Can I answer on the term “scientifically proven”?
This is a cliché used in all the arguments for defending certain methods, especially the Lovaas technique. As I pointed out, the term “scientifically proven” is used in a misleading way. We're not talking about 52% but about 47%. That's the magic number the 1987 Lovaas study reported. These are figures that, as you said, are much more exact than they can be for studies with low effect sizes that are related to methodological problems pointed out over the 30 years following that study. Therefore, it hasn't been scientifically proven that the ABA method is that effective.
There have been at least three meta-analyses, including those conducted by Patricia Howlin, by Ospina, and the Australian study conducted by Spreckley. Those meta-analyses are different from one another. Spreckley's analysis is about the absence of effect. Patricia Howlin discusses an unpredictable effect on some people. The third study talks about a low effect.
We should also point out that there are now a few randomized controlled trials for studies on communication. We're talking about studies that are conducted for two hours a week instead of for eight to ten hours, and their effect sizes are similar.
The cost-benefit ratio of the ABA method is currently lower than that of the other communication techniques. We must admit, however, that effects are low in all cases.