High consumption of meat, particularly red and processed meat, increases risk of CVD, diabetes and some cancers(Reference Clark, Springmann and Hill1–Reference Al-Shaar, Satija and Wang8). Further, red meat is a major contributor to environmental harms such as greenhouse gas emissions(Reference Clark, Springmann and Hill1,Reference Poore and Nemecek9–Reference Boehm, Ver Ploeg and Wilde13) , air and water pollution(Reference Clark, Springmann and Hill1,Reference Poore and Nemecek9) , biodiversity loss(Reference Clark, Springmann and Hill1,Reference Gerber, Steinfeld and Henderson14) and deforestation(Reference Gerber, Steinfeld and Henderson14,15) . Reducing red meat consumption is therefore an important strategy for reducing chronic disease risk and mitigating environmental damage(Reference Tilman and Clark16).
Despite growing recognition of the health and environmental harms of red meat, American adults consume an average of 284 g/week (about 0·6 pounds) of unprocessed red meat alone (i.e. not including processed red meats such as bacon), nearly three times the maximum level recommended for optimising human and planetary health(Reference Willett, Rockström and Loken2). More than half of Americans say they are willing to eat less red meat(Reference Leiserowitz, Ballew and Rosenthal17). Yet red meat consumption is projected to increase over the next decade(Reference Bentley18), perhaps in part because many Americans are unaware of the health and environmental harms of red meat(Reference Leiserowitz, Ballew and Rosenthal17,Reference Neff, Edwards and Palmer19) . Given this willingness to change, coupled with lack of awareness about red meat’s impacts, educating consumers about the harms of red meat could reduce red meat consumption.
A growing body of research has shown that communicating about a products’ health harms, for example, through product warning labels and mass media campaigns, is an effective strategy for reducing unhealthy behaviours including cigarette smoking(Reference Noar, Hall and Francis20,Reference Brewer, Hall and Noar21) , alcohol consumption(Reference Clarke, Pechey and Kosīte22) and sugary drink consumption(Reference Clarke, Pechey and Kosīte22–Reference Schwartz, Schneider and Choi25). Similarly, a recent systematic review found that providing information about the health effects of meat consumption is an effective strategy for reducing intentions to eat meat as well as meat consumption(Reference Harguess, Crespo and Hong26). Emerging literature also suggests the promise of communicating about products’ environmental harms as a strategy for changing consumer behaviour. For example, a randomised experiment with undergraduate students in the UK found that sending students 2 weeks of daily messages about the environmental effects of meat production reduced students’ red and processed meat consumption compared with a no-message control group(Reference Wolstenholme, Poortinga and Whitmarsh27). What remains unknown is which specific health and environmental harms hold the most promise for motivating consumers to reduce their red meat intake. Identifying the specific harms that most discourage red meat intake is important because messaging campaigns may not be able to communicate about all harms (e.g. due to space constraints), and because prior studies of tobacco and sugary drink messages suggest advantages to shorter, simpler messages(Reference Noar, Kelley and Boynton28–Reference Grummon, Hall and Mitchell32). Also unknown is whether consumers’ reactions to health and environmental harms of red meat vary by demographic characteristics, information that could help to tailor messages to specific groups.
To inform communication efforts, we examined consumers’ responses to health and environmental harms of red meat in an experiment with US parents of young children. Parents are a critical group to study in dietary communication interventions because their behaviours influence both their own health and the dietary habits of their children(Reference Savage, Fisher and Birch33). Parents of young children (i.e. under age five) are especially important, given that dietary habits in early childhood affect diet and health later in childhood and into adolescence(Reference Nicklaus and Remy34,Reference Singh, Mulder and Twisk35) . Moreover, US parents are nearly 40 % less likely than non-parents to have reduced their red meat intake compared with 3 years ago(Reference Neff, Edwards and Palmer19), suggesting red meat reduction campaigns may be especially beneficial for this group. Thus, the specific objectives of this study were to examine which health and environmental harms of red meat parents are aware of and which are most likely to discourage red meat consumption. Additionally, to provide insight on populations that might respond more strongly to messages about red meat’s harms, we examined demographic predictors of awareness of health and environmental harms and of the extent to which these harms discouraged participants from wanting to consume red meat.
