This article addresses some questions about the determinants and nature of response quality in self-completion surveys of young people. Every young person aged 10-15 in Understanding Society is asked to complete a self-administered paper questionnaire. Data from waves 1 to 4 (2009-2013) are analysed. We find lower data quality amongst children who live in households with lower socio-economic status. We find some support for both increasing social desirability bias and the panel conditioning effect. We find no evidence to support a hypothesis that increased difficulty in contacting the household would be associated with reduced data quality in the youth survey.