Obviously, the judges are not randomly assigned. But we’re able to control for a whole of judicial characteristics, including political party (which is the normal selection story), age, experience, gender, etc. To the extent that selection is correlated with these controls, we can do a decent job of accounting for it. We also have some robustness checks associated with the story that these free-market oriented judges take the programs. So yes, we deal with selection the best we can — which is not to say completely! One of the more interesting selection stories we are still working on is the idea that perhaps it is more intellectually curious judges that opt-in and therefore are better at dealing with complex economic evidence. I think that story would predict that these judges would do better in both simple and complex cases — which is not what we observe. Anyway, the short answer is yes. The longer answer with details you might be looking for is in the paper. Another way to deal with this would be to expand the data set back far enough in time so we could have a natural experiment with the same judges before and after LEC treatment, e.g. with judge fixed effects. Not enough data for that now. But maybe in the future. It takes a long time to collect this stuff with the level of detail in the decisions we are working with.