The Unconscious Effect of Descriptive Norms on Food Waste Behaviour Joly Bogers
10852026 Master’s Thesis
Graduate School of Communication Master Persuasive Communication
Dr S. Mollen 29-01-2016
Abstract
Previous research indicates that descriptive norms can influence environmentally friendly behaviour, without people being aware of this influence. However, research on descriptive norms and food waste behaviour is scarce. The present study investigated to what extent descriptive norms messages affect household food waste related behavior, compared to financial, environmental, social responsibility and information-only messages and whether people are indeed unaware of the influence of descriptive norms. The first study, consisting of an online survey and experiment among 48 young parents, found that descriptive norm beliefs were more predictive of intention to cook in proportions (i.e. using a cup, measuring cup or scale when sizing quantities of potatoes, pasta and rice), than other relevant beliefs (i.e. financial, environmental and social responsibility), even though participants rated the descriptive norm message as least influential in motivating them to cook in proportions. In contrast, Study 2, an online experiment among 371 young parents, did not find that a descriptive norm message resulted in stronger intentions to cook in proportions, interest in buying a tool to measure food proportions and parents’ actual cooking in proportions behaviour, compared to the other messages. The results of Study 1 indicate that there is a relationship between descriptive norms and food waste behaviour and that descriptive norms may indeed influence people unconsciously. However, further research should be conducted to investigate whether descriptive norm messages affect actual food waste behaviours.
Keywords: descriptive norms, food waste behaviour, cooking in proportions, young parents
The Unconscious Effect of Descriptive Norms on Food Waste Behaviour
Worldwide one-third of the food produced gets thrown away. The total amount of food waste in the Netherlands, in 2012, was estimated at 1.7 to 2.6 billion kilos (Soethoudt & Timmermans, 2013). Each year the Dutch consumer disposes of 50 kilos of food, this adds up to €150,- worth of foods that ends up in the dumpster, per person (Milieu Centraal &
Voedingscentrum, 2013). High levels of food waste have a large environmental impact, such as higher emissions of greenhouse gases and less natural resources (Garnett, 2011; Stuart, 2009). Food waste is therefore becoming a large problem for society.
Of all food wasters (e.g., consumers, agriculture, supermarkets, food industry), consumers contribute to this problem the most, as they are responsible for 38% of the total food disposal (Lipinski, Hanson, Lomax, Kitinoja, Waite, & Searchinger, 2013). Among consumers, young families with school aged children are one of the biggest food wasters (Parizeau, von Massow, & Martin, 2015; WRAP, 2007). It is therefore imperative to
investigate how young parents’ food waste behaviour can be changed. Unfortunately, so far little research has been done into the development and evaluation of interventions to reduce consumers’ food waste (Parizeau et al., 2015; Van Dooren & Mensink, 2014). The research in the field of household food waste are mainly focused on the amount of food waste and
reasons not to waste food (Parizeau, von Massow, & Martin, 2015; Quested, Parry, Easteal, & Swannell, 2011; Quested, Marsh, Stunell, & Parry, 2013; Stefan, Van Herpen, Tudoran, & Lähteenmäki, 2013) and only few studies have examined the effect of interventions targeting food waste (Barr, 2007; Comber & Thieme, 2013; Council & Lyndhurst, 2008; Gray & Toleman, 2006; Nye & Burgess, 2008; Tucker & Douglas, 2007; Whitehair, Shanklin, & Brannon, 2013). Most of these studies investigated interventions targeting several behaviours related to food waste prevention (e.g. waste reduction, recycling, reuse). Unfortunately, most of the research was based on surveys, qualitative research, correlational research or previous
literature (i.e. reviews) and had no control group. Only one study by Whitehair et al. (2013) used an experimental design and showed that a message in a dining facility, communicating not to waste food, reduced students’ food waste with 15%. To determine a causal relationship between an intervention and behaviour, experimental research is essential. Furthermore, with an experimental design it is possible to control for confounding variables. The goal of the current study therefore is to investigate how consumers can be motivated to reduce food waste within their household, using an experimental design.
Currently, campaigns that target consumers’ food waste behaviour mainly focus on the financial and environmental aspects of food waste. For example, the ‘Food is to eat!’ (‘Eten is om op te eten!’) campaign from Milieu Centraal highlights the financial and environmental benefits of disposing less food and the ‘Why50kilo?’ (‘Hoezo50kilo?’) campaign of The Netherlands Nutrition Centre Foundation emphasises the financial aspect of less food waste (www.milieucentraal.nl; www.voedingscentrum.nl). However, it is as of yet unclear whether this is in fact the most effective approach to reduce household food waste.
Another study into environmentally friendly behaviour highlights that even though people may perceive messages that stress financial, environmental and social responsibility advantages of energy saving to be effective, in fact a descriptive norm message, indicating what other people do, appeared to be most effective in changing energy saving behaviour (Nolan, Schultz, Cialdini, Goldstein and Griskevecius, 2008). Thus, people were unaware of the persuasive influence of descriptive normative information. Considering these findings, it is possible that an intervention focussing on descriptive norms may be more effective in reducing household food waste than messages focussing on financial and environmental motivations, as is done in current campaigns.
So far, only one study has been conducted that has (conceptually) replicated the study by Nolan et al. (2008) (Croker, Whitaker, Cooke, & Wardle, 2009). Prior findings were
partially replicated in this study. More specifically results showed an unconscious influence of normative information for male participants only. A shortcoming of this study is that
intended, but not actual food choice was measured. Thus, there is need for stronger scientific evidence of the unconscious influence of social norms.
In sum, there is need for experimental research investigating which message is most effective in reducing food waste and stronger scientific evidence of the unconscious influence of social norms is required. To address these gaps in the literature, the present study will conceptually replicate the study by Nolan et al. (2008) in the context of food waste. The following research question is proposed:
RQ: To what extent do descriptive norms messages affect household food waste related behavior, compared to financial, environmental and social responsibility messages and are people aware of the influence of descriptive norms?
Theoretical framework Food waste problem
Food waste is food losses occurring at the end of the food chain (i.e., retail and final consumption), which relate to retailers' and consumers' behaviour (Parfitt, Barthel, & Macnaughton, 2010). Food waste can be categorized into avoidable and unavoidable, based on the edibility of the foods. Unavoidable food waste refers to inedible food parts, such as bones, coffee ground and vegetable peels. Avoidable food waste on the other hand includes all discarded foods that were edible prior to disposal. This refers to food types and materials that could have been consumed if they had been stored or prepared properly (WRAP, 2008, 2009).
There are multiple behaviours that result in avoidable forms of food waste (Lyndhurst, 2011). For this reason, food waste behaviour is not viewed as a single behaviour, but as a set
of behaviours that increase the amount of food wasted (Jensen et al., 2012; Quested et al., 2013). These behaviours are related to different aspects of food's journey into and through the household: planning, shopping, storage, preparation and food consumption (Quested et al., 2013). The Waste & Resources Action Programme (WRAP) studies were designed to explore what kind of food is wasted, why it is wasted and what could be done to change consumers attitudes and behaviour. These studies found nine consumer behaviours that contribute to food waste reductions, for instance, making a shopping list, planning meals in advance and
portioning rice and pasta (Lyndhurst, 2011; WRAP 2007; WRAP 2009). The current study will focus on the portioning of rice, pasta and potatoes, because preparing too much food is one of the main reasons why people waste their food (Van Dooren & Mensink, 2014). Furthermore, portioning of rice, pasta and potatoes by using a measuring cup or scale can be quite easily implemented by consumers. In the current research this is called cooking in proportions.
