SUMMARY OF CHAPTER 9
Two important single-case research designs are the case study and the single-case experiment, or N = 1 design. The case study method can be an important source of hypotheses about behavior, can provide an opportunity for clinical innovation (for example, trying out new approaches to therapy), can permit the intensive study of rare phenomena, can challenge theoretical assumptions, and can provide tentative support for a psychological theory. The intensive study of individuals that is the hallmark of the case study method is called idiographic research, and it can be viewed as complementary to the nomothetic inquiry (seeking general laws or principles) that is also characteristic of psychology. Problems arise when the case study method is used to draw cause-effect conclusions, or when biases in the collection of, or interpretation of, data are not identified. The case study method also suffers from a lack of generalizability. How do we generalize on the basis of studying a single individual? Moreover, the "dramatic" results obtained from some case studies, though they may give scientific investigators important insights, are frequently accepted as valid by nonscientists who are not aware of the limitations of this method.
Behaviorism is an approach to the study of psychology that emphasizes the study of observable behavior under strictly controlled conditions. The behaviorism of B. F. Skinner is called the experimental analysis of behavior. Applied behavior analysis seeks to apply principles derived from an experimental analysis of behavior to socially relevant problems. The major methodology of this approach is the single-case experiment, or N = 1 design. Although there are many kinds of N = 1 designs, the most common are the ABAB design and the multiple-baseline design.
An ABAB design, or reversal design, allows a researcher to confirm a treatment effect by showing that behavior changes systematically with conditions of No Treatment (baseline) and Treatment. Methodological problems arise in this design when behavior that changed during the first treatment (B) stage does not reverse when treatment is withdrawn during the second baseline (A) stage. When this occurs, it is difficult to establish that the treatment, rather than some other factor, was responsible for the initial change. One may encounter ethical problems when using the ABAB design if a treatment that has been shown to be beneficial is withdrawn during the second baseline stage.
A multiple-baseline design demonstrates the effectiveness of a treatment by showing that behaviors across more than one baseline change as a consequence of the introduction of a treatment. Baselines are first established across different individuals, across behaviors in the same individual, or across situations. Methodological problems arise when behavior does not change immediately with the introduction of a treatment or when a treatment effect generalizes to other individuals, other behaviors, or other situations.
Problems of increasing or decreasing baselines, as well as excessive baseline variability, sometimes make it difficult to interpret the outcome of single-case designs. The problem of excessive baseline variability can be approached by seeking out and removing sources of variability, by extending the time during which baseline observations are made, or by averaging data points to remove the "appearance" of variability. Increasing or decreasing baselines may require the researcher to obtain other kinds of evidence for the effectiveness of a treatment--for example, measures of social comparison or subjective evaluation. Finally, the N = 1 design is often criticized for its lack of external validity. However, because treatments typically produce substantial changes in behavior, these changes can often be easily replicated in different individuals. The use of single "groups" of subjects can also provide immediate evidence of generality across subjects. Treatments that are clearly aversive to the participants should only be used after careful consideration of the ethical implications, including the risk/benefit ratio.
The fact that the N = 1 design usually is not appropriate for testing the possible interactions of variables highlights the importance of selecting the research methodology that is most relevant for answering the particular question under investigation.