I think most sociological interventions don't follow the scientific method because you need a control group. You can't just use random people you haven't helped elsewhere as a control, because you need to identify all of the external variables that you can so as to properly assess the effectiveness of your method (ruling out changes external to your intervention).

There aren't too many people comfortable with intentionally letting others suffer for the sake of the scientific method, and those who do have a hard time getting past ethics review boards.

Instead we have this reality where we throw money at problems and claim credit whenever something good happens, and claim we did good work when nothign changes.