Or rather, how I would like to see policy being made. A few people have reposted links to Ben Goldacres submission to the UK cabinet office on the use of randomized controlled trials in policy making recently. Which has got me thinking about the whole evidenced based policy mindset again. And then in the paper this morning was this: Child poverty costs NZ $10b a year.
Which is a fine place to lay out an example of how I wish policy was made. Really.
First, identification of the problem. In this case, a study is bringing child poverty and it's associated costs to the publics attention. The first thing to do is to have a look at the source of the report, in this case an independent researcher by the name of John Pearce. Is he a crackpot? Or is he someone with a relevant background and experience who has done the relevant work? In this case the Children’s Commissioner’s Expert Advisory Group (EAG) appear to rate him, so, if we are interested in doing something about the problem, then we should pay attention. Does the work stack up against other works of a similar nature here or overseas? More weight should probably be given to the problem if similar conclusions are being reached by local studies - for the sake of this example, I'm assuming it does stack up with similar work (likely but I can't be bothered going and finding sources to back me up on this atm, it's beside the point)
What's next? A problem has been identified. It's not good for people and it's costing us money. Get some of the people who have done the work in a room with people from the appropriate ministries. In this case, I would guess the Ministries of Work and Income, Housing and Finance. Lock them in a room with a statistician until they have identified at least 3 or 4 areas that can be tested and compared, 3 or 4 possible policies that could alleviate poverty, and a series of concrete measures. Things like, how much less is poverty costing us in a particular area, how does the number of people below the poverty line compare, before and after - taking into account population changes.
Implement the measures in a the different areas, 1 treatment per area. Wait. Years if need be. Build something into legislation that will protect the trials if need be. Compare the results against the measures proposed at the beginning and then implement the policy that produced the best results.
It's probably one of the reasons I'll never be good at politics. I wouldn't want to push for one particular policy over another. I'd rather push for a trial composed of a number of different measures and select the best based on the results.