Methods
Prior to data collection, we pre-registered the sample size, hypotheses and analysis plan (https://aspredicted.org/q5e9b.pdf). The only deviations from this plan were that we corrected for multiple comparisons using Bonferroni’s method rather than Tukey’s method because Tukey’s method cannot be applied to mixed models, and that we conducted two unplanned exploratory analyses, examining: (1) predictors of awareness of health and environmental harms and (2) predictors of harm-induced discouragement separately for health v. environmental harms.
Participants
In January 2020, we recruited a convenience sample of 544 US adults using the survey research firm Dynata as part of a study of parents’ responses to experimental stimuli and survey questions. Participants were eligible if they were aged 18 years or older and had a child aged 6 months to 5 years. Online convenience samples can provide efficient and generalisable findings for experiments like the one used in this study(Reference Jeong, Zhang and Morgan36).
Procedures
Participants provided informed consent, completed an online survey programmed in Qualtrics and received incentives from Dynata (e.g. gift cards, reward points).
Measures
A flow of survey questions is shown in Fig. 1. First, participants answered questions about their usual red meat consumption(Reference Neff, Edwards and Palmer19,Reference Thompson, Midthune and Kahle37) and general perceptions that red meat is bad for health and for the environment (e.g. ‘How bad or good for your health do you think eating red meat is?’). Next, they responded to questions about their awareness of, and discouragement in response to, specific health and environmental harms of red meat. The order of presentation of health and environmental harms was randomised such that half of participants answered questions about health harms first and half answered questions about environmental harms first.
Flow chart of survey questions
We assessed awareness of harms using a select-all-that-apply question adapted from previous studies(Reference Rohde, Noar and Mendel38–Reference Carl, Taillie and Grummon40), ‘Before today, had you ever heard that eating red meat can contribute to the following harms?’ Then, we listed the eight health or eight environmental harms, displayed in random order. Participants could also select ‘I haven’t heard of red meat contributing to any of these harms before;’ this option was always displayed last.
Next, we assessed the extent to which each harm discouraged participants from wanting to eat red meat using an item adapted from previous studies(Reference Rohde, Noar and Mendel38–Reference Carl, Taillie and Grummon40), ‘How much does knowing that eating red meat contributes to these harms discourage you from wanting to eat red meat?’ We assessed perceived discouragement because meta-analytic evidence indicates that perceived message effectiveness is predictive of actual message effectiveness(Reference Noar, Barker and Bell41). Participants rated perceived discouragement in response to each harm on a 5-point scale, from ‘Not at all’ (coded as 1) to ‘Very much’ (coded as 5).
We selected harms to present based on recent literature linking red meat production and consumption with health and environmental harms (Table 1). The eight health harms were type 2 diabetes(Reference Feskens, Sluik and van Woudenbergh42,43) , weight gain(43–Reference Schlesinger, Neuenschwander and Schwedhelm45), heart disease(43,Reference Guasch-Ferré, Satija and Blondin46) , stroke(Reference Kim, Hyeon and Lee47,Reference Bernstein Adam, Pan and Rexrode Kathryn48) , colon cancer(43,49,50) , prostate cancer(49), stomach cancer(49) and early death(43,Reference Etemadi, Sinha and Ward51) . The eight environmental harms were climate change(Reference Gerber, Steinfeld and Henderson14,15) , more greenhouse gases(Reference Clark, Springmann and Hill1,Reference Poore and Nemecek9–Reference Willits-Smith, Aranda and Heller12) , water shortages(Reference Clark, Springmann and Hill1,Reference Aleksandrowicz, Green and Joy11,Reference Mekonnen and Hoekstra52) , water pollution(Reference Clark, Springmann and Hill1), air pollution(Reference Poore and Nemecek9), plants and animals going extinct(Reference Clark, Springmann and Hill1,Reference Gerber, Steinfeld and Henderson14) , clearing of forests(Reference Gerber, Steinfeld and Henderson14,15) and worse land quality(Reference Clark, Springmann and Hill1,15,Reference Foley, Ramankutty and Brauman53) . All participants rated their awareness and discouragement for each of the eight health harms and each of the eight environmental harms. Within each type of harm (health v. environmental), specific harms were presented in random order.
