Behavioral Economics and Food Choice

Individual behavior plays a central role in the disease burden faced by society. Many major health problems in the United States such as obesity are exacerbated by unhealthy behaviors. In our research, we apply ideas from behavioral economics, which integrates concepts from psychology and economics, to the problem of changing health behaviors. In our research we use several of the decision biases that ordinarily lead people to self-harming behavior, to promote healthy behaviors instead. To date, we have been applying this approach to areas such as smoking cessation, weight loss and medication compliance. We were approached by Aramark to collaborate on projects to test the applicability of this approach to changing food choice. Successful pilots in this area would greatly contribute to ongoing discussions nationally on curbing the obesity epidemic. Our plan was to structure interventions to take advantage of the fact that individuals put disproportionate value on present relative to future costs and benefits, known as present-biased preferences. Present-biased preferences can be made to steer people toward healthier options if they are given immediate rewards for healthy behaviors with even small rewards, if they are immediate. Our first project with Aramark used price discounts on zero-calorie beverages as a means to make the benefits of healthier beverage choice more immediate and tangible. At 4 Aramark hospital cafeteria sites, we discounted the price of zero-calorie beverages by 10% and sought to determine the impact on consumers choice of these beverages. At this time, the intervention has just completed, and data analysis is ongoing. Besides financial incentives, as we used in our first pilot project, conveying information can also make the value of future costs and benefits more immediate. It is this principle which we plan to test with the second pilot.

This pilot intervention will use a quasi-experimental, factorial design to test the impact of calorie information presented on posters in different formats on beverage choice (zero-calorie beverage versus other drinks). At each of 4 hospital cafeteria sites, we will conduct 3 separate, consecutive interventions in which we post the following displays for bottled beverages: (A) calorie information, (B) calories plus calories as exercise equivalents, (C) calories as exercise equivalents. Each intervention will last 3 weeks with a 1 week "washout" period (no display) in between interventions. Therefore, the interventions will run for a total of 11 weeks. The order of interventions will be randomized at each site to address ordering effects. Data on bottled beverage sales (zero-calorie vs. non-zero-calorie) will be collected and analyzed at the cafeteria-level. This includes point-of-sale data, inventory, and stock-keeping-units (SKU) data of zero-calorie and regular beverages sold weekly at each site, before, during and after the pilot. In order to make appropriate comparisons across cafeterias, already-conducted site-specific demographic market research analysis on customers (in aggregate) will also be considered. No individual-level consumer data will be collected, obtained, or analyzed in this study.

The unit of analysis for these studies are hospital cafeterias and cafes which are operated by Aramark, the food services company. The cafeterias and cafes are located in a variety of hospitals located nationwide. Through discussions with Aramark, these cafeterias have volunteered to participate in this study. Representatives from Aramark have been in contact with representatives from the hospitals regarding their participation. We expect 4 Aramark sites will participate, which in total will include 6 cafeterias and 3 convenience stores. During the intervention, customers at each site will see the calorie information displays but will be under no obligation to purchase any of the beverages involved in the study.

The 11 week intervention is set to begin in early February. Data on beverage sales at each site will be collected in the weeks and months leading up to the intervention and in the weeks and months following the intervention.

There will be no individual-level data collection, only aggregate monitoring of cafeteria beverage inventory. As such there is almost no risk to human subjects, their privacy, or confidentiality.

Potential risks to humans include altered food choice that negatively affects their health. Though we will be promoting healthier options, there is a very small possibility that such promotion paradoxically may influence individuals to seek out less healthy items. There is also a small risk of such promotions such as exercise labeling to effect individuals psychologically in unexpected ways. The likelihood of both of these is very small and the seriousness of these risks also are minor.