The first theoretical contribution is answering the calls of Rothgerber (2014), and De Backer and Hudders (2014), of not treating flexitarians as one group. This research therefore acknowledges that the different motives for becoming flexitarians may result in different attitudinal consequences. Since flexitarians are constantly experiencing a dilemma about the consumption of meat, the ambivalence towards meat might be especially interesting, and be able to explain why behavioral consequences differ. The second theoretical contribution is investigating whether results obtained from studies in vegetarianism and veganism (Hoffman et al., 2013; Radnitz et al., 2015) could also be applied to flexitarians. More specifically, to see whether different food motives are associated with certain dietary restrictions. In this thesis the

dietary restriction would mean less frequent meat consumption.

(10)

10

Infrastructure in the Netherlands ("Adviesraad: we moeten minder vlees eten", 2018). They published that Dutch people should reduce their meat intake to contribute to bettering the environment. One of the tasks the council assigned to the government is to improve the campaigns to reduce meat intake. When one knows which food motive of flexitarians is associated with the highest amount of ambivalence and meat eating frequencies, the government knows which motive is most worth targeting given a fixed budget. The second practical contribution is directed at non-profit organizations, such as ProVeg Nederland. On their website they state that they are willing to help people who are interested in a vegetable lifestyle ("Viva Las Vega's wordt ProVeg International", 2018). If flexitarians experience high amounts of ambivalence towards meat, ProVeg could help them to find worthy meat alternatives. The last practical contribution is for marketing managers who may find the large number of flexitarians attractive potential customers. Based on their previously purchased products, customers could be clustered based on motivational distinctions, and thereby more accurately targeted.

The thesis is structured as follows: first, academic literature will be reviewed and hypotheses will be formed. Next, the methodology that has been applied, will be presented. Subsequently, the results will be presented and discussed. The thesis ends with a conclusion.

2. Theory development

(11)

11

2.1 Flexitarians

The terms ‘semi-vegetarian’ and ‘flexitarian’ are used interchangeably in the literature. Semi-vegetarians are defined as individuals who exclude some types of meat, but not others. While flexitarians are individuals who limit their meat intake, but still eat meat (Rosenfeld, 2018). This research will focus on flexitarians, since no distinction in the arguments between different types of meat is made in this thesis. The terminology also differs in the amount of meat that flexitarians consume. For example, de Backer and Hudders (2014) distinguish between not eating meat at least one or two days a week to not eating meat at least three days a week. In this thesis, flexitarians are defined as individuals who restrict their meat consumption by having weekly at least one meatless day, but still eat meat (De Bakker & Dagevos, 2012).

2.2 Ambivalence towards meat

Ambivalence is defined as perceiving both advantages and disadvantages towards an object simultaneously, in other words having both positive- and negative attitudes together (Povey et al., 2001). Therefore, ambivalence towards meat is defined as having mixed feelings about meat, both a negative- and a positive attitude simultaneously (Berndsen & Van der Pligt, 2004). One way to look at ambivalence is by looking at both the positive- and negative attitudes separately (Kaplan, 1972; Thompson, Zana, & Griffin, 1995). Attitude is defined as an evaluation of an attitude object (Wood, 2000). In this case, the positive- and negative attitude are about evaluations of meat.

(12)

12

However, it will not always be the case that the negative- and positive attitude are equally strong in power. In the following examples, an explanation is given how the ambivalence will change as the negative- and positive attitude are not equal in strength. When someone does not have a negative- or a positive attitude towards meat (not at all), the ambivalence is low. It is even on the lowest point when the opposing attitude is ‘extremely’, because there is a large discrepancy between the two attitudes. These persons do not experience a positive and a negative attitude simultaneously, hence very little ambivalence occurs. As the two attitudes move towards extremely, the average ambivalence increases (see table 2). This is because the individual experiences stronger mixed attitudes simultaneously. Therefore, the intensity with which the emotions are experienced increases (Thompson et al., 1995). One exception of this rule is when ‘extremely’ and ‘slightly’ are combined. The ambivalence is low, because the person has limited mixed feelings about meat, because of the large discrepancy between the negative- and the positive attitude.

2.3 Motives

Individuals emphasize different levels of importance for various aspects of food. For example, individuals may value the sensory appeal of food or may value the fact that the food they consume does not harm the environment. Lots of these food motives exist and there is still no consensus reached about the exact number of food motives (Steptoe et al., 1995; Lindeman & Väänänen, 2000). Five food motives will be elaborated upon in this thesis.

The first motive is the animal welfare motive. People with a high animal welfare motive stress importance on the rights of animals and the way they are treated (Lindeman & Väänänen,

Table 2 – Ambivalence towards meat Ambivalence Positive attitude

Not at all Slightly Quite Extremely

Negative attitude Not at all Low Low Low Lowest

Slightly Low Medium Medium Low

Quite Low Medium High High

(13)

13

2000). When making food choices, these people keep the welfare of animals in mind. The second motive is the health motive. The definition of the health motive differs in the literature. Some authors include personal health and weight loss within the definition of the health motive (Radnitz et al., 2015; Hofmann et al., 2013), while others only refer to personal health (Steptoe, Pollard, & Wardle, 1995; De Backer & Hudders, 2014). Personal health and weight loss load on different components, with only moderate correlations. Therefore, people with a high health motive stress importance on vitamins, fibers, proteins, and the nutrition of food (Steptoe et al., 1995). The third motive is the environmental motive. This motive has recently gained importance. People with a high environmental motive stress importance on environmentally preparing, -packaging, and -producing of foods. There is some overlap between the constructs of the animal welfare motive and the environmental motive (Lindeman & Väänänen, 2000). This could explain why some authors used both motives to refer to the ‘ethical motive’ (Hoffman et al., 2013). However, the term ‘ethical motive’ is not consistently used, since some researchers use the term to only point to the animal welfare motive (Radnitz et al., 2015). Consequently, both motives will be used separately. The fourth motive is the familiarity motive. People with a high familiarity motive stress importance on eating habits, familiarity of food, and food they ate in childhood (Steptoe et al., 1995). The last motive is the sensory appeal motive. The sensory appeal motive is sometimes also referred to as ‘hedonic aspects’, which points to the taste, versatility, variety, and flavor of food (Berndsen & Van der Pligt, 2004). While others refer to the sensory appeal motive by looking at the taste, texture, and smell of food (Pieniak, Verbeke, Vanhonacker, Guerrero, & Hersleth, 2009). In this thesis, the sensory appeal motive is defined as the taste, texture, smell and look of food (Steptoe et al., 1995). This is because it is the most straightforward definition, as well as providing a more reliable measure in contrast to previously mentioned sensory appeal motive measures.