Health and environmental harms of red meat shown in experiment and supporting evidence
Finally, the survey assessed standard demographics. Survey measures appear in online supplementary material, Supplemental Exhibit 1.
Analysis
First, we calculated the proportion of participants who reported they were aware of each harm and the mean discouragement ratings for each harm. We also calculated the proportion of participants who were aware of at least one harm, and the mean number of harms for which participants indicated awareness, both overall and separately for health and environmental harms. Next, we assessed whether likelihood of reporting awareness of harms was higher for health compared with environmental harms using mixed effects logistic regression, regressing awareness (coded as 0/1 for each harm) on an indicator variable for whether the harm was a health or environmental harm. These models treated the intercept as random to account for repeated measures within participants.
To shed light on the populations who were most aware of red meat’s harms, analyses also examined demographic predictors of the total number of harms for which participants indicated awareness (summed across all sixteen health and environment harms). These analyses used negative binomial regression with robust standard errors. We regressed the total number of harms for which participants indicated awareness on the following potential predictors: age, gender, race/ethnicity, educational attainment, income, political leaning and usual red meat consumption. We also used this approach to examine predictors of awareness separately for health harms v. environmental harms.
Next, analyses assessed whether health or environmental harms were more effective at discouraging participants from wanting to eat red meat. These analyses used a linear mixed model, regressing harm-induced discouragement ratings on an indicator variable for whether the harm was a health or environmental harm, treating the intercept as random. We then assessed the extent to which each specific harm elicited discouragement using a linear mixed model with indicator variables for each of the sixteen harms (excluding one as the referent), again treating the intercept as random. We used z-tests to conduct pairwise comparisons of predicted mean discouragement for each harm, applying Bonferroni’s method to adjust for multiple comparisons. These comparisons allowed us to determine which of the harms (if any) were more discouraging than the others while adjusting for repeated measures within participants.
To examine which population groups reported more discouragement in response to health and environmental harms of red meat, we also examined demographic predictors of average discouragement from wanting to eat red meat. First, we averaged discouragement ratings across all sixteen harms. We then used ordinary least squares linear regression to assess predictors of average discouragement. These analyses assessed the same demographic predictors as for awareness and additionally examined general perceptions that red meat is bad for health and for the environment. Exploratory analyses used the same approach to examine predictors of average harm-induced discouragement separately for health harms v. environmental harms.
Analyses were conducted in 2021 using Stata MP version 16.1 (StataCorp LLC).
Results
Participants’ average age was 33·8 (sd 8·0) years (range: 19–80). About two-thirds were White (69 %), 18 % were Latino(a), 8 % were another race/ethnicity and 5 % were Black (Table 2). Slightly more than half (57 %) of participants identified as female and 22 % had a high school education or less.
Participant characteristics, n 544 US parents of young children
Missing data ranged from 0·0% to 1·5%.
For each of the sixteen harms, fewer than one-third of participants indicated awareness that red meat contributed to that harm (Table 3). About one-third (33 %) of participants were not aware of any of the sixteen harms; 46 % were not aware of any of the health harms, and 51 % were not aware of any of the environmental harms. The specific harms with the highest level of awareness in the sample were heart disease (28 % reported awareness), weight gain (27 %), climate change (22 %) and increased greenhouse gas emissions (21 %). Participants were least aware that red meat contributes to stomach cancer (11 %) and prostate cancer (8 %). In mixed effects logistic regression, participants were similarly likely to report awareness of harms regardless of harm topic (health v. environment, OR = 1·02, P = 0·77).
