Johns Hopkins Center for a Livable Future

Challenge

Climate change is a critical public health threat that requires urgent action. The global food system, from production to waste, accounts for about one-third of all greenhouse gas emissions (GHGE). Meat from cattle (33%), dairy (19%), and other animal meats (9%) contribute over 60% of the projected global temperature rise from food systems by 2030. Reducing red meat and dairy consumption in high-income countries like the U.S., where intake far exceeds dietary guidelines, is essential. Universities offer an influential hub to drive this change by influencing student behavior through their dining services. Johns Hopkins Center for a Livable Future implemented climate labels in dining hall menue to encourage students to choose lower-GHGE foods, but their effectiveness, particularly in university settings. Targeting Generation Z students, who are more inclined towards plant-based diets, the center’s study aimed to assess the impact of climate change menu labels on promoting sustainable food choices and reducing climate anxiety. My role was to collect recipes and all ingredients for the study, and develop a model for calculating greenhouse gas (GHG) emissions of recipes over a four-week period in two dining halls at a private university in Maryland.

Approach

Note

View the code on GitHub (private repository—please email me for access).

Working with a colleague, I extracted over 650 menu items containing 26,277 ingredients from all recipes using Python libraries. I classified these ingredients according to the Cool Food Calculator categories for accurate greenhouse gas emissions (GHGE) calculations. Emissions for each ingredient were calculated based on its category and then aggregated at the recipe level. Together, we visualized the GHGE data using a strip plot to identify natural breaks in the emission distribution, which provided more precise thresholds than traditional methods like tertiles or k-means clustering. I then helped define the thresholds for climate impact labels (green, yellow, red), categorizing foods based on their relative climate impact.

Note

View the code on GitHub (private repository—please email me for access).

I designed the data collection analysis, collected, cleaned, transformed, and analyzed dining ingredient purchasing and student survey data before and after the study. This included preparing tables for publication on demographics and descriptive statistics relevant to food insecurity, diet quality, and types of food consumption. I managed data collection and validation processes, working closely with dining team members to make sure the accuracy of variables and data collection methods. I conducted preliminary exploratory analyses to understand survey respondent characteristics and outcome variables, and wrote methods draft for publication.

Results

By quickly gathering over 600 recipes, I saved the Center hundreds of hours in staff time and resources, resulting in a more efficient and effective research process. Key study findings included:

• Minimal differences between baseline and intervention periods, except for a notable increase in purchases due to a special event.

• Plant protein intake increased slightly post-intervention, while vegetable intake decreased. Overall diet quality scores showed no significant change.

• One-third of students noticed the climate labels, but the majority felt neutral about them. Many students reported that the labels did not influence their food choices or concerns about climate impact.

Distribution of greenhouse gas emissions across diets

Comparison of greenhouse gas emissions by food subgroups

← →

Distribution of greenhouse gas emissions across diets Comparison of greenhouse gas emissions by food subgroups