From (Powell et al., 2024)

Evaluators can break the Janus dilemma and make the best use of causal maps in evaluation by considering causal maps not primarily as models of either beliefs or facts but as repositories of causal evidence. We can use more-or-less explicit rules of deduction, not to make inferences about beliefs, nor directly about the world, but to organise evidence: to ask and answer questions such as:

• Is there any evidence that X influences Z?

• . . . directly, or indirectly?

• . . . if so, how much?

• Is there more or less evidence for any path from X to Z compared to any path from W to Z?

• How many sources mentioned a path from X to Z?

• . . . of these, how many sources were reliable?

We also argue that this is a good way of understanding what evaluators are already doing: gathering and assembling data from different sources about causal connections in order to weigh up the evidence for pathways of particular interest, like the pathways from an intervention to an outcome.