Awareness and discouragement of the health and environmental harms of red meat consumption, n 544 US parents of young children
In multivariate analyses examining predictors of the number of harms for which participants reported awareness, participants aged 26–34 years reported awareness of about 0·3 fewer harms of red meat compared with those aged 18–25 years (B = –0·33, P = 0·029, Table 4). Participants who identified as female reported being aware of fewer harms than those who identified as male (B = –0·26, P = 0·023). Likewise, those who identified as politically moderate (B = –0·31, P = 0·013) or conservative (B = –0·34, P = 0·011) reported awareness of fewer harms than those who identified as liberal. Participants who identified as Black reported awareness of more harms than those identifying as White (B = 0·57, P = 0·006), but identifying as Latino(a) or as another race/ethnicity (compared with identifying as White) was not associated with awareness. Higher educational attainment and higher income were generally associated with being aware of more harms. Usual red meat consumption was not associated with being aware of more health and environmental harms of red meat (B = 0·04, P = 0·558).
Associations between participant characteristics and the total number of health and environmental harms of red meat for which participants reported awareness, n 544 US parents of young children
Bs are unstandardised regression coefficients from negative binomial regressions, regressing the total number of health and environmental harms for which participants reported awareness on participant characteristics. Models estimated robust standard errors. Bold coefficients are statistically significant, P < 0·05.
* Referent group was male. The one non-binary participant was excluded from analysis due to small cell size.
In analyses of awareness of health harms only, participants who identified as Black (compared with White) and those who had higher educational attainment reported awareness of a greater number of health harms. By contrast, participants who identified as politically moderate (compared with liberal) reported awareness of fewer harms (online supplementary material, Supplemental Table 1). In analyses of awareness of environmental harms only, being age 26–34 years (compared with 18–25), identifying as female (compared with male) and identifying as politically conservative (compared with liberal) were associated with awareness of fewer harms (online supplementary material, Supplemental Table 2). Higher education, higher income and higher usual red meat consumption were associated with awareness of more of the environmental harms of red meat.
In mixed effects regressions of harm-induced discouragement, health harms elicited slightly more discouragement than environmental harms, but the magnitude of the difference was small (mean discouragement 2·73 v. 2·69 on the 1–5 Likert scale; B = 0·04, P = 0·010). Harms with higher awareness generally elicited higher discouragement (Fig. 2). Among specific harms, heart disease elicited the highest mean discouragement (mean = 2·82 on the 1–5 Likert scale), followed by early death (mean = 2·79), plants and animals going extinct (mean = 2·75), stroke (mean = 2·75) and weight gain (mean = 2·75) (Table 3). Prostate cancer (mean = 2·60) and worse land quality (mean = 2·65) elicited the lowest discouragement. After adjusting for multiple comparisons, the only significant differences in discouragement between harms were that early death was more discouraging than prostate cancer (difference in predicted means = 0·20, adjusted P = 0·002), and heart disease was more discouraging than both prostate cancer (difference = 0·23, adjusted P < 0·001) and worsening land quality (difference = 0·17, adjusted P = 0·024).
Health and environmental harms of red meat by discouragement and awareness, n 544 US parents of young children. 
, health harms; 
, environmental harms
In multivariate analyses examining predictors of average discouragement ratings, older participants generally reported lower discouragement than younger participants (Table 5). Participants who identified as Black reported higher levels of average discouragement compared with White participants (B = 0·56, P = 0·019). Those who identified as politically moderate (B = –0·40, P = 0·002) or conservative (B = –0·64, P < 0·001) were less discouraged by the health and environmental harms of red meat than those who identified as politically liberal. Participants who reported higher general perceptions that red meat is bad for health also reported higher average discouragement (B = 0·18, P = 0·006). By contrast, general perceptions that red meat is bad for the environment were not associated with average discouragement (B = 0·07, P = 0·329). Finally, participants who reported higher usual red meat consumption reported higher average discouragement (B = 0·36, P < 0·001). Gender, education, income and other race/ethnicities (i.e. Latino(a) and other race/ethnicity) were not associated with average discouragement (all Ps > 0·05). The pattern of results was similar when examining harm-induced discouragement separately for health v. environmental harms (online supplementary material, Supplemental Tables 3 and 4).
Associations between participant characteristics and average discouragement from wanting to eat red meat in response to health and environmental harms of red meat across, n 544 US parents of young children
Bs are unstandardised regression coefficients from ordinary least squares regressions regressing participants’ average discouragement ratings (across all sixteen health and environmental harms) on participant characteristics. Bold coefficients are statistically significant, P < 0·05.
