Louise Morpeth explains why new thinking about ‘drift’ as a normal aspect of service implementation is persuading researchers to ask themselves not "Why is fidelity so hard?" but "How much infidelity is tolerable?".

“Policy makers, educational leaders, and teachers value research when it supports their existing practices and beliefs, but often reject it if it does not, regardless of its quality…”

Randomly allocating children to care in order to test its effectiveness continues to be a contentious proposition, particularly in the US; Oregon researchers have been able to remedy the scarcity of impact evidence by using randomized methods to test the value of giving foster parents expert support.