https://orcid.org/0000-0003-2787-8086
ABSTRACT
Meat consumption has detrimental environmental effects. Research shows that social norms are important when it comes to meat consumption. However, social norm interventions have shown mixed effects regarding their effectiveness for decreasing meat consumption. Therefore, an experiment was conducted (n = 279) with a 2 (baseline meat consumption: above- vs. below-average) x 3 (social norm feedback: descriptive norm only, descriptive plus injunctive norm, no feedback) x 3 (time: T0 [baseline], T1 [+1 week from baseline], T2 [+2 weeks from baseline]) mixed-factorial design. Results showed that reported changes in meat consumption at T1 and T2 relative to T0 were not different after receiving social norm feedback (i.e. descriptive norm only or descriptive plus injunctive norm) compared to receiving no feedback. Irrespective of the social norm feedback condition, participants reporting above-average meat consumption at baseline reduced their consumption, whereas those reporting below-average meat consumption at baseline increased their consumption over time. A plausible explanation for these findings may be statistical regression to the mean. Further understanding is needed of how social norm interventions may be used to reduce meat consumption.
Food systems have a large impact on the environment and contribute to environmental problems like biodiversity loss, water shortage, and climate change. Especially meat consumption plays a large role in causing these environmental problems (Poore & Nemecek, Citation2018). Switching to a plant-based diet is therefore often posited as a good way to counter environmental problems caused by food systems. For example, just halving dairy and meat consumption could already lower greenhouse gas emissions with 20–40% (Westhoek et al., Citation2014), and switching to a fully vegetarian diet could save up to 55% greenhouse gas emissions from food by 2050 (Tilman & Clark, Citation2014). One way to change the food systems is bottom-up, by having consumers choose more plant-based options. Even though it is a challenging task to motivate meat eaters to lower their meat consumption and adopt a more plant-based diet (Hoek et al., Citation2017), interventions can be used to incite such behavior change (Buttlar et al., Citation2021; Chang, Citation2021). Such interventions for adopting a more plant-based diet can make use of different focal factors for behavior change. A recent review classifies these factors into personal factors (e.g. knowledge, skills, habits), socio-cultural factors (e.g. social norms, culture), and external factors (e.g. political, economic factors), referring to social norms as one of the most important predictors of meat consumption (cf. Stoll-Kleemann & Schmidt, Citation2017).
Social norms are unwritten and shared rules for behavior in certain situations, that are accepted by the majority of a group (Cialdini & Goldstein, Citation2004). Research shows that these social norms influence behavior. On one side, conforming to the behavior of others has been identified as a part of the cause of many environmental problems, as a majority of people act in ways that harms the environment (e.g. driving, air travel pollution, meat consumption; Richter et al., Citation2018). On the other side, people’s innate tendency to conform to norms could be used for the benefit of the environment (Richter et al., Citation2018). That is, using social norms in interventions can encourage people to change their behavior in the desired direction (Cialdini et al., Citation1991; Goldstein et al., Citation2008; Klöckner, Citation2013). Given the promise of using social norms for the benefit of the environment, many interventions have been developed which provide social norm information as a primary tool for inciting environmentally friendly behavior, such as towel reuse in hotels, consumption of organic foods, energy conservation in public toilets, and curb side recycling (e.g. Bergquist & Nilsson, Citation2016; Goldstein et al., Citation2008; Schultz, Citation1999; Schultz et al., Citation2008).
When looking at reducing meat consumption specifically, research also shows the potential of social norms interventions, given that findings highlight perceived social norms as an important predictor of reducing meat consumption (e.g. Cheah et al., Citation2020; Schenk et al., Citation2018). Social norms may be a specifically potent intervention technique as people's desire to conform to norms may cause a more widespread adoption of a plant-based diet. A recent study on how behavioral factors affect a global diet change showed that social norms are the main factor in a widespread dietary change (Eker et al., Citation2019). Social norms have the potential to cause a positive feedback loop – the more people eat plant-based foods, the more it becomes the social norm to eat plant-based foods, which then causes even more people to eat plant-based foods. Whereas social norm interventions in particular have been suggested to be successful in lowering meat consumption (Godfray et al., Citation2018; Stea & Pickering, Citation2019), a recent literature review concluded that research concerning social norms interventions for reducing meat intake is scarce (Kwasny et al., Citation2022). Therefore, the current study investigates the potential of social norm interventions for reducing meat intake.
To this end, we will conceptually replicate an important and often cited (3.904 times; Google Scholar, June 9, 2022) study on social norms in the environmental domain which focused on electricity use (Schultz et al., Citation2007). This study provided insight into when norms have positive effects on behavior, when they may have negative effects and how negative effects may be prevented, thereby explaining prior mixed findings in this domain. This study has therefore provided an important contribution to the existing literature on social norms. However, a conceptual, partial replication by Verkooijen et al. (Citation2015) was unable to detect the same effects of social norms (albeit in a different context: (un)healthy snacking), as the behavior change in the experimental groups did not differ from changes observed in the control groups. This partial replication, however, did not include all experimental conditions. Therefore the question remains how effective social norm interventions are and what the best method is to study these interventions. By performing a methodologically sound conceptual replication of research into effects of social norm feedback we hope to add to the current evidence base on social norm theory and provide insights on the effectiveness of social norm interventions for decreasing meat consumption.
Theory
While social norm interventions appear to be a promising approach to change behavior, results from prior research are mixed. While some studies confirm desirable effects of the social norm approach (e.g. Burger & Shelton, Citation2011; Neighbors et al., Citation2004; Stea & Pickering, Citation2019; Thomas et al., Citation2017) other studies fail to observe behavior change or find mixed effects within studies (e.g. Granfield, Citation2005; Mollen et al., Citation2013; Thomas et al., Citation2016; Werch et al., Citation2000). Moreover, some studies have even observed a so-called boomerang effect, i.e. that social norm interventions lead to undesirable effects (e.g. Clapp et al., Citation2003; Richter et al., Citation2018; Wechsler et al., Citation2003).
