What she says is what she does: Survey versus behavioural measures of food choice
How often did you eat a piece of white bread in the past four weeks? 1 slice per week? More? Are reported memories of food intake the same as actual food behaviour? Much of the burden of non-communicable diseases (NCDs) in developing and developed countries is driven by what people eat and drink, so influencing dietary behaviour for the prevention and management of NCDs is therefore a key challenge for policymakers. But measuring dietary changes can be a challenge with food surveys open to bias. As part of my PhD in behavioural economics at The Max Planck Institute in Bonn and University of Cologne, I wanted to see whether measuring what people do, rather than what they say they do, would provide a more credible channel to understand whether a program may be said to affect dietary behaviour.
We investigated the food choice behaviour of women taking part in a Low Carb, High Fat (LCHF) lifestyle intervention in under-resourced communities in the Western Cape. I had the opportunity to work with Eat Better South Africa (EBSA) and the women who had participated in the program from Ocean View and Atlantis. EBSA is a non-profit organisation, created to address health and socioeconomic issues faced by underprivileged South Africans, particularly women. Past programs were six weeks long and involved weekly 2-hour educational sessions at a central community hall to teach participants about nutrition, NCDs, shopping on a budget, cooking and how to access healthier foods.
Measuring behaviours of interest (e.g. what people chose to eat following receiving some advice) is critical to demonstrate the mechanism through which such a program may succeed or fail. Previously, the effectiveness of the EBSA program was evaluated qualitatively through focus group discussions. This is in alignment with much of the scientific literature, where epidemiological nutrition studies, such as the Nurses Health Study, use diet assessment tools that require subjective responses, such as the Food Frequency Questionnaire (FFQ), 24 food recall or food diaries. It is understood that these subjective measures may suffer from bias and noise due to people’s inattention to what they eat, inability to recall fully, social desirability bias and a lack of incentive for accuracy. Food choice behaviour is typically not observed and our question was whether this was a more objective measure, or at least, how this may compare with subjective surveys.
We designed a behavioural measure of food preferences and compared it with a Food Frequency Questionnaire about what participants had eaten over the past four weeks. The behavioural component was a grocery shopping activity at a major supermarket frequented by our participants, where they could spend a R250 SMS voucher. Participants took home whatever food they bought and sent photos of their groceries and the receipt. Eligible participants were drawn from the communities where EBSA operates already and plans to operate in the near future. Adult women could take part in the study if they had taken part in the EBSA program, or were eligible to take part in future programs (a control group). Thus two groups of women were included and their responses compared. About a hundred women took part in our study.
The Covid-19 lockdown introduced additional complexity to field research in July/August 2020. We needed to remove risks of interpersonal contact so all our interviews were conducted over the phone. Remote data collection has its own challenges of course – for example, load-shedding interruptions to WiFi and potential ambivalence towards interacting with an unknown researcher over the phone. However, what made my experience of remote data collection encouraging was the generosity of the women I talked to and the support from the EBSA team when we reached out to communities for recruitment.
We organised the quality of the women’s purchases according to the EBSA program’s traffic lights lists. According to EBSA, RED is to be avoided, ORANGE for occasional consumption, and GREEN to be eaten liberally. The results showed that the EBSA group made fewer RED choices and more GREEN choices in their grocery shopping compared to the Control group of similar women. Moreover, the behaviour we observed was reflected in their survey responses, validating the FFQ in our sample. The EBSA group reported lower consumption of sugar, refined carbohydrates, refined seed oils and junk foods and greater consumption of affordable sources of fish, organ meats, eggs and traditional fats. Data were collected during one of the strictest lockdowns globally, which impacted participants’ employment and food security. Despite the adverse conditions, we observed a marked difference in food choice behaviour by the EBSA group in line with the LCHF program advice. This speaks to the sustainability of the nutrition education model with women who had taken part, and by inference attempted to follow an LCHF diet.
While only an approximation of usual eating habits, the study showed that the inexpensive FFQ is representative of women’s revealed food preferences. While behavioural economists will likely continue to take a skeptical view of survey measures, the results from this research suggest that when it comes to food: what she says is what she does.
Archer, E., Lavie, C. J., & Hill, J. O. (2018). The Failure to Measure Dietary Intake Engendered a Fictional Discourse on Diet-Disease Relations. Frontiers in Nutrition, 5.
Hallsworth, M. (2017). Rethinking public health using behavioural science. Nature human behaviour, 1(9), 612-612.
Lusk, J. L., & Schroeder, T. C. (2004). Are Choice Experiments Incentive Compatible? A Test with Quality Differentiated Beef Steaks. American Journal of Agricultural Economics, 86, 467–482.
Medic, N., Ziauddeen, H., Forwood, S. E., Davies, K. M., Ahern, A. L., Jebb, S. A., … Fletcher, P. C. (2016). The Presence of Real Food Usurps Hypothetical Health Value Judgment in Overweight People. ENeuro, 3(2).
Prince, S. A., Adamo, K. B., Hamel, M. E., Hardt, J., Connor Gorber, S., & Tremblay, M. (2008). A comparison of direct versus self-report measures for assessing physical activity in adults: a systematic review. The International Journal of Behavioral Nutrition and Physical Activity, 5, 56.
Pujol-Busquets, G., Smith, J., Larmuth, K., Fàbregues, S., & Bach-Faig, A. (2020). Exploring the Perceptions of Women from Under-Resourced South African Communities about Participating in a Low-Carbohydrate High-Fat Nutrition and Health Education Program: A Qualitative Focus Group Study. Nutrients, 12(4), 894.
Roberto, C., & Kawachi, I. (2015). Behavioral economics and public health.
Webster, C. C., Murphy, T. E., Larmuth, K. M., Noakes, T. D., & Smith, J. A. (2019). Diet, Diabetes Status, and Personal Experiences of Individuals with Type 2 diabetes Who Self-Selected and Followed a Low Carbohydrate High Fat diet. Diabetes, Metabolic Syndrome and Obesity: Targets and Therapy, Volume 12, 2567–2582.
About the author
Sofia Monteiro is a behavioural economist. She is a research fellow at the Max Planck Institute in Bonn, and a PhD Candidate at the University of Cologne. She holds a Bachelor in Psychology and Economics and a Masters in Applied Economics from the University of Cape Town. Sofia’s research focus is to understand cognitive and behavioural barriers to healthy decision making.
As an economist, Sofia is concerned that the burden of diabetes is a growing global problem, not only for patients and families, but also for health insurance providers and the wider economy. As a researcher she understands that health-related behaviour is difficult to shift, and measuring and tracking behaviour in the field is often a challenge. Sofia is personally motivated by the impact that the LCHF lifestyle has had on her family’s health. She believes in empowering people with nutrition education and using behavioural insights to help people navigate complex decision environments.