Determinants of food waste behaviour
Several studies have investigated consumers’ motives to reduce food waste (Bos-Brouwers, Scheer, Nijenhuis, Kleijn, & Westerhoff, 2013; Parizeau et al., 2015; Quested et al., 2011; Quested et al., 2013; Stefan et al., 2013). The study by Parizeau et al. (2015) suggests that societal responsibility is a primary determinant for food waste behaviour, whereby people perceive food waste as a societal issue. Relatedly, other studies have found that food shortages elsewhere are an important motivator to reduce food waste (Bos-Brouwers et al., 2013; Quested et al., 2011). Another commonly found determinant to reduce food waste is related to economic savings that result from disposing less food. Furthermore,
environmental concern related to the negative environmental outcomes of food waste, is an important reason for people to decrease their food waste. Finally, determinants such as guilt and efficient household management were also found as common motivators of food waste
behaviour (Bos-Brouwers et al., 2013; Parizeau et al., 2015; Quested et al. 2011; Stefan et al., 2013).
Only one study to date has included norms as a determinant for food waste, but this determinant did not significantly predict consumers food waste (Stefan et al., 2013).
However, this study focused on injunctive norms (i.e., what others say we should do) and not descriptive norms (i.e., what others actually do). In contrast, the study by Nolan et al. (2008) that compared the effectiveness of descriptive norm messages to messages containing other reasons to conserve energy, such as financial, social responsibility and environmental reasons, showed that descriptive norm messages produced the greatest change in energy saving
behaviour. In addition, there is a substantial amount of evidence in experimental studies that descriptive norms can have a powerful impact on different types of environmentally friendly behaviours, such as recycling (Schultz, 1999), energy conservation (Goldstein, Cialdini, & Griskevicius, 2008, Nolan et al, 2008), reducing bottled water consumption (van der Linden, 2015), and reducing littering (Cialdini, Reno, & Kallgren, 1990; Cialdini, Kallgren, & Reno, 1991; Kallgren, Reno, & Cialdini, 2000; Reno, Cialdini, & Kallgren, 1993).
In sum, descriptive norms are an important predictor of environmentally friendly behaviors. It is therefore possible that an intervention to reduce food waste that focuses on descriptive norms may be more effective compared to interventions that focuses on
environmental or financial reasons to reduce food waste. This study aims to provide insight in the most effective way to frame a message that targets food waste behaviour.
Social norms
Social norms are "rules and standards that are understood by members of a group, and that guide and/or constrain human behaviour without the force of laws" (Cialdini & Trost, 1998, p. 152). In the literature two different types of norms are distinguished:
descriptive and injunctive norms (Cialdini et al., 1990; Cialdini et al., 1991). Descriptive norms specify what most people do in a particular situation, and motivate action by informing people of what is generally seen as adaptive or effective behaviour in that situation (Reno et al., 1993). An example is telling people most of others use a scale or measuring cup when sizing quantities of potatoes, rice or pasta. On the other hand, injunctive norms specify what most people approve or disapprove of within a certain culture, and motivate action by
promising positive or negative social sanctions for normative or counter normative behaviour, respectively (Reno et al., 1993). In the context of food waste, the injunctive norm can for instance be that most of other people approve of using a scale or measuring cup when sizing quantities of potatoes, rice or pasta. According to Cialdini et al. (1990, 1991) the type of norm that is salient at a particular time will direct an individual's immediate behaviour.
Underlying process of normative influence
The way in which behaviour is established can be explained in light of the reflection-impulsive model by Strack and Deutsch (2004). This model states that social behaviour is controlled by two systems of information processing: a reflective and an impulsive system. The reflective system elicits behaviour that is caused by a decision process, whereby knowledge about facts and values are weighted and integrated. On the other hand, the impulsive system generates social behaviour through associative clusters called behavioural schemata. These behavioural schemata are bonded representations of a particular behaviour and the consequence of that behaviour in a certain situation. For example, seeing a pan will activate the behavioural schemata for cooking. The reflective system requires a high amount of cognitive effort, whereas the impulsive system acts fast and requires little cognitive capacity and may induce automatic behaviours outside consciousness (Strack & Deutsch, 2004). Thus, some behavioural decisions occur automatic and reflexive. This may also be the case with the influence of descriptive norms on behaviour. Cialdini (2003) speculated that
because descriptive norms are based on observed behaviour of other individuals, it is relatively easy to adjust to the norm without much cognitive analysis. He described
descriptive norm information as a decision-making heuristic of "social proof", serving as a time- and effort-saving shortcut that can stimulate effective behaviour is different situations (Cialdini, 2009). Based on this Kredentser, Fabrigar, Smith, and Fulton (2012) hypothesized that descriptive norm messages are more effective under low elaboration, meaning that the cognitive abilities were limited, than injunctive norm messages. The results showed that descriptive messages indeed lead to stronger intentions to join a university program, than injunctive norm messages under low elaboration. These results can be explained through the mechanism of cognitive processing; when cognitive resources are limited, descriptive norms can function as a quick heuristic to make a decision (Kredentser et al., 2012). This mechanism is in turn related to the impulsive system whereby distractions or arousal do not interfere with its operation (Strack & Deutsch, 2004).
There is a substantial amount of research showing that people are not always aware of others’ behaviour in a social environment as a causal antecedent for their own behaviour (Aarts & Dijksterhuis, 2003; Croker et al., 2009; Nolan et al., 2008; Nolan, Kenefick, & Schultz, 2011). Aarts and Dijksterhuis (2003) for instance found that the goal to visit a certain environment can automatically activate norms appropriate in that context and stimulate behaviour that is in line with these norms. For example, they found that when the goal of going to the library was activated, participants’ speed of responding to concepts related to normative behaviour associated with being in a library (e.g. silent, whisper) increased and respondents voice intensity decreased (Aarts & Dijksterhuis, 2003). The study by Nolan et al. (2008) also demonstrated that people can be unaware of the persuasive impact of descriptive norms on energy conservation behaviour. They conducted two studies to investigate the persuasiveness and detectability of descriptive normative influence. The first study consisted
of a survey about energy conservations and showed that descriptive normative beliefs were more predictive of energy conservation intentions than financial, environmental and social responsibility beliefs, even though respondents rated descriptive norms as least important in their decision to conserve energy. The second study was a field experiment in which
households received one of the five experimental messages: descriptive norm, environment, social responsibility, financial or information-only control. Results showed that the
descriptive norm message produced the greatest change in energy conservation behaviour compared to the other messages, even though respondents rated the normative message as least motivating (Nolan et al., 2008).
Two explanations for the unconscious effect of descriptive norms on social behaviour can be found. The first, as stated above, is that descriptive norms can serve as a heuristic cue for behaviour that is accurate in a certain situation (Cialdini, 2009; Jacobson, Mortensen, & Cialdini, 2011; Kredentser et al., 2012). Another explanation why people underestimate the impact of social norms may be their alternative naïve explanations for their behavior. The term naïve psychology refers to a person’s own conception of behaviour and mental processes (Heider, 1958). Five studies by Pronin, Molouki, and Berger (2007) showed that people see others as more susceptible to social influences than themselves. People also tend to display an introspection illusion, whereby people place more weight on their own thoughts and beliefs in their decision to conform to a certain behaviour than other people’s behaviour, but this does not apply for others (Pronin et al., 2007). Thus, people tend to think that they act in a way because of inner beliefs and not because of the behaviour of others, but on the other hand think that other people are affected by social influences.
