The lead article in this issue by Johnson and colleagues uses a single-subject design (SSD). Like all research designs, SSDs have strengths and limitations1 in comparison to randomized controlled trials that are typically considered the “reference standard” design for intervention research.2 One advantage is applicability in the clinic. A single-subject participant may have characteristics that more closely match a specific patient and the intervention and results may then directly inform clinical practice. Randomized controlled trials have samples that may not fit as well to a specific patient's characteristics, thus making the applicability more challenging. Of course, this strength is also the weakness in terms of applicability, as results from an SSD are more challenging to apply to patients with a wider range in the characteristics of interest. Designs are chosen to answer specific questions and to address certain constraints of implementation. This makes SSDs a reasonable alternative for implementation in a clinical setting and by clinician/researcher teams.
Johnson and colleagues assembled such a team and implemented a variation of SSDs: a multiple baseline multiple probe design.3 In this design, durations of baselines, the number of data points in baselines, and the number of measurement points during intervention all varied by subject and the specific task. While this presents challenges for interpretation, it also provides detailed information regarding specific points when specific patients may benefit from an intervention. This design is used typically for interventions with tasks that have multiple components such as self-feeding behaviors. In these tasks, components include elements such as trunk and head control, limb projection, grasp, and successful hand-to-mouth actions. Baseline measures might differ in number and timing to capture what is typical for that subject or specific action within a task. Probes (measurement points) during interventions could also vary on the basis of each participant's responses to intervention.
Rigorous design considerations and implementation are as critical to valid single-subject research as they are to research using other designs. This includes the analysis of results. Results are often visually displayed and interpreted, but statistical analyses should also be implemented. Johnson and colleagues used the split middle trend line analysis method in baseline and intervention separately for 2 of their 3 outcomes to assist the visual inspection of trends in the data. This descriptive analysis could be extended to a statistical method using the celeration approach,4 but this requires more data points than were available in this study. In the celeration line analysis, the baseline trend line is extended into the intervention phases and the slope of change and the number of data points above and below the celeration line can be statistically analyzed. In addition, there are effect size statistics that are specific to SSDs5 and could be applied to the data in Table 5. The important issue and the relevance to this study is that design and analytical decisions must be informed and described with the statistics matched to the design. This study was conducted by an excellent team of clinicians and researchers and provides rigorous and thoughtful results for clinical application.
1. Fetters L, Tilson J. Evidence Based Physical Therapy. Philadelphia, PA: FA Davis; 2012.
2. Straus SE, Richardson WS, Glasziou P, Haynes RB. Evidence Based Medicine. 3rd ed. New York, NY: Churchill Livingston; 2005.
3. Horner RD, Baer DM. Multiple-probe technique: a variation on the multiple baseline. J Appl Behav Anal. 1978;11(1):189–196.
4. Nourbakhsh MRO, Ottenbacher KJ. The statistical analysis of single-subject data: a comparative examination. Phys Ther. 1993;74(8):768–776.
5. Campbell JM. Statistical comparison of four effect sizes for single-subject designs. Behav Modif. 2004;28(2):234–246.