I have some experience of EMD with real world results.
In my experience if you have two sites that are equally valued by Google, but one is an EMD and the other is not, the EMD will rank higher. I think everyone knows this though. Additionally, if you have two websites that are both EMD, but one is a dot com and the other is something else, the dot com will rank higher. Also, in a low competition niche, a site with a relatively low amount of good content, but with EMD, will easily rank on the first page and only need a little work to get to the top.
Does this mean a EMD will always rank higher than competition, obviously not, but also, it is definitely a factor in the algorithm. Google stated not too long ago that they were devaluing the EMD part of the algorithm slightly. Although this makes sense in part, as a EMD does not really mean the sites content is going to be any good, in my own experiments, EMD does still count for a lot.
Did your testing include testing the emd against just having the keyword in the url? I have not seem much measurable difference between
in my experience. Now I have not done any head to head testing but the benefit from having the keyword in the url seems pretty strong but the advantage over it having it in the domain as opposed to just in the url seems less clear.