Using dynamic social norms and temporal distance to change
intentions to reduce meat consumption: The influence of temporal
distance in framing normative messages.
July 17th, 2019
MSc Marketing Management Thesis
Using dynamic social norms and temporal distance to change intentions to
reduce meat consumption: The influence of temporal distance in framing
normative messages.
University of Groningen, Faculty of Economics and Business Department of Marketing
July 17th, 2019 Aron Huub Albers s2473887
Abstract
Climate change is a well-known problem all around the world, one of its main causes is the
production and consumption of meat. To increase pro-environmental behavior and in particular to reduce the amount of meat consumed research has been focused on using normative message framing. However, this research is mainly done by using static social norms, not much research has been done on the effect of dynamic social norms. This study aims to identify the effects of dynamic social norm messages and temporal distance on the intention to reduce meat consumption. The moderating effect of temporal distance on the effect of dynamic social norms on the intention to reduce meat consumption is also examined. The tests were conducted through the use of an online survey. The main effect of temporal distance on the intention to reduce meat consumption was found. In the short temporal distance condition the intention to reduce meat consumption was higher than in the long temporal distance condition. The interaction effect between temporal
Table of contents
Abstract
1. Introduction 6
2. Literature Review 7
2.1 Intentions to reduce meat consumption 7
2.2 Social Norms and Message framing on the intention to reduce meat consumption 8 2.2.1 Descriptive Norms versus Injunctive Norms and intentions to reduce meat consumption 8
2.2.2 Static Norms versus Trending and Dynamic Norms 11
2.3 Temporal distance 14
2.3.1 The effectiveness of using temporal distance to reduce the intention to eat meat 15 2.4 Interaction between normative message framing and temporal distance on intentions to reduce
meat consumption 17
2.5 Conceptual model 19
3. Methodology 20
3.1 Design and participants 20
3.2 Procedure and manipulations 21
3.3 Measures 23
3.4 Data Analysis 23
4. Results 24
4.1 Descriptive statistics 24
4.2 Reliability and validity of theoretical constructs. 26
4.3 The influence of dynamic social norms and temporal distance on intentions towards reducing
meat consumption. 27
4.3.1 Kruskal-Wallis H test 28
4.3.3 Contrast tests 29
1. Introduction
In the history of mankind meat has always been a scarce food supply (Latvala et al., 2012). Meat has always been valued highly due to being a source of high energy and protein (Fiddes, 1991; Twigg, 1983). Nowadays the meat consumption grows at a very fast rate and therefore so does the meat production in the world (Fiala, 2008). Due to the fast growth in meat production, meat become becomes cheaper and thus more accessible for more people (Latvala et al., 2012). Next to that, meat consumption is embedded in many cultures and it is still spreading more into other cultures (Latvala et al., 2012). All these shifts in the meat market make that meat consumption could rise with more than 70% from 2000 till 2030 Fiala (2008). Meat production could be doubled over the next 30 years Steinfeld et al. (2006). This trend is happening in two ways as both the average amount of meat consumed per capita and the total amount of meat consumed grows worldwide. That is more people start to eat meat and more people start to eat more meat (Charles et al., 2017). This growth is seen for all types of countries in different types of economic stages, it is a worldwide trend. To take an example of this, in Spain, China and Brazil the meat consumption has in the past 60 years more than doubled (Charles et al., 2017). This trend does not seem to stop anywhere soon as it is proven that meat consumption is a function of income and increases when income increases (Mathijs, 2015). As the income in highly populated countries like China increases the meat consumption will increase.
The current growth in meat production and consumption lays a heavy burden on the environment of the planet. The meat industry needs enormous amounts of water, land and energy (Elferink et al., 2007). Next to that the production and consumption creates a lot of pollution and emissions like water spillover, greenhouse gases and pollution of the biosphere. Overall the meat industry produces 18% of the greenhouse gases on a yearly basis (Pelletier, Tyedmers, 2010). As the meat industry is rapidly growing pollution of the environment will grow. The meat industry is partly responsible for the current climate change (Howard-Grenville et al., 2014).
A solution to reduce the impact is to switch to a diet in which less meat is consumed. Reducing meat consumption is a pro-environmental behavior, a pro-environmental behavior is defined as “behavior that harms the environment as little as possible, or even benefits the
environment” (Steg and Vlek, 2008, p.309). Reducing meat consumption would increase the demand for plants and reduce the demand for meat. Currently 32% of all grains produced worldwide are used to feed livestock, this could then be used to feed humans instead. Next to that the ground currently used for livestock can then also be used for growing plants (Pelletier, Tyedmers, 2010; Elferink et al., 2007).
can be divided in the descriptive norm, the behavior of significant others and the injunctive norm, what significant others think people ought to do (Deutsch and Gerard, 1955). Social norms can be made salient by the use of message framing.
However, this awareness does not seem to change behavior a lot as the meat consumption worldwide still grows. One of the possible explanations for the lack of change in the behavior of people is that they see climate change as a threat in the distant future. This distant future perception creates a psychological barrier called temporal distance. Temporal distance is a psychological
distance created by the temporal distance of a past or future event (Trope and Liberman, 2010). This distance gives people is this case the idea that they themselves will not be affected by the changing climate but that only future generations will. The way people perceive the messaging of climate change is influenced by temporal distance.
In this study message framing will be used to see what types of message could change the intention to reduce meat consumption. Dynamic social norms and temporal distance message framing in combination with climate change will be used to affect the intention to reduce meat consumption.
2. Literature review
Chapter two will present an overview of academic research in the field of intention, different types of social norms, the construal level theory, and, in more detail temporal distance and communication of climate change. Lastly, the conceptual model will be presented and explained.
2.1 Intentions to reduce meat consumption
Reducing meat consumption is an important behavior to focus on as meat production causes problems such as an overabundance of water, land and energy (Elferink et al.,2007). Furthermore, the meat production creates pollution and emissions like water spillover and 18% of the greenhouse gases.
sentences as “I intend to do X” or “I am planning to do X”. When looking at these definitions it can be concluded that intentions seem to consist of two elements. Intentions say something about the direction, thus whether to do something or to not do it and they explain the intensity, the level of time and effort, one is willing to take (Sheeran, 2002). These two elements together make an intention. Intention to will therefore in this study be used as the proxy of behavior. Therefore, the present study will focus on intention rather than behavior to reduce meat consumption
2.2 Social Norms and Message framing on the intention to reduce meat consumption
It is important to examine effective behavioral intervention strategies targeted at reducing one’s meat consumption. Message framing and more specifically normative message framing might be an effective way to encourage intentions to reduce meat consumption. Normative message framing makes use of social norms. Information about certain behavior or social acceptance can be shared through the use of social norms. In order to understand how social norms work they first need to be conceptualized. “A norm can refer either to what is commonly done that is, what is normal or what is commonly approved-that is, what is socially sanctioned.” (p202; Cialdini et al., 1991). A norm can further be described as a set of obtained representations which are then aggregated to form a common item which can then be a certain norm (Kahneman, Miller, 1986). Thus forming a certain group of representations into a group or class.
