I’m equivocal about the professed desire to develop what look to be ‘more of the same’ assumption-laden statistical methods to handle classes of such designs.
I agree entirely with the fundamental motivation to move beyond simple ‘eye-ball’ analysis of such data, but I prefer a more ‘nuanced’ approach to analysis - which directly addresses the kind of data at hand, the nature of the research questions, and the form of knowledge-claim one wishes to make. My preferred analysis approach is constructed around these themes rather than a reflex grab toward the nearest SPSS macro or ‘what others recommend’. But that’s just me!
An added complexity to SCD analysis is that they are very prevalent in many K-12 schools in the US (often used in conjunction with curriculum-based measurement tests), which is why graphical methods dominate the analysis and even the IES recommends visual inspection [
http://ies.ed.gov/ncee/wwc/documentsum.aspx?sid=229 ], despite the many known problems with solely relying on eyeballing data. So, if teachers cannot do the analysis (or have a computer program to do it), methodologists' suggestions are likely to be ignored. Personally, I think the methods Borckardt et al. (2008) suggested have the most promise of being accessible to teachers and being able to somewhat quantify change since they have point-and-click software:
http://clinicalresearcher.org/software.htm
Borckardt, J. J., Nash, M. R., Murphy, M. D., Moore, M., Shaw, D., & O'Neil, P. (2008). Clinical practice as natural laboratory for psychotherapy research: A guide to case-based time-series analysis. American Psychologist, 63(2), 77-95. doi: 10.1037/0003-066X.63.2.77