2.4 Hypotheses

(14)

14

2.4.1 Animal welfare motive

Meat is bothersome. Besides the killing of animals to get meat, animal rights are often neglected with the production of meat: the physical bruising of animals, the crowded lofts, and the worrisome transportation are serious problems that may raise concern among consumers (Grandin, 2014). The knowledge of indirectly killing an animal and the circumstances that these animals live in, may lead to a negative emotional state for the person who eats meat (Loughnan et al., 2010). Moreover, among people who want to become vegetarian, people who have an animal welfare motive associate meat with negative emotions, such as disgust and emotional distress. Distress was reduced, when the proportion of meat consumption was also reduced (Ruby, 2012). Lastly, more ambivalent persons associate meat with more immoral aspects (killings animals) than less ambivalent persons (Berndsen & Van der Pligt, 2004).

I suggest that the animal welfare motive is related to a negative attitude towards meat. This is because animals’ rights are neglected with the production of meat and because individuals that hold a high animal welfare motive associate meat with disgust and emotional distress (Grandin, 2014; Ruby, 2012). Hence, people higher on the animal welfare motive would have a more negative attitude towards meat. I hypothesize:

Hypothesis 1: An animal welfare motive is positively related to a negative attitude towards meat.

2.4.2 Environmental motive

(15)

15

who recognize that the environment is important still have difficulty restraining from meat because of the sensory appeal of meat (Tucker, 2014).

Consequently, I suggest that an environmental motive is related to a negative attitude towards meat. I assume that flexitarians with a high environmental motive know that the production of meat is bad for the environment, since publications show this negative impact of meat consumption on the climate (Austgulen et al., 2018; Godfray et al., 2010). Hence, a higher environmental motive would lead to a more negative attitude towards meat. I hypothesize:

Hypothesis 2: An environmental motive is positively related to a negative attitude towards meat.

2.4.3 Health motive

Meat consumption may be bad for one’s health. Meat consumption is associated with higher risks of several forms of cancer, diabetes mellitus, and cardiovascular diseases. Additionally, animals are given high levels of many antibiotics and therefore human bacteria may become resistant for antibiotics when lots of meat is consumed (Micha et al., 2010; Walker, Rhubart-Berg, McKenzie, Kelling, & Lawrence, 2005). This offers an explanation for why health vegetarians hold negative attitudes towards meat and are therefore reporting a somewhat disgust towards meat (Rothgerber, 2014). Some authors also argue that meat may be beneficial for one’s health. This is because meat is an important source of protein and encompasses important nutrients, such as: iron, zinc, and vitamin B12 (McAfee et al., 2010). Furthermore, more ambivalent individuals associate meat consumption with higher risk for one’s health than less ambivalent persons (Berndsen & Van der Pligt, 2004).

I argue that the health motive is related to a negative attitude towards meat. This is because meat consumption may have negative consequences for one’s health and may therefore lead to a more negative attitude towards meat when the health motive increases (Rothgerber, 2014; Walker et al., 2005). The argument that meat is beneficial for one’s health probably would not hold for individuals from Western cultures with a health motive, since lots of meat-substitutes and supplements are available to also get the important proteins and nutrition’s. Therefore, people who are well informed on health, would not associate meat with these health-related benefits, because the cons are more catastrophic than the potential pros. I hypothesize:

(16)

16

2.4.4 Familiarity motive

Among flexitarians, the second and third most mentioned reasons for eating meat, are: ‘habit’ and ‘this was how I was brought up’ (Mullee et al., 2017). Accordingly, there is a positive relationship between early childhood food preferences and current food practices (Unusan, 2006). Unusan reasons that this might be because the attitudes that are formed in early childhood prevail into adulthood. Thus, that familiarity with food is associated with positive attitudes towards that food, otherwise it is likely that he or she might have changed their diet during the years. Meat is far more familiar for people than meat-substitutes, even among heavy users of meat-substitutes. Familiarity with meat increases consumer acceptance and lowers the acceptance of meat-substitutes (Hoek et al., 2011). Lastly, people are inclined to like things they encounter as familiar, because it helps to avoid a conflict between attitudes and behavior (Kähkönen & Tuorila, 1999).

I suggest that the familiarity motive is related to a positive attitude towards meat. Increased familiarity increases liking and consumer acceptance. Meat can be said to be familiar for flexitarians (Mullee et al., 2017), therefore a higher familiarity motive will lead to an increasingly positive attitude towards meat, because flexitarians are more likely to accept and like meat (Hoek et al., 2011; Kähkönen & Tuorila, 1999). I hypothesize:

Hypothesis 4: A familiarity motive is positively related to a positive attitude towards meat.

2.4.5 Sensory appeal motive

(17)

17

I therefore propose that the sensory appeal motive is related to a positive attitude towards meat. The positive attitude towards meat is likely increasing as a result of increased consumer acceptance and higher levels of desirability of meat (Tucker, 2014) as one’s sensory appeal motive increases. I hypothesize:

Hypothesis 5: A sensory appeal motive is positively related to a positive attitude towards meat.

The five hypotheses are visualized in the conceptual model of figure 1.

3. Method

First, the procedure for reaching the flexitarians will be explained. Next, the measures for the variables are shown, including the formula for indirect ambivalence towards meat. Lastly, the data analysis plan is discussed.

3.1 Procedure

(18)

18

could fill in the survey. This method is known as ‘snowball sampling’. The link for Qualtrics was accompanied with a text which stated the target audience. The text described that the researcher was looking for people who (1) do not eat meat for at least one day a week and (2) are not vegan or vegetarian.

3.2 Measures

The measures for the main variables, exploratory variables, and the control variables will be shown.

3.2.1 Main variables.

The measures of the main variables are displayed in table 3.

Table 3 - Main variables

Variable Source Items Scale

Food motives It is important to me that the food I eat

on a typical day: 1 ‘Not at all important’. 5 ‘Extremely important’ Animal welfare motive Lindeman & Väänänen, (2002)

- Has been produced in a way that animals have not experienced pain. - Has been produced in a way that animals' rights have been respected. Environmental

motive

Lindeman & Väänänen, (2002)

- Has been prepared in an environmentally friendly way.