* Referent group was male. The one non-binary participant was excluded from analysis due to small cell size.
Discussion
Results of this study suggest that messages describing the health and environmental harms of red meat could be a promising strategy for discouraging red meat consumption among US parents of young children. The majority of respondents were not yet aware of the specific health and environmental harms of red meat assessed in this study, and one-third were not aware of any of the sixteen harms examined. These results suggest a major opportunity to educate consumers and motivate positive behaviour change. Expectancy disconfirmation theory posits that when consumers receive negative information about a product that conflicts with their prior expectations (e.g. being informed about the harms of red meat when they had previously not known these harms), their attitudes towards the product will become more negative(Reference Van Raaij54–Reference Burton, Howlett and Tangari56). This theory would suggest that correcting consumers’ misperceptions about the health and environment risks of red meat could motivate them to reduce their red meat consumption. In line with this prediction, one study found that warning messages about the health harms of sugary drinks led to larger changes in parents’ attitudes and purchase intentions when the messages were displayed on beverages parents had perceived as healthier compared with beverages parents already understood to be unhealthy(Reference Moran and Roberto57).
Regression analyses revealed that participants who were 26–34 years old (compared with 18–25 years), identified as female (compared with male) and identified as politically moderate or conservative (compared with liberal) reported awareness of fewer harms of red meat. By contrast, participants who identified as Black (compared with White) and those with higher educational attainment and higher income reported being aware of more harms. These results suggest that it may be beneficial to tailor awareness-raising campaigns to particular groups with lower awareness, such as parents who identify as female, are politically moderate or conservative or have lower educational attainment or income. However, given that the majority of participants were unaware that red meat contributes to the health and environmental harms assessed in this study, educational efforts are likely to benefit parents from all demographic groups.
Participants’ usual red meat consumption was not related to their awareness of red meat’s health harms, but higher red meat consumption did predict higher awareness of the environmental harms of red meat. The reason for this association is unclear. One explanation is that higher red meat consumers are more likely to pay attention to information about the environmental consequences of red meat consumption because this information is particularly relevant to them, but have not yet acted on their awareness by reducing their red meat consumption. Regardless of the explanation, this finding highlights that interventions would likely benefit from incorporating a variety of strategies to reduce red meat consumption, including increasing the accessibility, availability and attractiveness of non-meat options(Reference Ruby58).
Several participant characteristics predicted higher discouragement from wanting to eat red meat in response to health and environmental harms. For example, consumers aged 18–25 years reported higher discouragement in response to environmental harms of red meat compared with those aged 26–34 and 35–44 years, perhaps because young adults have stronger interest in environmental sustainability and greater concern about climate change than older adults(Reference Saad and Jones59–61). Younger adults also reported higher discouragement in response to health harms than older adults. This pattern of results differs somewhat from prior research finding that younger adults were less likely than older adults to report health reasons for not eating meat(Reference Pribis, Pencak and Grajales62). Our results could potentially reflect a growing openness among young adults towards reducing their red meat consumption or consuming a plant-forward diet(63), regardless of the precise motivation for making dietary changes. Participants who reported higher red meat consumption also reported higher discouragement in response to health and environmental harms of red meat. This finding is encouraging, as it suggests that messages about the harms of red meat might have the greatest impact on those who stand to benefit the most from reducing their red meat intake. Additionally, participants who had stronger general perceptions that red meat is bad for health reported being more discouraged, on average, in response the specific health and environmental harms examined in this study. By contrast, general perceptions that red meat is bad for the environment were not associated with average discouragement ratings. These findings might suggest that strengthening the public’s general perception that red meat is bad for health could increase the public’s receptivity to messages about specific health or environmental harms. However, the observed associations between participants’ characteristics and their average discouragement ratings should be interpreted with caution because we cannot rule out the possibility that some demographic groups (such as young adults or those who hold stronger general perceptions that red meat is bad for health) might respond more strongly to any type of message presented in an online survey, even messages not about red meat. Studies that experimentally compare red meat messages to control messages are needed to establish whether characteristics like age, meat consumption, and general perceptions about red meat influence the effectiveness of red meat reduction messages on consumer behaviour.