Social norms have shown that they are able to influence behavior by promoting a certain type of behavior (Neighbors, Larimer, and Lewis, 2004). Types of behavior that can be promoted include the promotion of healthier food choices (Robinson, Fleming, and Higgs, 2014) and a diversity of pro-environmental behaviors (Schultz, 1999).
There are different types of social norms. First, a distinction has been made between
descriptive and injunctive norms, (2.2.1), and, between trending, dynamic and static norms (2.2.2). In message framing/behavior change interventions these different types of norms have been used to encourage pro-environmental intentions in different ways, as explained below.
2.2.1 Descriptive Norms versus Injunctive Norms and intentions to reduce meat consumption
Sheeran, 2003) (Cialdini, Kallgren and Reno, 1991) (Cialdini, Kallgren and Reno, 1990). An example of such a descriptive norm in combination with meat consumption can be: “Most people I know are eating meat”.
A descriptive norm can be added to a message, which is more effective in changing intention and behavior than when a message without the descriptive norm is used (Cialdini, 2003). Thus a environmental message in combination with a descriptive message can increase the
pro-environmental behavior more than only a pro-pro-environmental message. For example, an experimental study among hotel guests showed that a descriptive normative message towards re-using towel use (“JOIN YOUR FELLOW GUESTS IN HELPING TO SAVE THE ENVIRONMENT BY RE-USING YOUR TOWEL ”, p474) was more effective to increase towel re-use than an environmental message only (“HELP SAVE THE ENVIRONMENT...”, p473; Goldstein, Cialdini and Griskevicius, 2008). Indeed, other research also supports that different types of pro-environmental behavior can be encouraged by using descriptive norms in messages, including reducing energy consumption (Nolan, Schultz, Cialdini, Goldstein and Griskevicius, 2008) and healthier and more sustainable food choices (Robinson, Fleming, and Higgs, 2014).
Although previous research has shown that making a descriptive norm salient by providing a normative message can be effective to encourage pro-environmental behavior, it has also been argued that there is a risk to use them in message framing. More specifically, using the descriptive norm to promote pro-environmental behavior can backfire in case the descriptive norm makes the “undesired”behavior (i.e., environmentally harmful behavior) salient. As in that case the majority of people are performing the “undesired” behavior (Cialdini, Reno, and Kallgren, 1990; Aitken et al., 1994; Oceja and Berenguer, 2009).
This backfire effect of descriptive norms has been shown especially relevant in relation to pro-environmental behavior because in many occasions the majority is performing the “undesired” behavior. For example, Cialdini, Reno, and Kallgren (1990), in which (1) littering was the descriptive norm (littered environment) and the other in which anti-littering was the descriptive norm (clean environment). They found that when the descriptive norm towards littering (i.e., the “undesired” behavior) was salient, people littered more than when the descriptive norm towards anti-littering (i.e., the “desired” behavior) was salient. Thus showing the “undesired” descriptive norm (littering) can actually promote this behavior.
Finally, a study in which was measured how many times people leave the light on in a room, when there is nobody left in that room. There was (1) in which the light was already on when entering the room was the descriptive norm (i.e., the “undesired” behavior) and the other in which the light was off when entering the room was the descriptive norm (i.e., the “desired” behavior). Oceja and Berenguer, (2009) found that when the descriptive norm towards the “undesired”
behavior was made salient people left the light on more often when they left the room, compared to when the descriptive norm towards the “desired” behavior was salient.
This backfire effect is relevant for this study. When writing a descriptive norm message about meat consumption the “undesired” behavior (i.e., eating meat instead of reducing meat
consumption) will be made salient as the majority of people still eat meat. Even though this backfire effect occurs, this does not mean that social norms cannot be used when the “undesired” behavior is the behavior of the majority. There are ways to counter the backfire effect, one of these ways involves the use of injunctive norms. What injunctive norms are, how they work, how they might solve the backfire effect and what their limitations are will be discussed in the coming paragraphs.
Injunctive norms present the norms of how people, according to societal standards, should behave in a certain situation (Deutsch and Gerard, 1955). Injunctive norms are formed by what significant others think people ought to do (Cialdini et al., 1990; Cialdini et al., 1991). People tend to follow these norms as they believe that acting accordingly will be rewarded and that acting
counternormative will be resulting in some sort of societal punishment (Burger et al., 2010). An example of such an injunctive norm is: “People, who are important to me, think that we should eat less meat”.
Injunctive norms can just as descriptive norms be used in message framing. An injunctive norm can be framed in a message and this message can influence intention or behavior. An injunctive norm can be used when the backfire effect will occur when using the descriptive norm. The
injunctive norm is, in the case of pro-environmental behavior, often already in favor of the “desired” behavior, even when the majority is performing the “undesired” behavior. For example, most people do not mind if people, or they do even think that it is a good idea to, eat less meat in order to reduce climate change. However when looking at the actual behavior people do not eat less meat, which is the descriptive norm. People thus know what the “desired” behavior is but they do not always act upon it.
norm (a man walked by and picked up a piece of litter) made the “desired” behavior salient. In the occasions that the “desired” behavior was made salient by the use of the injunctive norm, both in the littered and clean environment the littering was lower than in the control condition.
Another example is that of Cialdini et al., (2003), in which entrants of a petrified wood parc were shown a text about stealing wood from a park. Half of the entrants saw a text with an injunctive norm message, asking not to steal any wood (i.e., the “desired” behavior), the other half were shown a descriptive norm message, stating that many past visitors stole wood (i.e., the “undesired”
behavior). The results show that when the injunctive norm message made the “desired” behavior salient a lot less wood was stolen compared to when the descriptive norm message made the “undesired” behavior salient.
An injunctive norm does not always change behavior more towards the “desired” behavior compared to the descriptive norm. Some studies show that injunctive norms have either, an
insignificant effect compared to the descriptive norm or even an insignificant effect compared to the control group. Robinson et al. (2013) show in study two of their paper that when the descriptive norm message, about eating healthy food, was made salient it significantly increases eating healthy food compared to when the injunctive norm message was made salient and to the control condition.
In another study about energy conservation participants were exposed to two messages one message containing a descriptive norm and the other one an injunctive norm. The norms in the message could be both in favor, both against or one in favor and one against energy conservation. The results showed that only when the descriptive norm message made energy conservation salient (i.e., the “desired” behavior) then the injunctive norm message also had a positive influence. When the descriptive norm message made the disapproval of energy conservation (i.e., the “undesired” behavior) salient then the injunctive norm message did not have any influence on intentions. This is an example that an injunctive norm message cannot always have a positive influence when the descriptive norm message is promoting the “undesired” behavior (Smith et al., 2012).
The injunctive norm message about pro-environmental behavior cannot always counter the backfire effect of the descriptive norm message. To make sure the backfire effect will not occur in this study, other ways to counter it should be explored. Next to the use of the injunctive and descriptive norm messages, to manipulate intention and behavior, there are other types of norms that can change intention and behavior. To change the way that norms work they can be changed by making the norms either trending, dynamic or static. These types of norms will be discussed in section 2.2.2.