Present study
The present study investigates if the results of the study by Nolan et al. (2008) can be replicated in the context of food waste. By conceptually replicating these findings there will
be evidence for the unconscious influence of social norms on food waste behaviour.
Moreover, there will be stronger scientific evidence that people are, in general, unaware of the persuasive influence of social norms. Furthermore, this study aims to provide insight into which message (i.e., descriptive norm, financial, environmental, social responsibility, control) will be most effective in promoting behaviours that will reduce food waste.
The first study will investigate which type of message people perceive to be most effective in influencing them to cook in proportions and examine the extent in which participants’ beliefs about cooking in proportions (i.e., descriptive norm, financial,
environmental and social responsibility) are related with their self-reported intention to cook in proportions. The goal of this study is to examine whether participants’ perceived
effectiveness of the message aligns with the beliefs that are related to participants’ intention to cook in proportions. An important difference with the study as conducted by Nolan et al. (2008) is that instead of measuring the perceived extent to which various factors motivate people to perform the desired behaviour, the actual stimulus material (the different messages) will be shown to the participants and will be assessed. This adjustment improves the previous study because the obtained results will be closer to reality. The different motivations, as measured by Nolan at al. (2008), are much more distant to an actual message persuading people to perform a behaviour, than exposing people to different persuasive messages
including these motivations, as is done in this study. Thus this improvement will increase the power of the relationship between descriptive norms and food waste behaviour. The second study further examines the influence of descriptive norms on actual behavior, using an experimental design. This study will investigate which message will result in the greatest behavioral change and will again investigate if this aligns with participants’ perceived persuasiveness of the message.
Taken together, several studies have shown that people underestimate the persuasive influence of others' behaviour on their own actions (Aarts & Dijksterhuis, 2003; Croker et al., 2009; Kredentser et al., 2012; Nolan et al., 2008). Based on these findings, the following hypotheses are formulated:
H1: The descriptive norm and control messages will be perceived as less motivating, than financial, environmental and social responsibility messages in promoting cooking in proportions.
H2: Descriptive norm beliefs will have a stronger relationship with self-reported intention to cook in proportions, than financial, environmental and social
responsibility beliefs.
H3: The descriptive norm message will be more effective in stimulating people to cook in proportions, than the financial, environmental, social responsibility and control messages.
Study 1
The goal of this first study was to investigate which type of message (i.e., descriptive norm, financial, environmental, social responsibility, control) young parents perceive to be most effective in influencing them to cook in proportions and examine the extent to which participants’ beliefs about cooking in proportions predicts their intention to cook in proportions. To address these questions, an online experiment and survey among young parents was conducted, investigating their intention to cook in proportions, beliefs about cooking in proportion and evaluations of the different messages.
Method Participants & Design
The first study consisted of two parts. The first part of the study was an online survey in which participants had to indicate their intentions to cook in proportions and beliefs regarding cooking in proportions (Beliefs: descriptive norm, financial, environmental, social responsibility). The second part was an online experiment with a within-subjects design, in which participants were exposed to five different messages in a random order to avoid order effects (Type of message: descriptive norm vs. financial vs. environmental vs. social
responsibility vs. control) and evaluated each message.
The sample consisted of young parents with school aged children (up to 12 years old). The recruitment text clearly stated that only young parents with school aged children (up to 12 years old) who cook at least occasionally could participate in the study. This was done to ensure that only participants from the target group would participate in the research. Furthermore, the recruitment text stated that twenty euros would be raffled among the participants. The recruitment text and the link was posted on different Facebook pages, such as the personal Facebook page of the researcher, Facebook pages of friends and family of the researcher and the Facebook page of the Life Span Lab from Tilburg University. People were requested share the recruitment text and link with their friends and or followers.
A total of 113 parents participated in this online survey. Fifty-one participants were excluded from the study because they did not complete all measures (45.13%). Of the
remaining 62 participants, 14 people indicated that they never throw away any food or already cook in proportions, these people were also excluded from further analysis because their behaviour could likely not be changed (22.58%). The final sample of this study (N = 48) consisted of 6 males (12.50%) and 42 (87.50%) females. The age of participants ranged from 19 to 48 years old (M = 34.25, SD = 8.22). Most participants were married couples, of Dutch origin and were well-heeled. This study was approved by the Ethical Committee of
Communication Science of the University of Amsterdam, The Netherlands. Participants characteristics are summarized in in Table 1.
Table 1
Demographic Characteristics Measured in Study 1
Demographic characteristics M (SD) Age 34.24 (8.22) Household size 2.50 (.95) N (%) Gender Female Male 42 (87.5) 6 (12.5) Household composition Single 0 (0)
Single with children 8 (16.7)
Married with children 36 (75.0)
Married without children Other 3 (6.3) 1 (2.1) Education Vmbo 2 (4.2) Havo 4 (8.3) Vwo 0 (0) MBO 11 (22.9) HBO 18 (37.5) WO 13 (27.1) Income Belowe average 12 (25.0) Average 13 (27.1) Above average 23 (47.9) Ethnicity Netherlands 44 (91.7)
Suriname 0 (0) Netherlands Antilles 0 (0) Turkey 1 (2.1) Morocco 0 (0) Other 3 (6.3) Don’t know 0 (0) Ethnicity father Netherlands 44 (91.7) Suriname 1 (2.1) Netherlands Antilles 1 (2.1) Turkey 1 (2.1) Morocco 0 (0) Other 1 (2.1) Don’t know 0 (0) Ethnicity mother Netherlands 45 (93.8) Suriname 0 (0) Netherlands Antilles 0 (0) Turkey 1 (2.1) Morocco 0 (0) Other 2 (4.2) Don’t know 0 (0) Procedure
A link in an e-mail or post on a social network page directed participants to an online survey in Qualtrics. A short cover-story explained to participants that the current study was concerned with questions regarding the cooking behaviour of young parents. Before being able to continue to the actual study, participants had to indicate agreement with the informed consent form. First, participants answered questions relating to their intention to cook in proportions and their beliefs. Second, participants were exposed to each of the five messages
(descriptive norm, financial, environmental, social responsibility and control message) and rated the extent in which the message motivated them to cook in proportions. Finally,
demographic data was collected, participants were thanked for their contribution and had the opportunity to reflect their ideas of the purpose of the research and leave comments for the researcher. This was included to check whether there were participants that needed to be excluded from analysis, because they knew the actual purpose of the research or had
otherwise unreliable data. Upon completion of the online experiment participants immediately received a debriefing, explaining the actual purpose of the research.
Stimuli
The stimulus materials used in this study were five different messages based on the messages used in the research by Nolan et al. (2008). The messages were modified to the target behaviour of the current research, cooking in proportions, and translated into Dutch (Figures 1 to 5 in Appendix A).
