Off The Record #10: Exploring avenues to better nutrition

How priming and framing food options affect consumption behaviours in Kenya

By Ruth Canagarajah, David Wunschel, and Josphat Korir

Photo by Lotte Löhr on Unsplash

Good nutrition is hard. Whether you are tempted by sweets, tend to skip breakfast, or never eat your greens, all of us are at different stages on our journeys to better health. In this, you are not alone: poor nutritional practices affect billions of people on a daily basis. While structural issues are a large part of why 795 million people go hungry every day, human behaviour is also an important factor in why 2 billion people lack vital micronutrients in their diets, and 1.4 billion people are overweight or obese, both causing serious long-term negative health effects which affect productivity often on national levels. For many people, behavioural factors — including a lack of information, engagement or self-control — drive unhealthy consumption, even when nutritional options are available and accessible.

Behavioural science seeks to understand the factors which influence decision-making, and test potential interventions or “nudges” which can act as levers toward changing a given behaviour. This process lends itself well to promoting healthier eating, and indeed features this common question in behavioural science: how can we tweak the choice architecture of food options to nudge people towards healthier choices?

For example, the “Nudge to Nobesity II” study demonstrated that items placed at the beginning or end of a menu are ordered up to twice as much as items placed in the middle. Similarly, name changes can influence people’s diets: when the American bakery-café chain “Panera Bread” re-imagined and re-labelled its “Low Fat Vegetarian Black Bean Soup” into “Cuban Black Bean”, sales increased by 13%, according to a study conducted by the World Resources Institute’s (WRI) Better Buying Lab.

Why are we not as healthy as we could be?

Choosing less nutritious foods is not always due to systemic issues, such as the lack of access or finances. In Nairobi, where Busara operates, fresh fruits and vegetables are cheap and available largely due to Kenya’s rich agricultural diversity. Despite this, many Nairobians still seem to rely on diets that tend to be carbohydrate-heavy.

Patterns of food consumption and nutrition behaviour are grounded in a complex environment. A blend of multiple factors, from biological to anthropological to economic, influence how and why people develop and change their food consumption patterns (Nestle et al. 1998). In Nairobi, this preference towards less nutritional options can be for many reasons: they may appear to be more filling or have better relative pricing compared to healthier options. With rising incomes, traditional staple foods like millet, sorghum and maize have become less prevalent whereas wheat products, meat and processed foods are on the rise.

In order to understand what levers exist in improving nutritional consumption, Busara adapted and contextualised two existing studies on reframing healthier alternatives and introducing mixed indulgences. In this quick and dirty hypothesis-testing study, we targeted low-income individuals in Nairobi in order to understand what works — and what doesn’t — when promoting healthier eating. Specifically, we sought to answer three questions:

• What are the typical diet patterns among lower-income residents of Nairobi?
• Will consumers choose healthier foods if those options are framed more appealingly or with more indication of satiety?
• Does bundling healthy and unhealthy options change consumer habits when receiving an emotional prime?

We invited 221 people from Kibera, a low-income informal settlement in Nairobi, to take part in a survey and two experiments.

Our Research Design

Experiment #1

In our first experiment, we contextualised the WRI study to Kenyan food menus and explored whether simply re-framing healthier menu options could nudge consumers to choose options that differed from typically popular dishes. First, we held focus groups with Busara field officers to decide on typical lunch items in order to standardise our menu designs. We randomised our participants into three groups: a control menu group and two treatment menu groups. Each group then received a typical lunch menu with three healthy and three unhealthy dishes. The unhealthy dishes — such as fried chicken, chips, fried goat meat or white ugali (a very popular type of cornmeal porridge) — were always framed the same way. The healthy dishes — such as fish, brown ugali, bean stew or managu (a local vegetable) — were framed differently for the two treatment groups.

The “Control” menu showed straightforward descriptions of the dishes that are found in all low-income restaurants, for example Brown ugali, managu and fish. The “Treatment 1” menu reframed the healthy dishes in a way that appealed to taste and provenance: Brown ugali and managu served with grilled Coastarian fish. The “Treatment 2” menu reframed the healthy dishes as nutritious and satiating: Satisfying brown ugali with managu and fish. All unhealthy dishes remained consistently framed in simple terms. The dish names appeared in Swahili, with the English translations below:

Experiment #2

In the second experiment, we wanted to see: “Are mixed bundles of Kenyan snacks chosen more frequently in comparison to natural unhealthy bundles, especially when there is a messaging prime?” This concept is based on “mixed indulgences”, used to gently guide consumption changes by mixing hedonistic and utilitarian options. Mixed indulgences allow the consumer to gradually choose healthier options instead of forcing them to scrap unhealthy preferences entirely.

The bundles we chose are reflective of natural snack pairings that Kenyans consume. We started by giving each participant Kenya Shillings 500 ($5) as a reimbursement for participating in the experiment. All snack bundles, healthy and mixed, were to be purchased from the KSH 500 that participants had, thus forcing an actual consumption choice. A third of the participants also received the following prime that emphasises multiple gain framings of eating healthier:

“When you and your family eat healthy foods, you gain. You gain energy throughout the day. This helps you work better. When you work better, you gain satisfaction. When you gain satisfaction, you return home happy and ready to spend time with your family. Eat healthy, gain more days.”

