Skip to main content

Understanding the rise of vegetarianism

Date

This study demonstrates the potential of microsimulation modelling as a powerful tool for capturing the complex social and demographic factors shaping meat consumption in England.

Project overview

High levels of meat consumption pose significant health risks, including increased incidence of chronic diseases, and major environmental concerns such as greenhouse gas emissions. Efforts to reduce meat intake have seen limited success, often overlooking the impact of social factors such as peer influence, cultural contexts, and lifestyle behaviours. This research applies spatial microsimulation modelling to generate a population-level dataset that captures the social and lifestyle factors associated with meat consumption across England. By integrating survey data with census information, this innovative approach provides further understanding of meat-eating behaviours at a population level.

Data and methods

Using data from the National Diet and Nutrition Survey (NDNS; 2012-2019) and British Social Attitudes survey (BSAS; 2014), we constructed synthetic populations to simulate meat consumption patterns across the diverse demographic and social contexts in England. Logistic regression and clustering analysis identified social and lifestyle factors, including dietary habits, physical activity, religion, political alignment, and sociodemographic factors, that were significantly associated with meat-eating behaviour. Following this, the Iterative Proportional Fitting (IPF) spatial microsimulation model aligned demographic information from the survey data, including sex, ethnicity, household size, and age, with population census data to ensure the simulated populations accurately reflected local demographics. Based on these demographic constraint variables, the model extrapolated the identified social and lifestyle factors to match the population-level data, producing a synthetic population dataset. This allowed for the evaluation of these variables at a population level and provided a granular understanding of meat-eating patterns across England.

Key findings

This research highlighted the effectiveness of a microsimulation approach in uncovering detailed insights into meat consumption behaviours across England, producing a novel dataset for future research. The model successfully simulated meat consumption patterns at the population level, as well as lifestyle and social factors (Figure 1). Model validation was robust, with a correlation coefficient of 0.996 between census and microsimulated data, and the frequency distribution of variables in the census and synthetic dataset closely matched, demonstrating high fidelity in representing real-world distributions relating to age, sex, ethnicity and household size (Figure 2). The findings from this study suggest that microsimulation is a valuable tool for creating population level datasets from smaller surveys which may be used to better inform public health strategies on a larger scale.

Figure 1: Map illustrating variations in meat-eating behaviours across urban London and rural Cornwall (yellow, green and blue scale). Surrounding maps represent key social and demographic factors.

Quote from project partner

“The work carried out within this project has been a great demonstration of the utility provided by the microsimulation approach.  This will provide a fantastic foundation upon which to build further simulation models within this topic.” Tony Craig, Head of Department, Social, Economic and Geographical Sciences, The James Hutton Institute.

Figure 2: (A) Scatter plot displaying age, sex, ethnicity and household size constraint variables simulated counts from the dataset versus census. (B) Frequency distribution of variables in the census and synthetic dataset showing high similarity.

Value of the research

This study generated novel synthetic population datasets using microsimulation, revealing population-level insights into the social and demographic factors associated with meat consumption.

Insights

  • Microsimulation modelling simulated meat consumption patterns in England, incorporating data from the NDNS and BSAS.
  • Social and lifestyle factors significantly associated with meat consumption such as shopping habits, physical activity, religion, and political alignment are integrated into the model, revealing their significant influence on meat-eating behaviours.
  • The dataset produced may help inform nutritional policies and campaigns to reduce meat consumption, promoting healthier, sustainable diets while considering peer influence, cultural norms, and local contexts.

Research theme

  • Health
  • Environment

Programme theme

  • Statistical Data Science
  • Mathematical and Computational Foundations

People

Emily Connell (LIDA, University of Leeds), Andrew Prestwich (School of Psychology, University of Leeds), Tony Craig (The James Hutton Institute, UK) and Jiaqi Ge (School of Geography, University of Leeds).

Funders

Consumer Data Research Centre