Better Bites: Healthy Food Within A Budget
Haley MacLeod, Denique Ferguson, Scott Trepper, Steve Layton
Indiana University, Bloomington, IN
Abstract
Those most affected by poor access to healthy foods are individuals with low socioeconomic status. We propose
Better Bites: a sharable device that allows individuals and families to compare items in their cart with similarly
priced healthier alternatives and makes use of existing store loyalty program data to provide personalized
recommendations based on an individual’s historical shopping data.
Overview
Obesity is a significant problem in the United States, impacting more than a third of the population1. Not just a
cosmetic problem, obesity greatly increases the individual’s risk for serious health concerns, including heart disease,
stroke, diabetes, cancer, and osteoarthritis2. Each year, obesity-related health costs reach almost 160 billion dollars
in the US3.
Individuals from lower-socioeconomic status groups are especially vulnerable to obesity and its associated health
risks. Healthy food is often more expensive than foods with refined grains, processed sugars, and saturated fats4.
Low-income individuals will often try to get the most for their money by buying inexpensive energy-dense foods
that are filling5, but lower in nutritional quality, leading to obesity6.
There are a number of existing applications that provide consumers with shopping information. Several smartphone
applications make use of barcodes, NFC or RFID tags to receive information about a product7,8. These types of
systems may not be appropriate for low-income individuals, for many do not have access to smartphones9. Several
researchers have also begun to look at interfaces or hardware built right into a shopping cart10-13. Our solution, Better
Bites, builds on these works, providing incremental improvements to one’s diet by recommending alternatives based
on cost and food categories (e.g. snacks, frozen meals, desserts, etc.)
Design Methods
In our design of Better Bites, we engaged in several design cycles. Our understanding of the needs of lowsocioeconomic individuals (specifically as they relate to food and diet) emerged from observations and interviews
with a variety of domain experts, as well as an in depth interview with one low-income individual who became the
basis for the persona we used to inform our design.
Participant Observation. We visited eight different grocery stores across different neighbourhoods to observe
shopping behaviour. These grocery stores included large chain supermarkets, local food co-ops, smaller independent
stores, and health food specialty stores. They were located in neighbourhoods with varying levels of income.
Domain Expert Interviews. We consulted with two nutritionists, one psychologist, a representative from a local
health food store, and a representative from the USDA’s Supplemental
Nutrition Assistance Program (SNAP) in order to get a better understanding
of four fundamental questions that underpin this topic: How can we
determine which foods are healthier than others? How can we convey
nutritional information to a shopper in a meaningful way? How can we
motivate shoppers to try new food items? What political issues in the food
industry impact shoppers’ food selection habits? The answers to these
questions greatly informed the design of our system.
Persona. We developed a persona based off an interview with an adult
female who has two children, and earns $900 per month in income. She
also receives nutrition assistance from the SNAP. She describes the
challenges of feeding her family within a limited budget. “You buy what
you can afford. So we buy a lot of generic food and ramen noodles which is Figure 1: Default Screen
like $1.58 and it's a whole bunch of them! But it's not healthy at all. And
bread, you buy a lot of cheap bread because you can eat a lot of peanut
butter and jelly. You buy food that can stretch. You make spaghetti, like a
giant pot of spaghetti, because that will stretch like over a day or two.”
Despite these challenges, she does express a desire to eat healthy. “If I had
more money, nutrition would be in there. Because I want to eat better. I
used to eat better and I was smaller. But I just can't focus on that as much
now. We did get some veggies and fruit but it was almost $5 for some
cherries. And it all goes so
quickly.”
System Design
To help low-socioeconomic status
individuals find healthier food
within their budget, we introduce Better Bites (Figure 1), a tablet-sized
display that sits on a standard grocery cart. Better Bites makes use of
existing store loyalty programs in order to provide personalized
recommendations based on an individual’s historical shopping data. The
individual swipes his/her store loyalty card to start using the system.
Figure 2: Answering a Question
Each time an individual uses the system, she will have the opportunity to Figure 3: Scan an Item
answer one quick question to help the system learn about her shopping
needs (Figure 2). An individual could choose to skip this question and start shopping right away, in which case the
system will rely solely on historical shopping data and generic nutrition information to make recommendations.
