Using machine learning to predict attitudes towards meat consumption

Individual dietary choices are a complex phenomenon influenced by various economic, social and psychological constraints. Despite increasing evidence that current global levels of meat consumption are unsustainable environmentally, detrimental to human health and a driver of farming conditions resulting in low animal welfare, studies show that while meat-reduction appears to be on the rise, rates of vegetarianism or veganism in the UK have not significantly increased in recent years. This study attempts to develop predictive classification models of attitudes towards meat consumption based on demographic criteria and the answers to 3 survey questions on attitudes towards meat consumption. The findings are that even when controlling for reported attitudes towards meat consumption and demographic factors, predicting individual dietary behaviour still has a high degree of uncertainty. This may be due to latent factors such as normative considerations and the perceived difficulty of adopting more plant-based diets, which were not captured by the survey instrument. If this is the case, these findings may have implications for educational and campaigning groups wishing to develop communication strategies of behaviour change prompts, so further research in this area is required.

A Jupyter notebook for the project can be found here.

Back to main page