Directed Graphs / Judea Pearl's methods

[curious about method]

Since we're talking about Bayesian stats, here's another method I'm curious about--the Judea Pearl variant of structural equation modeling (which I usually see represented as DAGs but I know that's not exactly the same thing). All my friends in public health love it, but it's not nearly as common over here. How much more complicated is the Judea Pearl version than the version outlined in Bollen (1989)? What are the skills and/or background you need for Pearl's that you don't necessarily need for Bollen's (and vice-versa)? Are there any introductory texts that are easy to hack your way through? I found the Morgan & Winship explanation useful, and the same with the Elwert (2013) version in Morgan's edited volume. But in both cases, I thought they didn't really get into the nuts and bolts of how to do it. Meanwhile, I found Pearl's book pretty tough for a beginner.

Thoughts? Is this something people use?

[maybe]

I had never heard of it, but it seems like a non-parametric version of SEM. Is that a fair representation? What uses did you have in mind? I am sorry I am not of much help, but it seem like I am even more of a novice on the topic.

[curious about method]

It being a non-parametric method for SEM is pretty close. It seems like it involves some similar terminology to branches of social network analysis (edge, in-degree, etc). In sociology, the references to the method seem to be mostly in Morgan & Winship and some theoretical work by Elwert, but I haven't been able to identify why this method is better than a regular SEM, what extra things you would need to learn in order to use it properly, or even how it works (excepting Pearl, which I just can't make my way through).

One thing it doesn't do as well--it doesn't handle measurement error the same way (I think).