Answering one question per shopping trip allows the system to slowly learn more about the individual (or family)
and make better recommendations based on that knowledge, without asking the individual to invest time to answer
all the questions at once. The answers to these questions, along with the historical shopping data, informs the
suggestions it makes to the shopper.
Figure 4: Results of Scanning Cheetos
After having answered (or skipped) the question, the individual is
prompted to scan an item (Figure 3). If, for example, the shopper were to
scan a bag of Cheetos, he/she would see the screen shown in Figure 4.
This shows that Cheetos have a health grade of “D”. We use a letter
grade system to convey the healthiness of food, since nutrient-density
scores have been demonstrated to be an effective way of communicating
a food’s healthiness quickly and intuitively, without relying on confusing
and overwhelming nutrition labels and ingredient lists14,15. This screen
also shows a graphic image of a plate, segmented to illustrate different
nutritional categories; in the example, Cheetos is represented graphically
as being primarily made up of grains. This plate image is based off the
USDA’s myPlate initiative1 to represent the appropriate portions of
meals.
Across the bottom of the
screen are three recommended alternatives. To generate this list of
alternatives, we draw from a large number of foods in the same category
that are healthier and comparable in price. In the example, Cheetos
would be compared to other snack foods. We then narrow that list down
to foods that are appropriate to this particular shopper, based on her
answers to questions and historical shopping data. Among those
products that match these criteria, three items are randomly selected and
displayed on the screen. The randomization process guards against any
criticism of systematic product promotion, as a shopper is likely to see
different alternatives displayed on screen each time he/she shops.
Figure 5: Side by Side Comparison
* http://www.choosemyplate.gov/.
Figure 6: Compare All
A shopper can tap on any of the alternatives to see a side-by-side
comparison (Figure 5) of that item with the item he/she originally
scanned (for example, comparing edamame beans with Cheetos). This
comparison shows the sodium, saturated fat, and fibre contents for each
item. We choose to display those three factors because our discussions
with nutritionists indicated that those are the biggest areas of dietary
relevance to Americans’ nutritional health, and our conversations with
grocery shoppers indicated that they were three areas they were
concerned about. Showing this information helps shoppers learn how to
make healthier choices, even when shopping without this cart. A
shopper can also compare all options at once, and see the price, health
grade and nutrition information for each item all in a row (Figure 6).
Evaluation
We conducted usability tests with three participants as part of
a pilot study. We report here on an additional three tests
(table 1) of our current prototype. All participants had an
annual household income of less than $20,000. Each
participant completed five tasks, and answered a series of
questions to evaluate their understanding of the system.
Based on the results of these tests, we have identified three
areas where our prototype needs improvement.
Age
Gender
Technology owned
P1
23
F
Desktop computer, smartphone
P2
23
M
Smartphone
P3
38
M
Laptop, smartphone, tablet
Table 1: Usability test participant information
Tone. Participants were concerned that the information was presented in an accusatory way. P1 said, “I don’t like it
telling me what I should buy. I’m not eating unhealthy everyday!” We envision this system being used at the
shopper’s discretion, where they only scan items they are interested in learning about. We also envision that
participants could gradually disengage from using the system as they begin to form new, healthier habits.
Familiarity. P3 relied on his prior knowledge of the items that our system had recommended to decide which he
would buy, and hardly referred to the displayed information at all. This surprised us because previous studies of
food-related decision heuristics indicated that familiarity played a less significant role in a decision than factors like
sensory appeal, price, convenience, and health16,17. A field evaluation with additional participants may demonstrate
different priorities. We may also be able to leverage this behaviour in future iterations of our design to nudge
shoppers towards healthier choices, by ensuring they are repeatedly exposed to healthier options while limiting their
exposure to less healthy options.
Limitations and Future Work
We conducted usability tests of our system, but the tasks we asked the participants to perform were related to a
relatively straightforward scenario. We believe that a more thorough evaluation is necessary to take into account the
many factors at play during a grocery-shopping trip.