Schultz et al. (Citation2007) offer a possible explanation for these mixed effects, based on the focus theory of normative conduct (Cialdini et al., Citation1991). In this theory, two types of norms are distinguished: descriptive norms and injunctive norms. The theory posits that – once salient – both types of social norms may affect behavior. While descriptive norms pertain to the behavior of a majority of people (e.g. “most people eat vegetarian at least twice a week”), injunctive norms describe what behavior most others approve or disapprove of (e.g. “most people approve of eating vegetarian at least twice a week”). Relevant with regard to descriptive norms is that they set a standard from which people do not want to deviate (Cialdini et al., Citation1991). In other words, people estimate the correctness of their behavior by how close or how far away they are from the descriptive norm (Cialdini et al., Citation1991). Descriptive norm feedback in a social norm intervention therefore serves as a point of reference for an individual’s behavior. It can act as a magnet for both the behavior above and below the communicated norm (Schultz et al., Citation2007). Thus, a social norm intervention may target and motivate people whose behavior is at an undesirable above-average level (e.g. people who consume an above-average quantity of meat) to decrease their undesired behavior. But it can also, inadvertently, have the opposite effect, i.e. the so-called boomerang effect, by driving people whose behavior is currently at a desirable below-average level (e.g. people who already consume a below-average quantity of meat) to increase the undesired behavior (Schultz et al., Citation2007). Together, this may explain why some studies find desirable effects, some find null or mixed effects, and others even find undesirable effects. Schultz et al. (Citation2007) proposed a potential solution to help prevent undesirable descriptive norm effects. Specifically, they argue that by supplementing the descriptive norm with the injunctive norm, the feedback not only signals to people that their behavior is below-average, but also that this is approved of by others. As a result, people may want to keep up this desired behavior rather than conform to the communicated descriptive norm, thereby preventing the boomerang effect (Schultz et al., Citation2007). In a longitudinal field experimental study on household energy consumption, Schultz et al. (Citation2007) empirically tested these predictions. First, they examined whether descriptive norm information indeed had a differential effect on behavior depending on whether participants consumed more or less energy than average – a desirable or undesirable (boomerang) effect, respectively. Second, they investigated whether adding injunctive norm feedback to descriptive norm feedback could prevent the undesired boomerang effect in the group that already performed at a desirable below-average level. As expected, their results showed that households that had an above-average energy consumption at baseline decreased their energy use over time after receiving descriptive norm feedback (i.e. information about their own energy consumption as well as the average energy consumption of households in their neighborhood). In contrast, households that had a below-average energy consumption at baseline and received descriptive norm feedback showed an increased energy consumption over time, indicating a boomerang effect. Importantly, this boomerang effect for households with below-average energy consumption was prevented by adding injunctive norm feedback that conveyed approval of the below-average energy consumption.
As the study by Schultz et al. (Citation2007) did not include control groups that received no social norm feedback, a potential alternative explanation for the observed descriptive norm effects (but not for the descriptive plus injunctive norm effects) over time could be regression to the mean. Regression to the mean is the statistical phenomenon that an extreme score at baseline tends to get closer to the mean at the second measurement (Barnett et al., Citation2004; Galton, Citation1886; Nesselroade et al., Citation1980). Participants with below-average scores at baseline tend to obtain higher scores over time, whereas participants with above-average scores at baseline tend to obtain lower scores over time. In other words, scores tend to naturally converge over time. Regression to the mean could similarly play a role in descriptive norm studies. So, if we would group participants into above-average and below-average meat consumption groups, provide them with descriptive norm feedback, and then track how their meat consumption changes over time, we might falsely conclude that participants in the above-average condition reduced their meat consumption because of the social norm message, and those in the below-average condition increased their meat consumption because of the message. Instead of the descriptive norm information, however, it is possible that regression to the mean has caused the scores to move naturally towards the mean over time. To be able to rule out regression to the mean as an explanation, the research design should include above-average and below-average control groups that receive no norm information.
In a partial conceptual replication of the study by Schultz et al. (Citation2007) in the domain of snacking behavior, Verkooijen et al. (Citation2015) therefore compared behavior change in above- and below-average descriptive norm feedback groups to behavior change in above- and below-average no feedback control groups. Including such control groups in the design allowed for testing whether the results in the experimental groups were in fact different from the results in the control groups – and are thus driven by the social norm feedback rather than being an artefact of a statistical phenomenon. However, results of this study showed that the descriptive norm intervention did not result in more behavior change compared to the no feedback control groups. More specifically, in two experiments they found that participants who consumed more snacks than average at baseline (i.e. Exp. 1: healthy snacks, i.e. fruit, Exp. 2: unhealthy snacks, e.g. cookies, crisps) reduced their intake at the follow-up measurement, irrespective of whether they received the descriptive norm feedback or no feedback. The opposite effect was found for below-average consumption in Experiment 1. That is, they found that participants who consumed below-average quantities of healthy snacks at baseline increased their healthy snack intake at follow-up, irrespective of whether they received descriptive norm feedback or no feedback. In Experiment 2, no increase in unhealthy snack intake was observed from baseline to follow-up: not in the descriptive norm condition and neither in the no feedback control condition. These results suggest that social norm interventions might not always be so effective, but provide an indication that regression to the mean might indeed be an alternative explanation for the observed changes in some social norm feedback research using longitudinal designs. As the social norm approach is often used to stimulate environmental behavior change (for a review, see Cialdini & Jacobson, Citation2021) and meat consumption behavior specifically (e.g. Raghoebar et al., Citation2020; Stea & Pickering, Citation2019) it is important to further investigate the replicability of previous findings, for both science and practice. We therefore test the effectiveness of a social norm intervention of meat consumption by conducting a conceptual replication with the inclusion of no feedback control groups.