We then escorted the participants to a table where they could choose between six food bundles that are very common in Nairobi. Again, there were three unhealthy options such as smokie plus mandazi (grilled sausage with fried dough) and three mixed bundles such as mandazi and a boiled egg. We then analysed the effect of priming on purchase of food bundles.

Additional Survey

Finally, a subsample of 40 respondents representative in terms of differing income levels within Kibera via residential area participated in a qualitative interview describing what they eat in a typical day and why. This allowed us to learn more about the typical diet patterns of Kibera residents and the food-related challenges they face.

What did we find?

1. Framings of satiety and taste/provenance didn’t have an effect on respondents’ food choices.

Respondents who saw the framed versions of the healthy dishes were slightly more likely to order them — but as our “Off The Record” studies such as this work with limited sample sizes, these effects were not significant. When the dishes were framed through taste and provenance, 66% ordered them. When they were framed as satiating, 64% ordered them. In the control group, 61% ordered the simply-worded healthy options. The numbers across treatment and control also suggest a natural proclivity toward the healthier options.

Among the dishes, one dish turned out to be overwhelmingly popular: half of the participants ordered brown ugali with managu and fish. This might be due to the fact that the majority of Kibera’s residents are Luo and Luhya, two ethnic groups whose home provinces are close to Lake Victoria and whose main staple foods are ugali and fish. This dish was especially popular in the group which saw the satiety framing. It was ordered by 43 out of 80 people.

2. Receiving an emotional prime didn’t change consumer choices when bundling healthy and unhealthy options.

Priming people on the benefits of a healthy diet right before giving them the option to choose between healthy and unhealthy snacks didn’t significantly change their decisions.

Participants who didn’t get the prime chose a mixed bundle 43% of the time, and participants who received a prime chose a mixed bundle exactly half of the time. While those who received the prime did choose mixed bundles more frequently, there is no statistical significance to suggest that this is a causal relationship.

Again, respondents showed a strong preference for a very specific choice. In both groups, 40% or more chose an unhealthy bundle consisting of Smokey (sausage) and Chapati (white flat bread). A possible explanation for this is that the prime is not strong enough to displace consumer preferences — crafted from long-standing decision-making — from natural pairings of foods to mixed indulgences.

3. Kibera residents have a carb-heavy diet that is largely influenced by their income.

The survey that followed the experiments provided more granular, qualitative insights into how Kibera residents make their food choices. Income is a very important factor in determining where people buy their food. We learned that generally:

• Low-income earners prioritise affordability and proximity of food stores over food quality.
• Middle-income earners seem to shop in places with higher hygiene standards.
• High income respondents mostly value convenience and high food quality.

In our sample, an average household of four spends approximately KSH 300 ($3) per day on food — this limits our respondents to a small set of foods they can choose from. The variety in our respondents’ diets is thus quite small. For breakfast, most have tea and carbohydrates in the form of bread, mandazi, chapati or sweet potatoes. Lunch usually involves carbs in the form of ugali or rice mixed with beans, maize, potatoes or cabbage. For dinner, the vast majority of people eats Ugali, mostly accompanied by vegetables and fish or meat.

Conclusion and main takeaway

While framing and priming have shown significant effects on Western populations, they had no significant effects on our respondents from Kibera. Although our OTR did contextualise the Western studies to the Kenyan food context, there are important reasons why the experiments yielded inconclusive answers. While one limitation could be the limited sample size, the other is that local contextualisation needs to go deeper through iteration. This means adapting the study, overhauling or questioning the elements that do not work, and adjusting the research approach until it better mirrors the contextual reality.

It is possible that the Kibera respondents see carb and meat-heavy foods as either the most economical and comfortable purchasing decision which requires different kinds of nudges aside from framing and mixed options. Alternatively, the fact that menus in low-income restaurants are often straightforward and simple, rather than full of adjectives describing satiety or other appealing framings, is a potential case study that demonstrates how the WRI study may not neatly contextualise to the Kenyan low-income environment.

If this study shows anything, it is the absolute importance of hyper-contextualising studies. This is front of mind to us, but we know this is an iterative process for each study. This is the main benefit of “Off The Record” formats: to be able to test and iterate your research design on a small sample size, before full implementation to scale.

A future version of this study can iterate on the current approach and do the following things for Experiment 1 on re-framing menu options.

1. Have multiple qualitative sessions with Kibera restaurants and food institutions to understand the most common preferences in consumption behaviour. Indeed, this could be conducted through simple ethnographic surveying.
2. Include a larger sample size to increase confidence in findings.
3. Link hypothetical purchases on the menu to actual purchasing decisions in order to tie preferences to consumption behaviour.

For Experiment 2 on mixed bundles, one way of iterating on the current approach is to compare the relative purchasing behaviours from multiple primes. One prime could be more emotive and related to family well-being and health. Another could be loss aversion, to provide a useful comparison to the gain framing. Again, a higher sample size could provide better insight into different consumption preferences with and without primes.

If this is something you’d like to do in a follow-up Off The Record, please reach out to us on Twitter or via email.

A final note

Our goal is to make research accessible in order to start, continue or inform conversations that help us to better understand human behaviour.

With this in mind, many “Off The Record” posts provides access to the full findings from our studies, freely available here.

As a commitment to Open Science, we keep this anonymised data live at this page for all our on-going research efforts.

This blog post is part of our larger “Off The Record” initiative where we share findings from small-scale research projects, designed to collect initial data and kick-start a conversation.