Household food waste across multiple food groups in Dodoma, Tanzania: A Multinomial Probit approach

Food waste (FdW) undermines food security both directly and indirectly by disrupting sustainable food systems. Understanding the root causes of FdW and its effects across multiple dimensions is crucial. Despite extensive global research, studies focusing on Household Food Waste (HFdW) in Tanzania remain limited. This study addresses that gap by analyzing the determinants of FdW generation across combinations of Food Groups (FGs) in Dodoma, Tanzania. A cross-sectional survey of 402 households was conducted to collect data on FdW-related behaviors and preferences. Principal Component Analysis (PCA) with varimax rotation was applied to reduce dimensionality, revealing that Cereals, Legumes & Pulses, and Roots & Tubers contributed most strongly to the retained components that capture the main variation in HFdW patterns. Each group was then dichotomized into high or low waste based on the median, and the resulting binary indicators were combined to create eight possible FdW patterns, representing all combinations of waste intensity across the three groups. These FdW patterns formed the dependent variable in a Multinomial Probit Regression Model (MPRM). The model revealed that demographic factors such as higher education, female-headed, and older households were associated with lower HFdW. Behavioural practices, including meal planning and leftover reuse, also reduced waste, while attitudinal factors such as greater awareness of FdW impacts further reinforced this effect. In contrast, weaker perceptions of money value were linked to higher FdW levels. The findings highlight the importance of behavioral and contextual factors in shaping HFdW. Policymakers should consider targeted strategies such as meal planning support, storage and handling education, and gender-responsive interventions to reduce FdW and enhance food security in Tanzania and other low-resource settings.