A static norm is a norm that tells something about the current state of the norm. Thus the way the norm is right now, it does not say anything about how the norm changes over time (Sparkman, and Walton, 2017). The previous research studies discussed in this paper, both were descriptive and injunctive norm messages where used, were all written in a static way. An example of a static descriptive norm is: “Most people who are important to me, eat meat”. An example of a static injunctive norm is: “Most people who are important to me, think I should eat meat”. As mentioned before a problem with a static descriptive norm message can be that the norm used is describing the “undesired” behavior, which then causes the backfire effect. One way to counter this backfire effect is by changing the norm from static into dynamic.
A norm that shows a change in behavior over time is a so called dynamic norm (Sparkman, and Walton, 2017). The opposite of this dynamic norm is the previously explained static norm. Sparkman and Walton (2017) show in their first experiment the way a dynamic norm can be used, all norms used were also a descriptive norm. The message in which the static norm was used stated: “Recent research has shown that 30% of Americans make an effort to limit their meat consumption. That means that 3 in 10 people eat less meat than they otherwise would.” (p1664; Sparkman and Walton, 2017). Their message in which the dynamic norm was used stated: “Recent research has shown that, in the last 5 years, 30% of Americans have now started to make an effort to limit their meat consumption. That means that, in recent years, 3 in 10 people have changed their behavior and begun to eat less meat than they otherwise would.” (p1664; Sparkman and Walton, 2017). The results show that participants who read the message in which the “desired” behavior (starting to eat less meat) was made salient by the use of a dynamic norm have significantly more interest in
reducing their meat consumption, compared to those who read the message that used the static norm, which showed that the “desired” behavior is in the minority.
Experiment 3 of Sparkman and Walton (2017) was a similar experiment to the first
The two experiments that were just explained both showed that the dynamic norm message can increasing intentions towards the “desired” behavior even when the “desired” behavior is messaged in a descriptive norm which is a minority. In the fifth experiment of Sparkman and Walton (2017), which was about water conservation, another pro-environmental type of behavior, they test whether the effect of framing a descriptive message in a dynamic way (residents change and start to save more water) also holds when the dynamic norm is the majority. The results show that when the message made the dynamic norm salient more residents saved water compared to when the static descriptive norm message (help save more water) was made salient and compared to the control group (no message). Dynamic norms thus seem to be able to solve the backfire effect of descriptive norms and they can also be used effectively when the descriptive norm is already in the majority.
Next to the static and dynamic norm there is the trending norm, a trending norm can be described as a norm in which the participation in a certain behavior is increasing. The trend can be messaged by showing both the current percentage of participation in combination with the increase in participation. Thus the change of a certain norm over time can be described as trending. This type of norm was introduced by Mortensen et al. (2018). When comparing a trending and a dynamic norm to each other, it appears to be that the trending norm is a form of the dynamic norm. Sparkman and Walton (2017) used a norm experiment 3 in which there is future growth, this is a dynamic norm. This norm is also in line with the description of a trending norm. However, the norm used in
experiment 1 by Sparkman and Walton (2017) is again a dynamic norm but it is not a trending norm. Therefore, from now on the term “dynamic” will also be used to describe the trending norm. As a trending norm is in my opinion a type of the dynamic norm.
Recent research by Mortensen et al. (2018) shows that changing the behavior of a group towards the “desired” behavior can be done by using dynamic norms. In their first experiment, they measured whether water saving different when making the dynamic descriptive norm message salient compared to the static descriptive norm and the control group. The control group read no message, the static norm group read: “Research from (previous year) has found that 48% of (University name) students engage in one or more of the following water conservation behaviors.” (Mortensen et al., 2018, p.202). In the dynamic norm condition there was one sentence added: “This has increased from 37% in (2 years previous).” (Mortensen et al., 2018, p.202). As the descriptive norm messaged is a minority norm it was expected and found that the static norm message caused the backfire effect, the participants started using more water after they read the message. The control group and the dynamic norm message group did not differentiate from each other in water usage. By making the dynamic descriptive norm salient the backfire effect was countered.
dynamic norm message was made salient the backfire effect was not only countered, the completion and donation rate were even higher compared to when the static descriptive norm was made salient and to the control group. The behavior in the control condition and the static norm condition did not differ from each other.
The research on dynamic norms by Sparkman and Walton (2017) and by Mortensen et al. (2018) are the first and thus far the only researches on dynamic norms (to my knowledge). The research seems promising as either the backfire effect is countered or the dynamic norm is more effective in changing behavior towards the “desired” behavior. This is the case when the descriptive norm is a minority and when the descriptive norm is a majority. Therefore, dynamic norms will be used in this study instead of static norms.
In this study the injunctive norm will also be a dynamic norm. An injunctive norm is often already in favor of the “desired” behavior, therefore no research has been done with dynamic injunctive norm messaging. It will be interesting to see whether the intention will still be influenced by the difference between a descriptive and injunctive norm when both norms are dynamic.
Nevertheless, it is still expected that making the descriptive norm salient will change intention more towards the “desired” behavior than when the injunctive norm is made salient. This is expected because when both norms are already the majority a situation in which the backfire effect is ruled out), making the descriptive norm message salient changes the behavior more towards the “desired” behavior than when the injunctive norm is made salient.
2.3 Temporal distance
for everybody. Each individual perceives these values different as they are bound to someone’s own opinion and past experiences. (Trope and Liberman, 2010; liberman, Trope and Stephan 2007; Spence et al., 2012; Liberman et al., 2002).
Intentions to reduce one’s meat consumption have been related to climate change (De Boer, Schösler and Boersema, 2013). Climate change can be perceived as distant for all the different psychological distances. The majority of the people seem to know that these effects will not only happen in spatially distant areas but also in areas in which they live. Most people believe that climate change is actually happening and doing something good for the environment is accepted by most people (Whitmarsh, 2011). The biggest psychological distance seems to be the temporal distance. As many big consequences of climate change are in the more distant future, people only see some small effects now, it is hard for them to imagine what the consequences will be for future generations (McDonald, Chai and Newell, 2015). Therefore, the present research will focus on this category of psychological distance specifically.
2.3.1 The effectiveness of using temporal distance to reduce the intention to eat meat
Temporal distance is a psychological distance created by the temporal distance of a past or future event. The way an event that either will happen or has happened, will be processed is processed is depending on the distance of the occurrence. This dimension brings up questions about what actions to take with respect to for example goals in the future, investing money, investing in a greener future and behaving pro-environmental. As mentioned before these goals and the way they are influenced by temporal distance are different for each and every individual (Liberman, Trope and Stephan, 2007).