There are additional aspects of this system we would be interested in exploring further as well. In the future, we see
this grocery cart device as being part of a larger system that allows for shoppers to review their shopping history and
reflect on trends and patterns within it. We would also intend to adapt this system for use independent of store
loyalty programs, in order that it may be used not only in large super markets but also in smaller independent stores
and convenience stores. Finally, we intend to explore the possibility of integrating coupon promotions with the
system to further motivate shoppers to make healthier choices.
Conclusion
Obesity is a common, serious, and costly problem, particularly for individuals of low-socioeconomic status. To
address this, we introduce Better Bites to help people find healthy food within their budget. Although further
evaluation is needed to fully understand how it changes the shopping experience and its potential influence on health
outcomes, initial testing shows that there is a strong desire for this type of system.
References
1.
Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of obesity in the United States, 2009-2010. US
Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for
Health Statistics; 2012.
2.
National Institute of Health. What Are the Health Risks of Overweight and Obesity? National Heart, Lung, and
Blood Institute. 2012.
3.
Finkelstein EA, Trogdon JG, Cohen JW, Dietz W. Annual medical spending attributable to obesity: payer-and
service-specific estimates. Health Aff (Millwood). 2009;28(5):w822–w831.
4.
Drewnowski A. The cost of US foods as related to their nutritive value. Am J Clin Nutr. 2010;92(5):1181–8.
5.
Basiotis PP, Lino M. Food insufficiency and prevalence of overweight among adult women. Nutr Insights.
2002;26:1–2.
6.
Hartline-Grafton HL, Rose D, Johnson CC, Rice JC, Webber LS. Energy density of foods, but not beverages, is
positively associated with body mass index in adult women. Eur J Clin Nutr. 2009;63(12):1411–8.
7.
Resatsch F, Karpischek S, Sandner U, Hamacher S. Mobile sales assistant: NFC for retailers. Proceedings of
the 9th international conference on Human computer interaction with mobile devices and services. 2007. p.
313–6.
8.
Resatsch F, Sandner U, Leimeister JM, Krcmar H. Do Point of Sale RFID-Based Information Services Make a
Difference? Analyzing Consumer Perceptions for Designing Smart Product Information Services in Retail
Business. Electron Mark. 2008;18(3):216–31.
9.
Smith A. Smartphone Ownership - 2013 Update. Pew Research Center’s Internet & American Life Project;
2013 Jun.
10. Black D, Clemmensen NJ, Skov MB. Supporting the supermarket shopping experience through a context-aware
shopping trolley. Proceedings of the 21st Annual Conference of the Australian Computer-Human Interaction
Special Interest Group: Design: Open 24/7. 2009. p. 33–40.
11. Kahl G, Spassova L, Schöning J, Gehring S, Krüger A. IRL SmartCart-a user-adaptive context-aware interface
for shopping assistance. Proceedings of the 16th international conference on Intelligent user interfaces. 2011. p.
359–62.
12. Kallehave O, Skov MB, Tiainen N. Persuasion in situ: shopping for healthy food in supermarkets. Proceedings
of PINC 2011 workshop at CHI. 2011.
13. Kalnikaite V, Rogers Y, Bird J, Villar N, Bachour K, Payne S, et al. How to nudge in situ: Designing lambent
devices to deliver salient information in supermarkets. Proceedings of the 13th international conference on
Ubiquitous computing. 2011. p. 11–20.
14. Drewnowski A, Fulgoni V. Nutrient profiling of foods: creating a nutrient-rich food index. Nutr Rev.
2008;66(1):23–39.
15. Drewnowski A. Concept of a nutritious food: toward a nutrient density score. Am J Clin Nutr. 2005;
16. Eertmans A, Victoir A, Vansant G, Van den Bergh O. Food-related personality traits, food choice motives and
food intake: Mediator and moderator relationships. Food Qual Prefer. 2005;16(8):714–26.
17. Scheibehenne B, Miesler L, Todd PM. Fast and frugal food choices: Uncovering individual decision heuristics.
Appetite. 2007;49(3):578–89.