All messages included a food proportion scheme and informed participants that you can prevent food waste by cooking in proportions. The control message only mentioned: ‘You can prevent food waste by cooking in proportions, by using a measuring cup, scale or cup when preparing their pasta, rice and potatoes’. Additionally, each manipulation contained specific information on cooking in proportions related to the experimental condition. More specifically, the descriptive norm message focused on what other young parents do and stated that: ’77 % of the young parents cook in proportions, by using a measuring cup, scale or cup when preparing their pasta, rice and potatoes’. The financial message, aimed at creating awareness of the amount of money one can save by cooking in proportions, communicated: ‘You can save up to 340 euros by cooking in proportions, by using a measuring cup, scale or cup when preparing their pasta, rice and potatoes’. The focus of the environmental message was on the benefits to the environment and indicated that: ‘You can prevent 134 kt CO2
emissions by cooking in proportions, by using a measuring cup, scale or cup when preparing their pasta, rice and potatoes’. Lastly, the social responsibility message communicated that: ‘You can have up to 105 kilos less food waste by cooking in proportions, by using a
measuring cup, scale or cup when preparing their pasta, rice and potatoes’. The images used and the format of the messages was kept constant across all messages.
Measures
Intentions. Participants were first informed about the meaning of cooking in proportions. Hereafter, participants’ self-reported intention to cook in proportions was
measured using four items based on the intention scale by Stok, de Ridder, de Vet and de Wit (2014). The four items were: ‘I am planning to cook in proportions next week’, ‘I am
intending to cook in proportions next week’, ‘I want to cook in proportions next week’ and ‘I am expecting to cook in proportions next week’. The answer options ranged from 1 (totally disagree) to 5 (totally agree), where a higher score indicates a higher intention to cook in proportions. Exploratory factor analysis indicated that the items load on one factor, explaining 96.94% of the variance (EV = 3.88). The four-item scale proved reliable as indicated by a Cronbach’s of .99 (M = 3.79, SD = 1.61).
Beliefs. Participants were also asked about their beliefs related to cooking in
proportions. These were descriptive norm, financial, environmental and social responsibility beliefs and were measured using four scales consisting of three items partially based on items of Nolan et al. (2008) and Schultz (2000). Descriptive norm beliefs items were: ‘How often do you think other young parents cook in proportions?', ‘How often do you think your neighbors cook in proportions?’ and ‘How often do you think residents of your village or city cook in proportions?’. These items were measured on 7-point Likert scale ranging from 1 (never) to 7 (almost always). Example items for the financial, environmental and social responsibility beliefs were: ‘How much money do you think you can save by cooking in proportions?’, ‘How
much do you think cooking in proportions will protect the natural environment?’ and ‘How much do you think cooking in proportions will benefit society?’, respectively. These items were measured on a 7-points Likert scale ranging from 1 (not at all) to 7 (extremely).
Exploratory factor analysis indicated that the descriptive norm beliefs, financial beliefs, environmental beliefs and social responsibility beliefs scales were all unidimensional, explaining respectively 76.95% (EV = 2.31), 98.13% (EV = 2.94), 92.03% (EV = 2.76) and 88.51% (EV = 2.66) of the variance. The descriptive norm beliefs items form a reliable scale, as indicated by a Cronbach’s of .84 (M = 3.60, SD = 1.21). The financial beliefs scale proved reliable as well, Cronbach’s = .99 (M = 4.86, SD = 1.78). Finally, both the environmental beliefs and the social responsibility beliefs items formed reliable scales (Cronbach’s environmental = .96, M = 4.1, SD = 1.92, Cronbach’s social responsibility = .93, M = 3.79, SD = 1.61).
Message evaluation. Immediately after participants read one of the five messages they were asked to evaluate this message. They had to do this five times, once for each message. The evaluation of the message was measured with a single item and was worded as follows: ‘To what extent does this message motivate you to cook in proportions?’. The answer options ranged from 1 (totally disagree) to 5 (totally agree), where a higher score indicates a more positive message evaluation.
Demographic variables. Participants were asked questions regarding their gender, age, education, income, household composition, household size, ethnicity, ethnicity father and ethnicity mother. These variables were also used in the study by Nolan et al. (2008) or have been associated with food waste behaviour in previous studies (Bos-Brouwers et al., 2013; WRAP, 2007).
To investigate whether the descriptive norm and control messages were perceived as less motivating, than financial, environmental, social responsibility messages in promoting cooking in proportions, a one-way within-subjects ANOVA was used to reveal differences across the five different evaluations of the messages. The within-subjects factor was ‘Message evaluation’ (evaluation of the five different messages). Specific differences between
conditions were further analysed by Bonferonni post-hoc tests. The assumption of Sphericity was tested for the within-subjects factor with Mauchly’s Test of Sphericity.
To examine whether descriptive normative beliefs predicted self-reported intentions to cook in proportions to a larger extend than financial, environmental and social responsibility beliefs of cooking in proportions, a multiple regression analysis was used. The variable ‘Intentions’ was entered as dependent variable and the beliefs (i.e., descriptive norm, financial, environmental and social responsibility) were added to the regression as
independent variables. The enter method was used, whereby all independent variables were entered into the equation in one step. For all statistical analyses, a significance level of 0.05 (two-tailed) was applied.
Results Control variables
Correlation analyses were run to check whether any of the demographic characteristics were related to the dependent measure intention to cook in proportions. None of the
demographic measures correlated with intention (see Table 2). Thus, none of these background variables were included in the multiple regression analysis.
Table 2
Pearson Product-moment Correlation Between Intention and Demographics Measures Intention Gender Age Household
composition
Household size
Income Ethnicity Ethnicity father Ethnicity mother Education Intention Gender .06 Age .05 .17 Household composition .02 -.07 .02 Household size .06 -.07 .39** .58** Income -.00 .35* .62** .21 .34* Ethnicity -.11 -.11 -.07 .03 -.07 -.04 Ethnicity father .03 -.10 -.37* -.23 -.27 -.34* .56** Ethnicity mother .33 -.10 -.14 .03 .07 -.14 .45** .69** Education .10 .20 .22 .22 .18 .55** .11 .00 .10 Note. *= p ≤ .05 **= p ≤ .01.
Message evaluation
The one-way within-subjects ANOVA yielded a significant main effect for message evaluation F(4, 44) = 3.61, p = .012, η2 = .25, indicating that there was a significant
difference in participants’ evaluations of the five messages. The descriptive norm message was evaluated as the least motivating, but the evaluation of the descriptive norm message (M = 3.10, SD = .24) only differed significantly from the evaluation of the financial message (M = 3.85, SD = .25), p = .006. No other differences between conditions were found. Mean scores, standard deviations and results of the post-hoc tests are shown in Table 3.These results confirm the first hypothesis to a great extent.
Table 3
Bonferonni Post-hoc Test Descriptive Norm Message and Other Messages on Message Evaluation Message evaluation Message type M SD p 1. Descriptive norm 3.10 .24 2. Financial 3.85 .25 .006* 3. Environmental 3.26 .25 1.00 4. Social reponsibility 3.46 .25 1.00 5. Control 3.58 .25 .191 Note. *= p ≤ .05
Descriptive norm beliefs and intention
A multiple regression analysis showed that descriptive norm beliefs indeed
demonstrated to be the only significant predictor of intentions to cook in proportions (β = .33; t = 2.45, p = .018). The other beliefs about cooking in proportions were not significant
predictors of intention: financial (β = .06 , t = .31, p = .757), environmental (β = .32, t = 1.62, p = .113), and social responsibility (β = .01, t = .05, p = .959). Descriptive norm beliefs
explained a significant part of the variance in scores on intention to cook in proportions R2 = .33, F(4, 43) = 5.34, p = .001. This implies that descriptive norm beliefs have a stronger relationship with intention to cook in proportions, than the other beliefs. Thus, hypothesis 2 can be confirmed. Results of the multiple regression analysis are shown in Table 4.