- Has been produced in a way which has not shaken the balance of nature.

- Is packaged in an environmentally friendly way.

Health motive. Steptoe et al., (1995)

- Contains a lot of vitamins and minerals.

- Keeps me healthy. - Is nutritious. - Is high in protein.

(19)

19 Familiarity

motive.

Steptoe et al., (1995)

- Is what I usually eat. - Is familiar.

- Is like the food I ate when I was a child. Sensory appeal motive. Steptoe et al., (1995) - Smells nice. - Looks nice.

- Has a pleasant texture. - Tastes good.

Ambivalence towards meat.

Direct measure. Conor & Sparks, (2002)

- I have conflicting thoughts about meat. - I have mixed feelings about meat. - My thoughts and feelings about meat are conflicting.

1 ‘Extremely disagree’ 5 ‘Extremely agree’ Indirect measure. Thompson et

al., (1995)

- Considering only the positive things about meat, and ignoring the negative things, how positive are those things? - Considering only the negative things about meat, and ignoring the positive things, how negative are those things?

1 ‘Not at all positive/ negative’ 4 ‘Extremely positive/ negative’ 3.2.2 Calculation of ambivalence.

Several formulas were available to measure indirect ambivalence towards meat. The first formula which was widely used, was the formula of Kaplan (1972). However, criticism about the outcomes of the formula arose. When the weaker component was held constant, people who differed in what may be called attitude polarization, which can be thought of as the difference between the stronger and weaker component, had the same ambivalence score (Thompson et al., 1995). For example, a person who had an extremely negative attitude towards meat and not a positive attitude towards meat at all, had the same ambivalence score as a person who did not have a negative attitude, and a slightly positive attitude towards meat. Therefore, I did not make further use of the Kaplan formula, as, according to the definition of ambivalence, the ambivalence score should be lower for the person who experienced an extremely negative attitude and no positive attitude at all, since this person experiences no mixed feelings about meat. Consequently, the formula of Griffin, which tackles the polarization problem, has therefore been used in this thesis:

(20)

20

3.2.3 Exploratory variables.

The measures for the exploratory variables are displayed in table 4.

Table 4 - Eating frequencies

Eating frequencies Authors Scale

1.Can you estimate how many times you ate meat last week?

Cong, Olsen, & Tuu, (2013)

1 ‘0’ 16 ‘>14’ 2.Can you estimate how many times during a regular

week you eat meat?

Cong, Olsen, & Tuu, (2013)

1 ‘1’ 15 ‘>14’ 3.How often do you buy organic meat or meat with a

label? (Reverse coded)

1 ‘Never’ 5 ‘Always* 4.How often do you buy meat without a label or meat

that is not organic?

1 ‘Never’ 5 ‘Always’

5.How often do you eat meat indoors? 1 ‘Never’

5 ‘Always’ 6.How often do you eat meat outdoors? (Reverse

coded)

1 ‘Never’ 5 ‘Always’ 7.How often do you replace meat by meat substitutes

indoors?

1 ‘Never’ 5 ‘Always’ 8.How often do you replace meat by meat substitutes

outdoors.

1 ‘Never’ 5 ‘Always’ 9.How willing would you be to consider reducing

your meat consumption sometime in the near future?

Lentz, Connelly,

Mirosa, & Jowett, (2018)

1 ‘Not at all willing’’ 5 ‘Extremely willing’ 10.Specifically, in the next six months do you intend

to reduce your meat consumption?

Lentz, Connelly,

Mirosa, & Jowett, (2018)

1 ‘Not at all intent’ 5 ‘Extremely intent’

*In comparison with the original scale by Siegrist and Hartmann (2018), ‘sometimes’ was changed to ‘about half of the time’.

3.2.4 Control variables

(21)

21

Cardello & Piasentier, 2015). According to the theory of planned behavior, one’s subjective norm and behavioral control influence the intention of behavior and, indirectly, one’s behavior. Hence why subjective norm and behavioral control were controlled for. The afore mentioned variables are displayed in table 5. Socio-demographics were also controlled for. Gender was used as a control variable, since men have a more positive attitude towards meat than women (Kubberød, Ueland, Rødbotten, Westad & Risvik, 2002). Furthermore, research has found that the higher one’s age, education level, and income, the higher one’s preference for meatless meals (Rimal, 2002). Therefore, the survey design controlled for an individual’s age, education level, and income. Research has also showed that a higher social class is associated with less meat consumption (Gossard & York, 2003). For this reason, the current state of the labor market and the living area of the respondent were also controlled for. Lastly, the number of people in a household and the extent to which people: did their own grocery shopping, cooked their own food, ate alone, ate with vegetarians/vegan, and ordered their food have also been controlled for. These questions were measured on a five-points Likert Scale which ranged from ‘never’ to ‘always’.

Table 5 - Control variables

Variable Source Items Scale

Nutrition involvement.

Borgogno et al., 2015.

- I pay close attention to nutrition information. - It is important to me that nutrition

information is available.

- I ignore nutrition information.

- I actively seek out nutrition information. - Calorie levels influence what I eat.

1 ‘Extremely disagree’ 5 ‘Extremely agree’ Subjective norm. Verbeke & Vackier, 2005.

- People/Institutions who are important to me, think I should buy/eat meat.

- People/Institutions who influence my decisions, think I should buy/eat meat.

- People/Institutions who influence my buying behavior, think that I should buy/eat meat.

1 ‘Extremely disagree’ 5 ‘Extremely agree’ Behavioral control. Verbeke & Vackier, 2005.

- I have a lot of confidence in myself with respect to buying and eating of meat. - I am very convinced about my capacities with respect to buying and preparing of meat.

