Social competence as a future direction for early intervention programmes.

The effectiveness of comprehensive programmes providing early intervention services to children with clearly documented handicaps has been evaluated extensively over the past 2 decades (see Guralnick & Bennett, 1987). From the outset, it has been apparent that numerous methodological and practical problems have made this evaluation effort a particularly difficult one. The diversity of subject populations, including their etiologies and the varying severity of disabilities, have combined with a host of programmatic factors such as the timing of intervention, its intensity, and the developmental model applied, to challenge our ability to arrive at a reasonable assessment of effectiveness. Questionable scientific practices relating to the possibly biased nature of observer judgments, inadequacies in interobserver reliability estimates, and non-standard and difficult-to-defend statistical approaches have allowed alternative interpretations of even some of the better controlled studies in the field (Guralnick, 1988). Despite these difficulties, two general evaluation approaches have emerged yielding very similar outcomes. The first approach has focused on global scatistical analyses of the diverse array of scudies using meta-analysis cechniques to generate estimates of the effects of early intervention (Casto & Mastropieri, 1986; Shonkoff & Hauser-Cram, 1987). The second approach has focused on specific populations of children, such as those with Down's syndrome, and has sought to provide a rational and logical analysis of findings, both experimental and developmental in nature (Guralnick & Bricker, 1987). Interestingly, both approaches have yielded information suggesting that early intervention can indeed be effective for numerous groups of children. There appears to be general agreement that, for children with cognitive and general developmental delays, the group that is the focus of my discussion , early intervention can improve cognitive development as measured by standard intelligence tests to the extent of onehalf to three-quarters of a standard deviation. Nevertheless, there has existed in the field for some time a general sense of discomfort with prevailing thinking about the efficacy of early intervention for