Table 4
Multiple Regression Analysis Beliefs on Intention to Cook in Proportions Intention Beliefs β t p Descriptive norm .33 2.45 .018* Financial .06 .31 .757 Environmental .32 1.62 .113 Social responsibility .009 .05 .959 Note. *= p ≤ .05 Discussion
Overall, these findings are to a great extent in line with the findings of Nolan et al. (2008) in the context of energy conservation. Findings from this study indicate that young parents think that a descriptive norm message, stating how many other young parents cook in proportions, will have less impact on their behaviour to cook in proportions, compared to a message that stimulates people to cook in proportions by highlighting the amount of money they can save. In contrast with these findings, the results of the survey in which beliefs and intentions to cook in proportions were assessed, demonstrated that beliefs regarding how often other people cook in proportions showed to be a predictor of intention to cook in proportions. These results are greatly in line with prior findings by Nolan and colleagues and provide an initial indication that descriptive norms may influence peoples’ food waste
nature of the first study it is not possible to make causal claims based on these findings. To further examine the influence of descriptive norms on cooking in proportions, a second study using an experimental design was conducted.
Study 2
The goal of Study 2 was to investigate if results found by Nolan and colleagues (i.e. a descriptive norm message produced the greatest change in energy conservation behaviour compared to the other messages, even though participants rated the normative message as least motivating) could be replicated in the context of food waste behavior among young parents with school aged children. Additionally, because most research related to food waste interventions is based on surveys (Barr, 2007; Gray & Toleman, 2006; Tucker & Douglas, 2007), there is need for experimental research in this area. This second study therefore examined which persuasive message induces the greatest change in cooking in proportions, and if this again misaligns with the perceived effectiveness of the message. The second study improves the first study in two ways. First, because of the experimental design of this study conclusions may be drawn regarding the causal direction of the relationship and second, not only self-reported intentions, but also actual cooking in proportions behaviour was measured.
It was expected that participants, like in Study 1, would rate the descriptive norm message as least influential to motivate parents to cook in proportions, compared to financial, environmental, social responsibility and information-only messages. However, the descriptive norm message is expected to be most effective at promoting young parents to cook in
proportions, compared to all other messages. Method Participants & Design
In this second study a 1 x 5 (Type of message: descriptive norm vs. financial vs. environmental vs. social responsibility vs. control) factorial between-subjects design with a posttest of behaviour one week later, was conducted.
Again the target group consisted of young parents with school aged children. The recruitment text clearly stated that only young parents with children up to 12 years old and who at least cook occasionally could participate in the study. The text also communicated that four times twenty euros will be raffled among participants who took part in both parts of the study. Participants were recruited by approaching different organisations, magazines and blogs for young parents and the personal network of the researcher. Emails with the
recruitment text and a link to the online experiment were sent to organisations, such as Milieu Centraal and The Netherlands Nutrition Centre Foundation. They helped to recruit
participants by promoting this research via their social media channels. Furthermore, the recruitment text was posted on different Facebook pages, such as ‘Amsterdam durft te
vragen’, ‘Ontmoetingscentrum Jonge Ouders’ and ‘Ouders van nu’, the blog ‘Ouders online’, and the school newsletter of OBS den Bongerd in Goirle.
The final sample of the first part of the study (T1: N = 371) consisted of 22 males (5.93%) and 349 females (94.07%). The age of participants ranged from 19 to 61 years old (M = 35.81, SD = 6.40). The second part of this study (T2: N = 232) consisted of 10 males
(4.31%) and 222 females (95.69%). Participants’ age ranged from 23 to 60 years old (M = 35.77, SD = 5.98). Most participants were married couples, of Dutch origin and were well-heeled. Participants characteristics are summarized in Table 5.For further information about the selection of the samples see Figure 1.
Table 5
Demographic Characteristics Measured in Study 2
Descriptive Norm Financial Environmental Social Responsibility Control
Demographic (N = 73) (N = 75) (N = 75) (N = 78) (N = 70) characteristics M (SD) M (SD) M (SD) M (SD) M (SD) Age 35.67 (6.23) 35.99 (6.84) 35.19 (6.02) 35.62 (6.35) 36.66 (6.62) Household size 2.80 (.79) 2.82 (.90) 2.57 (.81) 2.87 (.71) 2.54 (.94) N (%) N (%) N (%) N (%) N (%) Gender 68 (5) 68 (7) 72 (3) 71 (7) 70 (0) Female 68 (93.2) 68 (90.7) 72 (96.0) 71 (91.0) 70 (100) Male 5 (6.8) 7 (9.3) 3 (4.0) 7 (9.0) 0 (0) Household composition Single 2 (2.7) 2 (2.7) 1 (1.3) 0 (0) 0 (0) Single with children 5 (6.8) 5 (6.7) 4 (5.3) 8 (10.3) 8 (11.4) Married with children 65 (89.0) 67 (89.3) 69 (92) 69 (88.5) 58 (82.9) Married without children 1 (1.4) 1 (1.3) 1 (1.3) 1 (1.3) 2 (2.9) Income Belowe average 11 (15.1) 10 (13.3) 5 (6.7) 9 (11.5) 12 (17.1) Average 26 (35.6) 19 (25.3) 24 (32.0) 24 (30.8) 20 (28.6) Above average 36 (49.3) 46 (61.3) 46 (61.3) 45 (57.7) 38 (54.3) Ethnicity Netherlands 67 (91.8) 71 (94.7) 70 (93.3) 71 (91.0) 68 (97.1) Suriname 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Netherlands Antilles 2 (2.7) 0 (0) 0 (0) 0 (0) 0 (0) Turkey 0 (0) 0 (0) 1 (1.3) 0 (0) 0 (0)
Marocco 1 (1.4) 0 (0) 1 (1.3) 0 (0) 0 (0) Other 3 (4.1) 3 (4) 3 (4) 7 (9.0) 2 (2.9) Don’t know 0 (0) 1 (1.3) 0 (0) 0 (0) 0 (0) Ethnicity father Netherlands 66 (90.4) 71 (94.7) 68 (90.7) 65 (83.3) 62 (88.6) Suriname 0 (0) 0 (0) 0 (0) 1 (1.3) 0 (0) Netherlands Antilles 0 (0) 0 (0) 0 (0) 1 (1.3) 0 (0) Turkey 2 (2.7) 0 (0) 1 (1.3) 0 (0) 2 (2.9) Marocco 1 (1.4) 3 (4.0) 1 (1.3) 2 (2.6) 0 (0) Other 4 (5.5) 1 (1.3) 5 (6.7) 8 (10.3) 6 (8.6) Don’t know 0 (0) 1 (1.3) 0 (0) 1 (1.3) 0 (0) Ethnicity mother Netherlands 64 (87.7) 67 (89.3) 68 (90.7) 69 (88.5) 61 (87.1) Suriname 0 (0) 1 (1.3) 0 (0) 1 (1.3) 0 (0) Netherlands Antilles 2 (2.7) 0 (0) 0 (0) 0 (0) 0 (0) Turkey 2 (2.7) 0 (0) 1 (1.3) 1 (1.3) 1 (1.4) Marocco 1 (1.4) 1 (1.3) 1 (1.3) 1 (1.3) 0 (0) Other 4 (5.5) 6 (8.0) 5 (6.7) 6 (7.7) 8 (11.4) Don’t know 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Procedure
As in the first study, participants clicked on a link directing them to the first part of the online experiment in Qualtrics. A short cover-story explained to participants that the current study was concerned with questions regarding the cooking behaviour of young parents.
