Intention to Treat is a term used in experimental evaluation to indicate that analysis of the results will include all cases randomly allocated to the intervention group, whether or not they receive the intervention. It is based on the assumption that in real life children and families may not turn up to appointments, take their medication or follow instructions. So including data on the progress of children dropping out of a program indicates impact on outcomes in the real world. The approach also helps to neutralize the effects of 'crossover' (when children in the control group get onto the program) and drop-out. Some analysts are more interested in what is called 'treatment on the treated', that is measuring just the impact on children getting the intervention.