Several exigent societal issues could be mitigated by reducing global consumption of meat and animal products (MAP) and encouraging predominantly plant-based diets in their place. Authoritative enjoinments for such a dietary shift have highlighted its potential to improve public health [1
], reduce risks of zoonotic pandemics and antibiotic resistance [7
], curb environmental degradation and climate change [3
], and limit the preventable suffering and slaughter of approximately 500 to 12,000 animals over the lifetime of each human consuming a diet typical of their country [6
]. Nevertheless, MAP consumption in Western countries far exceeds nutritional recommendations [3
] and, worldwide, continues to rise substantially [4
]. Whereas existing research attention has often focused specifically on reducing consumption of red and processed meat [10
], in this paper, we focus more broadly on shifting consumption of all meats (i.e., edible animal flesh) and animal products (i.e., eggs and dairy) to plant-based diets. In contrast, shifting consumption from red and processed meat to poultry, fish, dairy, and eggs would be beneficial for individual health, but probably less so than making comparable shifts to healthy plant-based foods, such as vegetables, fruits, whole grains, and legumes, that are severely lacking in the standard Western diet [3
]. Furthermore, producing poultry, fish, dairy, and eggs causes considerable environmental and ecological damage [13
] and has severe animal welfare impacts [6
Developing simple interventions to encourage dietary shifts from MAP to healthy plant-based foods could therefore carry widespread societal benefits. Educational interventions that make appeals to individual health [10
], the environment [10
], or animal welfare [16
] may be effective. More subtle “nudge” interventions that may operate outside participants’ conscious awareness, for example by repositioning meat dishes to be less prominent in cafeterias, may also be effective [16
]. Although these types of interventions are promising, many existing studies have methodological limitations [17
]. These include the potential for social desirability bias that could artificially inflate apparent intervention effects [19
], measurement of outcomes only in terms of participants’ attitudes or intended behavior rather than actual MAP consumption, and small sample sizes. As a result, we are aware of very few specific interventions that are adequately well-evidenced to strongly support their widespread dissemination at this point.
We conducted a series of parallel-group, randomized controlled experiments designed to help resolve these methodological challenges of previous studies. Namely, our studies took stringent precautions against social desirability bias, used longitudinal designs, and measured food consumption outcomes using food frequency questionnaires. The intervention was a 20-min documentary that encourages dietary shifts from all meats and animal products to plant-based diets. We selected this documentary because its content reflects certain best practices for designing effective interventions in general, and its content also harnesses the specific psychology of MAP consumption [17
]. In general, providing educational information
can influence beliefs and intentions that may subsequently shape behavior [20
]. The public appears to be poorly informed about the aforementioned consequences of global MAP consumption; in fact, many individuals appear to deliberately avoid such information [21
]. Thus, providing information that helps remedy this knowledge gap may be effective. Additionally, portraying the desired behavior as aligning with social norms
(what others believe one should do, or what others actually do) can effectively shift behaviors, including food choices [22
]. Providing concrete suggestions for how to change one’s behavior (e.g., recipes) may help individuals to form concrete implementation intentions
for what they plan to do when faced with food choices [24
]. According to the Theory of Planned Behavior, providing such suggestions may increase individuals’ perceived ability to control their future behavior and their intentions to do so [20
]. Indeed, previous interventions to reduce consumption of meat and/or animal products that invoked these components have obtained preliminarily promising results. Such interventions have included, for example, providing leaflets, news articles, and videos [17
In addition to leveraging these general components of effective behavioral interventions, the documentary we studied was designed to also harness the unique social, moral, and affective psychology underlying MAP consumption [28
]. For example, although ethical concern about factory farming conditions is now a majority stance in several developed countries [30
], MAP consumption remains nearly universal. This discrepancy between people’s ethical views and their actual behavior, termed the “meat paradox” [31
], can induce cognitive dissonance. Previous interventions have successfully invoked this dissonance by using meat-animal reminders
, which are simple visual or verbal reminders of the connection between MAP and animals (e.g., photographs of meat dishes presented next to photographs of the animals from which they came) [32
]. Last, physical disgust and moral disgust
are closely intertwined and powerfully shape food choices [38
]. Experiencing physical disgust can amplify negative moral judgments, and conversely, experiencing moral disgust can induce physical disgust [40
]. Previous interventions to reduce consumption of meat and/or animal products have often invoked disgust by describing, for example, “crowded conditions [and] pens covered in excrement and germs” [41
]. We also selected this documentary because it has been disseminated in practice via social media advertising by a nonprofit, The Humane League. For example, in 2019, the nonprofit’s advertising generated 13 million visits to websites deploying documentary-driven interventions, including this documentary, resulting in 8 million minutes of viewing.
