OVERVIEW OF SINGLE-CASE RESEARCH

Introduction to Single-Case Studies

Single-Case Research Designs

Measurement in Single-Case Research

Statistical Tools for Single-Case Research

Is There a Need?

Why Bother With Single-Case Studies?

Why Bother With Statistical Analysis?

Practitioners in school psychology and counseling confront issues daily in which the "best" choice of action is unclear. The number of professional journals, print and electronic, in which research studies are published grows larger each year. Together, this is a paradox. Each year there should be a noticeable reduction in the number of situations in which appropriate action is "unclear". Research studies should be having that effect. They are not.

Some have argued that practitioners simply don't have time to read the professional journals. This has a ring of truth, but the explanation doesn't hold up to careful examination. We would find more time to read the journals IF the research reports provided answers (or even good clues) to enhance our level of practice.

Over the last several years, it has periodically been suggested that this situation would be improved if there were more use of the single-case research design. The argument is that the results would thus be more easily applied in actual practice. But, despite at least a decade of such suggestions, there is no evidence of increasing use of the design outside the boundaries of experimental analysis of behavior.

This program is written with belief that single-case research designs do have the potential for providing results with more apparent applicability in the practice of school psychology and counseling. It is also written with belief that typical research methods textbooks do not provide sufficient direction to encourage practitioners or university researchers to use the design.

Before answering the question, it will be helpful to define some terms. The designs to be described in this program were historically identified as single-subject research, and this designation remains in use today. It's obviously not a "bad" designation, but the problem is that it implies that the designs are limited to the study of one individual at a time. It is especially important for practitioners in school psychology and counseling to recognize that these designs can also be used with groups, with family units, and with couples.

Because the term, single-subject, can easily be misinterpreted, other designations have been suggested, including single-case, single-system, and so forth. No name is perfect, and each brings with it some unwanted "baggage". In preparing this program, it was very tempting to use the designation of single-system, mostly because it has a more contemporary ring. But, with the seminal work of Barlow and Hersen (1984), the designation of single-case appears most universal and will be used throughout this program. (What matters in the long run is what you do with it, not what you call it.)

Another terminology concern is in how to identify "what you do" with the case. Is it counseling, therapy, guidance, instruction, intervention, consultation, or something else? In real life applications, of course, it could be any or all of these. To simplify the presentation in this program, however, the term treatment will be used to identify whatever you do to effect change in the case. This is not intended as a value judgment among the other possibilities. In conducting a study you will of course provide your own designations and operational definitions. The term "treatment" is often used to describe the independent variable in research studies and appears sufficiently generic to cover the range of activities provided by practitioners in school psychology and counseling. (And, using just one term could reduce my chances of getting carpal tunnel syndrome).

Now to the question of why bother with single-case studies. The easiest answer is simply because the other alternatives have not provided the needed information. By far, the majority of published studies in our field use group comparisons. The argument for this is that it increases the external validity of the study, the likelihood that the results can be generalized to other persons. The negative, a LARGE negative, is that group data often mask individual differences.

Consider a study comparing two treatments. One treatment provides "average" success for all participants. The other works extremely well for one-half of the participants and doesn't work at all for the other half. In a typical study based on group analysis, the outcome would be no significant difference (nsd) in the two treatments. If this study made it to publication (most nsd studies do not get published), it would still be difficult to identify what made one of the treatments more successful with a subset of the sample.

Perhaps the strongest argument in support of your use of the single-case designs is the ease with which they can be implemented in your present practice. Borrowing a phrase from a tv commercial, "inquiring minds want to know". As a school psychologist, you recommended and/or implemented a specific treatment intervention for a student you evaluated. Did it work? As a school counselor, you have argued that time spent in group guidance activities enhances the overall instructional mission of your school. Do you have any evidence to support your belief?

The single-case designs presented in this program are intended to provide empirical data directly related to the services you provide. You may or may not want to publish your findings in professional journals. But, with little or no interruption of your everyday responsibilities, your use of these designs can enhance the overall quality of your practice.