Current study
The current study sets out to conceptually replicate the study by Schultz et al. (Citation2007) in the context of meat consumption. Even though the results by Verkooijen et al. (Citation2015) suggest that statistical regression to the mean might explain previously found effects of Schultz et al. (Citation2007) regarding descriptive norm feedback, this explanation cannot account for the findings of Schultz et al. (Citation2007) regarding descriptive plus injunctive norm feedback. More specifically, Schultz et al. (Citation2007) found that the energy consumption of households with a below-average consumption at baseline did not move toward the mean when the descriptive norm was supplemented with the injunctive norm. To gain further understanding of social norm feedback effects, the current study tests the effect of descriptive as well as injunctive norm feedback. More specifically, it includes above-average and below-average groups who receive descriptive norm feedback, descriptive norm plus injunctive norm feedback, or no feedback. We formulated the following expectation: People whose meat consumption is above the mean at baseline will reduce their intake after exposure to descriptive norm feedback. In contrast, people whose meat consumption is below the mean at baseline will increase their intake after exposure to descriptive normative feedback (i.e. boomerang effect). This increase is not expected when descriptive norm feedback is supplemented with injunctive norm feedback. No changes in meat consumption are expected in the above-average and below-average no feedback control conditions.
Method
Design and participants
The study employed a 2 (baseline meat consumption: above- vs. below-average) x 3 (social norm feedback: descriptive norm only, descriptive plus injunctive norm, no feedback) x 3 (time: T0 [baseline], T1 [1 week from baseline], T2 [2 weeks from baseline]) mixed-factorial design. The between-participants factors were baseline consumption and social norm feedback. The within-participant factor was time. Data collection took place in November 2018. Participants in this experiment were adults (≥ 18 years old) living in the Netherlands who were not restricted in their meat consumption. People who did not eat meat for any reason could not participate. Participants were recruited using convenience and snowball sampling through the network of one of the researchers and through the university’s participant pool. In total, 339 participants completed the study. However, several participants had to be removed from the dataset due to an error causing them to receive the wrong social norm feedback (n = 27). Furthermore, some participants were excluded because they were under the age of 18 (n = 5), because they did not pass the manipulation check (n = 23), because they failed to complete at least two of the three meat consumption questionnaires each week (n = 4) or because they did not complete the final questionnaire (n = 1). The final sample therefore comprised 279 participants (71.7% female), with a mean age of 27.47 (SD = 10.49; range 18–79). As for education, 4.8% of the participants had a lower education level (e.g. primary school, lower vocational education), 32.0% had middle education level (e.g. secondary general education), and 63.2% of the participants had a higher education level (i.e. 25.3% had a higher professional education level and 37.9% had a Master’s degree level or equivalent). The required sample size was not predetermined but we attempted to recruit as many participants as possible within the available time frame. However, after data collection, a power analysis in G*power 3.1.9.4 was performed to check whether our obtained sample size was sufficient to detect a medium effect (f = .25; based on Schultz et al., Citation2007), given the lowest observed correlation in meat consumption between measurements (r = .55), an alpha of 0.05, and power of 0.95. This analysis showed that a minimum of 66 participants would be required to detect interactions with a medium effect size. Given that the sample consisted of 279 participants our study had sufficient power to detect a medium effect, and even a rather small effect (f = .12).
Procedure
Data were collected online, over a period of three weeks (T0, T1, T2). After first completing a short questionnaire assessing demographic characteristics (age, gender, and educational level), participants received three e-mails each week (at T0, T1, and T2 on Mondays, Tuesdays, and Wednesdays) with a link to a questionnaire to assess daily meat consumption. At the end of the baseline week (T0, on Saturday), participants in the descriptive norm only and the descriptive plus injunctive norm feedback conditions received an additional e-mail that contained the social norm feedback (see “Manipulation”). Participants in the no feedback control conditions did not receive this e-mail. For all participants, the final questionnaire (on Wednesday at T2) ended with questions assessing several control variables (see “Measures”). In exchange for participation, participants received course credit, or they could sign up for a raffle to win one of five web shop gift vouchers of €10. The study was approved by the Ethics Review Board of the Faculty of Social and Behavioral Science of the University of Amsterdam and informed consent was obtained from all participants at the beginning of the experiment. The questionnaires and materials used for this study, as well as the dataset and syntax can be found on Open Science Framework (osf.io/j65r2/).
Manipulation
Participants were randomly assigned to the descriptive norm only condition, the descriptive plus injunctive norm condition, or the no feedback control condition. In the descriptive norm only condition and the descriptive plus injunctive norm condition, participants received either “above-average” or “below-average” feedback, depending on their own meat consumption (frequency per day) at baseline (T0) relative to the average meat consumption of all participants who completed the baseline measurement (M = 1.32, SD = 0.67; 0–4 scale). In the final sample, 83 participants received descriptive norm only feedback (nabove-average = 42 and nbelow-average = 41), 87 participants received descriptive plus injunctive norm feedback (nabove-average = 49 and nbelow-average = 38), and 109 participants received no feedback (nabove-average = 63 and nbelow-average = 46). The social norm feedback consisted of information about the average meat consumption of all participants at baseline: “The results of last week show that the average meat consumption of Dutch residents was: 1.32 times per day.”. In addition, participants in the descriptive norm only conditions received descriptive norm feedback: “Based on your answers, it appears that your meat consumption is above-/below-average compared to the average Dutch resident.”. Participants in the descriptive plus injunctive norm conditions received the descriptive norm feedback plus the injunctive norm information: “Based on your answers, it appears that your meat consumption is above-/below-average compared to the average Dutch resident. How you are doing: 
”. The injunctive norm information was displayed by either a positively or negatively valenced emoticon to convey the feeling of approval of others of below-average consumption and disapproval of others of above-average consumption, in line with Schultz et al. (Citation2007).
