tained within 15% of initial outcome levels) was 12 weeks, with the longest assessment occurring at one year after intervention (Koegel et al., 1998).
Though these are very positive findings, evaluating studies, and their results, requires cognizance of the prevailing scientific trend, adopted by many journal editors, that favors publication of studies that report successful, rather than unsuccessful, interventions. Thus, the results summarized above, represented as percentages of published comparisons, represent possible outcomes when these procedures are carefully implemented and progress monitored; they do not reflect the number of unsuccessful interventions, which are not reported. As described above, research concerning problem behaviors in individuals with developmental disabilities has generally been strong and plentiful. However, there are relatively few studies directly addressing issues for young children with autistic spectrum disorders. In many cases, interventions that were successful with other populations may be appropriate for young children with autistic spectrum disorders (Wolery and Garfinkle, 2000). Studies testing this assumption with appropriately described and diagnosed children are crucial before it can be accepted. Using the guidelines established by this committee, published research concerning positive behavior approaches to young children was relatively strong in measurement of generalizability and in internal and external validity (see Figures 1–1, 1–2, and 1–3 in Chapter 1). Limitations in the existing studies are not due to a generally poor quality of research, but to changes (and differences) in standards of reporting and research designs in applied behavior analysis and those of the more general, educational and clinical guidelines for treatment evaluation (see Chapter 1). These limitations in these studies were particularly apparent in the selection and description of subjects, random assignment to treatment conditions, and independence of evaluation. As for other areas, these limitations also related to differences in the contexts in which methods were developed. For behavioral interventions that addressed such targets as dangerous self-injury in institutionalized adolescents with profound mental retardation, random assignment, accurate diagnosis, and independence of evaluation may have been of less concern than developing an immediately implementable effective individualized program. However, in order to evaluate treatments for milder difficulties in young children with autistic spectrum disorders, provision of standard, descriptive information about subject selection, subject characteristics and other aspects of research design is crucial in determining what approaches will be most effective for which children.
With these caveats in mind, consistent findings across reviews of published studies indicate several conclusions about current positive behavioral interventions and supports: