Stop Consuming Meat for Me or for Us?
An experimental study investigating the effects of health and environmental information on food consumption and the moderating role of pre-existing values with the use of Experience Sampling Methodology
Master’s Thesis
Demi van der Plas 10748067
Graduate School of Amsterdam
Research Master’s Program Communication Science Supported by Digital Communication Methods Lab Fund
Supervisor: Mw. dr. Marijn Meijers 31-01-2020
Abstract
In 2018, people in the Netherlands have increased their meat consumption for the first time in ten years, even though there is evidence pointing towards the negative effects of meat on human health and the environment. The current study tests to which extent the effects of health and environmental information on meat consumption are moderated by pre-existing values with the use of Experience Sampling Methodology. The results show that providing information about the consequences of (not) eating meat on one’s health or the environment is sufficient enough to lower the frequency of meat consumption. Furthermore, only health information is effective in lowering the quantity of meat consumption, compared to environmental information. Additionally, people with high levels of self-transcendence appear to be self-licensing by decreasing their frequency of meat consumption but increasing their quantity of meat consumption. No such effects have been found for people with low levels of self-transcendence.
Key words: Information provision, framing, environment, health, meat consumption, pre-existing values, value orientations
In 2018, people in the Netherlands have increased their meat consumption for the first time in ten years (Dagevos, Verhoog, van Horne, & Hoste, 2019). This is an alarming
development, as the production and consumption of meat have negative effects on both the environment and human health (Godfray et al., 2018). The production of meat contributes largely to all greenhouse gas emissions (GHGEs; Godfray et al., 2018) and excessive meat intake is associated with colorectal cancer, cardiovascular diseases, diabetes and weight gain (Bouvard et al., 2015; Rohrmann et al., 2013; Vergnaud et al., 2010; Wang et al., 2016; Wolk, 2017). Hence, there is an urgent call for a change.
One of the strategies that is put forward as a solution for mitigating the increased pressure on the environment and negative effects on human health is to switch to a more plant-based diet and, consequently, to consume less meat (Godfray et al., 2018; Springmann et al., 2018). However, stimulating meat-eaters to consume less meat and more plant-based is a challenging task because meat-eaters show an overall negative attitude and a low motivation towards consuming less animal products (Hoek, Pearson, James, Lawrence, & Friel, 2017). This is likely due to the fact that people do not have knowledge about the relationship between meat consumption and the environment (Lentz, Connelly, Mirosa, & Jowett, 2018; Macdiarmid, Douglas, & Campbell, 2016; Tobler, Visschers, & Siegrist, 2011) and the belief that meat-based diets are healthier than diets without meat (Stoll-Kleemann & Schmidt, 2017).
The provision of information about the impact of meat consumption could potentially be an effective strategy in changing people’s meat intake. Information provision is used in order to enhance people’s knowledge and in general, information provision is believed to improve individuals’ knowledge, change their attitudes, and redirect behavior with regards to food choices (Verbeke, 2008). Studies show that the use of information provision is effective in increasing people’s attitude about reducing one’s meat consumption and it also increases
their intention to do so (Harguess, Crespo, & Hong, 2019). Yet, no study hitherto has investigated whether increasing people’s knowledge by providing information about the impact of meat consumption can influence actual meat intake. However, studying these effects on behavioral outcomes is important to circumvent intention-behavior gap that is visible in sustainable behavior, where intentions only account for 27% of the variance in actual behavior (Bamberg & Möser, 2007).
Hence, this study investigates the effectiveness of providing health or environmental related consequences in relation to meat consumption as both of these types of information about the consequences have shown to positively increase consumers’ attitude about
sustainable products and increase their motivation to purchase such products (Harguess et al., 2019; Kareklas, Carlson, & Muehling, 2014; Yadav, 2016).
Moreover, the use of information provision is generally more effective if it is matched with consumer’s characteristics (Verbeke, 2008), such as values. Values are ‘guiding
principles in one’s life, which put them at the core of many decisions people make in their daily lives’ (Graham & Abrahamse, 2017, p. 99; Schwartz, 2012). In line with this, pre-existing values are related to food choices (e.g. Allen & Ng, 2003; de Boer, Schösler, & Aiking, 2017; Kalof, Dietz, Stern, & Guagnano, 1999) and sustainable behavior (e.g. De Groot & Steg, 2007; Nordlund & Garvill, 2002; Schultz et al., 2005), and are a viable
consumer characteristic which is rather unexplored in the literature. Literature on information provision shows that when information is aligned with people’s values, more positive
responses are elicited than when this is not the case (De Dominicis, Schultz, & Bonaiuto, 2017; Verbeke, 2008).
Until now, no study has investigated whether the different types of information (e.g. health or environment) and differences in pre-existing values can be effectively combined in order to decrease people’s meat consumption. This study aims to fill these gaps with the
following research question: “To what extent are the effects of health and environmental information on meat consumption moderated by pre-existing values?”. The study contributes to a solution for mitigating climate change by reducing meat intake and strives to provide policy makers and other agencies with insights how to effectively design messages that reduce meat consumption among the public.
Theoretical Framework Information Provision
The use of information provision is one of the most used approaches in sustainability communication (Moser & Dilling, 2009). The knowledge-deficit model of Schultz (2002) assumes that individuals are principally willing to change their behavior, provided that they understand why they should do so. Providing information is essential as it increases
individuals’ knowledge of how and/or why they should make a change and are therefore more willing to change their behavior (Schultz, 2002). In line with this, studies show that existing knowledge about the impact of meat consumption on health and the environment has been associated with the intention to reduce one’s meat intake and a general willingness to adopt a more vegetarian-based diet (Mullee et al., 2017; Pohjolainen, Tapio, Vinnari, Jokinen, & Räsänen, 2016). Furthermore, studies show that the use of information provision can be used to increase people’s knowledge and was effective in increasing intention or willingness to consume less meat as well (Cordts, Nitzko, & Spiller, 2014; Graham & Abrahamse, 2017). Enhancing people’s knowledge by means of information provision in combination with (self)monitoring of meat consumption also reduced one’s willingness to eat meat (Carfora, Caso, & Conner, 2017a, 2017b; Loy, Wieber, Gollwitzer, & Oettingen, 2016). While these studies have tested the effects of information provision on the attitudes towards and intentions of reducing meat consumption in an experimental setting, only one study took into account meat consuming behavior through means of a daily food diary (Carfora et al., 2017a). Even
though information provision decreased meat consumption, no baseline meat consumption was measured and adding this to a study can positively reinforce the effectiveness of information provision. Therefore, it is hypothesized that the provision of information can decrease meat consumption (H1).
Message Framing: Health vs. Environment
To provide information in a way that is relatable to the receiver, a person’s motivations for engaging in a particular behavior can be considered (Barr & Gilg, 2007). Hereby, the relevance of the information that is presented is enhanced and can therefore contribute to a greater change in behavior in a way that is desired (Hawkins, Kreuter,
Resnicow, Fishbein, & Dijkstra, 2008). Enhancing relevance of the information can be done through the means of message framing (Graham & Abrahamse, 2017).
Message framing describes the act of presenting information in a specific way to the audience, which can affect how the message is used and interpreted (Chong & Druckman, 2007). One particular way of framing is emphasis framing, which focuses on stressing a subset of considerations which are qualitatively different, yet potentially relevant (Chong & Druckman, 2007). Studies have shown that health and environmental motives are the most important reasons for reducing meat consumption compared to other motives including animal welfare (Beardsworth & Keil, 1991; Santos & Booth, 1995; Tobler et al., 2011; Zur & Klöckner, 2014). In order to stimulate people to consume less meat, it can therefore be
beneficial to use health and environmental frames to provide information about the impact of meat consumption.
Both environmental and health frames have been used to increase people’s attitude and intention to buy more sustainable (Harguess et al., 2019; Kareklas et al., 2014; Yadav, 2016). However, there have only been two experimental studies that explicitly compared the
yielded mixed results. Kareklas et al. (2014) found that a combination of health and
environmental information was more effective in influencing sustainable buying intentions than health information alone, and as effective as providing environmental information only. In contrast, Yadav (2016) found that even though both health and environmental motives were equally important for the intention to behave more sustainable, highlighting health information impacted the intention more than environmental information. Though, some correlational studies suggest that health frames are more powerful in instigating sustainable behavior, because health motives were consistently more important to behave sustainable compared to environmental motives (Hoek et al., 2017; Magnusson, Arvola, Hursti, Åberg, & Sjödén, 2003).
Decreasing meat consumption can be considered a sustainable behavior, but there are no experimental studies that compare the effectiveness of health or environmental frames on actual meat intake. Correlational studies suggest that using health information to stimulate people to reduce their meat consumption might be more effective than environmental information, as health motives are also the most important reason for reducing meat
consumption (Lea, Crawford, & Worsley, 2006; Tobler et al., 2011). To address the gap in the literature, this study wishes to experimentally compare the effectiveness of health and
environmental frames on meat consuming behavior.
A possible explanation why health frames are argued to be more effective is provided by the social dilemma theory (Dawes, 1980). In a social dilemma, individual and collective interests are at odds with each other and individuals usually make decisions based on short-term gains and individual interests, as collective interests often involve sacrifices in the short run. Collective interests are ultimately beneficial for all individuals. In spite of the eventual benefit, individuals do not see a personal, short-term interest in this and are less inclined to behave accordingly, thereby creating a social dilemma (Aertsens, Verbeke, Mondelaers, &
van Huylenbroeck, 2009; Aschemann-Witzel, 2015; van Dam & van Trijp, 2013). In relation to decreasing meat consumption, the collective interest is related to the environment and the preservation of it, while short-term individual interests might be related to liking the taste of meat or a satisfier for being hungry.
As such, health frames emphasize individual benefits of the behavior and carry (relatively) short-term, personal gains, while environmental frames emphasize collective interests and carry long-term, collective gains. Because of the different focus on interest, health frames might thus be more effective, as the focus is on the individual. Therefore, the expectation is that health information decreases meat consumption to a greater extent than environment information (H2).
Pre-Existing Values
While there is evidence that health frames appear to be more persuading, some studies suggest that the effectiveness of these individualistic (e.g. health) or collectivistic (e.g.
environmental) frames depends on pre-existing values (De Dominicis et al., 2017; Graham & Abrahamse, 2017; Schultz & Zelezny, 2003; von Borgstede, Andersson, & Hansla, 2014). Values are defined as ‘important life principles and goals that drive a person’s actions’ and are classified along two dimensions: self (a continuum from transcendence to self-enhancement) and change (a continuum from openness to change to conservatism; Schwartz, 1994). In the current study, the dimension of self is of particular importance.
Self-transcendence can be described as having goals that are not related to oneself and that it promotes ‘the interests of other persons and the natural world’, while self-enhancement focuses on goals that promotes one’s self-interest, including wealth, power and influence (Schwartz, 1994b). Therefore, self-transcendence is linked to collectivistic interests as both focus on goals that overstep the self, while self-enhancement is associated with individualistic interests, where the focus is more towards the self.
The self-transcendence/enhancement value dimension is related to sustainable behavior, such that self-transcendence values positively correlate with behaving sustainable, and self-enhancement values negatively correlate with behaving sustainable (Karp, 1996; Nordlund & Garvill, 2002; Schultz & Zelezny, 1999, 1998; Stern, Dietz, Abel, Guagnano, & Kalof, 1999). This can be reasoned by the fact that sustainability issues are typically framed in a way that is more congruent with self-transcendence values, such as focusing on the things one can do to save, help or protect the environment and sacrificing something as an individual (De Dominicis et al., 2017; Schultz & Zelezny, 2003; von Borgstede et al., 2014). This is more appealing to people with self-transcendence values, while people who endorse values that are more concerned about oneself (e.g. self-enhancement values), might be less likely to behave sustainable (De Groot & Steg, 2007). Using an environmental frame might therefore be more suitable for people with transcendence values, as opposed to people with self-enhancement values.
In order to stimulate people with self-enhancement values to behave sustainable, it is suggested to present information in a value-congruent way. Presenting value-congruent information postulates that individuals who score higher on self-enhancement values will act more sustainable when an issue is framed as something that directly affects them, thus enhancing the motivation to act upon it (Schultz & Zelezny, 2003), by for example using health information. Based on existing literature on value-framed information, value-congruent information is more effective for promoting sustainable behavior than presenting information in a value-incongruent way for individuals with self-enhancement values (De Dominicis et al., 2017; Graham & Abrahamse, 2017; Nilsson, Hansla, Heiling, Bergstad, & Martinsson, 2016; von Borgstede et al., 2014). People with self-transcendence values were not significantly more persuaded by self-transcendence frames and do not necessarily benefit from value-congruent information (Graham & Abrahamse, 2017; von Borgstede et al., 2014).
According to Schultz and Zelezny (2003), self-enhancing frames are the lowest common denominator, such that people with high self-enhancement values and people with high transcendence values are both persuaded by those frames. When making self-enhancement values more salient, people with high self-self-enhancement values are persuaded because the message is aligned with their values and focuses more on individualistic concerns. Additionally, people with high self-transcendence values are equally persuaded because even though their values are focused on things outside oneself, it does not mean that they do not care for themselves. A self-enhancing frame therefore serves as an extra motive to act upon the behavior for people with self-transcendence values (De Dominicis et al., 2017). De Dominicis et al. (2017) took the idea of Schultz and Zelezny (2003) of self-enhancement values being the lowest common denominator and argued for a concentric structure of concerns (Figure 1), such that individualistic concerns (which are more important for individuals with self-enhancement values) are included within collectivistic concerns. This concentric structure visually shows that self-enhancement values and the corresponding concerns are included in everyone, while self-transcendence values and the corresponding concerns are only included in some individuals.
Figure 1. Concentric structure of concerns based on individuals’ values.
Biospheric concerns (self-transcending) Social concerns (self-transcending) Individual concerns (self-enhancing)
Based on this, it is postulated that relating meat consumption to health consequences might motivate more individuals than focusing on environmental consequences, as a health frame (i.e., self-enhancing frame) appeals more to self-enhancement values, which appeals to more people (i.e., both people with self-enhancement and self-transcendence values). In contrast to an environmental frame, which appeals to people with self-transcendence values only. In line with this reasoning, it is hypothesized that a message with an environmental frame (i.e., self-transcending frame) only leads to a decrease in meat consumption in individuals with high self-transcendence values, while a message with a health frame (i.e., self-enhancement frame) leads to a decrease in meat consumption in both individuals that have high self-enhancement values or high self-transcendence values (H3).
Methods
This thesis investigated the moderating effects of enhancement and
self-transcendence values on the relationship between health and environmental information on meat consumption with the use of Experience Sampling Methodology (ESM). ESM is an intensive technique that asks participants to report on thoughts, feelings, and/or behaviors on multiple occasions over time (Bolger & Laurenceau, 2013). It has several advantages for a longitudinal design such as the current study. First of all, the self-reports are more reliable because their meat consumption could be logged during the same day. Furthermore, it is less obtrusive, because the participants’ own phone was used to fill in the surveys and did not take up much of the participants’ time. Lastly, ESM allows for the scheduling and automatic notification of sending surveys to participants. Both of these functions have been useful in this study for tracking daily meat consumption, while also exposing theto intervention messages which were about the health or environmental impact of meat consumption. To develop effective intervention messages, a pre-test has been conducted in which the stimuli was chosen.
Pre-test
The stimuli of this study were intervention messages either related to health or environmental information. These intervention messages were created in such a way that the health information appealed to health consequences of (not) eating meat in a (positive) negative way, while the environmental information described environmental consequences of (not) eating meat in a (positive) negative way (see Appendix A for all messages). In order to use messages that were perceived as being health- or environment-related in the main study, a set of ten messages for each type were pretested. The pre-test was conducted among Dutch meat-eaters (N = 34) ranging from twenty to fifty-five years old (M = 26.12, SD = 7.60). The sample consisted of 58.8% females (n = 20) and 41.2% males (n = 14). The participants had a mean education of a bachelor’s or equivalent. Participants were randomly exposed to ten health messages and ten environment messages. Participants rated whether they believed the arguments were about the impact on health or environment (1 = completely disagree, 7 = completely agree) and whether the arguments were perceived as a weak or strong argument to reduce meat consumption (1 = very weak, 7 = very strong). They could also indicate whether they thought the argument was interesting or entertaining to read (1 = completely disagree, 7 = completely agree). These questions were included in order to keep the messages as equal as possible across these variables.
The goal of the pre-test was to select three health and three environment messages to be used in the main study. Based on mean scores and paired sample t-tests, three health messages and three environment messages that differed significantly from each other on their scores of impact on health/environment, but did not differ significantly from each other on the variables: strength, entertaining, and interesting, were chosen for further analyses (see Table 1 for means and standard deviations). The three health messages and the three environment messages were averaged for each of the variables and were analyzed with paired samples
t-tests again. The stimuli were found to be successful, the three health messages were significantly more perceived to be about health (M = 6.56, SD = 0.54) than the three
environment messages (M = 2.42, SD = 1.54), t(33) = 14.24, p < .001, while the environment messages were significantly more perceived to be about environment (M = 6.45, SD = 0.60) than the three health messages (M = 2.45, SD = 1.67), t(33) = -12.00, p < .001. Furthermore, there were no significant results in perceived argument strength scores in the health messages (M = 5.11, SD = 1.46) and environment messages (M = 4.98, SD = 0.95); t(33) = .57, p = .571, or in ratings of interest (Mhealth = 5.17, SDhealth = 1.29; Menvironment = 5.34, SDenvironment = 1.11); t(33) = -1.11, p = .274. Unexpectedly, environment messages were significantly more entertaining than the health messages (Mhealth = 3.16, SDhealth = 1.35; Menvironment = 3.71, SDenvironment = 1.21); t(33) = -3.51, p < .001.
Table 1
Overview of Means and Standard Deviations of Pre-Tested Messages1
Message
Manipulation Characteristics
Health Environment Interesting Entertaining Strength
M SD M SD M SD M SD M SD E1 2.35 1.77 6.44 0.66 4.88 1.63 3.18 1.47 4.41 1.54 E2 2.26 1.60 6.53 0.71 5.12 1.39 3.03 1.57 4.91 1.24 E3 2.79 1.81 6.47 0.75 5.03 1.29 3.12 1.07 4.76 1.48 E4 3.85 1.96 6.50 0.62 5.09 1.33 3.12 1.68 4.56 1.42 E5 2.47 1.81 6.53 0.56 5.38 1.65 3.56 1.69 5.18 1.45 E6 2.59 1.69 6.32 0.81 4.26 1.62 3.29 1.59 3.76 1.65 E7 2.91 1.83 6.41 0.56 5.15 1.40 3.56 1.78 4.82 1.55 E8 2.53 1.76 6.29 0.97 5.53 1.48 4.53 1.75 4.82 1.51 E9 2.76 2.15 6.44 0.56 5.41 1.54 4.38 1.79 4.68 1.72 E10 4.56 2.06 6.41 0.78 5.15 1.52 3.44 1.56 4.62 1.58 H1 6.71 0.46 2.12 1.70 5.18 1.53 2.71 1.32 5.26 1.76 H2 6.62 0.55 2.32 1.74 4.79 1.74 3.09 1.44 4.65 1.67 H3 6.44 0.66 2.68 1.84 5.29 1.38 3.32 1.57 5.12 1.45 H4 6.29 0.76 2.71 1.77 4.41 1.74 3.29 1.53 3.85 1.86 H5 6.44 0.71 2.21 1.61 4.85 1.54 3.15 1.50 4.47 1.71 H6 6.26 0.90 2.71 1.84 4.82 1.73 3.47 1.64 4.53 1.64 H7 5.15 1.46 4.85 1.91 5.62 1.07 3.91 1.51 4.38 1.42 H8 6.44 0.66 2.74 1.81 4.65 1.54 3.41 1.64 4.41 1.52 H9 6.53 0.96 2.56 1.91 5.03 1.57 3.44 1.67 4.94 1.65 H10 6.26 1.14 2.26 1.58 4.91 1.49 3.38 1.60 4.47 1.54
1 Messages that have been selected in the main study can be found in bold. For an overview of the textual messages, see Appendix A.
Main study
The main study tested the hypotheses experimentally by investigating to what extent the effectiveness of health and environmental information on meat consumption was
moderated by pre-existing values by means of a longitudinal ESM study over a three-week period.
Participants and Design
The sample of the study consisted of Dutch residents of 18 years old and older, who consume meat on a regular basis and had a smartphone with which they could participate in the consumption tracker. People were recruited via the participant pool of the University of Amsterdam (participants either received €5 or 1 Research Credit), via Facebook advertising (participants opted for €5) and via personal communication (participants opted for €5). Table 2 shows the timeline of the study and demonstrates that the first week was used as a baseline tracker for participants’ daily meat consumption, while the second and third week were used as intervention weeks in which health and environmental messages were sent, next to the daily meat consumption tracker. To ensure enough data points per participant, it was decided that they should have at least completed four out of seven surveys in each week2.
The full model used a two-factorial within-subjects design with two conditions (framing: health vs. environment). Two moderators were included: self-enhancement values (low vs. high) and self-transcendence values (low vs. high). In order to counterbalance the order of the experimental interventions, participants were randomly assigned to either receiving health or environment messages first. The partial model used an experimental between-subjects design. Participants were randomly assigned to one of two experimental conditions (framing: health vs. environment) and the same moderators were included:
self-2 Due to the high attrition rate that arose from this decision, two models have been used to perform analyses on. In the remainder of this thesis, the term ‘full model’ will be used to describe participants who completed the baseline, week 1 and week 2 meat consumption tracker, while the term ‘partial model’ will be used to describe participants who completed the baseline and week 1 meat consumption tracker.
enhancement values (low vs. high) and self-transcendence values (low vs. high). For both models, the dependent variable was daily meat consumption in frequency and quantity.
A total amount of 95 participants completed the baseline week consumption tracker (Mage = 22.53, SD = 6.67; 77.5% female; the majority received an education in upper secondary education; 65.8%). The partial model consisted of 82 participants (Mage = 22.90, SD = 7.53; 78.0% female; the majority received an education in upper secondary education; 62.2%), while the full model consisted of 51 participants (Mage = 22.35, SD = 7.03; 78.0% female; the majority received an education in upper secondary education; 64.7%)
Procedure
All participants were first presented with an informed consent form, ethical
information, and a survey with inclusion questions about their general meat consumption and having access to a mobile phone. Hereafter, participants answered questions about their pre-existing values, identification with being healthy and environmentally friendly, demographic questions, and a question for their email address to which the invitation to the MyPanel application could be sent to. Approximately one week before the first week of data collection, participants were sent an invitation to the MyPanel app. This invitation consisted of a
personal password with which they could log into the app. The MyPanel app is a mobile application which makes use of the Experience Sampling Methodology to send people push notifications every time a questionnaire is ready to be answered by the participants. The data was collected over a period of three weeks with the use of this app. In the three weeks, a daily push notification was sent to all participants at 8 p.m. to fill in their daily meat consumption. These surveys were available for four hours. Additionally, in the two intervention weeks, all participants received an experimental intervention at 8 a.m. on three days (see Table 2 for the timeline3). These interventions consisted of the pre-tested messages (e.g., “ Did you know that
eating less meat contributes to a healthy lifestyle, because you are less at risk for a heart stroke or other cardiovascular diseases?” or “Did you know that producing meat costs a lot of (fresh)water? To produce 100 grams of beef, 1.500 liters of water is necessary”) and included a question about these messages (e.g., “What did the text you have just read,
describe?”) to ensure participants’ exposure to them. These interventions were also available for four hours. On those days, participants had to fill in two surveys; the survey including the message and the meat consumption tracker. Screenshots of the push message and the
MyPanel environment can be found in Appendix B. Table 2
Timeline
Measures
Self-transcendent vs. self-enhancement values. The moderator variable ‘pre-existing values’ was measured by items of the Portrait Value Questionnaire (PVQ) by Schwartz et al. (2012) and was adapted for this study by only using the items for enhancement and self-transcendence in order to decrease survey fatigue and to shorten the survey. The PVQ has been tested and validated in several studies, and following the study of Giménez & Tamajón (2019), the statements described persons in the first person and used a Likert scale ranging from “Strongly Disagree” (1) to “Strongly Agree” (7). Statements included: “For me, it is important to take care of people who are close to me”, “For me, it is important to have
Week 1
16-22 December Week 2 2-8 January Week 3 9-15 January
Measurement Baseline Experimental Week 1 Experimental Week 2 Meat consumption tracker Monday Tuesday Wednesday Thursday Friday Saturday Sunday Thursday Friday Saturday Sunday Monday Tuesday Wednesday Thursday Friday Saturday Sunday Monday Tuesday Wednesday Intervention Thursday Saturday Monday Thursday Saturday Monday
ambitions in life”, and “For me, it is important to be wealthy”. All statements can be found in Appendix C. Both scales for self-transcendence and self-enhancement were reliable for the partial model (aST = .81; aSE = .78), as well as for the full model (aST = .78; aSE = .73). In
order to create dichotomous variables for self-enhancement and self-transcendence values, a median split was executed (Partial model: MdnSE = 4.22, MdnST = 5.93; full model: MdnSE = 4.33, MdnST = 5.93).
Meat consumption. During the three weeks of data collection, participants were asked to report their meat consumption on a daily basis via the MyPanel app. The app sent out a push notification at 8 p.m. to let the participants know a new questionnaire was ready to be filled in (‘Hello! A new questionnaire is ready for you.’). In this questionnaire, they first saw a welcoming text (‘Hello and welcome to the meat consumption tracker of today.’). After that, they answered the questions: “Have you consumed meat during breakfast?”, “Have you consumed meat during your lunch?”, “Have you consumed meat during your diner?”, and “Have you consumed meat during another moment of the day (excluding breakfast, lunch, and diner)?”. These questions could be answered with “Yes” or “No”, and when participants answered “Yes”, they received an additional question for the specific eating moment(s): “Please indicate how much meat you have consumed during your
breakfast/lunch/diner/another moment of the day (excluding breakfast, lunch, and diner). This was an open-ended question (numeric value only), and the participants received guidelines: “Meats for bread (beleg) = approx. 15 grams. Meats as part of diner = approx. 75-100 grams.” Participants had to answer the questionnaire before midnight (00:00). A reminder was sent automatically at 10 p.m. if they did not fill in the available survey yet. The meat consumption per participant was calculated by summing the mean for breakfast, lunch, diner, and snack moments divided by the number of days participants reported their meat
frequency of meat consumption for the baseline and intervention week for the partial model, and in the average daily frequency of meat consumption for the baseline, the health
intervention week and the environment intervention week for the full model. With this, the difference between the baseline meat consumption and the intervention week, and between the baseline meat consumption, meat consumption during the health intervention and meat consumption during the environment intervention could be measured. The same calculation was done for the average daily quantity of meat consumption for the baseline and intervention week, and for the baseline, the health intervention week and the environment intervention week.
Control variables. Several control variables were measured in order to make sure the results were not influenced by them. First of all, gender was measured to check whether or not it correlated with meat consumption, as previous research suggests that women are less likely to consume meat to a high extent (Gossard & York, 2003; Westenhoefer, 2005). Age and level of education have been included as exploratory variables that might impact meat consumption. Furthermore, to check whether the effects of the interventions could be explained by identification with the topic (being healthy or environmentally friendly), variables that measured these identifications were also included in the first questionnaire.
Identification with being environmentally friendly. Identification with being
environmentally friendly was measured by combining several items (c.f. Van der Werff, Steg, & Keizer, 2013): ‘Acting environmentally friendly is an important part of who I am’, ‘I am the type of person who acts environmentally friendly’, ‘I see myself as an environmentally friendly person’, where participants could answer on a 7-point Likert scale ranging from “Strongly Disagree” (1) to “Strongly Agree” (7). For the partial model, the scale is highly reliable (a = .85). Participants scored average on the identification with environment (M = 4.60, SD = 0.92, min = 2, max = 6.33). For the full model, the scale is highly reliable as well
(a = .82) and participants scored slightly average on the identification with environment (M = 4.63, SD = 0.94, min = 3, max = 6.33).
Identification with being healthy. Identification with being healthy was measured
similarly as identification with being environmentally friendly (c.f. Van der Werff, Steg, & Keizer, 2013). The items included in this scale were: ‘Acting healthy is an important part of who I am’, ‘I am the type of person who acts healthy, ‘I see myself as a healthy person’, where participants could answer on a 7-point Likert scale ranging from “Strongly Disagree” (1) to “Strongly Agree” (7). For the partial model, the scale is highly reliable (a = .84). Participants score above average on the identification with health (M = 4.98, SD = 1.12, min = 2, max = 7). For the full model, the scale is highly reliable as well (a = .86). Participants score slightly average on the identification with health (M = 4.90, SD = 1.12, min = 2, max = 7).
Intervention check. To ensure that participants were indeed exposed to the intervention messages, participants were asked to describe what they had just read in the message. This was an open-ended question and was not used for analyzation purposes as its goal was solely to ensure people had been exposed to the intervention, which is checked by the number of participants who had filled in the survey.
Plan of Analysis
Due to the high attrition rate, the analyses were done on two sets of data. For the partial model, a 2 (meat consumption: baseline vs. week 1) x 2 (framing: health vs.
environment) mixed repeated measures ANOVA has been employed, including interaction terms for self-enhancement and self-transcendence values to assess the moderating role of pre-existing values on the relationship between information provision and meat consumption. For the full model, a 3 (meat consumption: baseline vs. environment week x health week) x 2 (order: health first vs. environment first) mixed repeated measures ANOVA has been
employed for meat consumption, including the interaction terms for self-enhancement and self-transcendence values as well. The ANOVAs were performed for meat consumption in frequency and in quantity.
Results Partial Model: Preliminary Analyses
Randomization check. To check whether the randomization of gender across the conditions was successful, a Chi-square analysis was performed with the variables condition (health vs. environment) and gender. Gender appeared to be equally divided among the conditions, χ2 (1) = 0.29, p = .594, fC = .06. The randomization of age was successful too, an
independent samples t-test with condition as independent variable and age as dependent variable showed an equal distribution of age among the conditions, t(80) = 1.06, p = .294, CI = [-1.55, 5.06], d = .23. Lastly, the randomization of level of education was successful as well, a Chi-square analysis showed that level of education was equally divided among the conditions, χ2 (4) = 2.54, p = .637, fC = .18. As such, no demographic variables were
included as covariates in the main analyses. To check whether identification with being environmentally friendly and being healthy were randomly distributed, independent samples t-tests were performed with condition as independent variable and identifications as
dependent variables. Identification with being environmentally friendly was found to be equally distributed across the conditions; t(80) = -0.675, p = .501, CI = [-.55, .27], d = -0.15, as well as identification with being healthy; t(80) = -0.07, p = .948, CI = [-.51, .48], d = -0.01. Therefore, these variables were not included in the main analyses.
Partial Model: Main Analyses
The results of the 2 x 2 mixed repeated measures ANOVAs for the partial model can be found in Table 3. In order to answer the first hypothesis, a main effect of consumption was expected: consumption during the intervention is expected to be decreased compared to
consumption during the baseline. A significant main effect of consumption in frequency was found; F (1, 74) = 10.51, p = .002, ηp2 = .124. The frequency of meat consumption during the
intervention (M = 1.06, SD = 0.58) was significantly lower than the frequency of meat consumption during the baseline measure (M = 1.24, SD = 0.53). However, no significant main effect of consumption was found for quantity of meat consumption; F (1, 70) = 0.06, p = .814, ηp2 = .001. H1 is therefore partially confirmed.
The second hypothesis postulated that health information decreases meat consumption to a greater extent than environmental information, and therefore expects an interaction effect between framing (health vs. environment) and consumption. While no such an effect has been found for meat consumption in frequency, F (1, 74) = 1.10, p = .297, ηp2 = .015, the
interaction effect was significant for meat consumption in quantity, F (1, 70) = 4.89, p = .030, ηp2 = .065. Participants in the health condition consumed less meat in quantity after the
intervention (M = 55.94, SD = 36.43) than during the baseline (M = 61.78, SD = 41.32), while participants in the environment condition consumed more meat in quantity after the
intervention (M = 63.83, SD = 32.89) than during the baseline (M = 57.95, SD = 34.87). As such, H2 can be partially be confirmed.
In order to answer the third hypothesis, a three-way interaction effect between SE (self-enhancement) values, information provision (condition) and meat consumption: people with high levels of SE were expected to decrease their meat consumption in the health condition only, while people with high levels of ST (self-transcendence) were expected to decrease their meat consumption no matter in which condition they were in. There was no significant three-way interaction effect for consumption x framing x SE for consumption in frequency; F (1, 74) = 2.34, p = .130, ηp2 = .031; nor for consumption in quantity, F (1, 70) =
However, an unexpected significant three-way interaction effect was found for
consumption x framing x ST in meat consumption frequency; F (1, 74) = 4.24, p = .043, ηp2 =
.054. In Figure 2, it can be seen that the decrease in meat consumption in the health condition is stronger for participants with high levels of ST than for participants with low levels of ST. In the environment condition, there is a decrease of meat consumption in both individuals with high and low levels of ST, but it is not stronger for either one of these groups.
0,85 0,95 1,05 1,15 1,25 1,35 1,45 Baseline Week 1
Low ST levels High ST levels
Baseline Week 1
Health Condition Environment Condition
Me at C on su m p ti on (fr eq ue nc y)
Table 3
Mixed repeated measures ANOVAs for the partial model with meat consumption frequency and quantity as within-subjects factors and SE and ST as between-subjects factors
Source Meat consumption frequency Meat consumption quantity df F p ηp2 df F p ηp2 Between-subjects effects Framing 1 0.04 .838 .001 1 0.30 .586 .004 SE 1 0.86 .357 .011 1 0.19 .661 .003 ST 1 0.14 .708 .002 1 0.28 .600 .004 Framing x SE 1 2.69 .105 .035 1 0.27 .605 .004 Framing x ST 1 0.73 .396 .010 1 0.04 .836 .001 Framing x SE x ST 1 0.29 .620 .003 1 1.45 .242 .040 Error 74 70 Within-subjects effects Consumption 1 10.51 .002 .124 1 0.06 .814 .001 Consumption x Framing 1 1.10 .297 .015 1 4.89 .030 .065 Consumption x SE 1 0.08 .779 .001 1 0.45 .503 .006 Consumption x ST 1 1.98 .163 .026 1 0.90 .345 .013 Consumption x Framing x SE 1 2.34 .130 .031 1 0.40 .528 .006 Consumption x Framing x ST 1 4.24 .043 .054 1 0.72 .399 .010 Consumption x SE x ST 1 1.30 .257 .017 1 0.36 .549 .005 Consumption x Framing x SE x ST 1 0.07 .788 .001 1 1.27 .265 .018 Error (Consumption) 74 70
Note. df = degrees of freedom. The degrees of freedom differ per column as less participants completed questions about their meat consumption quantity. Text in bold is significant at p < .005.
Full model: Preliminary Analyses
Next, analyses will be done with the full model (baseline, week 1 and week 2). Because meat consumption is measured as a within-subjects variable and information provision is also treated as a within-subjects variable, there is no need to control for
demographic variables. However, correlations of identification with being environmentally friendly/healthy with meat consumption (in frequency and quantity) have also been checked for. Due to non-significant results, they were not included in the main analyses (see Table 4).
Table 4
Pearson Correlation Matrix Among Possible Covariates and DV's, all correlations p > .050 Identification
with
environment
Identification with health
Identification with environment .08
Identification with health .08
Meat frequency - Environmental week -.19 .02
Meat frequency - Health week -.24 .13
Meat quantity - Environmental week -.00 -.13
Meat quantity - Health week .05 -.77
Full Model: Main Analyses
The results of the 3 x 2 mixed repeated measures ANOVAs for the full model can be found in Table 5. Due to the within-subjects treatment of the information provision, some effects need to be interpreted differently.
For the first and second hypothesis, a main effect of consumption is expected: consumption during the intervention is expected to be decreased compared to the baseline. With pairwise comparisons within this main effect of consumption, it can be analyzed whether the health intervention decreased meat consumption to a greater extent than the environment intervention. A main effect of consumption in frequency was found; F (2, 86) = 4.66, p = .012, ηp2 = .098. Pairwise comparisons show that participants decreased their meat
consumption in frequency during the health intervention (M = 1.13, SD = 0.57), compared to the baseline meat consumption (M = 1.35, SD = 0.52, p = .017). However, no main effect of consumption in quantity was found; F (2, 80) = 0.13, p = .882, ηp2 = .003. As such, both H1
and H2 are partially confirmed.
The third hypothesis expected that people with high levels of SE decreased their meat consumption in the health condition only, while people with high levels of ST were expected to decrease their meat consumption in both conditions, thereby expecting an interaction effect between consumption and SE values. No such an effect was found for meat consumption in frequency; F (2, 86) = 0.12, p = .884, ηp2 = .003 nor for quantity; F (2, 80) = 1.09, p = .343,
Unexpectedly, an interaction effect between consumption in quantity and ST was found; F (2, 80) = 4.75, p = .011, ηp2 = .106. For participants with high levels of ST, meat
consumption is increased during the health intervention week (M = 70.77, SD = 57.22) compared to the baseline (M = 57.57, SD = 41.07) or environment week (M = 57.72, SD = 38.74) while participants with low levels of ST consumed comparable amounts of meat during the health week (M = 52.08, SD = 23.49), environment week (M = 56.96, SD = 25.56), and baseline (M = 59.40, SD = 34.53)4.