Measures
Meat consumption
During the three weeks of data collection, participants were asked to report their meat consumption on Mondays, Tuesdays, and Wednesdays. Similar to Verkooijen et al. (Citation2015), we measured the target behavior on three consecutive weekdays. We decided on Mondays, Tuesdays, and Wednesdays because people have typically fewer social gatherings on those days, and are therefore more likely to decide for themselves what is on the menu. E-mails with a link to the questionnaire were sent at the end of the day, around 8 PM. Participants were first asked to read the following text: “For this study, we are interested in how often people consume meat. With meat, we mean all kinds of meat, in all kinds of quantities. For instance, having a slice of chicken filet on your sandwich during breakfast or lunch or having a cracker with some pâté as a snack. But also, of course, having half a chicken or some diced ham in the pasta”. Then, participants were asked to answer the following four questions with “Yes” or “No”: “This morning during breakfast, I have eaten meat”, “This afternoon during lunch, I have eaten meat”, “This evening during dinner, I have eaten meat”, and “Today, I have eaten a snack that contained meat”. Participants received a reminder if they had not completed the questionnaire before 11 AM the next day. For the analyses, an average daily meat consumption (frequency) was calculated by taking the sum of the reported number of meals and snacks containing meat divided by the number of days participants reported their meat consumption. As a result, scores could range from 0 (no meat consumption) to 4 (meat consumption four times a day). Meat consumption scores were calculated separately for T0 (baseline), T1, and T2.
Manipulation check
To check whether participants had read the social norm feedback, a link to a questionnaire was included in the social norm feedback e-mail. In this questionnaire, participants were asked: “Was your meat consumption of last week above- or below-average compared to other Dutch residents?”. The answer options included: “Above-average”, “Below-average”, and “I don’t know”. Participants who did not answer this question, answered it incorrectly, or answered “I don't know”, were removed from the analyses (see “Design and participants”).
Control variables
Prior research has shown that the effectiveness of social norm feedback is influenced by the extent to which participants identify with the referent group (e.g. Stok et al., Citation2012; Terry & Hogg, Citation1996). For participants in the descriptive and the descriptive plus injunctive feedback conditions, identification with Dutch residents was therefore measured, with two items by Stok et al. (Citation2012): “I identify with Dutch residents” and “I feel a connection to Dutch residents”, both measured on a Likert-scale ranging from 1 (Strongly disagree) to 7 (Strongly agree). The two items were averaged to create one measure of identification (r = .83; M = 5.26, SD = 1.54). Additionally, all participants were asked to indicate to what extent they could decide for oneself what to eat. This question was included as it is likely that participants who do not decide for themselves what they eat, because they for instance do not prepare (all) meals themselves, may be less influenced by the social norm feedback. The ability to decide for oneself what to eat was measured with one statement, assessed on a Likert-scale ranging from 1 (Strongly disagree) to 7 (Strongly agree): “I decide for myself what I eat during the day.” (M = 6.17, SD = 1.16).Footnote1
Results
Randomization checks
To determine whether the demographic characteristics and control variables were equally distributed across the six feedback conditions, randomization checks were performed. Pearson’s chi-squared tests and one-way ANOVAs showed no significant differences between conditions regarding gender, χ² (5, 279) = 9.80, p = .081, age, F (5, 279) = 1.33, p = .251, educational level, χ² (35, 269) = 34.48 p = .493, identification with Dutch residents, F (3, 170) = 0.15, p = .928, and the ability to decide for oneself what to eat, F (5, 279) = 1.64, p = .148. Because the p-value for gender was marginally significant, the main analyses were also conducted with gender as control variable. However, this did not change the conclusion regarding significance of any of the main or interaction effects (tested at p < .05). Additionally, a one-way ANOVA confirmed that baseline meat consumption did not differ between the three social norm feedback conditions (i.e. descriptive, descriptive plus injunctive, no feedback), F (2, 279) = 0.40, p = .670.
Main analyses
A mixed-design ANOVA was conducted to examine the effects of social norm feedback on meat consumption. A three-way interaction between the baseline meat consumption, social norm feedback, and time on meat consumption was hypothesized. More specifically, it was expected that after receiving descriptive social norm feedback, participants with an above-average meat consumption at baseline (T0) would reduce their meat consumption over time (at T1 and T2), whereas participants with a below-average meat consumption at baseline would increase their meat consumption. This increase in meat consumption for the latter group was not expected when supplementing the descriptive norm feedback with injunctive norm feedback. No changes in meat consumption were expected in the no feedback control conditions. In contrast to these predictions, however, no three-way interaction effect between baseline meat consumption, social norm feedback, and time was found, F (4, 546) = 1.56, p = .184, ηp² = .01 (see and ). The test did reveal a significant main effect of baseline meat consumption, F (1, 273) = 242.75, p < .001, ηp² = .47, showing that, across all measurements (T0, T1, and T2 averaged), participants with an above-average meat consumption at baseline reported higher amounts of meat consumption than participants with a below-average meat consumption at baseline. There were no significant main effects of social norm feedback, F (2, 273) = 0.54, p = .585, ηp² < .01, or time, F (2, 546) = 0.88, p = .417, ηp² < .01. Furthermore, in contrast to our predictions, a significant two-way interaction effect between baseline meat consumption and time was found, F (2, 546) = 30.94, p < .001, ηp² = .10. Pairwise comparisons showed that participants with an above-average meat consumption at baseline (T0) decreased their meat consumption at T1 (Mdiff = −0.27, p < .001) and T2 (Mdiff = −0.25, p < .001), whereas participants with a below-average meat consumption at baseline (T0) increased their meat consumption at T1 (Mdiff = 0.20, p < .001) and T2 (Mdiff = 0.26, p < .001). This means that irrespective of the social norm feedback condition, participants moved toward the average over time. There was no significant two-way interaction effect between baseline meat consumption and social norm feedback, F (2, 273) = 0.07, p = .935, ηp² < .01, nor between social norm feedback and time, F (4, 546) = 1.42, p = .227, ηp² = .01.
