I am working with a small group of subjects (n=6), and I intend to employ multiple ABA single-subject designs for each of my subjects. I would appreciate it if someone could guide me about the statistical test/s that can be used to study the impact of the intervention.
Hello Moazzam,
There are a number of effect size (ES) indicators that have been suggested for SS (single subject) designs. In general, inferential statistical methods fare pretty poorly in this setting due to: (a) the correlated observations within and across phases; and (b) what is frequently a very small number of data points before the analyst declares "stability" to have been attained.
Here is a link you may find helpful as a starting point for ES values in SS designs: Article Effect size calculations and single subject designs
Good luck with your work.
The statistical approach can be used as supplemental to the observational and not a replacement. There's still a big debate on what is the most useful approach. One statistical approach that can be used is to compute and compare the mean and the standard deviation of each participant's response under each condition. Another approach is to compute the PND (which is the percentage of the responses in the intervention phase that are more extreme than the most extreme response in the controlled phase). The greater PND percentage, the stronger the intervention effect. Inferential statistics typically are not used.
- Badges
- Science topic
- Similar topics
- Political Science