I got useful information from these forums (thanks). This is the study path I used:
* Read the EJB 3.1 spec (ad nauseum).
* Read the O'Reilly EJB 3.1 book.
* Read several Oracle tutorials.
* The Enthuware test-prep tool. I took all those tests multiple times, but usually so long between as to forget the contents.
* Deployed many, many Stateless, Singleton and MDB beans to two different servers.
* Deployed (at least) one Stateful bean. It was enough, because of the overlap. Also, I see little utility for those in reality.
* Deployed a few Asynchronous beans. I should have done more with them.
* Used JBoss (abandoned for this purpose at one point because it did not support true Global Portable JNDI lookups).
* Used Netbeans / Glassfish.
As expected, my score was better than the most recent Enthuware test. It was up 10%. I strongly recommend that, and I recommend the multi-server deployment. After all, the point of these tests is to learn along the way. I had used JBoss prior to this on-the-job (and still would, despite this oversight with java:global).
I had been planning to take this test for a very long time, but finally "time-boxed" it only over the last couple of months. I did not take time off from work to do this, and I do have other homelife responsibilities. But, I did take some reading along on vacation.
No one does anything from a single motive.
--Samuel Taylor Coleridge