Figure 1. Frequency of daily meat consumption at T0 (baseline), T1 (1 week from baseline), and T2 (2 weeks from baseline).
Table 1. Frequency of daily meat consumption at T0 (baseline), T1 (1 week from baseline), and T2 (2 weeks from baseline)
Discussion
With the current study, we aimed to test the effectiveness of a social norm intervention for decreasing meat consumption. Based on the focus theory of normative conduct and in line with Schultz et al. (Citation2007), we expected that descriptive norm feedback (i.e. information on one’s behavior compared to what most others do) would reduce people’s meat consumption over time for those with an above-average meat consumption at baseline, and would increase people’s meat consumption over time for those with a below-average meat consumption at baseline (i.e. boomerang effect). We further predicted that the undesirable increase in meat consumption for those with a below-average meat consumption at baseline could be prevented by supplementing the descriptive norm feedback with injunctive norm feedback (i.e. information on what others approve of). Finally, in the no feedback control conditions, no changes in meat consumption over time were expected. The results, however, were not in line with these expectations. Even though social norm feedback containing the descriptive norm resulted in a decrease in meat consumption for people with an above-average baseline meat consumption and in an increase for people with a below-average baseline meat consumption, the same pattern was found in the descriptive plus injunctive norm feedback conditions, as well as in the no feedback control conditions. In other words, regardless of receiving (different types of) feedback, above-average baseline scores tended to decrease, whereas below-average baseline scores tended to increase over time. Rather than resulting from social norm feedback, these findings seem to be more in line with the alternative explanation brought forward by Verkooijen et al. (Citation2015), namely regression to the mean.
Implications, limitations, and future research
With regard to the findings on the descriptive norm feedback, the original research by Schultz et al. (Citation2007) reported changes in energy consumption in line with the descriptive norm feedback that communicated whether one’s behavior was above or below the average energy consumption behavior. However, both the current study as well as the research by Verkooijen et al. (Citation2015) found similar changes in the no feedback control groups. An alternative explanation for the reported changes in meat consumption over time in the current study could thus be regression to the mean. It is possible that the observed changes over time in some prior research – such as in the study conducted by Schultz et al. (Citation2007) – on descriptive norm messages are similarly due to regression to the mean. This does not hold, however, for the descriptive plus injunctive norm feedback, as Schultz et al. (Citation2007) found that an undesirable increase in meat consumption in the below-average descriptive norm group could be prevented by supplementing descriptive norm feedback with an injunctive norm message (i.e. to indicate approval of the below-average meat consumption). In the current study, the findings were different. Similar to receiving descriptive norm feedback and receiving no feedback, those reporting above-average meat consumption at baseline reduced their consumption, whereas those reporting below-average meat consumption at baseline increased their meat consumption over time. Thus, regression to the mean seems the most plausible explanation for the findings in the current study, but it cannot account for the effects found by Schultz et al. (Citation2007) regarding the descriptive plus injunctive norm feedback. Our study was designed for testing the replicability of the social norm effect in the context of meat consumption, but in future research it would be interesting to specifically test the regression to the mean effect. Furthermore, the current study underscores the importance of adding control groups when investigating effects of interventions, such as social norms feedback, in longitudinal designs. Such control groups are necessary for ruling out alternative explanations related to time, such as regression to the mean.
It should be noted that there are differences between the study by Schultz et al. (Citation2007) and the current study, which might explain differential findings and provide avenues for future research. For instance, in the original study by Schultz et al. (Citation2007) as well as in the current study, the injunctive norm feedback was provided by means of a positively or negatively valenced emoticon that expressed approval of below-average consumption and disapproval of above-average consumption. However, whereas the feedback in Schultz et al. (Citation2007) was handwritten (i.e. the emoticons were drawn on doorhangers that were placed on residents’ doors), in the current study this feedback was provided digitally (i.e. via an e-mail). Potentially, handwritten feedback provides a stronger signal of (dis)approval by implicitly communicating that one’s behavior is being watched and judged by another person. In contrast, the digital feedback in the current study could be perceived as automated, without someone monitoring the behavior, and might therefore be less successful in effecting behavior change. This idea that the feeling of being watched might enhance the effectiveness of norm messages is corroborated by previous experimental research, showing that cues suggesting that one is being watched (i.e. pictures of eyes) increases prosocial behavior (Bateson et al., Citation2006; Haley & Fessler, Citation2005). An interesting avenue for future research would therefore be to experimentally investigate whether descriptive plus injunctive norm messages are more effective when feedback on (dis)approval of the pro-environmental behavior is presented in a way that is perceived as more personal (vs. automated).
