Administrative data are increasingly being used to study inequalities in education. Yet a well-known difficulty with such resources is the limited information they hold. A commonly used proxy for children coming from a low-income background is their eligibility for free school meals, yet this is likely to be of little use in measuring academic achievement amongst middle and high-income groups. This study adds to the literature by showing how eligibility for free school meals -- averaged over the time a child has spent at school -- is the best available proxy for low income during childhood. In contrast, creating a continuous index combining free school meal eligibility with information on the neighbourhood in which they live represents the best way of comparing educational outcomes across children from low, average and high-income backgrounds.