What I've described below is the ideal, naturally as soon as politics gets involved in anything you can move away from the ideal rapidly and there's no way of getting politics out of policy formation.

Say you have a policy problem to solve or a policy goal to meet (reducing road fatalities, improving high school graduation rates etc.) and you have a policy you think will work to solve the problem, but you want to check your reasoning or develop a formalised explanation so you can convince another analyst or agency. One way to do this is develop an intervention logic.

The basic format of an intervention logic is a flowchart that outlines the causal relationship between your policy and the desired outcome "This policy will cause A, which causes B, which causes C, which results in outcome Z".

Its not a perfect system, it violates one of the cardinal rules of rationality since its generally used to justify a pre-reasoned position rather than reasoning from scratch and there's inevitably a certain amount of handwaving involved since the causal factors involved in most policy work are very hard to get a grip on, but at least it forces the person using it to state their assumptions and logic explictily.