In Study 1, we aimed to assess the documentary’s effectiveness using a study design that improved upon certain methodological limitations of previous work, described above. In Study 2, we aimed to adjudicate discrepancies between the results of Study 1 and those of previous studies by further examining the role of social desirability bias. In Study 3, we aimed to assess the documentary’s effectiveness in a different population and upon adding new components to the intervention, which were designed to increase participant engagement. To this end, in Studies 1 and 3, our primary outcome was participants’ total MAP consumption over the past week (henceforth “consumption”), reported approximately 2 weeks after random assignment and exposure to the documentary. In both studies, we secondarily assessed consumption of specific categories of MAP as well as consumption of healthy plant-based foods. We also assessed the extent to which the intervention’s effects might differ by participants’ demographic characteristics; such findings could be used to cost-effectively target dissemination. For example, previous work has suggested that sex, education, and political liberalism could moderate the effectiveness of interventions to reduce consumption of meat and/or animal products [42
]. Studies 1 and 3 took stringent precautions against social desirability bias. In Study 2, to further examine the potential for social desirability bias, our primary outcome was participants’ immediate intentions to increase, decrease, or not change their consumption, similar to many existing studies in the literature.
5. General Discussion
In a series of 3 randomized, controlled experiments, we investigated the effectiveness of a documentary that was professionally designed to reduce MAP consumption by harnessing many of the best practices in designing interventions to change behavior. The documentary’s content was designed to leverage general components of effective behavioral interventions (e.g., by providing educational information, invoking social norms, and providing implementation suggestions) and of the specific psychology governing MAP consumption (e.g., by invoking physical and moral disgust and using meat-animal reminders).
In a study designed to minimize social desirability bias (Study 1), we found that the documentary did not affect participants’ reported consumption 12 days after random assignment (
oz/week; 95% CI: [
]). However, in a second study with less protection against social desirability bias (Study 2), the documentary did substantially increase the percentage of participants who immediately intended to reduce consumption from 20% to 68%, a 3.42-fold change (95% CI: [2.49, 4.92]), which was in fact an even larger effect size than was seen in existing, comparably designed, studies and meta-analyses [17
]. Finally, we changed the study sample from members of a commercial study recruitment platform to participants who had previously volunteered to be contacted about nutrition studies, and we enhanced the intervention with evidence-based engagement items and opportunities to form concrete goals (Study 3). These changes did not markedly improve the intervention’s effects on subsequently reported MAP consumption (
oz/week; 95% CI: [
]), even though a majority of participants who viewed the documentary (60%) had pledged to reduce or eliminate their consumption of at least one meat or animal product. Additionally, the documentary was not substantially more effective among participants who were expected to be more receptive to its content: among participants who had expressed interest in participating in nutrition studies (i.e., all participants in Study 3) or among the demographic subset of these participants who were Democrats and had at least graduated 2-year college. The documentary also did not meaningfully affect participants’ responses to secondary consumption outcomes or the exploratory attitude outcomes.
5.1. Strengths and Limitations
This research has a number of methodological and conceptual strengths. To help mitigate widespread limitations of previous research regarding the potential for social desirability bias [17
], we introduced methodological innovations to help prevent and benchmark social desirability bias in studies of reported food outcomes. First, in the 2 longitudinal studies (Studies 1 and 3), we used naïve re-recruitment to conceal the purpose of the study. We also included decoy foods (e.g., sweetened beverages) in the outcome measurements. Awareness probes in these studies indicated that, with these methods, participants indeed remained largely unaware of the purpose of the experiment during follow-up data collection. In Study 2, as a benchmark of the severity of social desirability bias, we investigated “retrospective causation” by asking participants to report on their past-week consumption, mere minutes after they had viewed either the documentary or the control video. Any apparent intervention effects on reported consumption over the week prior
to random assignment would seem to necessarily reflect social desirability bias. We estimated that the documentary affected these retrospective reports of total consumption by
oz/week (95% CI: [
]). Although the confidence interval is wide, this spurious point estimate is in the vicinity of typical intervention effects seen in studies of interventions to reduce meat consumption, most of which had similar designs that are susceptible to social desirability bias [17
As an additional methodological strength, in Studies 1 and 3, we assessed reported consumption rather than only intentions, and we assessed consumption in terms of the numerical weight of food that participants reported consuming of specific categories of meats and animal products (e.g., chicken, beef). Such measures would allow intervention effects to be translated into direct measures of cost-effectiveness and societal impact, such as the estimated reduction in human all-cause mortality events, in the number of animals raised for consumption, and in greenhouse gas emissions [17
]. Last, in Study 3, we enhanced the intervention with engagement items that were based on strong evidence from the broader literature on lifestyle behavior-change interventions, including tailored content via the interim emails reminding participants of their stated pledges [72
]. These enhancements have rarely been studied in the context of reducing MAP consumption [17
This research also has limitations. Although our findings seem to rule out the possibility that this intervention has large effects, smaller effects may nevertheless be consequential if a large number of individuals were exposed to the intervention, or if effects accumulated over repeated exposures [74
]. Indeed, we intentionally selected an intervention that has already been widely disseminated at relatively low cost. However, it is difficult to precisely establish small intervention effects without very large sample sizes and high retention. Although Study 1 achieved high retention, Study 3 did not, which compromises statistical precision and increases susceptibility to missing data bias. Additionally, reported dietary data can be subject to substantial measurement error even when collected via measures such as 24-h recalls or food frequency questionnaires [75
]; such measurement error can further obscure small intervention effects. Although Studies 1 and 2 used somewhat longer durations of follow-up (approximately 2 weeks) than most existing studies [17
], even longer durations would help identify potential delayed adoption and would characterize how well effects are sustained over time. The documentary we studied was previously disseminated by a nonprofit, so it is possible that some individuals may already have viewed it prior to participating in our studies, although no participants mentioned having seen the documentary when they provided free-text feedback on the study. Finally, it is possible that the control video could have elicited effects of its own, although this seems unlikely given its irrelevant content.
5.2. Future Directions
The majority of previous studies of similar interventions have assessed outcomes in terms of participants’ reported intentions to change their consumption, using designs similar to our Study 2 [17
]. Indeed, in this context, the documentary we studied appeared to be remarkably effective, with an effect size stronger than those of any of the 100 studies included in our recent meta-analysis [17
]. That is, in Study 2, we estimated a 3.42-fold reduction in the percentage of participants who intended to reduce consumption. Critically, though, the documentary had little effect on reported consumption after approximately 2 weeks. This discrepancy between intentions and reported consumption reiterates that reported intentions may be a poor proxy for reported actual consumption, and further underscores the urgency of designing studies that minimize social desirability bias. Future longitudinal studies could consider adopting the same methodological innovations, detailed above, that we used to successfully reduce participant awareness.
As noted above, precisely estimating potentially small, but consequential, intervention effects requires large sample sizes as well as stringent precautions against biases and sources of random measurement error. Future studies of similar interventions might consider using designs other than traditional individual randomization. For example, interventions could be deployed at locations of food purchases (e.g., by posting educational flyers in a grocery store). This could facilitate measuring outcomes in terms of actual food purchases at that location and could also help generate social norms within local groups. Such designs can enable large sample sizes while reducing the potential for social desirability bias and measurement error [77
In conclusion, these findings suggest that some past studies of similar interventions may have overestimated effects due to methodological biases. The methodological innovations we introduced could help future studies mitigate these biases. Novel intervention strategies may be needed to effectively shift dietary consumption away from MAP.