Looking forward to presenting my findings and thoughts on family support funding‘s role in child welfare (and our approach to evidencing its effects) from my unfunded pilot research project to DfE colleagues today. Journal article is still under review!

This is a kind of quantitative alternative proposition to the RCT paradigm in the need for causal quants evidence (rather than throwing out quants altogether), which includes a new method for population models that deals with issues of comparability between LAs.

RCTs can be very useful for some types of service. We are trying to build better services on the basis of the best causal evidence that comes from RCTs in the current paradigm, & then scale these up. That makes RCTs a problem for services that don’t conform well to their design

This risks pushing out effective services because of a methodological issue. Instead, I argue, we should be looking at whole system and population causal effects to design services, and then use RCTs and appropriate evaluations to refine these, not build them from scratch.

The other benefit to this is it recognises that services are more than the sum of their parts. I liken this to an ecological landscaping project: planting many of the same tree might work well for one specific part of the ecosystem, but poorly for the rest.

Not all parts of the system work for everyone but in a well-designed ecosystem the features that work well got one will complement the features that work well for another.

Another analogy might be going to Ikea & trying to build a bookcase out of shelves; on the surface the function is the same but the bookcase has some ‘redundancies’ when its parts are considered in isolation, but the end result of the two is very different in cost and outcome!