(22)

22

3.3 Data analysis plan

First, the raw data is checked for outliers and missing data. Missing data points were replaced by the mean of the series. Respondents’ data is deleted when they did not fill in lots of statements. Subsequently, dummy variables were created for the categorical variables in order to prepare these variables for linear regression analysis. The largest group for each of these categorical variables is chosen as reference group, except for gender where ‘men’ is chosen as the reference group. Subsequently, factor analysis was conducted on the main variables to check whether the underlying dimensions in this dataset matched the variables in the literature. After the factor analysis, a reliability analysis was conducted on the underlying dimensions that the factor analysis discovered. The Cronbach’s Alpha is used for this, with a score above .6 considered acceptable (Malhotra & Birks, 2006). Based on the results of the factor analysis and Cronbach’s Alpha, the mean of the different variables was calculated. In this way there was one score for each construct. This procedure was slightly different for the exploratory variables ‘eating frequencies’. These questions had different scales, therefore the menu of SPSS was used to save the factor scores in the standardized form. For this reason, the other variables in these models were also standardized. Furthermore, the assumptions of linear regression and the model fit of the estimated models are reviewed. Lastly, regression analyses are executed to test the hypotheses. The results are interpreted by means of p-values. In addition, regression analyses are executed to examine the influence of the motives and both forms of ambivalence on ‘normal meat eating frequency’ and ‘less future meat consumption’, and to investigate the moderating influence of ambivalence on the relationships between ‘the motives’ and ‘normal meat-eating frequency/less future meat consumption’.

4. Results

(23)

23

4.1 Descriptive statistics of the sample

The survey was completed by a total of 238 participants. Fifty-six of them were not part of the ‘flexitarian’ population, because they identified themselves as full-time meat-eaters or because they identified themselves as vegetarians or vegans. Thus, 182 flexitarians remained. Due to thirty-three incomplete responses, the final dataset consists of 149 flexitarians. The sample descriptive statistics are shown in table 6. The flexitarian population in this dataset consist mostly of females, is relatively young, is well educated, and either in education or a paid job.

Table 6 - Descriptive statistics of the sample

Variable Mean Standard deviation

Age M = 28.2 S.D. = 10.4

Income M = 30818.8 S.D. = 26569.1

Category Number of respondents (%)

Gender 1. Man 52 (34.9%)

2. Female 97 (65.1%)

Education 1. High school 2 (1.3%)

2. Intermediate vocational education 14 (9.4%)

3. Higher vocational education 44 (29.5%)

4. University 89 (59.8%)

Current situation 1. Paid job 80 (53.7%)

2. Following education 66 (44.3%)

3. Unemployed 3 (2%)

Living area 1. Big city 71 (47.6%)

2. Suburbs big city 4 (2.7%)

3. City 59 (39.6%)

4. Village 14 (9.4%)

(24)

24

4.2 Scale development

Factor analysis and the Cronbach’s Alpha are used for scale development. This is done respectively for the main variables and for ‘eating frequencies’.

4.2.1 Scale development main variables

The animal welfare motive, the environmental motive, the health motive, the sensory appeal motive, the familiarity motive, and ambivalence towards meat are used in one factor analysis. All the outcomes of the factor analysis are shown in appendix A.

4.2.2 Criteria factor analysis

To test whether the hypothesized dimensions of the main variables also exists in this dataset, a factor analysis is conducted. Factor analysis helps with this, because it reduces the large number of items into fewer constructs by finding common variance. The Kaiser-Meyer-Olkin Measure, Bartlett's test of Sphericity, and the communalities are checked to see whether it is appropriate to carry out factor analysis. The values are mentioned in appendix A and compared with established criteria (Malhotra & Birks, 2006). The Kaiser-Meyer-Olkin Measure had a value of .720 which was above the criteria of .5. According to this criteria factor analysis could be performed, since the data is likely to factor well based on correlation. According to the Bartlett's Test of Sphericity, factor analysis is adequate to perform, since the test reaches significance (p-value: 0.001). The hypothesis that the variables are uncorrelated, is thus rejected. Communities should be above .4. This is not the case for ‘tastes good’, since the communality is .395. The item is kept in the factor analysis for two reasons. (1) The extraction is very close to .4. (2) Every measurement of the sensory appeal includes taste. Therefore, not including taste could violate validity.

4.2.3 Number of factors in the factor analysis

(25)

25

a five-factor solution, while the eigenvalues criteria points to a six-factor solution. The difference between the five and six factor solution is that the health motive is split, so that the first three items load on the fifth factor and the last three items on the sixth factor. Since this six-factor solution is in contradiction with existing literature and considering that the sixth component in the factor is just above an eigenvalue of one, a five-factor solution is preferred.

4.2.4 Loadings factor analysis

High cross-loadings are not present in this factor analysis, with the highest cross loading being .22. All items load on their expected factor, except for the animal welfare- and environmental motive. These motives load on the same factor. The existing literature acknowledges this but accepts that these motives can be used separately (Lindeman & Väänänen, 2000). Both the five and the six-factor solution put both motives into the same factor and the correlation between the two motives is high (corr: .64). For this reason, the two motives are combined into the ‘ethical motive’. A small adjustment in the hypothesis is required. Hypothesis 1 and hypothesis 2 will be replaced by a new hypothesis 1. This new hypothesis is: the ethical motive is positively related to a negative attitude towards meat.

4.2.5 Reliability main variables

In this thesis one form of reliability is checked: the internal consistency. The Cronbach’s Alpha (CBA) is used to measure the internal consistency. The higher the CBA the higher the internal consistency of the items within the factor and the more reliable the factor is. All variables meet the threshold of .6 (Malhotra & Birks, 2006). The items within the factors are thus reliable enough to form a new factor. The Cronbach’s Alpha are also displayed in appendix A.

4.2.6 Scale development eating frequencies

The ten questions of eating frequencies are put in one factor analysis. All the outcomes of the factor analysis are put in appendix B.

4.2.7 Criteria factor analysis

(26)

26

the factor analysis is conducted. The mean of question 1 is 5.07 (S.D. = 2.03). The mean of question 2 is 4.09 (S.D. = 2.02). For both questions, three observations are kept out of the factor analysis, because these observations are more than two standard deviations away from the mean. The Bartlett's test of Sphericity, the KMO and the communities were all above the previously mentioned thresholds to execute factor analysis.

4.2.8 Number of factors in the factor analysis

The criteria that determined the number of factors points to different solutions. The eigenvalues criteria points to a four-factor solution, the % of variance explained points to a seven-factor solution, and the cumulative variance explained points to a three-factor solution. A four-factor solution is chosen, because the eigenvalues criteria is often considered as leading in these cases and because the % of variance explained is not considered that meaningful when there are only 10 items in the factor analysis, since 5% can be easily reached.

4.2.9 Loadings factor analysis

Question number 7 (How often do you replace meat by meat substitutes indoors?) is not used in constituting a new factor, because of high cross-loadings.3 Four new variables are formed based on the four-factor solution: ‘normal meat-eating frequency’, ‘less future meat consumption, ‘not buying organic meat’, and ‘eating less meat outdoors’. The factor loadings were saved as Z-scores by SPSS.

4.2.10 Reliability

The Cronbach’s Alpha is used to measure the internal consistency of the variables. For the calculation of the CBA, the marker items (loadings higher than .5) for each newly formed factor are used. The outcomes of the Cronbach’s Alpha are shown in appendix B. Three out of the four variables meet the threshold of .6 (Malhotra & Birks, 2006). The factor ‘eating less meat outdoors’ is not reliable enough to use in subsequent analyses. Lastly the Cronbach’s Alpha is inspected for the control variables. All the variables met the threshold of .6 (Malhotra & Birks,

3

(27)

27

2006). Nutrition involvement, subjective norm, and behavioral control have a Cronbach’s Alpha of respectively .819, .893, and .718.

4.3 Descriptive results

The descriptive statistics and inter-correlations of the variables are shown in table 7

Table 7 - Descriptive statistics and inter-correlations

M S.D. 1 2 3 4 5 6 7 8 9 10 1.Health motive 3.33 .65 1 2.Sensory motive 3.43 .73 .15† 1 3.Familiarity motive 2.24 .87 -.01 .07 1 4.Ethical motive 3.13 .87 .33** .21** -.01 1 5.Positive attitude 2.69 .77 -.09 .22** .15† -.16† 1 6.Negative attitude 3.01 .74 .03 .04 -.23** .26** -.09 1 7.Direct ambivalence 3.41 .88 .04 .12 -.03 .15† -.01 .21** 1 8.Indirect ambivalence 1.92 .91 .02 .16† .11 .01 .63** .27** .15† 1

9. Normal meat-eating frequency4 -.02 .02 .18** -.26** .17* -.21* .05 .10 1

10.Less future meat consumption .14† .03 -.07 .40** -.18** .32** .25** -.04 .005 1

Notes. † p < 0.10, * p < 0.05, ** p < 0.0

4

The descriptive statistics of normal meat eating frequency and less future meat consumption will be given in paragraph 4.7.1 and 4.7.4.

(28)

28

4.4 Regression analysis

Four models are estimated. Positive attitude is the dependent variable in model A0 and model A1. Model A0 is the base model which consists of the control variables and model A1 is the full model where the main variables are added. Negative attitude is the dependent variable in model B0 and B1. Model B0 is the base model which consists of the control variables and model B1 is the full model where the main variables are added. Five hypotheses are tested. The outcomes of the regressions are shown in table 8.

4.4.1 Assumption checks for linear regression

One of the assumptions of linear regression is that the dependent variable is normally distributed. Shapiro-Wilks Test of Normality is used to investigate whether the dependent variable is normally distributed. For both negative- and positive attitude, the data is not normally distributed (sig: 0.001). Negative attitude is negatively skewed with a value of -.315 and positive attitude is positively skewed with a value of .145. To make the dependent variable normally distributed, the data is transformed by means of log transformation. The formula that was used to transform the negatively skewed negative attitude was: 4 (maximum value) + 1 - Negative Attitude. After the log transformation, another Shapiro-Wilks Test of Normality was conducted, but the data for the negative- and positive attitude remains not normally distributed (sig: 0.001). The absolute value of the skewness even increased for both the negative attitude (from .412 to -.315) and the positive attitude (from -.692 to .145). Due to the increasing skewness after log transformation, the dependent variables are not log-transformed.

(29)

29

Furthermore, linear regression assumes that the relationship between the independent and dependent variable is linear. To check this assumption, scatter plots are inspected. The scatterplots confirm the assumption that the relationships between the independent and dependent variables are linear. For example, no U- form or inverted U-form are detected. According to this assumption, linear regression can be executed.

To conclude, the assumption that the dependent variable is normally distributed is not met. Therefore, the results of an ordered probit (ordinal regression) will be compared with a linear regression in the footnotes in chapter 4.2.3 and chapter 4.5.4.

4.4.2 Model fit

Model A0 with the control variables explains 17.2% of the variance of positive attitude. The model is significant (F = 1.838, p = .035), with three out of the fifteen individual variables reaching at least the marginal significance. With the inclusion of the main variables in model A1, 25.1% of the variance of positive attitude is explained. To investigate whether the additional variables are included in model A1 are of importance, the values of the adjusted R square are discussed. The adjusted R square is substantially higher in model A1 (Adj R square: .141) than in model A0 (Adj R square: .078). Since the R square is increasing, one can conclude that useful variables are added in model 2. The overall model is also significant (F = 2.275, p = 0.004). Six out of the nineteen variables reach at least marginal significance of which three main variables. To conclude, the model fit of model A1 is good.

(30)

30

4.4.3 Hypothesis testing

To analyze whether an ethical motive influences a negative attitude, a regression of negative attitude on an ethical motive is executed. The results show that an ethical motive has a significant positive effect on a negative attitude (β = .22, t = 3.01). Therefore, there is evidence found to confirm hypothesis 167. In the footnotes the regressions of negative attitude on an animal welfare motive and on an environmental motive are reported.89

To analyze if a health motive influences a negative attitude, a regression of negative attitude on a health motive is executed. According to the results, a health motive does not influence a negative attitude (β = -.01, t = -.11). This means that there is no evidence found to confirm hypothesis 3.

A positive attitude is regressed on the familiarity motive, to analyze if a familiarity motive influences a positive attitude. A familiarity motive has a positive marginal significant effect on a positive attitude (β = .14, t =1.88). Therefore, there is evidence found to confirm hypothesis 4 at a marginal level of significance.

To analyze if a sensory appeal motive influences a positive attitude, a regression of a positive attitude on a sensory appeal motive is executed. The results show that a sensory appeal motive has a significant positive effect on a positive attitude (β = .23, t = 2.57). This means that there is evidence to confirm hypothesis 5.

6

The original hypothesis 1 and 2 are combined, the reason for this is stated in chapter 4.2.4.

7

Since the dependent variables are not normally distributed, an ordered probit (ordinal regression) is executed. The significance labels for the relationship between the motives and negative attitude stayed the same. The significance labels for the relationship between the motives and positive attitude also stayed the same.

