Abstract
One- and two-way analysis-of-variance procedures are shown logically to be appropriate for testing hypotheses in successive treatment reversal designs for one-subject and N-subject experiments, respectively. The applicability of these designs is demonstrated through analyses of typical data.
Full text
PDF![193](https://cdn.ncbi.nlm.nih.gov/pmc/blobs/d5a8/1310751/962ae270e8ee/jaba00068-0096.png)
![194](https://cdn.ncbi.nlm.nih.gov/pmc/blobs/d5a8/1310751/02bb6f1901bf/jaba00068-0097.png)
![195](https://cdn.ncbi.nlm.nih.gov/pmc/blobs/d5a8/1310751/f53cf86b2deb/jaba00068-0098.png)
![196](https://cdn.ncbi.nlm.nih.gov/pmc/blobs/d5a8/1310751/4755ea3d6dde/jaba00068-0099.png)
![197](https://cdn.ncbi.nlm.nih.gov/pmc/blobs/d5a8/1310751/3ac05f33d7fa/jaba00068-0100.png)
![198](https://cdn.ncbi.nlm.nih.gov/pmc/blobs/d5a8/1310751/cbcfd287f92b/jaba00068-0101.png)