Figure 1. Flowchart showing how the sample of Study 2 is selected at Time 1 and Time 2.
Before being able to continue to the actual study, participants had to indicate agreement with the informed consent form. First, participants filled in their demographic data. Secondly, participants were randomly exposed to one of the five different messages encouraging parents to cook in proportions. Participants were asked to read this message carefully. The messages were identical to the ones used in Study 1. Hereafter, they had to indicate their intentions to cook in proportions and their interest in buying a tool to measure food proportions. Lastly, participants were asked to fill in their email address so they could receive the link to the second part of the experiment. Seven days after people filled in the first part of the
experiment, they received an email with the link to the second part and the request to fill in this questionnaire the same day. People who did not participate in the second part of the experiment that day, were sent a reminder email a day later repeating the request to fill out the questionnaire. This second part included questions about the cooking behaviours (measuring pasta, rice and potatoes) of the last seven days, message evaluation, involvement with food waste and a manipulation check. At the end people were thanked for their cooperation and had the opportunity to reflect their ideas of the purpose of the research and leave additional comments. This opportunity was included to check whether there were participants that needed to be excluded from the analysis, because they knew the actual purpose of the research or had unreliable data. Upon completion of the online experiment participants received a debriefing, explaining the actual purpose of this research.
Stimuli
The stimulus materials were the same as in the first experiment, consisting of the five different messages promoting young parents to cook in proportions (see also Figures 1-5 in Appendix A).
Measures
Intentions. Participants’ self-reported intention was measured with the same scale as in the first experiment. The constructed scale did not have a normal distribution: the skewness was -.416 and the kurtosis -1.004. In attempt to create a more normal distribution a
logarithmic transformations, exponential transformation and a power transformation was applied. However, these attempts did not improve the distributions and therefore it was decided to remain the initial intention scale for the analyses. Furthermore, the sample size of this study was large enough (e.g. 30+), therefore the violation of the assumption of normal distribution should not be too problematic (Pallant, 2005). Exploratory factor analysis
indicated that the items load on one factor, explaining 93.64 % of the variance (EV = 3.75). The four-item scale proved reliable as indicated by a Cronbach’s of .98 (M = 3.57, SD = 2.14).
Interest in buying a tool to measure food proportions. Participants first read a description of the tool (i.e. ‘A measuring cup especially made for measuring pasta and rice’) and a note that they could buy it at a reduced price. Following this participants were asked to indicate if they wanted to buy the tool and be redirected to the webshop where this product could be purchased. This was measured with a single item: ‘Are you interested in buying the tool to measure food proportions?’. The answer options were: ‘No, I already have one at home’, ‘No, I don’t want one’ and ‘Yes, send me to the website’. If participants clicked on the last answering option, they were send to the web shop of the Netherlands Nutrition Centre Foundation upon completion of the Time 1 measures. There they could buy the tool at a reduced price. The three answering options were recoded: ‘No, I already have on at home’ = 99 (missing), ‘No, I don’t want one = 0 (71.91%) and ‘Yes, send me to the website’ = 1 (28.09%).
Cooking in proportions. At Time 2 the actual number of times that participants cooked in proportions the past seven days was measured with seven items. The first item checked whether participants cooked at all in the past seven days. When participants indicated that they did not cook, the follow-up questions about cooking in proportions were skipped. The other six items were split into three product categories: potatoes, pasta and rice
(couscous, quinoa, etc.). Whereby participants were first asked to indicate how many times they prepared a certain product and then were asked to specify how many times they
measured the product when they prepared the meal. Example items are: ‘How many times in the past seven days have you prepared pasta for your meal?’ and ‘How many times in the past seven days have you measured your pasta by using a measuring cup, a scale or cup while
preparing your pasta?’. If participants did not prepare a product, the measurement question was automatically skipped and they were send to the next product question. The answer options ranged from 0 times to 14 times. Scores were recoded into a percentage of the number of times a product was measured out of the total number of times a particular product was prepared. This was calculated by number of times measured divided by number of times cooked. This was done for all three products, resulting in three new variables: potatoes measured (N = 218, M = 16.59, SD = 34.39), pasta measured (N = 214, M = 47.35, SD = 48.57) and rice measured (N = 173, M = 60.93, SD = 48.24). Four outliers with invalid data (for example, a score of 200 %) were excluded from further analyses (2 outliers on potatoes measured and 2 on pasta measured).
Message evaluation. At Time 2 participants were asked to evaluate the message they had read at Time 1. The evaluation of the message was measured with a single item based on Nolan et al. (2008) and is as follows: ‘To what extent did the information in the message you read last week about cooking in proportions motivate you to cook in proportions more often?’. The answer option ranged from 1 (not at all) to 5 (extremely), where a higher score indicates a more positive message evaluation (M = 2.84, SD = 1.82).
Involvement with food waste. Participants’ involvement with the issue of food waste was measured using three items based on Zaichkowsky’s (1994) personal involvement
inventory scale. Items were: ‘I think reducing food waste is important’, ‘I think reducing food waste is relevant’ and ‘I think reducing food waste is needed’. Items were measured on a 7-point Likert scale ranged from 1 (totally disagree) to 7 (totally agree), where a higher score indicates high level of involvement. Exploratory factor analysis indicated that the scale was unidimensional, explaining 76.27% of the variance (EV = 2.29). The three items form a reliable scale as indicated by a Cronbach's of .84 (M = 6.18, SD = .97).
Demographics. Participants were asked about their gender, age, income, household composition, household size, ethnicity, ethnicity father and ethnicity mother.
Control variables. Besides demographics, some measures were included as well that could be potential confounders. In the first part of the experiment, participants were asked three questions about their past behaviour, these included: ‘How often have you cooked potatoes, pasta and rice in the past seven days’, ‘How often have you measured potatoes, pasta and rice in the past seven days’ and ‘How often have you discarded potatoes, pasta and rice in the past seven days’. The answer options ranged from 0 times to 14 times. The first two variables were recoded into one percentage of potatoes, pasta and rice measured in the past. This was calculated by how often they cooked divided by how often they measured these products. Resulting in the variables: past measuring behaviour (M = 41.39, SD = 30.44) and past discarding behaviour (M = 2.56, SD = 1.61). In addition, participants were asked about their use of eating boxes (i.e. boxes with ingredients and recipes delivered at home) of Hello Fresh or Albert Heijn. Response options were: yes (5.66%) or no (94.34%).