8

Animal welfare motive (β = .14, t = 2.14).

9

(31)

31

Table 8 - Regression analyses

Variable Positive attitude Negative attitude

A0 A1 B0 B1 Constant 3.51 (.78)** 3.01 (.84)** 3.01 (.78)** 2.67 (.82)** Gender -.05 (.14) -.03 (.14) .07 (.14) .05 (.13) Age -.00 (.00) -.01 (.01) .01 (.01)† .02 (.01)† Education level .13 (.05)* .14 (.05)** -.06 (.05) -.07 (.05) Living area -.12 (.06† -.12 (.06)† -.12 (.06)† -.15 (.06)*

Current situation in labor market .10 (.12) .10 (.12) -.08 (.12) -.06 (.12)

Income -.00 (.00) -.00 (.00) -.00 (.00) -.00 (.00)

Household size .01 (.04) -.01 (0.03) -.01 (.04) -.00 (.03)

Own grocery shopping -.21 (.10)* -.20 (.01)* .03 (.10) .04 (.09)

Cook own food -.01 (.10) -.01 (.10) -.09 (.10) -.08 (.10)

Eat alone -.07 (.08) -.04 (.08) .09 (.08) .06 (.07)

Eat with vegetarians/vegans -.01 (.07) -.02 (.07) .12 (.07)† .08 (.07)

Ordered their food .08 (.10) .02 (.10) -.12 (.10) -.10 (.10)

Nutrition involvement -.07 (.08) -.06 (.09) -.06 (.08) -.09 (.09)

Subjective norm 0.01 (.07) -.01 (.07) -.01 (.07) .03 (.07)

Perceived behavioral control .15 (.09) .11 (.10) .02 (.09) .03 (.09)

Ethical motive -14 (.08)† .22 (.08)**

Health motive .02 (.12) -.01 (.11)

Familiarity motive .14 (.08)† -.19 (.07)**

Sensory motive .23 (.09)* .03 (.09)

(32)

32

4.5 Influence on ambivalence

The goal of this thesis is to examine whether the motives influence the ambivalence towards meat. Ambivalence towards meat occurs when people have simultaneously a negative- and a positive attitude towards meat. In this thesis, four models are used to test the hypothesis that the motives influence ambivalence towards meat. The baseline models are shown in appendix C and the models with the main variables are shown in table 11. Two forms of ambivalence are inspected: the direct ambivalence towards meat and indirect ambivalence towards meat. Both forms of ambivalence are discussed separately.

4.5.1 Direct ambivalence

People with a high direct ambivalence towards meat report that they have conflicting feelings and thoughts about meat, while people with a low direct ambivalence towards meat report that they barely have conflicting feelings and thoughts about meat.

4.5.2 Descriptive statistics direct ambivalence

Direct ambivalence is not normally distributed. The descriptive statistics of direct ambivalence are shown in table 9. In total, 82 respondents explicitly state that they are ambivalent towards meat, since they have conflicting feelings and thoughts towards meat. Remarkably, only 25 respondents indicate that they do not have conflicting feelings and thoughts towards meat. To conclude, only 16.7% of the flexitarians explicitly state they are not ambivalent towards meat.

4.5.3 Model fit direct ambivalence

Model C0 with the control variables explains 17.2% of the variance of direct ambivalence. Model C0 is shown in appendix C. This model is significant (F = 1.842, p = 0.035) and four out of the fifteen individual variables reaches at least marginal significance. With the inclusion of the main variables in model C1, 19.5% of the variance of direct ambivalence is

Table 9 - Descriptive statistics direct ambivalence

Mixed feelings/thoughts Respondents (%)

(33)

33

explained. This model is significant at a marginal level of significance (F = 1.648, p = 0.054) and three out of the nineteen variables reach at least marginal significance. However, the adjusted R square decreases from model C0 to C1 (Adj R square C0: .079; Adj R square C1: .077). This means that the additional main variables included in model C1 are not considered useful. Consequently, the model fit of model C1 is considered weak.

4.5.4 Parameter estimates direct ambivalence

To analyze whether an ethical motive influences direct ambivalence towards meat, a regression of direct ambivalence on an ethical motive is executed10. The results show that an ethical motive does not have a significant effect on direct ambivalence (β = .04, t = .43)11

. Therefore, there is not significant evidence to suggest that the ethical motive influences direct ambivalence. The results of the regression analysis are shown in table 11.

To assess whether a health motive influences direct ambivalence, a regression of direct ambivalence on a health motive is executed. According to the results, a health motive does not influence direct ambivalence (β = .03, t = .27). This means that there is no evidence to suggest that the health motive influences direct ambivalence.

Direct ambivalence is regressed on the familiarity motive, to analyze if a familiarity motive influences direct ambivalence. A familiarity motive does not have a significant effect on direct ambivalence (β = -.10, t = -1.16). Therefore, there is no evidence to suggest that the familiarity motive influences direct ambivalence.

Finally, to analyze whether a sensory appeal motive influences direct ambivalence, a regression of direct ambivalence on a sensory appeal motive is executed. The results show that a sensory appeal motive does not have a significant effect on direct ambivalence (β = .11, t = 1.257). Therefore, there is no significant evidence to suggest that the sensory appeal motive influences direct ambivalence.

10

Regarding the assumption checks: (1) The variables direct ambivalence and indirect ambivalence are not normally distributed, but log transform did not fix this issue, therefore the dependent variables are not log

transformed. (2) All the VIF values were far below the value of 10, therefore no multicollinearity took place. (3) The scatter plots seem to point to a possible weak linear relationship between the motives and ambivalence.

11

(34)

34

4.5.5 Indirect ambivalence

Indirect ambivalence is measured by two separate questions that inquired about the respondents’ positive- and negative attitudes towards meat. Respondents with a high direct ambivalence towards meat reported that they had both a positive- and a negative attitude towards meat. While respondents with a low ambivalence towards meat either indicated that they did not have positive- and negative attitudes towards meat or reported that one of their attitudes was way higher than the other attitude.

