Food choice data from online pre-order systems and pupils’ selection of school meals

Dietary intake among children, aged 4-10 years, in the UK, falls short of several
nutritional recommendations, including saturated fat, free sugar and dietary fibre. The
relationship between inadequate diet and public health issues, such as childhood obesity, is
well established. Understanding children’s food choices is essential to trying to address these
challenges, especially in real-world settings such as primary schools. Many schools use
Online Pre-order Systems (OPS), where children and parents select school meals in advance
and remotely. This thesis aims to examine OPS and their food choice data, in order to gain a
deeper understanding of how pupils choose their meals from the options available each day.
Food choice data from one primary school were examined, cleaned and prepared for
analysis. Using these data (11,437 selections, 155 pupils), cluster and time-series analyses
emphasised pupils’ selections of meat/fish-based meals, and that meal selections remained
stable across repeated three-week menu cycles. Data from eight additional schools (with the
same catering provision) and across two academic years, were cleaned, prepared, harmonised
and merged. Analysis of data from the largest school (2021/22 78,604 selections, 545 pupils;
2022/23 72,139 selections, 576 pupils) found significant associations (albeit with small effect
sizes) between meal options and year group, free school meal (FSM) status (pupils in key
stage 2) and pre-order location. Younger pupils were more likely than older pupils to
pre‑select their meals at home. Time‑series analysis again indicated that meal selections
remained stable across four menu cycles of the two academic years.
A categorisation system for school meals was subsequently developed with categories
based on the feature ingredient and food format of the meals. Pupils’ selection rates for
categories were examined across four menu cycles, and there was a level of consistency
observed, in particular with the rank order of meals according to feature ingredient
categories. School meals featuring poultry accounted for more than half of all meal
selections, when on offer; conversely, meals featuring eggs comprised less than 10% of
selections when available.
The categorisation was applied across all eight schools and subsequent cluster
analysis determined pupils’ patterns. For feature ingredients, the largest patterns in both years
were distinguished by selections of meals categorised with red meat as the feature ingredient.
Patterns of food format categories were dominated by pizza and coated fish, that were
consistent for both academic years. Notably, cluster membership was associated with the
school with a moderate effect (whereas factors such as pupil age, FSM status were small),
indicating that individual schools have an important role in pupils’ meal selections. This may
relate to schools’ distinctions with regards to food practices, policies or overall ethos.
Thesis findings across multiple components, emphasise the stability of pupils’ meal
selections, and future work should investigate the relevance of familiarity in children’s food
choices. Furthermore, child age associated with several factors, including the selection of
meal categories and order mode, should be explored. Further work is also recommended to
examine the role of schools in pupils’ meal selection, as highlighted in the final study. Taken
together, this thesis highlights OPS food choice data as a valuable resource, that can be
utilised to provide insights to children’s meal selections in school, and have the clear
potential to inform future school‑based interventions and policies.