The way people process information about temporal distance construal is not the same in every situation. An event in the distant future is better processed when the message is of a high-level construal, a more abstract message. However, when the event is happening in the near future a more concrete, low-level construal message will be processed better (Spence et al., 2012). For example in the case of a flooding, which causes an environmental disaster, the message for the near and far future should be different. For the near future the message should be something like
“agricultural disaster due to flooding of the Rijn”. In case of an event in the distant future the
goals should be messaged in a more concrete near future goal (Liberman et al., 2002; Carver and Scheier, 1990).
Due to the fact that the processing is different, messages over different temporal distance should be framed differently. This is also because people tend to feel more confident, become more risk-taking and are better in procrastinating gratification when an event is in the distant future (Gilovich et al., 1993). People thus change their feelings about an event purely based on the temporal distance even when the event would be the same. People tend to be more confident about positive outcomes in the more distant future. This makes people less willing to change when the event is in the far future. This effect is further strengthened by the fact that people perceive and measure future benefits and costs in an inconsistent manner. Future benefits seem to be weighted more than costs (Spence et al., 2012).
The emission of greenhouse gasses is a kind of risk-taking behavior as it is partly causes climate change (Pelletier, Tyedmers, 2010). On an individual level the effects of one person are close to none, only when the overall impact of for example a country is measured there is some effect noticeable (Moser, 2010). This causes the social dilemma theory, which is caused when people have to change their personally preferred behavior to a less preferred behavior for the good of society (Bock, Zmud, Kim and Lee, 2005).
An empirical example of this is the research done by Chandran and Menon (2004). They measure the moderating role of negative valence (the “badness” of a situation) versus positive valence (the “goodness” of a situation) on the effect that temporal distance has on the intention to act precautious. The negative valence message is about worsening health if precautious actions are not done, while the positive valence message is about enjoying health/preventing diseases by doing something precautious. In case of negative valence, the intention to act precautious is higher when short temporal distance is used compared to long temporal distance. For the positive valence it is the other way around. Contradicting is the result of Nan, Zhao, Yangl. (2014) as they did not find any evidence that message framing where either present or future risks of smoking were made salient changed anything in the intentions to smoke. Negative valence is used in this study (negative consequences of climate change due to meat consumption are messaged) it is expected that
intention to reduce meat consumption is higher in the short temporal distance condition than in the long temporal distance condition.
It is expected that in this study the short temporal distance message will change intention more towards reducing meat consumption than a long temporal distance message does. This is expected because distant future risks are perceived in a more positive way compared to risk in the near future, the valence of the message (which is negative in this study) will influence the
short temporal distance) and the message will be concrete which works better in the short temporal distance condition (compared to a long temporal distance condition).
2.4 Interaction between normative message framing and temporal distance on intentions to reduce meat consumption
No research has yet been done on the moderating role of temporal distance on the effectiveness of normative messaging. There could however very well be an interaction effect between temporal distance and normative messaging. Both normative and temporal message framing have an effect on behavior. First, as discussed in section 2.2 it is expected that the use of a descriptive norm message will change intention more towards reducing meat consumption than an injunctive norm message does. Secondly, as discussed in section 2.3 it is expected that the use of a short temporal norm message will change intention more towards reducing meat consumption than a long temporal norm message does, as a negative valenced and concrete message is used. The expected moderating effect of temporal distance on normative messages on the intention to reduce meat consumption will be discussed in this section.
First the short temporal distance condition. In this condition the difference in effect of different dynamic normative messages, that are about the current generation, on the intention to reduce meat consumption is measured. All the examples that show that the descriptive norm message changes behavior more towards the desired behavior compared to the injunctive norm message were in the short temporal distance span (Smith et al., 2012: Robinson et al., 2013 ;
Mortensen et al., 2018; Sparkman and Walton, 2017). As the temporal distance is short and the norm is dynamic the expected difference, that the dynamic descriptive norm message changes intentions more towards the “desired” behavior than the dynamic injunctive norm message, applies here.
The second condition is the long temporal distance condition. In this condition the difference in effect of different dynamic normative messages, that are about the future generations, on the intention to reduce meat consumption is measured. The expected effect is different from the short temporal distance condition. In the long temporal distance condition the dynamic normative
Cialdini et al., 1990; Cialdini et al., 1991). However now this behavior is not presentable. This makes the descriptive norm a thought of how people will or should behave just as the injunctive norm, which is always thought as it is the way people think one should behave and not actual behavior (Deutsch and Gerard, 1955; Cialdini et al., 1990; Cialdini et al., 1991). Because both norms are a thought the expected result is that the dynamic descriptive norm and the dynamic injunctive norm do not differ in their influence on the intention to reduce meat consumption in the long distance condition.
The expectation is that in the short temporal distance condition both the dynamic descriptive norm message and the dynamic injunctive norm message increases the intention to reduce meat consumption more than in the long temporal distance condition. (Gilovich et al., 1993; Spence et al., 2012; Chandran and Menon, 2004)The expected difference between the short temporal distance condition and the long temporal distance condition can partly be explained by the social dilemma theory. People need to change their behavior because it is good for society. However that does not mean that they want to change, they therefore face the social dilemma theory. If the social norm is that of the current generations than that could help people overcome the barriers to change. As it is their current society that creates the pressure to change and they also benefit from it themselves. However when the social norms presented are of future generations, then the societal pressure that causes the dilemma seem to fade away. As it is not the current society but the future society that demands the change (Joireman, 2005).
Graph 1.
2.5 Conceptual model
The research objective of this paper is to examine whether and how dynamic descriptive- and dynamic injunctive norms influence the intention to eat meat. The social norms will be about the influence of the meat industry on climate change and the current trend in the behavior of people on this topic. This effect will be moderated by using a present and future temporal distance. The hypotheses are visualized in the conceptual model below (Figure 1).
H1: A dynamic descriptive normative message will be more effective in changing the intention to reduce meat consumption than a dynamic injunctive normative message.
H2: A short temporal distance message will be more effective in changing the intention to reduce meat consumption than a long temporal distance message.
H3: Temporal distance will moderate the influence of dynamic normative messages on the intention to reduce meat consumption.
H3(a): When the message is framed as short temporal distance, the influence of dynamic descriptive normative messages will be more effective in increasing the intention to reduce meat consumption, than a long temporal distance dynamic descriptive normative message.
H3(c): When the message is framed as short temporal distance, the influence of dynamic descriptive normative messages will be more effective in increasing the intention to reduce meat consumption, than dynamic injunctive normative messages.
H3(d): When the message is framed as long temporal distance, the influence of dynamic descriptive normative messages in increasing the intention to reduce meat consumption will not be different than the influence of dynamic injunctive normative messages.
Figure 1, The Conceptual Model
3. Methodology
3.1 Design and participants
The research study in this paper will be conducted by using an online experiment, this online experiment will be done by the use of the website www.qualitrics.com. In this online experiment three effects will be examined. First, the main effect of a dynamic descriptive norm (compared to a dynamic injunctive norm), by the use of message framing, on the intention to reduce meat
The participants were contacted through the use of social media and by face-to-face contact. In total there were 215 participants. The study uses a 2 (descriptive norm vs injunctive norm) x 2 (short temporal distance vs long temporal distance) between subject design. With the independent variable being the dynamic descriptive or injunctive norm and the present or future temporal distance. The moderator for the two dynamic norms is the temporal distance variable. The intention to reduce meat consumption is the dependent variable.