4.5.6 Descriptive statistics indirect ambivalence

Indirect ambivalence is not normally distributed. The descriptive statistics relating to indirect ambivalence are shown in table 10. Thirty respondents have an ambivalence score that was low (below 1.5). These respondents either had a large discrepancy between the negative- and the positive attitude or they did not experience a positive- and a negative attitude. The largest group of flexitarians (58 respondents) had an ambivalence score of 1.5. Two choice option could lead to an ambivalence score of 1.5. (1) The respondents could have a slightly positive attitude and a moderately negative attitude towards meat or (2) a moderately positive attitude and a slightly negative attitude towards meat. These respondents experience some ambivalence towards meat. While 57 respondents had an ambivalence score equal to or higher than 2.5. This means that these respondents had at least a moderately positive- and negative attitude towards meat. To conclude, according to the formula of indirect ambivalence, only thirty respondents did not have ambivalent feelings towards meat.

4.5.7 Model fit indirect ambivalence

Model D0 with the control variables explains 14.2% of the variance of indirect ambivalence. Model D0 is shown in appendix C. This model is not significant (F = 1.471, p =

Table 10 - Descriptive statistics indirect ambivalence

Ambivalence score Respondents (%)

≤ 1 30 (20.1%)

1 ≤ 2 62 (41.7%)

2 ≤ 3 50 (33.5%)

3 ≤ 4 7 (4.7%)

(35)

35

0.125). Furthermore, four out of the fifteen individual variables included reach at least marginal significance. With the inclusion of the main variables in model D1, 17.6% of the variance of indirect ambivalence is explained. This model is not significant (F = 1.451, p = 0.115), with again four out of the nineteen variables reaching at least marginal significance. The adjusted R square increases a little from model D0 to D1 (Adj R square D0: .046; Adj R square D1: .055). This means that the main variables that are added in model D1, are considered useful, but not to a great extent. To conclude, the model fit of model C1 is not optimal.

4.5.8 Parameter estimates indirect ambivalence

To investigate whether an ethical motive influences indirect ambivalence towards meat, a regression of indirect ambivalence on an ethical motive is executed. The results show that an ethical motive does not have a significant effect on indirect ambivalence (β = -.00, t = -.05). Therefore, there is no evidence found to suggest that the ethical motive influences indirect ambivalence. The results are shown in table 11.

To analyze whether a health motive influences indirect ambivalence, a regression of indirect ambivalence on a health motive is executed. According to the results, a health motive does not have an effect on indirect ambivalence (β = .10, t = .97). Therefore, there is no evidence to suggest that the health motive influences indirect ambivalence.

Indirect ambivalence is regressed on the familiarity motive, to analyze whether a familiarity motive influences indirect ambivalence. A familiarity motive does not have a significant effect on indirect ambivalence (β = .10, t = 1.22). Therefore, there is no evidence to suggest that the familiarity motive influences indirect ambivalence.

(36)

36

Table 11 Regression analyses - Ambivalence towards meat

Variable Direct ambivalence Indirect ambivalence C1 D1 Constant -.00 (.08) .00 (.08) Gender -.03 (.09) .03 (.09) Age .05 (.12) .15 (.12) Education level .02 (.09) .19* (.09) Living area .08 (.09) -.21* (.10)

Current situation in labor market .09 (.10) .09 (.10)

Income .04 (.12) -.25* (.12)

Household size -.23* (.09) -.08 (.09)

Own grocery shopping -.21† (.12) -.24† (.12)

Cook own food .17 (.12) -.02 (.12)

Eat alone -.13 (.10) -.09 (.10)

Eat with vegetarians/vegans .12 (.09) -.06 (-.09)

Ordered their food .02 (.09) -.05 (.09)

Nutrition involvement -.11 (.10) -.13 (.10)

Subjective Norm .11 (.09) -.01 (.09)

Perceived Behavioral Control -.21* (.09) .09 (.09)

Ethical motive .11 (.09) .13 (.09)

Health motive .03 (.10) .10 (.10)

Familiarity motive .04 (.09) -.00 (.09)

Sensory motive -.10 (.09) .11 (.09)

(37)

37

4.6 Hypotheses

The hypotheses with the corresponding results are shown in table 12.

Table 12 - Hypotheses

Hypothesis Result Influence on

ambivalence score

H1: An ethical motive is positively related to a negative attitude towards meat.

Accepted. None.

H3: A health motive is positively related to a negative attitude towards meat.

Rejected. None.

H4: A familiarity motive is positively related to a positive attitude towards meat.

Accepted at a marginal level of significance.

None.

H5: A sensory appeal motive is positively related to a positive attitude towards meat.

Accepted. None.

4.7 Additional analysis

Up until this point, this research has investigated the motives and attitudes of flexitarians. In addition to this, I will now explore the actual meat-eating behaviors and the reported future meat consumption of flexitarians. The two most important factors of the previously mentioned factor analysis are discussed. These factors are: ‘normal meat-eating frequency’ and ‘less future meat consumption’. In accordance with previous research, ambivalence is used as a moderator (Conner et al., 2002). Ambivalence towards meat is used as a moderator on the relationship between the different movies and normal meat-eating frequency/less future meat consumption, because ambivalence could have a disruptive effect on the relationships between the familiarity- and sensory appeal motive and behavior/less future consumption, while it could strengthen the effect on the relationships between the ethical- and health motive and behavior/less future consumption.

Six models are estimated for the additional analyses12. Normal meat-eating frequency is the dependent variable in model E0, E1, and E2. Model E0 is the base model which consists of the control variables and model E1 is the full model where the main variables are added. Model

12

(38)

38

E2 is the moderated model where the interactions between the different motives and indirect ambivalence are added. Less future meat consumption is the dependent variable in model F0, F1, and F2. Model F0 is the base model which consists of the control variables, model F1 is the full model where the main variables are added, and model F2 is the moderated model where the interactions between the different motives and direct ambivalence are added. Model E0, E1, F1, and F2 are shown in appendix D and model E2 and F2 are shown in table 13.

4.7.1 Descriptive statistics normal meat-eating frequency

The factor ‘normal meat frequency’ consists out of three main questions. Flexitarians that score one standard deviation below the mean consumed meat in the last week on average 3.2 times, consume meat normally 2.3 times per week, and when these flexitarians do eat meat, they do so less than half of the time at home. Flexitarians who score average on last week’s meat consumption, consumed meat 4.9 times last week, normally consume meat 3.9 times per week, and when these flexitarians eat meat, they do so approximately half of the time at home (M = 3.26, S.D. = .90). Flexitarians who score one standard deviation above the mean consumed meat last week on average 6.6 times, normally consume meat 5.5 times per week, and when these flexitarians eat meat, they do this often at home.

