On Tue, 19 Oct 1999, Paul Brandon wrote:
> > There is such a thing as Single-subject design,where the individual
> > is used as his or her own control.
> > Take this example:
> > Pre- have a headache X (IV) drink a cup of cactus tea Y (dv)
> > given by the witch doctor headache gone
> >
> > Now a between groups design may not provide such evidence.
> >
> >
> > Michael Sylvester
> > Daytona Beach,Florida
>
> This is not really a single case design -- I see no baseline; just a treatment.
> At a minimum you'd need two conditions:
>
> A -- drink a cup of non-cactus tea given by the witch doctor when headache.
>
> B -- drink a cup of cactus tea given by the witch doctor when headache.
>
> A -- drink a cup of non-cactus tea given by the witch doctor when headache.
>
> This would be the minimal A-B-A single case design.
> It would be better to add a second 'B' condition to make it a reversal.
> Even better would be to do a multiple baseline study across situations.
>
> After a few replications (both direct and systematic), we might have some
> confidence in either the tea or the witch doctor. You'd need the multiple
> baseline design to separate the witch doctor from the tea.
>
> The definitive reference on single case experimental designs is:
> Strategies and tactics of behavioral research
> J.M. Johnston, H.S. Pennypacker.
> 2nd ed.
> Hillsdale, N.J.: L. Erlbaum Associates, 1993.
>
or referred to as the One-shot case. Just curious as to how you would
control for interaction and expectancy effects-which may not be cofined
and restricted and assessed by only measuring the DV?
the absence of the IV can also be viewed as a level of the IV.
Michael Sylvester
Daytona Beach,Florida