3.2 Procedure and manipulations
The respondents will be randomly distributed across the conditions. The online survey will be available on both their smartphone and computer. Accessibility to the survey should thus not pose a problem. The full survey can be found in Appendix A. The survey starts with basic information about the survey. It states that the survey will take on average 7 minutes, that participants can drop out of the experiment at any moment and that participations is fully anonymous. Respondents have to click on “agree” to confirm that they will take the survey in full honesty and that they agree to the
mentioned terms. Participants were not told what the experiment is about as this could influence the results. Before starting the message framing, the respondents were asked for some demographic information. This was asked first so that the respondents did not immediately have to fully focus on the text.
After the introduction and the demographic questions the participants had to read the message about climate change. Before the participants had to answer the 15 questions of the survey, they first had to read a partly general message. The introduction of the text was the same for all the participants. First descriptions of climate and climate change were given to have no confusion on what these concepts mean. After these descriptions the current climate change in the Netherlands was explained and the consequences are mentioned. In this section there were 2 versions. The two versions of the texts were written in such a way that similar words and word counts were used. However, the keywords to describe when the consequences were felt, were different. Group 1 and 2 were shown a text in which the consequences of climate change are already felt by current
generations. The text: ”A large amount of greenhouse gasses is produced during the production of meat. The global warming will be worsened and be experienced by current generations, unless people start to change their behavior. Some of the consequences that will be experienced by current
generations are:” The term “current generations” is used to indicate the short temporal distance. In
“future generations” was used to indicate long temporal distance. The consequences mentioned were the same for all groups.
The text than continues with a dynamic norm message about eating meat. There are two versions of the text with both similar text and word count, again only the keywords were different. Group 1 (dynamic descriptive norm x short temporal distance) was presented with a text in which the keywords indicated that there is a trend in eating less meat in the Netherlands. This is the descriptive trend. The text: “The amount of Dutch inhabitants that chose to reduce or even stop eating meat, to stop the effects of climate change for current generations, has grown with 70%. Because of this, the total amount of kilograms of meat eaten in the Netherlands has shrunk by 10%, the shrinkage will continue.”. Group 2 (dynamic injunctive norm x short temporal distance) was presented with a text in which the keywords indicated that there is a trend in thinking there should be eaten less meat in the Netherlands. This is the injunctive trend. The text: “The amount of Dutch inhabitants that thinks
that eating meat should be reduced or even stopped, to stop the effects of climate change for current generations, has grown with 70%. Because of this, the Dutch inhabitants think that the total
amount of kilograms of meat eaten in the Netherlands should shrunk by 10%, the shrinkage should continue.”. Group 3 (dynamic descriptive norm x long temporal distance) was presented with a text in which the keywords indicated that there is a trend in eating less meat in the Netherlands. This is the descriptive trend. The text: “The amount of Dutch inhabitants that chose to reduce or even stop
eating meat, to stop the effects of climate change for future generations, will grow with 70%.
Because of this, the total amount of kilograms of meat eaten in the Netherlands will shrunk by 10%, the shrinkage will continue.”. Group 4 (dynamic injunctive norm x long temporal distance) was presented with a text in which the keywords indicated that there is a trend in thinking there should be eaten less meat in the Netherlands. This is the descriptive trend. The text: “The amount of Dutch inhabitants that thinks that eating meat should be reduced or even stopped, to stop the effects of climate change for future generations, will grow with 70%. Because of this, the Dutch inhabitants think that the total amount of kilograms of meat eaten in the Netherlands should shrunk by 10%, the shrinkage should continue.”.
The text is based in the text of Brügger, Morton and Dessai (2016), in their experiment they measure the effect of message framing by using spatial distance. The text was used as an indication, no text was directly used. The 70% growth is chosen to make sure that the majority of the people is in the dynamic norm, this is done to make sure that the back-fire effect could not occur.
After the messages were shown, the participants had to fill in questions, all questions are on a 7-point Likert scale with 1 being fully disagree and 7 being fully agree. They had to answer
injunctive norm of eating meat and the dynamic descriptive norm of eating meat, these questions were asked so that the manipulation check could be conducted. At the end of the survey participants were thanked for participating and there was a textbox entry for comments.
3.3 Measures
To obtain the required information the survey conducted several measures see Appendix A. Intention to reduce meat consumption. To measure intention to reduce meat consumption, respondents were asked to fill in 3 items on whether they want to eat less meat. This was measured using a 7-point Likert scale, with 1 being totally disagree and 7 being totally agree based on Ajzen (1991).
Dynamic descriptive norm. 3 items were asked about the current trend in reducing meat consumption. This was done on a 7-point Likert scale with 1 being totally disagree and 7 being totally agree based on Nicolaij and Hendrickx (2003).
Dynamic injunctive norm. 3 items were asked about the current trend in the opinion of people in reducing meat consumption. This was done on a 7-point Likert scale with 1 being totally disagree and 7 being totally agree based on Nicolaij and Hendrickx (2003).
Temporal distance. Temporal distance was measured using a 3-item 1 to 7 Likert scale. In which respondents are asked whether they think that climate change is already happening. With 1 being totally disagree and 7 being totally agree based on Brügger et al. (2016) and Spence et al. (2012).
Besides the dependent variable, the manipulation variables some other variables are measured as well. In order to get a better overview of the sample.
Gender. Gender is measured, there were three options male, female and other.
Age. Age is measured, the participants could fill in their age in years in an open entry box, it was measured on an interval scale.
Highest received educational level. The demographic variable educational level is included. Participants had the following options primary school, secondary school, MBO, HBO University Bachelor, University Master and University PhD.
3.4 Data Analysis
To check these manipulations a factor and reliability analysis was done. All variables were measured by the use of 3 questions each on a 1 to 7 Likert scale. Three different factors were made with all a cronbach’s alpha above 0.6 (Tavakol and Dennick, 2011), the factors were based on a rotated VARIMAX matrix see Appendix D. The dynamic injunctive norm variable was made out 3 items, however the dynamic descriptive norm variable and temporal distance variable were both made out of 2 of the 3 questions, as dropping 1 question would increase the cronbach’s alpha (Appendix F).
Checking whether the manipulations hold is done by performing an one-way ANOVA. The temporal distance item (M = 5.1790, SD =1.33119) did not significantly differ in the short temporal distance condition compared to the long temporal distance, F (1,175) = 1.394, p = 0.239. The dynamic descriptive item (M = 4.4019, SD = 1.53538) did not significantly differ depending on whether the descriptive or injunctive norm was used, F (1,175) = 0.675, p = 0.413. Lastly, the dynamic injunctive item (M = 5.0587, SD = 1.12505) did not significantly differ depending on whether the descriptive or injunctive norm was used, F (1,175) = 0.067, p = 0.796.
