A Systematic Review of Instructional Comparisons in Single-Case Research

Abstract

Comparison studies conducted to determine which instructional interventions are most efficient for teaching discrete behaviors to individuals with disabilities are potentially valuable, although some threats to internal validity may be more likely in these studies. Studies included in this review typically met common internal validity standards, such as reliability measurement, but often did not include controls specific to comparison designs. Comparisons often included young children with autism and were frequently conducted by researchers in self-contained classroom settings. Systematic instruction was effective in nearly all comparisons, although many included undifferentiated data (i.e., both interventions were equally effective), and within-participant replications were often inconsistent (i.e., outcomes varied across comparisons for a single participant). Results suggest implementers should conduct high-fidelity instruction with corrective and instructive feedback and should choose intervention variations based on participant preference. We recommend researchers include control sets or time-lagged introductions, counterbalance behavior sets, and measure differential acquisition over time.

Replicating stimulus‐presentation orders in discrimination training

“…There were larger differences between these conditions and the sample-first condition. Specifically, of the 13 comparisons where multiple conditions were efficacious, nine comparisons showed that the sample-first condition required at least 10% more sessions to reach mastery than the other conditions (Ledford et al, 2019; see Online Supporting Information for more details). This finding aligns with Vedora et al (2019) but does not replicate the efficiency advantage of the sample-first condition in Petursdottir and Aguilar (2016) and Schneider et al (2018).…”

An evaluation of prompting procedures in error correction for children with autism

“…For example, some intraverbal studies demonstrated quicker skill acquisition through picture prompts (e.g., Ingvarsson & Hollobaugh, 2011) while others supported the use of echoic prompt (e.g., Kodak et al, 2012). In the present study, we found active student response with picture prompt was more efficient in facilitating the acquisition of English‐to‐Chinese translations for five of the eight comparisons using the total sessions as the efficiency measure based on the criterion of 10% difference as proposed by Ledford et al (2019). However, when using total trials as the measure, all comparisons across participants produced differentiated results based on the same criterion.…”

“…Similar results were also obtained during her performance in Comparison 2 where she had a faster acquisition with the picture‐prompt condition (i.e., six sessions of 83 trials) than error correction with echoic prompt (i.e., eight sessions of 110 trials). As such, using Ledford et al's (2019) criterion of 10% difference, both comparisons for Lam yielded differentiated effects between the numbers of sessions and trials in two conditions.…”

“…sessions as the efficiency measure based on the criterion of 10% difference as proposed by Ledford et al (2019).…”

“…Thus, the active student response with picture prompt might be more efficient than the echoic‐prompt condition, even though the differences in total sessions seemed to be minimal in some comparisons. In addition, as both Kay et al (2020) and Ledford et al (2019) pointed out, the difference between the two procedures, even if not socially significant, may accumulate over time and result in meaningful difference and larger timesaving in the future. However, even though Carroll et al (2015) found the total trials matched the total duration as efficiency measures, because we did not measure session durations, we do not know how well the total trials corresponded to the total duration in this study.…”