Manipulation check. To check whether participants read and believed the
information in the message, a manipulation check was conducted. In the second part of the experiment, participants were asked a question about the unique content of the experimental (not control) messages. The question for the descriptive norm message was: ‘What percentage of young parents cooks in proportions in your opinion?’ (1 = 12%, 2 = 33%, 3 = 52%, 4 = 77%, 5 = 85 %). The financial message question stated: ‘How much money do you think you can save up by cooking in proportions?’ (1 = up to €30, 2 = up to €80, 3 = up to €160, = up to €230, 5 = up to €340). The question for the environmental message was: ‘How much kt CO2 emissions do you think can you prevent by cooking in proportions?’ (1 = up to 110 kt per year, 2 = up to 134 kt per year, 3 = up to 167 kt per year, 4 = up to 198 kt per year, 5 = up to 212 kt per year). Lastly, the social responsibility message questions worded: ‘How many kilos
less food could be wasted by cooking in proportions you think?’ (1 = up to 55 kilos, 2 = up to 85 kilos, 3 = up to 105 kilos, 4 = up to 145 kilos, 5 = up to 175 kilos). It was expected, that the participants in the specific experimental message condition would give the correct answer to the question more often, than participants in the other conditions. Scores were recoded into: incorrect answers = 0 and correct answer = 1.
Statistics
To examine the first hypothesis, whether participants indeed rate the descriptive norm message as less influential in motivating young parents to cook in proportions, compared to the financial, environmental, social responsibility and information-only messages, a one-way ANOVA was conducted, with ‘Message evaluation’ as dependent and ‘Condition’ as the independent variable.
The third hypothesis, that a descriptive norm message will be more effective in stimulating young parents to cook in proportions, compared to all other messages, was examined with three different dependent variables: intention, interest in buying a tool to measure food proportions and actual cooking in proportions. To examine the effect of the type of message on intention, an ANOVA was performed with ‘Condition’ as the independent and ‘Intention’ as the dependent variable. To investigate if there was an effect of type of message on buying the tool to measure food proportions, a Chi-square test was used. Specific
differences were further examined by means of separate Chi-square tests between the specific experimental message and the control message. To test whether participants in the descriptive norm condition measured their potatoes, pasta and rice more often than participants in the other conditions, a MANOVA was conducted, with ‘Potatoes measured’, ‘Pasta measured’ and ‘Rice measured’ as the dependent variables and ‘Conditions’ as the independent variable. For all statistical analyses, a significance level of 0.05 (two-tailed) was applied.
Results Manipulation check
Four chi-square tests were conducted to examine if the manipulations were successful. It should be noted that if the assumption that 5 cells have expected count less than 5 has been violated, the Likelihood Ratio (LR) was used. A Chi-square test with the descriptive norm manipulation question showed that participants in the descriptive norm conditions gave the correct answer more often than participants in other conditions, LR(4) = 10.70, p = 0.030 (correctdescriptive norm: 8.88%, correctfinancial: 4.55%, correctenvironmental: 0%, correctsocial responsibility: 0%). However, no differences between conditions were found for the other three manipulation questions. There was no difference between the conditions on the financial manipulation question, LR(4) = 5.60, p = 0.231 (correctdescriptive norm: 2.22%,correctfinancial: 0 %,
correctenvironmental: 4.35%, correctsocial responsibility: 7.84%), the environmental manipulation question, X2(4) = 2.81, p = 0.591 (correctdescriptive norm: 31.11%, correctfinancial: 25.0%, correctenvironmental: 28.26%, correctsocial responsibility: 29.41%) and the social responsibility manipulation question, X2(4) = 1.62, p = 0.806 (correctdescriptive norm: 33.33%, correctfinancial: 22.73%, correctenvironmental: 30.43%, correctsocial responsibility: 25.49%). In conclusion, the manipulation only partially succeeded.
Randomization check and control variables
To explore whether participants were indeed randomly assigned to the five different conditions, ANOVAs and Chi-square tests were conducted. It should be noted that if the assumption of equal variances in the population has been violated, Welch’s F test was used. ANOVAs revealed that age, F(4, 366) = .52, p = .719, η2 = .006, past measuring behaviour, Welch’s F(4, 182) = .59, p = .62, past discarding behaviour, F(4, 366) = 1.53, p = .192, η2 = .016, income, F(4, 366) = 1.01, p = .400, η2 = .011, and involvement with food waste, F(4,
227) = .49, p = .746, η2 = .008, were equally distributed across the five conditions. Chi-square tests also showed that household composition, LR(16) = 15.71, p = 0.473, ethnicity
participant, LR(20) = 19.94, p = 0.462, ethnicity father, LR(24) = 26.73, p = 0.317 and ethnicity mother, LR(20) = 16.35, p = 0.695, were equally divided across the different conditions. On the other hand, further tests showed that variables gender, LR(4) = 11.69, p = 0.020 and household size, Welch’s F(4, 179) = 2.51, p = .044, were not randomly distributed across the five conditions.
In addition, correlation analyses were conducted to reveal which variables are related to the dependent variables message evaluation, intention to cook in proportions and cooking in proportions behaviour. Correlation analyses demonstrated that message evaluation
correlated with two variables, namely household size (r = -.21, p = .002) and income (r = -.25, p < .001 ). Intention significantly correlated with three control variables, gender (man were less inclined to cook in proportions than woman (Mman = 3.51, SD = 1.86; Mwoman = 4.59, SD = 1.94), r = -.13, p = .011), past measuring behaviour (r = .40 p < .001) and involvement with food waste (r = .22 p = .001). The dependent variable, potatoes measured correlated with three control variables, namely past measuring behaviour (r = .45 p < .001), income (r = -.15 p = .024) and involvement with food waste (r = .15 p = .027). Pasta measured only correlated significantly with past measuring behaviour (r = .46 p < .001). Finally, rice measured
correlated with past measuring behaviour (r = .43 p < .001) as well. All correlations are shown in Table 6.
Overall, based on these analyses six covariates may be relevant to include in the analyzes, namely gender, household size, age, income, past behaviour measured and involvement with food waste. Controlling for these covariates, however, did not affect the results of the different analysis, it was therefore decided not to include these variables and report the more parsimonious model without covariates.
Table 6
Pearson Product-moment Correlation Between Independent variables and Covariates
Measures Age Gender House-
hold com- position House- hold size Past Measuring behaviour Past discarding behaviour
Income Ethnicity Ethnicity father
Ethnicity mother
Eat box Message evaluation Intention Pota-toes meas-ured Pasta meas-ured Rice meas-ured Age Gender .12* Household composition -.01 .16** Household size .01 .16** .18**
Past measuring behaviour -.11* .04 .03 -.03
Past discarding behaviour -.04 -.03 .00 .01 .01
Income .02 .06 .29** .15** -.02 .02 Ethnicity -.07 .04 .03 .06 .11* -.05 -.07 Ethnicity father -.04 -.02 -.02 -.05 .10 -.04 -.06 .52** Ethnicity mother -.05 .00 -.01 .01 .07 -.02 .01 .53** .49** Eat box .01 -.08 -.02 -.01 -.06 .02 -.08 -.02 .01 -.07 Message evaluation .05 .05 -.03 -.21** .13 -.01 -.25** -.02 .05 -.06 .05 Intention -.13* .06 -.01 -.08 .40** -.06 -.07 .07 -.01 -.06 .03 .30** Potatoes measured -.11 -.08 -.01 .01 .45** -.08 -.15* .08 -.06 -.04 .07 .10 .28** Pasta measured -.06 .01 .05 -.08 .46** -.11 -.03 -.12 .00 -.11 -.05 .27** .38** .34** Rice measured .11 .13 .09 -.09 .43** -.10 .07 .02 -.05 -.08 .10 .16* .35** .26** .60** Note. *= p ≤ .05 **= p ≤ .01.
