The present-paper introduced the logics and practical methods of single subject designs by comparing them with group comparison designs. It was pointed out that the single subject designs assert that controlling variables for a subject's behavior should be identified using individual analysis. Those characteristics were summarized as (a) acquisition of steady state data resulting from repeated measurement of target behaviors, and (b) identification of the functional relationship between experimental variables and dependent variables by successive comparison between baseline and experimental condition. The methods, advantages and problems in the principal single subject designs, such as AB design, reversal designs, multiple baseline design and alternating treatments design were explained. Several problems in the group comparison designs were pointed out and the possibilities for applying the single subject designs to research on acupuncture were discussed.

The present-paper offered the basic ideas and methods of analytical evaluation for the single subject data. Firstly, the reasons that emphasize clinical significance rather than statistical significance of independent variables were discussed, and the visual inspection method that is used most often for single subject data was introduced. Secondly, it was pointed out that the time-series analysis method is appropriate for single subject time-series data that have the possibility of the serial dependency, and C statistic was introduced as a useful analytical technique. Thirdly, the usefulness of nonparametric tests that don't require the normal distribution (distribution free) was discussed. Fourthly, it was suggested that the exploratory data analysis is fitted for the single subject designs and its use was recommended. Finally, the reasons were emphasized that while statistical tests for single subject data should be identified and developed from now on, it is important that the visual inspection method is adopted at the same time.

The present-paper discussed about the internal and external validity in single subject designs. Firstly, intrasubject replication was introduced as a method to increase the internal validity of single subject data. Secondly, it was pointed out that direct replication and systematic replication are effective in evaluating the external validity, and the experimental design for the direct replication and the systematic replication within a single experiment was introduced. Thirdly, the methods for dealing with individual differences which came out in repeating the replication were discussed. Finally, the theoretical differences between group comparison designs and single subject designs on the external validity were pointed out, and the merits of single subject designs in the clinical research of acupuncture were discussed.

Suppressive effects of thermal stimulation of the tail on the jaw opening reflex were examined by the reversal design of single subject designs (SSD) in an urethane anesthetized rat. The results were evaluated by visual inspections of the graphs of moving averaged data and by C statistic as a simple time series analysis. The digastric evoked EMGs elicited by electrical stimulation of the gingiva (1.5x threshold intensity, 0.2Hz) had stable latency and biphasic wave form. Clear suppression of the JOR during thermal stimulation of 54°C for 30s was found by the visual inspection. The calculated C statistics based of 12 data of that of before and during thermal stimulation showed significant trends (p<0.05).
On the other hand, no significant trends were observed in baseline period and the data of 32°C thermal stimulation, and they were in agreement with the results of visual inspections.
These results indicate the noxious thermal stimulation activate the endogenous pain inhibitory mechanisms then induces analgesic effects, and the similar mechanisms were suggested to be participated in acupuncture or moxibustion stimulation-induced analgesia. Moreover the fact that evaluations by the C statistics agreed with the visual inspection suggests the validity of C statistic for evaluating the time series data in the SSD.