4.7.2 Model fit normal meat-eating frequencies

(39)

39

variables and interactions that are added in model E1 and E2 are considered useful. To conclude, the model fit of model E2 is good.

4.7.3 Parameter estimates normal meat-eating frequencies

To analyze whether the motives and indirect ambivalence influence normal eating frequency, a regression of normal meat-eating frequency on the ethical-, health-, sensory appeal-, the familiarity motive, and indirect ambivalence is executed13. The results are shown in table 13.

The results show that an ethical motive has a significant negative effect on normal meat-eating frequency (model E1: β = -.24, t = -3.00; model E2: β = -.24, t = -2.97). According to the results, a health motive does have a positive marginal significant effect on normal eating frequency in model E1 (β = .18, t = 1.80) but does not have an effect on normal eating frequency in model E2 (β = .16, t = 1.63). Furthermore, the results show that a sensory appeal motive does not have a significant effect on normal eating frequency (model E1: β = .04, t = .50; model E2: β = .02, t = .29). The results also show that the familiarity motive does not have a significant effect on normal meat-eating frequency (model E1: β = .13, t = 1.62; model E2: β = .11, t = 1.39). Lastly, the results reveal that indirect ambivalence does not have a significant effect on normal meat-eating frequency (model E1: β = .06, t = .81; model E2: β = .07, t = .88).

To analyze whether indirect ambivalence towards meat moderates the relationship between the motives and normal meat-eating frequencies, a regression of normal meat-eating frequency on the interactions between indirect ambivalence and the motives is executed. The results show that indirect ambivalence does not moderate the relationship between the ethical motive and normal meat-eating frequencies (β = .05, t = .71). Furthermore, the results show that indirect ambivalence moderates the relationship between the health motive and normal meat-eating frequencies (β = .17, t = 2.13). More specifically: indirect ambivalence strengthens the positive, but insignificant or marginally significant relationship between the health motive and normal meat-eating frequencies. The results also show that indirect ambivalence does not moderate the relationship between the sensory appeal motive and normal meat-eating frequencies (β = -.06, t = -.79). Lastly, the results reveal that indirect ambivalence moderates the

13

Regarding the assumption checks: (1) Normal meat frequency is normally distributed, since the Shapiro-Wilk Test of Normality is not significant (.08 > .05) (2) All the VIF values were far below the value of 10, therefore no multicollinearity took place. (3) The scatter plots point to linear relationships or to possible weak linear

(40)

40

relationship between the familiarity motive and normal meat-eating frequencies at a marginal level of significance (β = -.14, t = -1.82). More specifically: indirect ambivalence weakens the positive, but insignificant relationship between the familiarity motive and normal meat-eating frequencies at a marginal level of significance.

4.7.4 Descriptive statistics less future meat consumption

The factor ‘less future meat consumption’ consists of two main questions. Flexitarians who scored one standard deviation below the mean are somewhat to moderately willing to consider consuming less meat in the near future and are somewhat intending to reduce their meat consumption in the next six months. Flexitarians who score the average, are between moderately willing and willing about considering reducing their meat consumption (M = 3.61, S.D. = 1.08) and score between moderately willing and willing (M = 3.00, S.D. = 1.3) regarding intending to consume less meat in the next six months. Flexitarians who score one standard deviation above the mean are more than willing to consider reducing their meat consumption in the near future and intend to reduce their meat consumption in the next six months.

4.7.5 Model fit less future meat consumption

(41)

41

4.7.6 Parameter estimates less future meat consumption

To analyze whether the motives and direct ambivalence influence less future meat consumption, a regression of less future meat consumption on the ethical-, health-, sensory appeal-, the familiarity motive, and direct ambivalence is executed14. The results are shown in table 13.

The results show that an ethical motive has a significant positive effect on less future meat consumption (model E1: β = .34, t = 4.14; model E2: β = .32, t = 3.81). According to the results, a health motive does not have a significant effect on less future meat consumption (model E1: β = .02, t = .17; model E2: β = .04, t = .40). Furthermore, the results show that a sensory appeal motive does not have a significant effect on less future meat consumption (model E1: β = -.04, t = -.45; model E2: β = -.03, t = -.30). The results also show that the familiarity motive does not have a significant effect on less future meat consumption (model E1: β = -.11, t = -1.37; model E2: β = -.13, t = -1.55). Lastly, the results indicate that direct ambivalence does have a significant effect on less future meat consumption (model E1: β = .22, t = 2.81; model E2: β = .25, t = 3.00).

To investigate whether direct ambivalence towards meat moderates the relationship between the motives and less future meat consumption, a regression of less future meat consumption on the interactions between direct ambivalence and the motives is executed. The results show that direct ambivalence does not moderate the relationship between the ethical motive and less future meat consumption (β = -.09, t = -1.32). Furthermore, the results show that direct ambivalence does not moderate the relationship between the health motive and less future meat consumption (β = -.08, t = -.89). The results also show that direct ambivalence does not moderate the relationship between the sensory appeal motive and less future meat consumption (β = .03, t = .36). Lastly, the results reveal that direct ambivalence does not moderate the relationship between the familiarity motive and less future meat consumption (β = .06, t = .78).

14

Referenties

GERELATEERDE DOCUMENTEN

› Of the different motives, the ethical motive positively influences less future meat consumption. › Direct ambivalence positively influences less future

consumption appeal (relative to a green consumption appeal) on a customer its brand attitude and how is this effect moderated by supermarket brand

In the control group, the product packaging was shown in its original state (i.e. without an eco-label). A total of 5 choice sets was shown to each respondent. It is

If people who are positively self-affirmed do not abandon their intentions to reduce meat consumption, the defensive response was most likely driven by psychological threat..

The first two are policing functions, needed to make sure that PCN- packets admitted into the PCN-domain belong to a flow that has been admitted and to ensure that the

The slope of the magnetic field dependence of the SMR signal changes sign at the transition between the helical and conical spiral states and shows another discontinuity when

15005-EEF: Dijkstra, P.T., Price Leadership and Unequal Market Sharing: Collusion in Experimental Markets.. Tuinstra, Fee Structure, Return Chasing and Mutual Fund Choice:

If students can access the academic support programmes provided to them at an earlier stage, they may be able to integrate academically and socially into the tertiary