Including all children in large-scale educational studies is a pressing concern. Omitting certain types of children from studies can lead to skewed findings that promote inaccuracies about learning levels or educational quality. Increasingly, assessments are a method for investigating the quality of education systems, but national assessments are typically conducted in classrooms and may fail to accurately represent a country's full range of children. Alternatively, households can be a site for testing or collecting data on children's learning. Analysing the quantitative data from two studies in India that use both household-based and school-based methods alongside data from Government of India sources, the issues with child representation in school-based assessments are examined, and the benefits of household-based research as an alternative are explored.