Main analyses Message evaluations
The ANOVA revealed there was no effect of condition on the evaluation of the message, F(4, 227) = .49, p = .747, η2 = .008. This means that there was no difference
between the evaluations of the descriptive norm (M = 2.78, SD = 1.92), control (M = 2.54, SD = 1.88), financial (M = 3.00, SD = 1.74), environmental (M = 2.96, SD = 1.83) or the social responsibility (M = 2.96, SD = 1.79) message. Therefore, in Study 2 no evidence was found for Hypothesis 1.
Effect of condition on intention to cook in proportion
The ANOVA demonstrated there was no effect of condition on intention to cook in proportions, F(4, 366) = .28, p = .893, η2 = .003. Thus, the descriptive norm message (M = 4.68, SD = 1.86) did not have a stronger effect on intention to cook in proportions, compared to the other messages (Mfinancial= 4.56, SD = 1.82, Menvironmental= 5.12, SD = 1.78, Msocial responsibility = 4.38, SD = 2.04, Mcontrol = 4.52, SD = 1.94). Hypothesis 3 was therefore not supported.
Effect of condition on interest in buying a tool to measure food proportions
Results of a Chi-square test indicated there was a significance difference between the conditions on interest in buying a tool to measure food proportions, X2(4) = 20.00, p < .001. Participant in the social responsibility (46.15%) and environmental condition (38.18%) were more willing to buy the tool, than participant in the descriptive norm (12.77%), financial (24.59%) or control condition (17.31%). Further analyses showed that the social
responsibility condition significantly differed from the control condition, X2(1) = 9.99, p = .002, and the environmental condition differed from the control condition as well, X2(1) = 7.44, p = .006. It seems that an environmental or social responsibility message is most
effective in arousing interest to buy a tool to measure food proportions. Once again, hypothesis 3 was not supported.
Effect of condition on cooking in proportions behavior
To investigate whether participants in the descriptive norm conditions measured their potatoes, pasta and rice more often than participants in the other conditions, a MANOVA was used. Results showed that there was no effect of conditions on cooking in proportions, F(3, 142) = 75.17, p = .718; Wilks’ Lambda = .94; η2 = .02. When the results for the dependent variables were considered separately, no significant differences between the conditions on potatoes (Welch’s F(4, 106) = .56, p = .695), pasta, (F(4, 144) = .51, p = .729, η2 = .014) and rice measured (F(4, 144) = .12, p = .975, η2 = .003) were found. Means and SDs per product measured in each condition are shown in Table 7.
Table 7
Means and Standard Deviations Product Measured per Condition
In sum, the results of Study 2 demonstrated that a descriptive norm message was not perceived as less effective in motivating people to cook in proportions, compared to the other messages. Therefore, Hypothesis 1 was not supported. Furthermore, a descriptive norm message neither had a stronger effect on intention to cook in proportions nor activated people Potatoes Pasta Rice
measured measured measured
Message M SD M SD M SD Descriptive norm 10.48 30.12 44.62 48.95 60.22 49.00 Financial 15.52 33.31 43.10 49.50 55.17 48.82 Environmental 22.06 39.30 52.94 47.76 61.76 49.33 Social responsibility 22.12 38.94 38.46 49.61 62.50 48.61 Control 8.62 26.96 53.45 49.88 56.90 49.50
more to buy a tool to measure food proportions, compared to the other messages. The environmental and social responsibility message did appear to stimulate people more to buy the tool. Lastly, descriptive norms did not stimulate participants to cook in proportions more often, than the other messages. Hence, Hypothesis 3 was not confirmed.
Discussion
Inconsistent with the findings of Study 1, the descriptive norm message was not perceived as less motivating, than the other messages. Furthermore, a descriptive norm
message did not have a stronger effect on intention to cook in proportions, interest in buying a tool to measure food proportions and cooking in proportions behaviour, than a financial, environmental, social responsibility or control message. Results even showed that a social responsibility message or environmental message may be more effective in activating people to buy a tool to measure food proportions. Thus, both Hypothesis 1 and 3 could not be
confirmed. Overall, the results of the second study are inconsistent with the findings of Nolan and colleagues, who did found an unconscious effect of descriptive normative information on energy saving behaviour.
A potential reason for this may be that the manipulation only seemed to have partially succeeded. In most conditions, except for the descriptive norm condition, people did not correctly answer the question related to the content of the message more often, than people who were not exposed to the message. Although it should be noted, that even though young parents in the descriptive norm conditions had a higher recall of the content of the descriptive norm message, compared to the people in the other conditions, this recall was still quite low (only 8.88% of the people gave the correct answer). An explanation for this finding could be that the manipulation check was performed seven days after people were exposed the
check, it is possible that young parents were not able to recall all the information
communicated in the message. In order to prevent that people would discover the actual purpose of the research, it was decided to do the manipulation check at Time 2 instead of Time 1. Potential solutions for this partially succeeded manipulation are further explained in the general discussion.
General discussion
The goal of the current study was to investigate the extent to which a descriptive norm message would affect household food waste related behaviour, compared to a financial, environmental, social responsibility and control messages, and whether people are actually aware of the influence of descriptive norms. First, it was hypothesized that descriptive norm and control messages would be perceived as less motivating, compared to financial,
environmental and social responsibility messages in promoting cooking in proportions. In contrast, it was expected that descriptive norm beliefs would have a stronger relationship with self-reported intention to cook in proportions, compared to financial, environmental and social responsibility beliefs. In other words, while it was expected that people would rate messages that describe the cooking behaviour of others as not influential for their own behaviour, that there would be a relationship between beliefs people have about what others do and what they themselves intend to do. Finally, it was hypothesized that a descriptive norm message would be more effective at stimulating young parents to cook in proportions, than the other messages.
First, to address the first two hypotheses, an online experiment and survey among young parents was conducted, investigating their evaluations of the different messages, beliefs about cooking in proportion and intention to cook in proportions. In line with
how many other young parents cook in proportions, as least influential in motivating them to cook in proportions. In contrast to the results regarding perceived message effectiveness, survey results showed that beliefs of how often other people cook in proportions revealed to be the only predictor of parents’ self-reported intention to cook in proportions. Thus, the first two hypotheses were supported. The first study therefore provides evidence of the
unconscious influence of descriptive norms on food waste behaviour.
The second study investigated the influence of a descriptive norm message compared to financial, environmental, social responsibility and information-only messages on perceived message effectiveness, intentions to cook in proportions and actual behaviour. For this
purpose, an experiment with a post-measurement of behavior one week later was conducted. In contrast to Study 1, the results of Study 2 showed that the descriptive norm message was not perceived as less motivating, compared to the other messages. Furthermore, a descriptive norm message did not have a stronger effect on intention to cook in proportions, interest in buying a tool to measure food proportions and parents’ behaviour to cook in proportions, than a financial, environmental, social responsibility or control message.
In sum, while the findings of the first study support the idea that people are in fact unaware of the influence of descriptive norms on their cooking in proportions behavior, these findings could not be corroborated by the second experimental study in which the influence of descriptive norm messages on message effectiveness, intentions and actual behaviour was investigated. The following sections will discuss potential explanations for the contradicting findings between the first and second study.
In agreement with the results of the study by Nolan et al. (2008), in Study 1,
participants when asked to rate five different messages on perceived persuasiveness rated a descriptive norm message as least influential in motivating them to cook in proportions,
