I guess I have to make this clearer:  I am not opposed to Applied Behavior Analysis.  A lot of very good education is done using its methods.  What I object to is the denial of the inner reality of the autistic person.  Richard Bromfield wrote so moving on this topic recently that he forces me to break my rule and link to the Huffington Post (still not a credible source for autism information or opinion):

Any understanding of this disorder that suggests the child with autism has an internal emotional world has been neglected. For example, this past summer I discovered that since its founding in 1979, the Journal of Autism and Developmental Disabilities, a premier journal "devoted to all aspects of autism spectrum disorders [...including] clinical care ... and treatment for all individuals," had published 2,262 articles, not one of them focusing on psychotherapy or counseling with a child with Asperger's.

The bias is undeniable, even if understandable and defensible. Early research announced loud and clear that autistic children lack a theory of the mind, a capacity for symbolic thought, rich and varied affect and so on. That fundamental knowledge was aimed at extinguishing the belief that any child with autism, however high-functioning or mild the form, might benefit from play and talk therapy. The notion that a clinician could sufficiently modify the techniques of traditional play and relationship-based therapy to befit the child's special needs and deficits never occurred as a possibility. The conventional thinking in autism judged any kind of therapy that addressed the child's feelings and inner experience to be psychoanalytic, misguided and, some went so far as to say, negligent.

I've been watching a lot of B.F. Skinner videos on YouTube lately (beats reruns), and the extent to which the real founder of educational behaviorism was willing to deny that people are more complex than starving pidgeons is pretty shocking:

B.F. Skinner considered free will an imaginary internal force.

Behaviorists only attempt to deal with things that are easy to measure in experiments-- they tend not to believe that other things actually exist.  This means that they have an automatic advantage in experiments that gage the efectiveness of interventions-- the intervention designed only to deal with things that can be seen in a experiment will tend to outperform interventions that take a more holistic approach.

But that has more to do with the tools used for measurement than anything else.

There is a very real problem with behaviorist interventions: they get less effective the longer they are applied.  They work extremely well in the short-term, but effects tend to wear off.  I think it's important to see behaviorism and psychiatric medication in much the same way:  they create a window in which actual education or therapy can begin, but they don't replace intrinsically motivated learning or talk therapy.

In the worlds Skinner imagined, a world Beyond Freedom and Dignity, there would be nothing wrong with trying to change the identity of a little gay boy to make him straight, as Ivar Lovaas, Great God of ABA, did, because there really is no inner reality.  Only behavior matters-- everything else is imaginary. 

In the autism world, experts and educators tend to ignore the experience of autistic people.  I think this comes at least in part from the extent to which behaviorists have been allowed to define autism.  You have been trained to ignore how we feel and what we think.