Communication interventions such as product warnings and mass media campaigns that describe the health harms of cigarettes(Reference Noar, Hall and Francis20,Reference Brewer, Hall and Noar21) , alcohol(Reference Clarke, Pechey and Kosīte22) and sugary drinks(Reference Clarke, Pechey and Kosīte22–Reference Schwartz, Schneider and Choi25) have been shown to generate small but meaningful reductions in purchases and consumption of these products, suggesting that communicating about the harms of red meat could help curb red meat intake. We found that both health and environmental harms elicited similar levels of perceived discouragement. Likewise, a variety of health and environmental harms were similarly discouraging to consumers. These results suggest that message developers have many promising options for topics to address in meat reduction messages. The limited differences in mean discouragement observed between the specific harms also suggest that communication campaigns could easily rotate among these harms, a strategy that could help prevent messages from becoming ‘stale’ and losing efficacy over time(Reference Hammond64).
The strengths of this study include the comprehensive set of health and environmental harms tested and the experimental comparison of how much each harm motivated participants to reduce their red meat consumption. Limitations include the use of a convenience sample of parents and the relatively young age distribution of the sample. Although prior studies indicate that online convenience samples can provide similar experimental results as probability samples(Reference Jeong, Zhang and Morgan36,Reference Berinsky, Huber and Lenz65,Reference Weinberg, Freese and McElhattan66) , future research should confirm our findings with non-parents and with a wider range of ages. Additionally, although we did not query whether participants were vegetarian or vegan, about 13 % of our sample reported eating red meat less than 1 time per week during the past 30 d. Future studies may wish to examine awareness and discouragement specifically among non-vegans/vegetarians or among high red meat consumers, for whom messages may be most relevant. Another limitation is that we did not assess whether some consumers believe that red meat is beneficial for specific health and environmental outcomes; understanding whether these beliefs are widespread, and who is most likely to hold these beliefs, could help inform messaging campaigns. Additionally, this study focused on health and environmental harms because these are two of the key reasons that US adults report as motivating them to change their diet or reduce their meat consumption(Reference Neff, Edwards and Palmer19,67) . However, consumers might also be motivated by other harms of red meat production. For example, a recent meta-analysis suggested that interventions appealing to animal welfare (e.g. by portraying farm animals) hold promise for reducing meat purchases and consumption(Reference Mathur, Peacock and Reichling68). We also did not assess other potentially important aspects of message design, such as message framing(Reference Stea and Pickering69–Reference Bertolotti, Chirchiglia and Catellani71). Finally, while perceived message effectiveness is predictive of behaviour change(Reference Noar, Barker and Bell41), we did not assess behavioural outcomes. Future studies should experimentally evaluate the extent to which messages describing different types of harms of red meat reduce red meat purchases and consumption.
Conclusions
Reducing meat consumption is critical for mitigating climate change and reducing chronic disease burden(Reference Tilman and Clark16). Our study suggests that communication interventions describing how red meat consumption affects both human and planetary health hold promise for informing US consumers and motivating reductions in red meat consumption.
Acknowledgements
Acknowledgements: The authors thank Marissa G. Hall for helpful feedback on survey measures. Financial support: This research was funded in whole, or in part, by the Wellcome Trust (216042/Z/19/Z). For the purpose of open access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. DG was supported by T32 HL 098048. MGS was supported by T32 DK 007703. Conflict of interest: There are no conflicts of interest. Authorship: A.H.G., L.M.J., C.A.C. and L.S.T. designed and conceptualised the study. A.H.G., D.G. and M.G.S. contributed to data curation. A.H.G. and D.G. analysed the data and created data visualisations. A.H.G. drafted the initial manuscript. E.B.R. provided study oversight. All authors provided critical feedback on the manuscript draft and approved the final manuscript. Ethics of human subject participation: This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving research study participants were approved by the Harvard Longwood Campus Institutional Review Board. Written informed consent was obtained from all subjects/patients.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980021003098
Open access