Table 5
Mixed repeated measures ANOVAs for the full model with meat consumption frequency and quantity as within-subjects factors and SE and ST as between-subjects factors
Source Meat consumption frequency Meat consumption quantity df F p ηp2 df F p ηp2 Between-subjects effects Order 1 0.21 .651 .005 1 1.59 .215 .038 SE 1 0.03 .856 .001 1 0.17 .681 .004 ST 1 3.84 .057 .082 1 0.12 .733 .003 Order x SE 1 0.01 .911 .000 1 2.14 .141 .051 Order x ST 1 2.89 .096 .063 1 0.11 .740 .003 Order x SE x ST 1 0.15 .699 .004 1 0.78 .384 .019 Error 43 40 Within-subjects effects Consumption 2 4.66 .012 .098 2 0.13 .882 .003 Consumption x Order 2 0.56 .573 .013 2 1.11 .335 .027 Consumption x SE 2 0.12 .884 .003 2 1.09 .343 .026 Consumption x ST 2 1.01 .370 .023 2 4.75 .011 .106 Consumption x Order x SE 2 0.20 .817 .005 2 0.60 .550 .015 Consumption x Order x ST 2 1.12 .330 .025 2 0.43 .651 .011 Consumption x SE x ST 2 0.46 .632 .011 2 1.04 .359 .025 Consumption x Order x SE x ST 2 0.54 .584 .012 2 2.02 .140 .048 Error (Consumption) 86 80
Note. df = degrees of freedom. The degrees of freedom differ per column as less participants completed questions about their meat consumption quantity. Text in bold is significant at p < .005. Order refers to whether participants first saw health or environmental messages and is included to control for order bias.
Conclusion and Discussion
In an attempt to stimulate people to consume less meat, this study investigated to which extent the effects of information provision about health and environmental
consequences on consuming meat were moderated by pre-existing values. It was expected that health information stimulated both people with high levels of ST and high levels of SE to consume less meat, while environmental information only stimulated people with high levels of ST to consume less meat. Even though this study found no support of a moderating role of pre-existing values on the relationship between information provision and meat consumption, some interesting results were found and can therefore contribute to the current body of research in this field.
Findings and Implications
The first important finding was found for both the partial model (baseline and week 1) and the full model (baseline, week 1 and week 2) and showed that providing information in general significantly lowered the frequency of meat consumption. These findings supported Hypothesis 1 and were also in line with literature which states that the provision of
information can be effectively deployed in order to decrease the intention to consume meat (Carfora et al., 2017a, 2017b; Cordts et al., 2014; Graham & Abrahamse, 2017; Loy et al., 2016). A such, this finding adds to the existing literature that using information provision can be effective in reducing the frequency of meat consuming behavior.
Furthermore, when comparing within-subjects, it appeared that health information was more effective in reducing the frequency of meat consumption than providing environmental information, compared to the baseline frequency of meat consumption. Alongside, the results of the partial model showed that health information was also more successful in lowering the frequency of meat consumption than environmental information. The combined findings of frequency and quantity meat consumption shows that while health and environmental
information both seem to be successful in lowering the frequency of meat consumption, only health information seems to lower the quantity of meat consumption. This supports
Hypothesis 2 and is also in line with literature that suggests that health information is more effective in instigating sustainable behavior (Yadav, 2016) and that health information would be more effective in reducing meat consumption (Lea et al., 2006; Tobler et al., 2011). This finding is relevant in practice as well, as future campaigns to decrease meat consumption should be focused on health information in order to both decrease the frequency and quantity of meat in people’s diets. It also shows that the common practice to frame sustainable issues in a way that is more congruent to self-transcendence values (De Dominicis et al., 2017; Schultz & Zelezny, 2003; von Borgstede et al., 2014) is not beneficial to instigating sustainable behavior and that the use of a self-enhancing frame
Moreover, some unexpected findings were related to self-transcendence values. For the partial model, participants with high levels of ST more strongly decreased their frequency of meat consumption in the health information condition as compared to the environmental information condition, while for participants with low levels of ST, no such effect was found. This is partly in line with literature, as it is previously shown that people with
self-transcendence values are not significantly more persuaded by value-congruent information (Graham & Abrahamse, 2017; von Borgstede et al., 2014). However, the current study showed that people with high levels of ST were even more persuaded by a self-enhancing frame, thereby countering literature that posit that people with self-transcendence values are equally persuaded by either self-enhancing or self-transcendence frames (De Dominicis et al., 2017; Schultz & Zelezny, 2003). One explanation for this result could be a difference in baseline frequency of meat consumption. Figure 2 shows that in the health condition,
with low levels of ST, while in the environmental condition, the participants with low levels of ST have a higher baseline meat consumption.
A last unexpected finding is an interaction effect for meat consumption in quantity and ST values. It indicated that participants consumed more meat during the health intervention compared to the baseline and environment intervention, while participants with low ST values ate comparable amounts of meat. This shows that while participants with high levels of ST consumed meat on a less frequent basis, they increased the amount of meat in quantity. The results are similar to the findings of Dagevos et al. (2019), which showed that people in the Netherlands had lowered their frequency of meat consumption, but increased their meat consumption in quantity in 2018. While the study of Dagevos et al. (2019) does not have a conclusive answer to this phenomenon, one likely explanation can be ‘self-licensing’. Self-licensing refers to a concept which postulates that people are rewarding themselves after doing something positive (Khan & Dhar, 2006). Translated to the current study, people might increase their meat consumption in quantity because they are rewarding themselves for eating meat less frequently.
Limitations & Future Research
One major limitation is related to the use of the MyPanel app. While it was a new and innovative way for tracking participants’ consumption, it was not yet fully developed which resulted in several issues. However, these could not be resolved by the researcher or by the company itself. Not all participants could get access to the project in the app or did not
receive push notifications and thus were not exposed to all surveys and intervention messages. This issue probably is the reason for the high attrition rate; a great number of participants who had initially started with the experiment did not finish the experiment and could therefore not be included in the analyses. Even though the analyses have been done twice with different samples, the number of participants in each sample was not high. As such, the small samples
might have impeded statistical significance of the results. Future studies could investigate in a more reliable way to track individuals’ meat consumption, for example by exploring
specialized food tracking apps or reverting to sending out multiple surveys through Qualtrics or similar programs.
Another limitation is related to the stimulation material. While the pre-test analyzed whether messages were more related to health or environmental consequences of (not) eating meat, it had not been analyzed whether the messages appealed more to the self (i.e., self-enhancement) or more to others (i.e., self-transcendence). Therefore, it cannot be certain that the health information was perceived as more self-enhancing or that the environmental information was perceived as more self-transcending and could therefore be a possible explanation for why no moderating effect of pre-existing values has been found. In future studies, this should also be pre-tested before using them in an experiment which wishes to compare those values.
A last limitation concerns the measurement of pre-existing values. This study did not take into account a continuum from self-enhancement to self-transcendence due to the fact that the PVQ produces separate variables for self-enhancement and self-transcendence. Therefore, it was chosen to dichotomize the variables for pre-existing values in this study. However, this equalizes all participants that fall under or below the median split so that people on the extreme end of the scale are rated equally to people who are closer to the median. This results in loss on information on those individuals (MacCallum, Zhang,
Preacher, & Rucker, 2002). Next to that, it also results in a loss of power, making it harder to find an actual effect (Aiken, West, & Reno, 1991). Both of these results are undesirable and future studies should find a way to have a continuous predictor from enhancement to self-transcendence.