Another, more prominent difference between the studies is the target behavior. Schultz et al. (Citation2007) investigated energy consumption, whereas both the current study and the research by Verkooijen et al. (Citation2015) focused on food consumption. Although both pro-environmental behaviors, energy consumption and food consumption have different behavioral attributes. Energy consumption is a less socially visible behavior, in the sense that it takes place in the privacy of one’s home. Eating behavior is much more visible, as many meals are consumed in the company of others (Yates & Warde, Citation2017). Whereas research shows that social norms are generally more important for socially visible behaviors (Lapinski & Rimal, Citation2005), the influence of social norm messages may be greater for less socially visible behaviors. People may be more certain about their existing social norm perceptions of behavior that is more apparent (such as meat consumption) but less so about social norms regarding less visible behavior (electricity use). As a result, one’s perception of the social norm around energy consumption may be shaped by a social norm message to a greater extent compared to one’s perception of the social norm regarding meat consumption. Whether this is indeed the case requires further study.
Furthermore, because of the visibility of meat consumption, the communicated social norm message might not match with perceptions that people have of this norm and could therefore fall outside of the latitude of acceptance. If, for example, the social norm message states that on average meat is consumed twice a day, but what people observe is that most others eat meat four times a day, people may experience a large discrepancy between the message and their own perceptions. As a consequence, the effects of social norm messages might be attenuated or nullified. For future research into the effects of social norm messages on meat consumption it might therefore be interesting to take into account the behavior of the social network of people, to see whether this might influence the effectiveness of social norm messages. Related to this, in the current study, people identified quite (but not very) strongly with the referent group. Given that research shows that such a social identification is important when it comes to the effectiveness of social norm messages (Rimal & Lapinski, Citation2015; Stok et al., Citation2012), it would be interesting to test whether the effectiveness of social norm messages could be strengthened by using a referent group that people feel even more closely connected to. The potential directions for future research as suggested here would provide further insight into the effectiveness of social norm feedback and the specific conditions, such as the context and the type of behavior, that are needed for this feedback to be effective.
A limitation of the current study is that meat consumption was assessed with a subjective, self-report measure: participants had to report their meat consumption in retrospect at the end of the day. Using self-report could have caused a (socially desirable) bias in reporting meat consumption. To omit such a potential bias, future studies should aim for more objective measures of meat consumption. Moreover, only the frequency of meat consumption was measured – participants answered with “Yes” or “No” to the questions on whether they ate meat during breakfast, lunch, dinner, and snack time. A more precise measure of meat consumption may be used in future research. That is, in addition to the frequency of meat consumption, the quantity of meat consumption could be measured (e.g. by assessing intake in grams).
When it comes to stimulating more plant-based food consumption and less meat eating, it is probably most effective to design interventions that target multiple factors. As stated earlier, meat consumption is influenced by personal, socio-cultural, and external factors (cf. Stoll-Kleemann & Schmidt, Citation2017). Interventions that target several of these factors, thereby taking a more holistic approach to behavior change, are likely more successful than interventions targeting only social norms (cf. Michie et al., Citation2011). In the case of reducing meat eating, for instance, information provision and implementation intentions on meat-eating habits have proven to be effective (Harguess et al., Citation2020; Kwasny et al., Citation2022; Rees et al., Citation2018). Targeting such personal factors alongside a social norm message (a socio-cultural factor), and possibly an external factor such as increasing the accessibility of plant-based alternatives may likely improve the effectiveness of the intervention.
Conclusion
In sum, this study adds relevant insights to the social norm literature and has considerable practical implications. Even though perceived social norms are shown to be a strong predictor of (pro-environmental) behavior (Cialdini et al., Citation1991; Goldstein et al., Citation2008; Klöckner, Citation2013), this is no guarantee for social norm interventions to effectively influence the norm perception and subsequent pro-environmental behavior. As such, the current study complements the existing literature on the mixed effects of social norm interventions (e.g. Mollen et al., Citation2013; Richter et al., Citation2018; Stea & Pickering, Citation2019; Thomas et al., Citation2017; Werch et al., Citation2000). When promoting pro-environmental behavior such as meat consumption in practice, one should be aware that social norms interventions might not always result in desirable behavior change and that there are likely boundary conditions that influence the effectiveness of these interventions, such as the type of behavior (e.g. more vs. less visible pro-environmental behaviors) or the specific way the feedback is presented (e.g. handwritten vs. digital feedback). Future research should therefore aim to untangle the key ingredients for effective social norm interventions, as well as other types of interventions, to accelerate the transition into more sustainable eating patterns.
Financial support
The research was supported by a Dutch Research Council (NWO) grant awarded to the second author (grant number Veni.201S.075).
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Several additional variables were measured for exploratory purposes. First, following Verkooijen et al. (Citation2015) we asked whether participants were surprised about their meat consumption being above or below the average of Dutch residents’ meat consumption (measured at T0, after the manipulation check) as a means to focus the participants’ attention on the feedback. Second, in the descriptive and the descriptive plus injunctive feedback conditions, we measured resistance to the feedback, credibility of the feedback, and whether participants discussed the feedback with others (measured in the final questionnaire, after the final measurement of meat consumption at T2). Third, in the no feedback control conditions, participants were asked to estimate the average meat consumption of Dutch residents and to indicate whether they thought their meat consumption in the last three weeks was above or below this average (measured, after the final measurement of meat consumption at T2). However, none of these variables provided relevant additional insight or explanation in light of the main findings, so they are not further discussed here. Data on these additional variables can be found on Open Science Framework (osf.io/j65r2/).