There is no question that the traditional case study with detailed information about the progress of a case has provided useful information. In addition, qualitative research studies are now often recommended as tools which provide rich contextual information. There is no intent here to question the usefulness of such designs. When a certain treatment protocol is successful with a particular individual, there is every reason for the school psychologist and/or the school counselor to be pleased. But, are these procedures likely to work well with another person in need? And, an important related question: what is the probability that this successful outcome was the result of chance?

Consider a student/client/patient being seen for symptoms of anxiety. When treatment began, intense anxiety symptoms were being reported at the rate of one episode per day. By the second treatment session, the individual reported a significant decrease in the frequency of the anxiety symptoms, three days with no anxiety episodes. Could this success have occurred by chance?

One of the resources in the tool kit to address the "did it happen by chance" question is the binomial expansion, a simple yet powerful tool for clinical inference and research. In essence, the binomial expansion provides direct information about the probability that something you have observed would have occurred by chance alone.

In the example above, there were seven days (trials) and three successes. The individual feels better, but the binomial expansion indicates a .77 probability that this success rate could be the result of chance alone (Don't stop the treatment yet.)

When using the single-case designs, it is possible to avoid statistical analysis, using only a visual graphic presentation of the data. Some have, in fact, suggested that statistical manipulation of the single-case design data is at best unnecessary. After all, seeing is believing.

Or is it?

Consider the graph below. This is a report of a student's performance on an achievement test battery with six subtests.

Now consider the next graph, also a report of performance on the achievement test battery.

Now look carefully at the two graphs, side by side.

In fact, these two graphs describe identical performance. The only difference is the order in which tests were placed on the graph. But, for most of us, the initial impression from the "visual gulp" is not the same.

The next graph is in the typical form for visual portrayal in single-case research. In this example, there are four weekly measurements during the baseline (before treatment) period. This is then followed by four weekly measurements after the intervention begins. Measurement is with a 10-point scale for feelings of positive self-concept.

_________________________________________________________________________

This graph suggests a moderate degree of improvement after the treatment began. Now consider the graph below portraying another case with the same design.

_________________________________________________________________________

This graph suggests a more dramatic degree of improvement after intervention.

_________________________________________________________________________

Now look at these two cases, side by side, with the addition of the metric.

(Okay so you've already figured it out.)

_________________________________________________________________________

The two graphs are again describing identical performance, scores of 4 during the baseline followed by one score of 5 and then three scores of 6 during the treatment phase.

Before moving to consideration of some of the design alternatives, there is the other question. If a treatment was successful with one student, what are the chances that it will successful with another? This concern has frequently been raised as a limitation of the single-case design. Avoiding the temptation (actually a strong temptation) to go on at length about how typical group designs do not effectively deal with this question either, the premise here is that single-case designs can address issues of generalizability.

The extent to which findings can be generalized to other persons is an issue of the "external validity" of a research design. If single-case designs provide data on the treatment effect with just one student, how can you assume anything about the effect with another student?

First, the premise in the statement above is not necessarily accurate. The designs to be presented in the following section often provide data on more than "just one student".

Also, and perhaps even more important, is that the essence of external validity is found in replication. In contrast with typical studies using group data, the non-invasive nature of most single-case designs makes it far easier to repeat the study with additional sample(s).

It is also worth noting that, in the practice of school psychology and counseling, the generalizability question is not always the primary concern. To again borrow from a tv commercial, we measure our success one student at a time.

And, even when external validity is a primary concern, the single-case strategy may be the better choice. The history of psychology, in fact the history of science in general, is filled with examples of significant developments which came from research strategies comparable to the single-case designs.

Finally, there is no intent to suggest that single-case designs are the best for all research questions. What is "best" is contingent on what the question is. There is intent, however, to emphasize that the single-case designs are well suited to answer many questions in the practice of school psychology and counseling.

The nature of the designs allows you to gather data in situations in which it would be difficult, if not impossible, to conduct a study involving treatment and control groups. Creating comparability on extraneous variables between control and treatment groups is often a major challenge, particularly since random assignment may be difficult to accomplish in a school settings. The single-case designs allow each student to serve as her/his own control. And, the nature of the single-case designs allows you to focus on the impact on individuals, an impact which can be lost when merging data from a group of students.

_________________________________________________________________________

continue to research designs         introduction review questions

access statistical analysis tools          back to program guide