Conclusion
Concluding, the results of this study show that pre-existing values do not play a role in the effectiveness of information provision on decreasing meat consumption. Rather, the study indicated that providing information about the consequences of (not) eating meat is sufficient enough to lower the frequency of meat consumption on its own. Additionally, using health information is effective in both lowering the frequency and quantity of meat consumption and should therefore be a starting point for messages that are designed to persuade people into decreasing their meat consumption. However, the current study found possible self-licensing effects as well, and a suggestion for this is to make people aware of this (unconscious) self-licensing effect by including it in the messages, thereby possibly increasing the effectiveness of the messages.
References
Aertsens, J., Verbeke, W., Mondelaers, K., & van Huylenbroeck, G. (2009). Personal determinants of organic food consumption: A review. British Food Journal, 111(10), 1140–1167. https://doi.org/10.1108/00070700910992961
Aiken, L. S., West, S. G., & Reno, R. R. (1991). Multiple regression: Testing and interpreting interactions. Sage.
Allen, M. W., & Ng, S. H. (2003). Human values, utilitarian benefits and identification: The case of meat. European Journal of Social Psychology, 33(1), 37–56.
https://doi.org/10.1002/ejsp.128
Aschemann-Witzel, J. (2015). Consumer perception and trends about health and
sustainability: trade-offs and synergies of two pivotal issues. Current Opinion in Food Science, 3, 6–10. https://doi.org/10.1016/J.COFS.2014.08.002
Bamberg, S., & Möser, G. (2007). Twenty years after Hines, Hungerford, and Tomera: A new meta-analysis of psycho-social determinants of pro-environmental behaviour. Journal of
Environmental Psychology, 27(1), 14–25. https://doi.org/10.1016/J.JENVP.2006.12.002 Barr, S., & Gilg, A. W. (2007). A conceptual framework for understanding and analyzing
attitudes towards environmental behaviour. Geografiska Annaler: Series B, Human Geography, 89(4), 361–379. https://doi.org/10.1111/j.1468-0467.2007.00266.x Beardsworth, A., & Keil, T. (1991). Health-related beliefs and dietary practices among
vegetarians and vegans: a qualitative study. Health Education Journal, 50(1), 38–42. https://doi.org/10.1177/001789699105000111
Bolger, N., & Laurenceau, J.-P. (2013). Intensive longitudinal methods: An introduction to diary and experience sampling research. Guilford Press.
Bouvard, V., Loomis, D., Guyton, K. Z., Grosse, Y., Ghissassi, F. El, Benbrahim-Tallaa, L., … Straif, K. (2015). Carcinogenicity of consumption of red and processed meat. The Lancet Oncology, 16(16), 1599–1600. https://doi.org/10.1016/S1470-2045(15)00444-1 Carfora, V., Caso, D., & Conner, M. (2017a). Correlational study and randomised controlled
trial for understanding and changing red meat consumption: The role of eating identities. Social Science & Medicine, 175, 244–252.
https://doi.org/10.1016/J.SOCSCIMED.2017.01.005
Carfora, V., Caso, D., & Conner, M. (2017b). Randomised controlled trial of a text messaging intervention for reducing processed meat consumption: The mediating roles of
anticipated regret and intention. Appetite, 117, 152–160. https://doi.org/10.1016/J.APPET.2017.06.025
Chong, D., & Druckman, J. N. (2007). Framing heory.
https://doi.org/10.1146/annurev.polisci.10.072805.103054
Cordts, A., Nitzko, S., & Spiller, A. (2014). Consumer response to negative information on meat consumption in Germany. International Food and Agribusiness Management Review, 17(SpecialIssueA), 83–106.
Dagevos, H., Verhoog, D., van Horne, P., & Hoste, R. (2019). Vleesconsumptie per hoofd van de bevolking in Nederland, 2005-2018.
Dawes, R. M. (1980). Social Dilemmas. Annual Review of Psychology, 31(1), 169–193. https://doi.org/10.1146/annurev.ps.31.020180.001125
de Boer, J., Schösler, H., & Aiking, H. (2017). Towards a reduced meat diet: Mindset and motivation of young vegetarians, low, medium and high meat-eaters. Appetite, 113, 387– 397. https://doi.org/10.1016/J.APPET.2017.03.007
De Dominicis, S., Schultz, P. W., & Bonaiuto, M. (2017). Protecting the environment for self-interested reasons: Altruism is not the only pathway to sustainability. Frontiers in Psychology, 8(JUN), 1–13. https://doi.org/10.3389/fpsyg.2017.01065
De Groot, J. I. M., & Steg, L. (2007). Value orientations and environmental beliefs in five countries: Validity of an instrument to measure egoistic, altruistic and biospheric value orientations. Journal of Cross-Cultural Psychology, 38(3), 318–332.
https://doi.org/10.1177/0022022107300278
Giménez, A. C., & Tamajón, L. G. (2019). Analysis of the third-order structuring of Shalom Schwartz’s theory of basic human values. Heliyon, 5(6), e01797.
https://doi.org/10.1016/J.HELIYON.2019.E01797
Godfray, H. C. J., Aveyard, P., Garnett, T., Hall, J. W., Key, T. J., Lorimer, J., … Jebb, S. A. (2018). Meat consumption, health, and the environment. Science (New York, N.Y.), 361(6399). https://doi.org/10.1126/science.aam5324
Gossard, M. H., & York, R. (2003). Social structural influences on meat consumption. Human Ecology Review, 10(1), 1–9.
Graham, T., & Abrahamse, W. (2017). Communicating the climate impacts of meat
consumption: The effect of values and message framing. Global Environmental Change, 44, 98–108. https://doi.org/10.1016/J.GLOENVCHA.2017.03.004
Harguess, J. M., Crespo, N. C., & Hong, M. Y. (2019). Strategies to reduce meat
consumption: A systematic literature review of experimental studies. Appetite, 104478. https://doi.org/10.1016/J.APPET.2019.104478
Hawkins, R. P., Kreuter, M., Resnicow, K., Fishbein, M., & Dijkstra, A. (2008).
Understanding tailoring in communicating about health. Health Education Research, 23(3), 454–466. https://doi.org/10.1093/her/cyn004
Hoek, A. C., Pearson, D., James, S. W., Lawrence, M. A., & Friel, S. (2017). Shrinking the food-print: A qualitative study into consumer perceptions, experiences and attitudes towards healthy and environmentally friendly food behaviours. Appetite, 108, 117–131. https://doi.org/10.1016/J.APPET.2016.09.030
Kalof, L., Dietz, T., Stern, P. C., & Guagnano, G. A. (1999). Social psychological and structural influences on vegetarian beliefs. Rural Sociology, 64(3), 500–511. Kareklas, I., Carlson, J. R., & Muehling, D. D. (2014). “I eat organic for my benefit and
yours”: Egoistic and altruistic considerations for purchasing organic food and their implications for advertising strategists. Journal of Advertising, 43(1), 18–32. https://doi.org/10.1080/00913367.2013.799450
Karp, D. G. (1996). Values and their effect on pro-environmental behavior. Environment and Behavior, 28(1), 111–133.
Khan, U., & Dhar, R. (2006). Licensing Effect in Consumer Choice. Journal of Marketing Research, 43(2), 259–266.
Lea, E. J., Crawford, D., & Worsley, A. (2006). Consumers’ readiness to eat a plant-based diet. European Journal of Clinical Nutrition, 60(3), 342–351.
https://doi.org/10.1038/sj.ejcn.1602320
Lentz, G., Connelly, S., Mirosa, M., & Jowett, T. (2018). Gauging attitudes and behaviours: Meat consumption and potential reduction. Appetite, 127, 230–241.
https://doi.org/10.1016/J.APPET.2018.04.015
Loy, L. S., Wieber, F., Gollwitzer, P. M., & Oettingen, G. (2016). Supporting sustainable food consumption: Mental contrasting with implementation intentions (MCII) aligns intentions and behavior. Frontiers in Psychology, 7(APR), 1–12.
https://doi.org/10.3389/fpsyg.2016.00607
MacCallum, R. C., Zhang, S., Preacher, K. J., & Rucker, D. D. (2002). On the practice of dichotomization of quantitative variables. Psychological Methods, 7(1), 19–40. https://doi.org/10.1037/1082-989X.7.1.19
Macdiarmid, J. I., Douglas, F., & Campbell, J. (2016). Eating like there’s no tomorrow: Public awareness of the environmental impact of food and reluctance to eat less meat as part of a sustainable diet. Appetite, 96, 487–493.
https://doi.org/10.1016/j.appet.2015.10.011
Magnusson, M. K., Arvola, A., Hursti, U.-K. K., Åberg, L., & Sjödén, P.-O. (2003). Choice of organic foods is related to perceived consequences for human health and to
environmentally friendly behaviour. Appetite, 40(2), 109–117. https://doi.org/10.1016/S0195-6663(03)00002-3
Moser, S. C., & Dilling, L. (2009). Opportunities and challenges for improving
communication. Communicating Climate Change: Closing the Science Action Gap, 163– 174. Retrieved from
http://sciencepolicy.colorado.edu/admin/publication_files/2011.30.pdf
Mullee, A., Vermeire, L., Vanaelst, B., Mullie, P., Deriemaeker, P., Leenaert, T., …
Huybrechts, I. (2017). Vegetarianism and meat consumption: A comparison of attitudes and beliefs between vegetarian, semi-vegetarian, and omnivorous subjects in Belgium. Appetite, 114, 299–305. https://doi.org/10.1016/J.APPET.2017.03.052
acceptability towards environmental policy measures: Value-matching appeals. Environmental Science & Policy, 61, 176–184.
https://doi.org/10.1016/J.ENVSCI.2016.04.013
Nordlund, A. M., & Garvill, J. (2002). Value structures behind proenvironmental behavior. Environment and Behavior, 34(6), 740–756. https://doi.org/10.1177/001391602237244 Pohjolainen, P., Tapio, P., Vinnari, M., Jokinen, P., & Räsänen, P. (2016). Consumer
consciousness on meat and the environment — Exploring differences. Appetite, 101, 37– 45. https://doi.org/10.1016/J.APPET.2016.02.012
Rohrmann, S., Overvad, K., Bas Bueno-de-Mesquita, H., Jakobsen, M. U., Egeberg, R., Tjønneland, A., … Linseisen, J. (2013). Meat Consumption and Mortality - results from the European Prospective Investigation into Cancer and Nutrition. BMC Medicine, 17(11). https://doi.org/10.1186/1741-7015-11-63
Santos, L., & Booth, D. (1995). Influences on meat avoidance among students. Appetite, 24(3), 268. https://doi.org/10.1006/appe.1996.0046
Schultz, P. W. (2002). Knowledge, information, and household recycling: Examining the knowledge-deficit model of behavior change. New Tools for Environmental Protection: Education, Information, and Voluntary Measures, 67–82.
Schultz, P. W., Gouveia, V. V., Cameron, L. D., Tankha, G., Schmuck, P., & Franěk, M. (2005). Values and their relationship to environmental concern and conservation behavior. Journal of Cross-Cultural Psychology, 36(4), 457–475.
https://doi.org/10.1177/0022022105275962
Schultz, P. W., & Zelezny, L. (1999). Values as predictors of environmental attitudes: Evidence for consistency across 14 countries. Journal of Environmental Psychology, 19(3), 255–265.
with American Values. Human Ecology Review, 10(2), 126–136.
Schultz, P. W., & Zelezny, L. C. (1998). Values and proenvironmental behavior: A five-country survey. Journal of Cross-Cultural Psychology, 29(4), 540–558.
Schwartz, S. H. (1994a). Are There Universal Aspects in the Structure and Contents of Human Values? Journal of Social Issues, 50(4), 19–45. https://doi.org/10.1111/j.1540-4560.1994.tb01196.x
Schwartz, S. H. (1994b). Beyond individualism/collectivism: New cultural dimensions of values.
Schwartz, S. H. (2012). Unit 2 Theoretical and Methodological Issues Subunit 1 Conceptual Issues in. Psychology and Culture Article, 11, 12–13. https://doi.org/10.9707/2307-0919.1116
Schwartz, S. H., Cieciuch, J., Vecchione, M., Davidov, E., Fischer, R., Beierlein, C., … Konty, M. (2012). Refining the theory of basic individual values. Journal of Personality and Social Psychology, Vol. 103, pp. 663–688. https://doi.org/10.1037/a0029393
Springmann, M., Clark, M., Mason-D’Croz, D., Wiebe, K., Bodirsky, B. L., Lassaletta, L., … Willett, W. (2018). Options for keeping the food system within environmental limits. Nature, 562(7728), 519–525. https://doi.org/10.1038/s41586-018-0594-0
Stern, P. C., Dietz, T., Abel, T., Guagnano, G. A., & Kalof, L. (1999). A value-belief-norm theory of support for social movements: The case of environmentalism. Human Ecology Review, 81–97.
Stoll-Kleemann, S., & Schmidt, U. J. (2017). Reducing meat consumption in developed and transition countries to counter climate change and biodiversity loss: a review of influence factors. Regional Environmental Change, 17(5), 1261–1277.
https://doi.org/10.1007/s10113-016-1057-5
to adopt ecological food consumption behaviors. Appetite, 57(3), 674–682. https://doi.org/10.1016/J.APPET.2011.08.010
van Dam, Y. K., & van Trijp, H. C. M. (2013). Relevant or determinant: Importance in certified sustainable food consumption. Food Quality and Preference, 30(2), 93–101. https://doi.org/10.1016/J.FOODQUAL.2013.05.001
Van der Werff, E., Steg, L., & Keizer, K. (2013). The value of environmental self-identity: The relationship between biospheric values, environmental self-identity and
environmental preferences, intentions and behaviour. Journal of Environmental Psychology. https://doi.org/10.1016/j.jenvp.2012.12.006
Verbeke, W. (2008). Impact of communication on consumers’ food choices. Proceedings of the Nutrition Society, 67(3), 281–288. https://doi.org/10.1017/S0029665108007179 Vergnaud, A. C., Norat, T., Romaguera, D., Mouw, T., May, A. M., Travier, N., … Peeters,
P. H. M. (2010). Meat consumption and prospective weight change in participants of the EPIC-PANACEA study. American Journal of Clinical Nutrition, 92(2), 398–407. https://doi.org/10.3945/ajcn.2009.28713
Verkooijen, K. T., Stok, F. M., & Mollen, S. (2015). The power of regression to the mean: A social norm study revisited. European Journal of Social Psychology, 45(4), 417–425. https://doi.org/10.1002/ejsp.2111
von Borgstede, C., Andersson, M., & Hansla, A. (2014). Value-Congruent Information Processing: The Role of Issue Involvement and Argument Strength. Basic and Applied Social Psychology, 36(6), 461–477. https://doi.org/10.1080/01973533.2014.958226 Wang, X., Lin, X., Ouyang, Y. Y., Liu, J., Zhao, G., Pan, A., & Hu, F. B. (2016). Red and
processed meat consumption and mortality: dose–response meta-analysis of prospective cohort studies. Public Health Nutrition, 19(5), 893–905. https://doi.org/DOI:
