Abstract
Reducing meat consumption, especially in high-intake countries such as the United States, is crucial in mitigating the climate and biodiversity crises and improving public health and animal welfare. Choice-architecture interventions or nudges in the food domain, such as choice defaults (e.g., reduced default portion sizes), can be powerful levers of behavior change. However, evidence remains limited in large-scale, real-life settings, and little is known about potential effects on diner satisfaction and backfiring effects that reduce or even reverse the desired behavior. These uncertainties have posed substantial barriers to scalability and wider adoption by the food service industry. In our single-blinded, quasi-experimental, pre-registered field interventions in Stanford University dining halls with staff-served portions, a 25% reduction in the serving spoon size (Study 1, 24 days, 364 diners, made-to-order burritos) produced a non-significant trend of 18% less meat served per day without reducing overall diner satisfaction (p = 0.059, d = 0.64) but with a wide CI that included the null (- 49.2, 1.07). A more substantial 50% reduction in serving spoon size (Study 2, 29 days, 1802 diners, varying menu items) did not reduce the amount of meat served (p = 0.60, d = 0.20), triggered backfiring effects, and significantly decreased diner satisfaction. Combining the two studies, the intervention did not significantly reduce meat consumption. While the trends in our findings are consistent with the ‘norm range model’-i.e., that moderate portion reductions may decrease intake but drastic reductions may prompt compensatory eating-key differences and contextual nuances between the two studies help explain the mixed results. Future studies on the ‘norm range’ of default portion size nudges to reduce meat consumption across different menu items and food service models is suggested to increase our understanding of effective and scalable interventions that facilitate collective shifts towards more sustainable dietary behaviors.
Generated Summary
This research article details two quasi-experimental, single-blinded field interventions conducted in Stanford University dining halls, aimed at investigating the effect of portion-size default nudges on meat consumption and diner satisfaction. The studies employed a between-subjects design, where diners were unaware of the study’s objectives. The primary intervention involved reducing the size of serving spoons used by staff to serve meat portions at lunch stations, thereby reducing the default portion sizes. Study 1, a pilot, involved a 25% reduction in serving spoon size at a burrito bar. Study 2 tested a 50% reduction in serving spoon size with varying menu items at a different dining hall. Data collection included measuring the amount of meat served and diner satisfaction through surveys. The study’s goal was to assess the effectiveness of default portion size nudges as a strategy to reduce meat consumption, while also considering the impact on diner satisfaction and identifying potential backfiring effects. The researchers aimed to evaluate the effectiveness of default portion-size nudges in a real-world setting, addressing the need for more evidence on large-scale, real-life interventions and exploring potential effects on diner satisfaction.
Key Findings & Statistics
- Study 1 (25% Spoon-Size Reduction) Results:
- The total number of diners eating at the dining hall was n = 13729 during lunch and n= 6597 during dinner. The resulting sample of survey respondents was n = 364.
- The estimated effect of the intervention was 24.04 lbs (SE: 11.9, 95% CI: [-49.2, 1.07], p-value: 0.059, Cohen’s d: 0.64), meaning around 24 lbs less meat was served in total on intervention days vs. control days
- When normalized by the total number of diners at the dining hall, there was considerably more variability (Fig. 4-right panel). The effect was 0.023 lbs/person (SE: 0.022, 95% CI: [-0.065, 0.020], p-value: 0.27, Cohen’s d: 0.46), indicating no difference between conditions.
- Study 2 (50% Spoon-Size Reduction) Results:
- The total number of diners eating at the dining hall was n = 6692 during lunch and n= 6063 during dinner. The resulting sample of survey respondents was n = 1802.
- The estimated effect of the intervention was 5.82 lbs (95 CI: [- 28.35, 16.71], p-value: 0.60, Cohen’s d: 0.20), meaning around 6 lbs less meat was served in total on intervention days vs. control days.
- In the secondary weight measure, when normalized by the total number of diners at the dining hall, the effect was around – 0.03 lbs/person (95% CI: [- 0.11, 0.24], p-value: 0.46, Cohen’s d: 0.27).
- Backfiring Effects (H2):
- In Study 1, the effect on reducing meat consumption was not statistically significant. The reduction of 24 lbs (18%) in meat served approached significance (p = 0.059, d = 0.64) but included the null in the CI.
- When normalized by the total number of diners, the effect was not significant (p = 0.27, d = 0.46).
- In Study 2, the intervention did not significantly reduce meat consumption either. The amount of meat served was reduced by 13.15 lbs (95% CI: [- 34.30, 7.99], p-value: 0.21, Cohen’s d: 0.60), though the CI was wide.
- Diner Satisfaction:
- In Study 1, participants in the intervention condition reported being significantly hungrier (p-value: 0.0195, Cohen’s d: 0.29) than participants in the control condition.
- In Study 2, overall diner satisfaction significantly decreased on intervention days. The effect of intervention vs. control on fullness was similar, leading to a significant decrease in fullness by 7.45 points (95% CI: [- 10.17, – 4.74], p-value: <0.001, Cohen's d: 0.29).
- Complementing the decrease in satisfaction and fullness, hunger increased by 6.81 points (95% CI: [4.00, 9.61], p-value: <0.001, Cohen's d: 0.25), indicating diners reported feeling significantly hungrier on intervention vs. control days.
Other Important Findings
- The study found that the 25% serving spoon reduction in Study 1 produced a non-significant trend of 18% less meat served per day without reducing overall diner satisfaction (p = 0.059, d = 0.64).
- A more substantial 50% reduction in serving spoon size (Study 2) did not reduce the amount of meat served (p = 0.60, d = 0.20), and significantly decreased diner satisfaction.
- Diner satisfaction (H3): Across the three measures of satisfaction, hunger, and fullness, split by whether diners ate meat, the rating differences are very small, thus suggesting the intervention did not meaningfully detract from diners’ dining experience, especially in satisfaction and fullness.
- When controlling for meat type, the intervention did not meaningfully affect overall satisfaction, given differences of only 1.02 points (95% CI: [- 7.94, 9.97], p-value: 0.82, Cohen’s d: 0.01) and of 0.93 points for fullness (95% CI: [- 7.94, 9.79], p-value: 0.84, Cohen’s d: 0.01). However, unexpectedly, the intervention significantly decreased hunger by 10.57 points (95% CI: [- 19.66, – 1.48], p-value: 0.02, Cohen’s d: 0.12).
- Meat type significantly modified how hungry diners were on intervention vs. control days across all meat types, with diners reporting significantly lower levels of hunger following beef consumption compared to all other meat protein sources (p-value: <0.001).
Limitations Noted in the Document
- The lack of data on survey response rates specific to the intervention station’s diners.
- The relatively small number of data collection days.
- The measurement of backfiring effects was limited due to the lack of data on the specific study sample at lunch versus dinner.
- The study did not measure consumption of non-meat items but suggest this for future research.
- The study used meat served as a proxy for meat consumption and did not assess leftover meat.
- The study lacked sufficient data collection days and menu items to analyze the impact of more detailed characteristics of menu items.
Conclusion
The findings reveal that while default portion size nudges have the potential to influence meat consumption, their effectiveness is context-dependent and can lead to unintended consequences. The study’s mixed results underscore the importance of considering various factors when implementing such interventions. The 25% reduction in serving spoon size, as seen in Study 1, showed promise, suggesting that moderate portion reductions might be more successful in decreasing intake without negatively impacting diner satisfaction. The 50% reduction in Study 2, however, triggered backfiring effects and lowered diner satisfaction, highlighting the complexities of drastically altering portion sizes. The article suggests that the ‘norm range model,’ which posits that moderate reductions can decrease intake without compensatory eating, is relevant here. The research underscores the need for future studies to identify the boundaries of this ‘norm range’ to design effective interventions that minimize negative outcomes. Additionally, the study’s insights into diner satisfaction are important. In Study 1, diners reported being hungrier during the intervention, but overall satisfaction was maintained. Study 2, however, showed a decrease in both satisfaction and fullness, suggesting that the drastic reduction in portion size, along with the varying menu items, played a crucial role. The researchers also highlight the need for incorporating strategies to offset portion size reductions by increasing the availability of non-meat protein and plant-based options, ensuring that interventions do not inadvertently lead to compensatory behaviors or decreased satisfaction. The study concludes with a call for further research focused on exploring the impact of default portion size nudges across diverse food service settings. The goal is to facilitate collective shifts towards more sustainable dietary habits. The results of this research provide valuable insights for scaling cost-effective meat reduction strategies across various high-volume food service settings. These strategies include both commercial establishments and noncommercial ones like K-12 schools, healthcare facilities, and corporate dining.