Schools are places where children can learn behaviour, skills and attitudes that have lifelong relevance. In England, despite the continuing emphasis on attainment, there are clear moves to consider also the wider and non-cognitive outcomes of schooling -- such as pupils' development of trust, critical thinking and civic-mindedness. However, there is little existing evidence on how such non-cognitive outcomes can be improved through school-based interventions. This paper presents findings from a quasi-experimental design using 2722 pupils in 42 primary schools. A treatment group of schools participated in Philosophy for Children (P4C) for 18 months, whereas the other group of schools was a clean control. The outcomes compared were pupil self-reports with an instrument designed to assess "social and communication skills", "teamwork and resilience" and "empathy" and a number of other such constructs. Post-intervention comparisons show that pupils who received the P4C intervention were ahead of their counterparts in the comparison schools, and this was generally more so for those pupils living in relative poverty (FSM-eligible). Teachers reported that positive effects could be observed in pupils' confidence in questioning and reasoning, and pupils generally reported that they enjoyed the intervention. However, the differences are small, and it is not clear that the two groups were comparable at the outset. Nevertheless, there is promise that targeted school-based intervention such as P4C can improve pupils' non-cognitive outcomes, and there are lessons for how to conduct such studies and how to assess the wider outcomes of schooling.