How Long Should a Zero Baseline Be? (Poster 11)

• In Event: Methodological Innovations and Complexities in the Analysis and Meta-Analysis of Single-Subject Experimental Design Data

Abstract

How many zeros must there be in the baseline period to detect an effect in a single case design? Standard advice is to have a minimum of five or six baseline observations. Researchers/clinicians using SCDs sometimes have a behavior that doesn’t occur that they want to occur. How many baseline sessions must they observe before switching to treatment phase to have a change of having statistically significant results? This study simulates a single subject in an AB design, and replicates it 1000 times to find the power, standard error of the estimated effect size, and other measures to determine how many baseline zeros must be observed before switching to treatment. We looked at length of baseline, length of treatment phase, and size of effect as factors, and a Poisson model for the data.

Author

• David M. Rindskopf, Graduate Center - CUNY