4. Results
4.1 Descriptive statistics
The total amount of respondents in the online experiments is 215. 186 of these respondents
completed the survey. 10 more were excluded as they had filled in an age that is not possible (>150). Thus the total amount of the sample is 176. Most of the items were pre-set in a 7-point Likert scale.
As presented in Table 1, descriptive statistics are as follows. 56.3% of the respondents is male, 42.6% of the respondents is female and 2 respondents, 1.1%, filled in other. The average age is 27.85, with a standard deviation of 11.516. The population included a range of educational levels, of which 78.4% had HBO (applied university) or University, the highest levels of education in the
Table 1. Sample characteristics (N=176).
Variable Count Percentage
Gender # % Male 99 56.3 Female 75 42.6 Other 2 1.1 Age # % 18-20 19 10.8 21-25 99 56.3 26-30 30 17.0 31-35 5 2.8 36> 23 13.1
Highest received educational level # %
4.2 Reliability and validity of theoretical constructs.
The online experiment consisted of different items. These questions were set-up in a way that they should belong to 5 different factors. In order to see whether these items belong to the right factors a factor analysis has to be conducted. There are 3 items per factor, there are 5 factors, temporal distance, dynamic injunctive norm, dynamic descriptive norm, intention to reduce meat consumption and uncertainty about climate change.
A KMO and Bartlett’s test has been done. The KMO score should be higher than 0,6, when the score is higher than 0,6 than the sample is suitable for factor analysis (Kaiser, 1974) The KMO of the sample is 0.833. The Bartlett’s test measures whether the variables are unrelated. The value should be below 0.05, this would indicate that the sample is suitable for factor analysis (Kaiser, 1974). The value is below 0.001 thus this criterion also holds. (Appendix B) The communalities measure in percentage the amount of variance of a variable that is explained by all the extracted factors. This should be at least 0.4. All the values are above 0.4 with the lowest being 0.671 (Appendix C). The results showed that the variables for temporal distance and uncertainty about climate change almost measured the same. To make sure that the variables about the uncertainty about climate change will not intervene in the research, they are dropped from the data. There thus should be a total amount of 4 factors.
To choose the amount of factors there are different measures that can be taken into account. The first possible measure is only taking factors with an eigenvalue higher than 1.0. This would lead to 3 factors. Another criterion could be the total amount of variance explained, this should be above 60%, this is met after creating 3 factors. A third measure is all factors which explain at least 5% of the variance each, this is met for the first 6 factors. Lastly, looking at the “elbow” of the scree plot would indicate 2 or 4 factors.
4.3 The influence of dynamic social norms and temporal distance on intentions towards reducing meat consumption.
To test the influence/effect of dynamic social norms and temporal distance on the intention to reduce meat consumption, a two-way ANOVA was conducted. Six assumptions were checked before conducting the ANOVA (Cardinal and Aitken, 2006). The first assumption states that the dependent variable should be measured on a continuous level, which was met: that is, intention to reduce meat consumption was measured on a 7-point Likert-scale. Assumption two states that there are two independent variables which both consist of two or more categorical, independent groups, which was met (i.e., dynamic social norms: descriptive or injunctive experimental condition; temporal distance: future or present experimental condition). Assumption two is also met. Assumption three is also met, as this assumption states that there should be independence of observations.
Independence of observations means that there is no relationship between the observations neither between groups nor within the groups. The fourth assumption is met when there are no significant outliers, as outliers (impossible age) are deleted this assumption holds. The fifth assumption is that the dependent variable should be normally distributed. This was tested by using a Shapiro-Wilk test. This assumption was violated for the dependent variable (Appendix G), meaning that the data was not normally distributed. The last assumption is that of homogeneity of variance for all groups. This is measured by the use of a Levene test, this was not significant for all cases (0.839, 0.938 and 0.879). There is homogeneity of variance assessed by the use of the Levene test. Therefore, five of the six assumptions have been met. Even though the dependent variable is not normally distributed the ANOVA tests could still be used as it is fairly robust to deviations of normality (Maxwell and Delaney, 2004).
To test whether the three hypotheses hold two-way ANOVA tests were conducted. For testing H1: “A dynamic descriptive normative message will be more effective in changing the intention to reduce meat consumption than an injunctive normative message.” Respondents were put in two different groups: dynamic descriptive norm (n = 92) and dynamic injunctive norm (n = 84). The results are not as hypothesized. Participants in the dynamic descriptive norm condition (M = 4.3370, SD = 1.90077 ) did not show more intention in reducing meat consumption than the
participants in the dynamic injunctive norm condition (M = 4.3294, SD = 1.84652), F(1, 175) = 0.001, p = 0.979.
consumption in the short temporal distance condition (M = 4.6835, SD = 1.8412) compared to the long temporal condition (M = 4.0481, SD = 1.8534), F(1,175) = 5.148, p = 0.025.
For testing H3: “Temporal distance will moderate the influence of dynamic normative messages on the intention to reduce meat consumption.” Respondents were put in four different groups: dynamic descriptive norm*short temporal (M = 4.4884, SD = 1.92361, n = 43), dynamic injunctive norm*short temporal (M = 4.9167 , SD = 1.73548, n = 36), dynamic descriptive norm*long temporal (M = 4.2041 , SD = 1.89030, n= 49) and dynamic injunctive norm*long temporal (M = 3.8889, SD = 1.82099, n = 48). The hypothesized interaction effects were not significant, F(1,172) = 2.299, p = 0.079, η² = 0.039. However the p-value of 0.079 shows borderline significance, therefore the interaction effect will be further examined.
Table 2. Main and Interaction Effects between Dynamic Social Norms and Temporal Distance on Intentions to Reduce Meat Consumption
Results two-way ANOVA
df F p η²
Intention to reduce meat consumption
Dynamic norm 2 0,001 0,979 -
Dynamic norm*temporal distance 3 2,299 0,079 0,039
Temporal distance 2 5.148 0.025
Table 2
4.3.1 Kruskal-Wallis H test
Because assumptions four was violated a double check of the results of the ANOVA tests will be done. This double check will be done by using a Kruskal-Wallis H test (McKight, Najab, 2010). There are four assumptions that should hold in order to do a Kruskal Wallis H test. The first three
For testing H1: “A dynamic descriptive normative message will be more effective in changing the intention to reduce meat consumption than an injunctive normative message.” Respondents were put in two different groups: dynamic descriptive norm (Median = 4,6700, n = 92) and dynamic injunctive norm (Median = 4,330, n = 84). The median CWWS scores for testing the main effect between dynamic norms and the intention to reduce meat consumption were not significant, χ²(1) = 0.005, p = .943, hereby rejecting Hypothesis 1.
To test H2: “A short temporal distance message will be more effective in changing the intention to reduce meat consumption than a long temporal distance message.” Respondents were put in two different groups: short temporal distance (Median = 5.000, n = 79) and long temporal distance (Median = 4.3300, n = 97). The hypothesized effect was found, participants had significantly more intention to reduce meat consumption in the short temporal distance condition compared to the long temporal condition. The median CWWS scores for testing the main effect between the intention to reduce meat consumption and temporal distance were significant, χ²(1) = 5.392, p = .020, hereby accepting Hypothesis 2.
