I love the premise behind data driven giving, but it's really hard. In tech, we are all used to running experiments to determine whether things really work. But

I love the premise behind data driven giving, but it's really hard. In tech, we are all used to running experiments to determine whether things really work. But experimental design is not trivial and really benefits from expensive specialists. Bad analysis and experimental design has many silent failure modes in both the false positive and false negative directions. I have to imagine that these reports that OP talks about charities sending their donors are full of broken methodology. "We caused 10k kids to graduate high school!" when 9990 would have graduated anyways. If anyone knows of a way for a charitably minded data scientist to donate time and expertise, I'd love to hear about it.