References
- Barnett, A. G., Van der Pols, J. C., & Dobson, A. J. (2004). Regression to the mean: What it is and how to deal with it. International Journal of Epidemiology, 34(1), 215–220. https://doi.org/10.1093/ije/dyh299
- Bateson, M., Nettle, D., & Roberts, G. (2006). Cues of being watched enhance cooperation in a real-world setting. Biology Letters, 2(3), 412–414. https://doi.org/10.1098/rsbl.2006.0509
- Bergquist, M., & Nilsson, A. (2016). I saw the sign: Promoting energy conservation via normative prompts. Journal of Environmental Psychology, 46, 23–31. https://doi.org/10.1016/j.jenvp.2016.03.005
- Burger, J. M., & Shelton, M. (2011). Changing everyday health behaviors through descriptive norm manipulations. Social Influence, 6(2), 69–77. https://doi.org/10.1080/15534510.2010.542305
- Buttlar, B., Rothe, A., Kleinert, S., Hahn, L., & Walther, E. (2021). Food for thought: Investigating communication strategies to counteract moral disengagement regarding meat consumption. Environmental Communication, 15(1), 55–68. https://doi.org/10.1080/17524032.2020.1791207
- Chang, C. (2021). Effects of responsibility appeals for pro-environmental ads: When do they empower or generate reactance? Environmental Communication, 15(4), 546–569. https://doi.org/10.1080/17524032.2021.1876132
- Cheah, I., Shimul, A. S., Liang, J., & Phau, I. (2020). Drivers and barriers toward reducing meat consumption. Appetite, 149, 104636. https://doi.org/10.1016/j.appet.2020.104636
- Cialdini, R. B., & Goldstein, N. J. (2004). Social influence: Compliance and conformity. Annual Review of Psychology, 55(1), 591–621. https://doi.org/10.1146/annurev.psych.55.090902.142015
- Cialdini, R. B., & Jacobson, R. P. (2021). Influences of social norms on climate change-related behaviors. Current Opinion in Behavioral Sciences, 42, 1–8. https://doi.org/10.1016/j.cobeha.2021.01.005
- Cialdini, R. B., Kallgren, C. A., & Reno, R. R. (1991). Advances in experimental social psychology. Advances in Experimental Social Psychology, 24, 201–234. https://doi.org/10.1016/S0065-2601(08)60330-5
- Clapp, J. D., Lange, J. E., Russell, C., Shillington, A., & Voas, R. B. (2003). A failed norms social marketing campaign. Journal of Studies on Alcohol, 64(3), 409–414. https://doi.org/10.15288/jsa.2003.64.409
- Eker, S., Reese, G., & Obersteiner, M. (2019). Modelling the drivers of a widespread shift to sustainable diets. Nature Sustainability, 2(8), 725–735. https://doi.org/10.1038/s41893-019-0331-1
- Galton, F. (1886). Regression towards mediocrity in hereditary stature. The Journal of the Anthropological Institute of Great Britain and Ireland, 15, 246–263. https://doi.org/10.2307/2841583
- Godfray, H. C. J., Aveyard, P., Garnett, T., Hall, J. W., Key, T. J., Lorimer, J., Pierrehumbert, R. T., Scarborough, P., Springmann, M., & Jebb, S. A. (2018). Meat consumption, health, and the environment. Science, 361(6399), eaam5324. https://doi.org/10.1126/science.aam5324
- Goldstein, N. J., Cialdini, R. B., & Griskevicius, V. (2008). A room with a viewpoint: Using social norms to motivate environmental conservation in hotels. Journal of Consumer Research, 35(3), 472–482. https://doi.org/10.1086/586910
- Granfield, R. (2005). Alcohol use in college: Limitations on the transformation of social norms. Addiction Research & Theory, 13(3), 281–292. https://doi.org/10.1080/16066350500053620
- Haley, K. J., & Fessler, D. M. T. (2005). Nobody's watching? Subtle cues affect generosity in an anonymous economic game. Evolution and Human Behavior, 26(3), 245–256. https://doi.org/10.1016/j.evolhumbehav.2005.01.002
- Harguess, J. M., Crespo, N. C., & Hong, M. Y. (2020). Strategies to reduce meat consumption: A systematic literature review of experimental studies. Appetite, 144, 104478. https://doi.org/10.1016/j.appet.2019.104478
- Hoek, A. C., Pearson, D., James, S. W., Lawrence, M. A., & Friel, S. (2017). Shrinking the food-print: A qualitative study into consumer perceptions, experiences and attitudes towards healthy and environmentally friendly food behaviours. Appetite, 108, 117–131. https://doi.org/10.1016/j.appet.2016.09.030
- Klöckner, C. A. (2013). A comprehensive model of the psychology of environmental behaviour—A meta-analysis. Global Environmental Change, 23(5), 1028–1038. https://doi.org/10.1016/j.gloenvcha.2013.05.014
- Kwasny, T., Dobernig, K., & Riefler, P. (2022). Towards reduced meat consumption: A systematic literature review of intervention effectiveness, 2001–2019. Appetite, 168, 105739. https://doi.org/10.1016/j.appet.2021.105739
- Lapinski, M. K., & Rimal, R. N. (2005). An explication of social norms. Communication Theory, 15(2), 127–147. https://doi.org/10.1111/j.1468-2885.2005.tb00329.x
- Michie, S., Van Stralen, M. M., & West, R. (2011). Individual determinants of research utilization by nurses: A systematic review update. Implementation Science, 6(1), 1–12. https://doi.org/10.1186/1748-5908-6-1
- Mollen, S., Rimal, R. N., Ruiter, R. A. C., & Kok, G. (2013). Healthy and unhealthy social norms and food selection. Findings from a field-experiment. Appetite, 65, 83–89. https://doi.org/10.1016/j.appet.2013.01.020
- Neighbors, C., Larimer, M. E., & Lewis, M. A. (2004). Targeting misperceptions of descriptive drinking norms: Efficacy of a computer-delivered personalized normative feedback intervention. Journal of Consulting and Clinical Psychology, 72(3), 434–447. https://doi.org/10.1037/0022-006X.72.3.434