To test H3: “Temporal distance will moderate the influence of dynamic normative messages on the intention to reduce meat consumption.” Respondents were put in four different groups: dynamic descriptive norm*short temporal (Median = 4,8350, n = 43), dynamic injunctive norm*short temporal (Median = 5,3300, n = 36), dynamic descriptive norm*long temporal (Median = 4,3300, n = 49) and dynamic injunctive norm*long temporal (Median = 4,1650, n =48). The hypothesized interaction effects were not significant. The median CWWS scores for testing the interaction effect between the intention to reduce meat consumption and dynamic norms*temporal distance, χ²(3) = 6.750, p = .080, hereby rejecting Hypothesis 3. Again the p-value is borderline significant therefore in the next section a contrast analysis will be conducted.
4.3.3 Contrast tests
To examine the interaction effect a contrast analysis is conducted. A dynamic descriptive norm message in the short temporal condition (M = 4.4884, SD = 1.92361) does not lead to a higher intention to reduce meat consumption than a dynamic descriptive norm message in the long temporal condition (M = 4.2041, SD = 1.89030, F(1,172) = 0.541, p = 0.463, CImean-differences = [-1.047, 0.478]). Hypothesis 3a is therefore rejected. A dynamic injunctive norm message in the short temporal condition (M = 4.9167, SD = 1.73548) does lead to a higher intention to reduce meat consumption than a dynamic injunctive norm message in the long temporal condition (M = 3.8889, SD = 1.82099, F(1,172) = 6.354, p = 0.013, CImean-differences = [0.223, 1.833]). Hypothesis 3b is therefore accepted. The analysis result shows that in the short temporal distance condition a
reduce meat consumption than a dynamic injunctive norm message (M = 4.9167, SD = 1.73548, F(1,172) = 1.051, p = 0.307, CImean-differences = [-0.396, 1.253]). Hypothesis 3c is therefore rejected. In the long temporal distance condition a dynamic descriptive norm message (M = 4.2041, SD = 1.89030) does not lead to a higher intention to reduce meat consumption than a dynamic injunctive norm message (M = 3.8889, SD = 1.82099, F(1,172) = 0.704, p = 0.402, CImean-differences = [-0.426, 1.056]). Hypothesis 3d is therefore accepted. Comparing the means shows that in the long temporal distance condition the effect of the dynamic descriptive norm and the dynamic injunctive norm on the intention to reduce meat consumption do not deviate from each other. Additionally, the dynamic injunctive norm leads to higher intention to reduce meat consumption when in the short temporal condition compared to the long temporal condition. Below in graph 2 the intention to reduce meat consumption in the four conditions.
Graph 2
4.4 Results summary
The results of the ANOVA tests and the Kruskal-Wallis H tests show that hypothesis 1 can be rejected. Dynamic descriptive norms do not increase the intention to reduce meat consumption significantly more compared to dynamic injunctive norms. Hypothesis 2 is accepted, in case of short temporal distance messaging (present) the intention to reduce meat consumption is significantly higher compared to when long temporal distance messaging (future) is used. The results show that
reduce meat consumption than a dynamic descriptive norm message in the long temporal condition H3(b) is accepted, a dynamic injunctive norm message in the short temporal condition does lead to a higher intention to reduce meat consumption than a dynamic injunctive norm message in the long temporal condition. H3(c) is rejected a dynamic descriptive norm message in the short temporal distance condition does not lead to a higher intention to reduce meat consumption than a dynamic injunctive norm message in the short temporal distance condition. H3(d) is accepted, in the long temporal distance condition a dynamic descriptive norm message does not lead to a higher intention to reduce meat consumption than a dynamic injunctive norm message.
5. Discussion
The purpose of this study was to investigate the effect of using dynamic social norms in message framing on the intention to reduce meat consumption, using temporal distance in message framing on the intention to reduce meat consumption and to investigate whether temporal distance as a moderator changes the effect of dynamic social norms on the intention to reduce meat
consumption. Both the dynamic injunctive norm as this particular moderation effect have never been tested before.
5.1 Main results
It was expected that there will be a difference in the intention to reduce meat consumption when the dynamic descriptive norm was made salient compared to when dynamic injunctive norm was made salient. Hypothesis 1 claims that the dynamic descriptive norm will, when made salient, lead to a higher intention in reducing meat consumption compared to when the dynamic injunctive norms was made salient. The literature shows examples of both injunctive norms made salient that increase intention or behavior more towards the “desired” behavior Reno, Cialdini, and Kallgren (1993); Cialdini et al., (2003) and descriptive norms made salient that increase intention or behavior more towards the “desired” behavior Robinson et al. (2014). All found differences in intentions or behavior when a static descriptive norm was used compared to when a static injunctive norm was used. However in combination with the fact that a dynamic majority norm was used (no backfire effect) the assumption in the first hypothesis was made. However results show that in this study, the intention to reduce meat consumption is not significantly different when a dynamic descriptive norm was made salient compared to when a dynamic injunctive norm was made salient.
condition. The expectations are based on Gilovich et al. (1993) who state that future risk are processed more positively and Chandran and Menon (2004) who found that a negative valence message (which is used in this study) works better in short temporal distance conditions. The effect found is that the intention to reduce meat consumption is larger when the temporal distance is short compared to when the temporal distance is long, the result is significant and the hypothesis is accepted.
The third expected effect, which was also hypothesized as H3, is the interaction effect of temporal distance on the effect of social norms on the intention to reduce meat consumption. Not all expected interaction effects were found.
The expected result that dynamic descriptive norms increase intention to reduce meat consumption more in the short temporal distance condition compared to the long temporal distance condition was not found (H3(a)). A possible explanation for this result is that most people still eat meat. Therefore the descriptive norm of eating no or less meat is not seen in the everyday life of people. In the long temporal distance the descriptive norm is a prognosis and thus also not seen in the everyday life of people. It seems that because the descriptive norm cannot be made tangible in either condition there is no difference in the intention to reduce meat consumption.
The second part of H3 is H3(b), the expected result that the dynamic injunctive norm increases intention to reduce meat consumption more in the short temporal condition compared to the long temporal condition. The expected results were found. The current injunctive norm is already in favor of reducing meat consumption. It is therefore increasing the intention to reduce meat consumption. In the long temporal distance condition the injunctive norm is stating how future generation think that future generations should behave which has less effect (Gilovich et al., 1993; Spence et al., 2012; Chandran and Menon, 2004). This norm does therefore not apply to the current generation and as the social dilemma theory holds for this case people will increase their intention to reduce meat consumption less (Joireman, 2005).
