Up-vote for Popper!
Ask your friend: "How would you know if that treatment was not working?" And by 'not working' I mean (a) no change (b) a change for the worse (c) a change for the better, but not related to that treatment (d) a change for the better, but at a greater cost than benefit. The more clear the idea of how to know if that treatment was not working, the more likely it's science.
I don't know you, your friend or your son's friend. I don't know what any of you need, although I'm sure just being compassionate instead of always being right is a part of it. My Popper-inspired suggestion above is good for distinguishing science from non-science. It may not be helpful to what any of you need.
First, a short personal note to make you understand why this is important to me. To make a long story short, the son of a friend has some atypical form of autism and language troubles. And that kid matters a lot to me, so I want to become stronger in helping him, to be able to better interact with him and help him overcome his troubles.
But I don't know much about psychology. I'm a computer scientist, with a general background of maths and physics. I'm kind of a nerd, social skills aren't my strength. I did read some of the basic books advised on Less Wrong, like Cialdini, Wright or Wiseman, but those just give me a very small background on which to build.
And psychology in general, autism/language troubles in particular, are fields in which there is a lot of pseudo-science. I'm very sceptical of Freud and psychoanalysis, for example, which I consider (but maybe I am wrong?) to be more like alchemy than like chemistry. There are a lot of mysticism and sect-like gurus related to autism, too.
So I'm bit unsure on how from my position of having a general scientific and rationality background I can dive into a completely unrelated field. Research papers are probably above my current level in psychology, so I think books (textbooks or popular science) are the way to go. But how to find which books on the hundreds that were written on the topic I should buy and read? Books that are evidence-based science, not pseudo-science, I mean. What is a general method to select which books to start in a field you don't really know? I would welcome any advise from the community.
Disclaimer: this is a personal "call for help", but since I think the answers/advices may matter outside my own personal case, I hope you don't mind.