- Nesselroade, J. R., Stigler, S. M., & Baltes, P. B. (1980). Regression toward the mean and the study of change. Psychological Bulletin, 88(3), 622–637. https://doi.org/10.1037/0033-2909.88.3.622
- Poore, J., & Nemecek, T. (2018). Reducing food’s environmental impacts through producers and consumers. Science, 360(6392), 987–992. https://doi.org/10.1126/science.aaq0216
- Raghoebar, S., van Kleef, E., & de Vet, E. (2020). Increasing the proportion of plant-based foods available to shift social consumption norms and food choice among non-vegetarians. Sustainability, 12(13), 5371. https://doi.org/10.3390/su12135371
- Rees, J. H., Bamberg, S., Jäger, A., Victor, L., Bergmeyer, M., & Friese, M. (2018). Breaking the habit: On the highly habitualized nature of meat consumption and implementation intentions as one effective way of reducing it. Basic and Applied Social Psychology, 40(3), 136–147. https://doi.org/10.1080/01973533.2018.1449111
- Richter, I., Thøgersen, J., & Klöckner, C. A. (2018). A social norms intervention going wrong: Boomerang effects from descriptive norms information. Sustainability, 10(8), 2848. https://doi.org/10.3390/su10082848
- Rimal, R. N., & Lapinski, M. K. (2015). A re-explication of social norms, ten years later. Communication Theory, 25(4), 393–409. https://doi.org/10.1111/comt.12080
- Schenk, P., Rössel, J., & Scholz, M. (2018). Motivations and constraints of meat avoidance. Sustainability, 10(11), 3858. https://doi.org/10.3390/su10113858
- Schultz, P. W. (1999). Changing behavior with normative feedback interventions: A field experiment on curbside recycling. Basic and Applied Social Psychology, 25–36. https://doi.org/10.1207/s15324834basp2101_3
- Schultz, P. W., Nolan, J. M., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V. (2007). The constructive, destructive, and reconstructive power of social norms. Psychological Science, 18(5), 429–434. https://doi.org/10.1111/j.1467-9280.2007.01917.x
- Schultz, P. W., Tabanico, J. J., & Rendón, T. (2008). Normative beliefs as agents of influence: Basic processes and real-world applications. In W. D. Crano, & R. Prislin (Eds.), Attitudes and attitude change (pp. 385–409). Psychology Press.
- Stea, S., & Pickering, G. J. (2019). Optimizing messaging to reduce red meat consumption. Environmental Communication, 13(5), 633–648. https://doi.org/10.1080/17524032.2017.1412994
- Stok, F. M., de Ridder, D. T. D., de Vet, E., & de Wit, J. B. F. (2012). Minority talks: The influence of descriptive social norms on fruit intake. Psychology & Health, 27(8), 956–970. https://doi.org/10.1080/08870446.2011.635303
- Stoll-Kleemann, S., & Schmidt, U. J. (2017). Reducing meat consumption in developed and transition countries to counter climate change and biodiversity loss: A review of influence factors. Regional Environmental Change, 17(5), 1261–1277. https://doi.org/10.1007/s10113-016-1057-5
- Terry, D. J., & Hogg, M. A. (1996). Group norms and the attitude-behavior relationship: A role for group identification. Personality and Social Psychology Bulletin, 22(8), 776–793. https://doi.org/10.1177/0146167296228002
- Thomas, J. M., Liu, J., Robinson, E. L., Aveyard, P., Herman, C. P., & Higgs, S. (2016). The effects of liking norms and descriptive norms on vegetable consumption: A randomized experiment. Frontiers in Psychology, 7, 442. https://doi.org/10.3389/fpsyg.2016.00442
- Thomas, J. M., Ursell, A., Robinson, E. L., Aveyard, P., Jebb, S. A., Herman, C. P., & Higgs, S. (2017). Using a descriptive social norm to increase vegetable selection in workplace restaurant settings. Health Psychology, 36(11), 1026–1033. https://doi.org/10.1037/hea0000478
- Tilman, D., & Clark, M. (2014). Global diets link environmental sustainability and human health. Nature, 515(7528), 518–522. https://doi.org/10.1038/nature13959
- Verkooijen, K. T., Stok, F. M., & Mollen, S. (2015). The power of regression to the mean: A social norm study revisited. European Journal of Social Psychology, 45(4), 417–425. https://doi.org/10.1002/ejsp.2111
- Wechsler, H., Nelson, T. E., Lee, J. E., Seibring, M., Lewis, C., & Keeling, R. P. (2003). Perception and reality: A national evaluation of social norms marketing interventions to reduce college students’ heavy alcohol use. Journal of Studies on Alcohol, 64(4), 484–494. https://doi.org/10.15288/jsa.2003.64.484
- Werch, C. E., Pappas, D. M., Carlson, J. M., DiClemente, C. C., Chally, P. S., & Sinder, J. A. (2000). Results of a social norm intervention to prevent binge drinking among first-year residential college students. Journal of American College Health, 49(2), 85–92. https://doi.org/10.1080/07448480009596288
- Westhoek, H., Peter, J. P., Rood, T., Wagner, S., de Marco, A., Murphy-Bokern, D., Leip, A., van Grinsven, H., Sutton, M. A., & Oenema, O. (2014). Food choices, health and environment: Effects of cutting Europe’s meat and dairy intake. Global Environmental Change, 26, 196–205. https://doi.org/10.1016/j.gloenvcha.2014.02.004
- Yates, L., & Warde, A. (2017). Eating together and eating alone: Meal arrangements in British households. The British Journal of Sociology, 68(1), 97–118. https://doi.org/10.1111/1468-4446.12231