Lastly, the expected result that in the long temporal distance condition there is no difference due to the dynamic descriptive or the dynamic injunctive norm message in the intention to reduce meat consumption holds (H3(d)). As both norms are for future generations the norms become less applicable for the current generations. Therefore both norms are becoming a thought of the future norms, because you cannot see the actual behavior of future generations nor can you see the actual thoughts of future generations. The difference between the descriptive norm and the injunctive norm as described by Deutsch and Gerard, (1955); Cialdini et al., (1990); Cialdini et al., (1991) therefore fades away as both are just a thought of how it will be in the future.
5.2 Academic contributions
The differences between using a static descriptive norm condition and a static injunctive norm condition to influence intentions and behavior was tested in previous research (Reno, Cialdini, and Kallgren, 1993; Cialdini et al., 2003; Robinson et al. 2013). The difference between the effect on intentions and behavior by making a static descriptive norm or a dynamic descriptive norm salient has also been researched (Cialdini, 2003; Goldstein, Cialdini and Griskevicius, 2008; Nolan, Schultz, Cialdini, Goldstein and Griskevicius, 2008; Robinson, Fleming, and Higgs, 2014; Mortensen et al., 2018; Sparkman ad Walton 2017). However, using dynamic injunctive norms conditions to change intentions or behavior has not been done before. In this study the difference in the effect of the dynamic injunctive norm condition and the dynamic descriptive norm condition on intentions and behavior has been tested. This study combines the research on static social norms and dynamic descriptive norms and serves as a theoretical extension. From previous research it is known that both the descriptive norm and the injunctive norm sometimes change intentions and behavior more towards the “desired” behavior, depending on different situations (e.g. valence and majority or minority of the norm). This study shows that there is no difference between the descriptive and injunctive norm condition when both norms are presented as a dynamic norm.
The interaction effect between temporal distance and dynamic social norms show that in the short temporal distance condition the dynamic descriptive norm message and the dynamic injunctive norm message do not differ in their effect on intention to reduce meat consumption. It seems that even though the respondents read the manipulations they reflected the descriptive norm to what they see in society and that is different from the message. The difference in intention to reduce meat consumption between the short temporal distance condition and the long temporal distance
condition was not significant for the dynamic descriptive norm. Because the descriptive norm is something people notice in their everyday life and the norm was contradicting what people
experience in their everyday life. It thus is important for future research that the trend explained in the descriptive norm is something people can relate to in their life. It would particulaty be interesting to see how people would react if they see actual behavior instead of a text.
5.3 Limitations
Some of the data that was collected was unusable. There was a question that asked the participants how many days a week they ate meat. This was set to be measured as this could influence the intention to reduce meat consumption. If someone only eats meat 1 or 2 days a week, the intention to reduce this consumption compared to someone who eats meat almost every day, could be much lower. However, as this question was not set as a mandatory question to answer, not all respondents answered the question. The data is thus incomplete and could not be used. There is one other control variable of which the data collected could not be used. The question was about monthly income, this question was also not mandatory as some participants might not be willing to answer that question. Not all respondents filled in the question. For other research, when this would be measured again, the incomplete data could be avoided by displaying a certain income and letting people answer whether they earn more or less than that amount. Participants might be more willing to answer that instead of giving the actual number.
The consequences of climate change that are presented in the message framing are very concrete. The way to deal with the consequences is also presented in a concrete way. This could be one of the reasons that using short temporal distance in the message framing increases the intention to reduce meat consumption more than using long temporal distance. As mentioned by Spence et al. (2012); Liberman et al. (2002); Carver and Scheier (1990) a concrete message is better for short temporal distance than for long temporal distance.
manipulation was performed well. It could be that the manipulation was not performed correctly, or the manipulation questions were not correct. There were multiple respondents who filled in a note at the end of the survey. Some respondents noted that they thought that there were a lot of questions asking the same thing. It is possible that these respondents did not really notice the difference between the questions stated in a descriptive and in an injunctive manner. This could mean that the questions were not correct and that this is the reason why the manipulation check failed. However, the manipulation check questions were based on the work of Brügger et al. (2016); Spence et al. (2012); Nicolaij and Hendrickx (2003), therefore the descriptive and injunctive questions have proper theoretical background.
Another explanation why the manipulation failed is the order in which the manipulations are presented in the text. In the text before stating what the social norm is the consequences are already explained. When the consequences are for future generations it could be that the respondent loses his/her interest and the manipulation fails.
Lastly, it could for the dynamic descriptive norm be that the manipulation failed because respondents did not believe the manipulation. The respondents experience a different behavior in their everyday life than was described in the dynamic descriptive norm.
5.4 Future recommendations
Future research can, based on the results and limitations of this research, build further on finding ways in which intention, the proxy for behavior, can be influenced when it comes to meat
consumption. More research can show which kind of manipulations work and how they work best. The difference in the effect of dynamic injunctive and dynamic descriptive norms on
intentions and behavior should be further explored, as in previous studies it is shown the difference in effects of descriptive and injunctive norms on intentions and behavior can be significant (Reno, Cialdini, and Kallgren, 1993; Cialdini et al., 2003; Robinson et al. 2013) however in this study it is not. The finding in this study is that when a norm is stated in a dynamic way the difference between descriptive norms and injunctive norms is not detected, this should be further examined. If both dynamic social norms increase the intention or behavior more towards the “desired” behavior than the static social norms, than norms should be written in a dynamic matter. When the social norms are dynamic, than according to the results of this study, it would not matter if you chose a descriptive or injunctive norm.
It is shown that there is a significant difference in intention to reduce meat consumption when different temporal distances (present vs. future) are used. The main effect of temporal
consumption is still higher in the short temporal distance condition when the valence of the message is positive and when the consequences of climate change are explained in a more abstract way. Finding these results could lead to more effective messaging to change intentions and behavior about meat consumption because of its effect on climate change.
The sample should be more diverse and it should be larger. At the moment the sample mainly contains students and the power of the results will rise with a sample with a larger diversity and more participants can tell with more certainty when the intentions and behavior to reduce meat consumption change.
Future research should try and track people their real behavior. The link between behavior and intention is known to be fading when temporal distance increases (Azjen and Fishbein 1980). In this study found influence of the temporal distance and social norms on the intention to reduce meat consumption should also be measured for behavior. As there could be a difference between the perceived intention and the actual behavior (Ajzen, 1991). Participants of such tests should be exposed to different dynamic social norm conditions and temporal distance conditions. After which their meat consumption should be tracked for a certain period of time. After tracking the real consumption it is possible to see what kind of manipulations work best.
The moderation effect of temporal distance on the effect of dynamic social norms on the intention to reduce meat consumption was found to be non-significant. But the main effect of temporal distance on the intention to reduce meat consumption was significant. For further research the moderation effect of the other three types of psychological distance on the effect of dynamic social norms on the intention to reduce meat consumption should be examined.
The moderation effect shows that the dynamic injunctive norm in the short temporal
distance condition leads to a higher intention to reduce meat consumption compared to the dynamic injunctive norm in the long temporal distance condition. Future research should examine whether it is best to always use a dynamic injunctive norm in the short temporal distance condition when the descriptive norm is still in the minority and the injunctive norm is already in the majority.
5.5 Conclusion
