What it takes to tutor: A preregistered direct replication of the scaffolding experimental study by D. Wood et al. (1978)

Scaffolding, which can be summarized as the right instructional support at the right moment in time, is often mentioned as an important feature of effective teaching. The present study is a preregistered direct replication of the cornerstone experimental study by D. Wood et al. (1978), who found empirical evidence for the claim that a contingent use of verbal and nonverbal levels of instruction lies at the heart of scaffolding. D. Wood et al. (1978) found that children who had received contingent instruction were significantly more successful in completing a difficult construction task than children who had received one of the three different types of less contingent instruction. In our preregistered replication of D. Wood et al. (1978), we repeated the experiment and randomly assigned a large sample of 3-year-olds (N = 285) to one of the four instructional conditions: demonstration, verbal instruction, swing (a combination of demonstration and verbal), and contingent instruction to study the children's task performance in terms of outcome, efficiency, and autonomy. We replicated the instruction times and tutor accuracy of the original study (68.9% vs. 70% in D. Wood et al., 1978) but failed to replicate the effect of contingent instruction on children's independent performance when they were tested for problem-solving skills after instruction. We discuss the implications of these important null results for theory, practice, policy, and teacher education and for methodological considerations when studying effective instruction.