Skimming the study abstracts, I am stunned by the number of studies relying on a single child. A "large" study of the effectiveness of social stories included only five students. And then, the recorded improvements are observational and anecdotal. We have no idea if the stories really worked or not. I find little evidence the simple passage of time or the repetition of rules didn't contribute to any observed social skills.
From Focus Autism Other Developmental Disabilities, Fall 1998 vol. 13 no. 3 176-182:
The student was a 12-year-old boy diagnosed with autism, Fragile X syndrome, and intermittent explosive disorder.From the Journal of Applied Behavioral Analysis, Winter 2001; 34(4): 425–446:
This study investigated the effects of written text and pictorial cuing with supplemental video feedback on the social communication of 5 students with autism and social deficits. Two peers without disabilities participated as social partners with each child with autism to form five triads. Some generalized treatment effects were observed across untrained social behaviors, and 1 participant generalized improvements within the classroom.From the Journal of Positive Behavior Interventions January 2006 vol. 8 no. 1 29-42:
A multiple-probe design across participants was used to evaluate the effects of social stories on the duration of appropriate social engagement and the frequency of 4 social skills in 3 elementary-age students with autism. Two students demonstrated generalization to a classroom setting. These findings suggest that the use of social stories without additional social skill interventions may be effective in increasing the duration of social engagement and the frequency of specific social skills.Based on these journal articles, if you study one student, you get a 100% effectiveness. Three students earns 66% effectiveness. Five students, the generalized improvement is observed in 20% of the subjects. In other words, the larger the cohort, the lower the percentage of observed improvement. There are simply too many problems with small tests, including the fact it is nearly impossible to generalize about autism, which is broadly defined, with a study of five students or fewer. You certainly cannot generalize a study of one, two, or even three students. Statistically, that simply isn't possible.
I never understood (and still don't grasp) social stories, examples, role-playing or anything similar. Such things don't make sense to me. If "Johnny" or "Susie" is doing something wrong, tell Johnny or Susie, not me. I've learned to just nod and say "sure" because as I child I got tired of the repetition. "Sure" wasn't a lie, but it was confusing. Now, I just bluntly tell people — tell me what the rules are and don't confuse me with stupid pictures or stories about people I don't know.
I realize that's not a universal, but I seldom get how "other people" think or act, so why would I grasp that some fictional character's motivation is supposed to parallel mine?
When I try to understand characters in a book, it takes a lot of extra effort. I make notes to myself and look things up constantly to understand why a character might do something. When a character turns out to be deceptive, I seldom grasp that until much, much later than other people. Social stories aren't any easier than literature.