generalize to a broad group of individuals (random samples), some designs attempt to determine cause and effect relationships (true experiments), and some used to provide rich, detailed, descriptive, qualitative, and/or quantitative information. The purpose of single-case study is to present or describe a case (or a small number of cases) and not necessarily to represent the large diverse group. Nonetheless single-case studies have been important educational research methodologies (Yin, 1984). Single-case studies set forth to depict or analyze and/or explain the uniqueness of individuals in specific situations through personalized accounts of these situations. Wolery and Gast (2000) observe that teachers or researchers in education often do not have access to the large number subjects that are required for statistical analyses in group designs. Given this, single-case designs offer a tremendous opportunity for teachers to conduct research in the classroom. Single-case designs are useful in the classroom due to the limited size of the accessible population such as in special education where the number of students in resource rooms is generally small.
Single-case studies allow for many types of approaches to choose from. The most common methods of single-case studies include withdrawal designs, reversal designs, multiple-baseline designs, and alternating treatments designs (Gay & Airasian, 2000). There have also been different classifications of single-case studies depending on the objective of the researcher. For example, single-case studies have been classically classified by Yin (1984) as either: 1) exploratory (e.g., pilot studies for other research questions or to guide larger studies); 2) descriptive (providing narrative accounts of a specific case or type); or 3) explanatory (testing or explaining theoretical assumptions or conclusions). Wallace (1998) provided further elaboration on Yin's classes of single-case studies by defining the specific problems or areas of inquiry that are the focus of case studies. Wallace suggests that the specific feature of a single-case study allows it to: 1) put theoretical assumptions to direct tests; 2) provide illustrations of theoretical applications; 3) solve particular problems in practical applications of theoretical knowledge that cannot be solved in larger studies; and 4) to generate hypotheses for further investigation.
Adelman, Jenkins, and Kemmis, (1980) provided an explanation as to why single-case research studies are advantageous to educational research. First, the data gleaned from single- case research has excellent external validity when applied to similar cases and it is therefore appealing to educational practitioners. Secondly, readers of educational single-case studies can readily identify with the issues, facts, and concerns raised by these designs. Third, while they are not considered to be generalizable to larger general populations (one of the weaknesses of these designs is their lack of overall external validity) single-case studies allow for generalizations about a specific case in point, a specific type of occurrence, or to a specific class of subjects or observations. Fourth, single-case research studies can be designed as to represent an assortment of different perspectives allowing for researchers to offer evidence or support to alternative interpretations of the findings. Fifth, due to the potential for rich qualitative data collection single-case studies can provide a store of descriptive material that can be readily made available for reinterpretation by others. Sixth, the findings derived from single-case studies can be put to immediate use for a variety of purposes making single-case studies very applicable to practical interventions. And finally, single-case studies present the findings from research in an accessible form.
Nisbet and Watt (1984) add that the results from single-case research are often easily understood by a wide audience as the findings are frequently written in everyday nontechnical language. This is because in part, sophisticated statistical analyses are not required to understand the findings. Therefore, weak or small unimportant effects are not viewed as significant in single-case research. This makes the findings immediately intelligible and the results of single-case studies often speak for themselves. In addition, due to the descriptive nature of single-case research designs they are able to catch unique features of the data that may otherwise be lost in the statistical interpretation of large scale data. Often such distinctive characteristics can lead to the generation of hypotheses that might hold the key to understanding special situations not covered in larger group research. This allows for the incorporation of unanticipated events or uncontrolled variables into the findings. Finally, due to their relative simplicity single-case research designs can be performed and implemented by a single researcher or small team and often do not require a full research team.
Stenhouse, (1983) discussed the pitfalls of comparing group data that focuses on interpretations of averaged or mean data to individuals. Just because a statistic accurately describes a group does not indicate that it accurately describes any individual within the group (of course the opposite would also be true). However, group research draws conclusions or makes interpretations about individuals based on statistical findings. Single-case designs can be used to overcome the problem of using means of groups applied to an individual. For example in looking at the effects of a specific reading program that has been judged by prior research to be effective, a single-case design can look at the effects of the program on a specific individual or individuals with different demographic characteristics and describe similarities or departures from the program's reported effectiveness in group studies. The design can also be replicated in other individuals. Because some populations or characteristics of interest are rare, single-case research can apply the findings from group studies to examine participants from hard to find populations.
This brings up the question of variability in studies. In group studies inter-participant variability, the observation that inevitably occurs because different students perform differently on the same test, is attributed to or classified as error variance because source is not measured (this is why means are used). In reality, there are many sources of this variability, but researchers either cannot measure them or there are too many sources to measure. There is also the problem of intra-participant variability, the notion that the same person will perform differently on the same task at different times or under different situations. This can contribute to inter-participant variability. Both of these can be directly measured and described in single-case research. Inter-participant variability can be investigated by comparing the different subjects observed and measured on the same variables or interventions, whereas intra-participant variability can be measured by taking measurements in the same individual at different times (Birnbrauer, Peterson, & Solnick, 1974).
Multiple baseline designs have been used to look at sources of variability in single-case research. These include the multiple baselines across behaviors, participants, and settings (Gay & Airasian, 2000). For example, a multiple baseline across behaviors study could investigate talking and aggressive behaviors; a multiple baseline study across subjects could investigate reading comprehension over different students; and a multiple baseline study across settings could examine a student's aggression in class, at recess, and in the cafeteria.
It is rare to find replication of some effect, say a new reading program, within a group study. Group study replications occur at different times or between different researchers and groups. However, single-case research allows for two types of replication (Jackson, 2008). The first type of replication is intra-participant replication through the use of reversal designs or withdrawal designs (e.g., ABAB designs). In this case the intra-participant variability is not treated as error but as a dependent measure (this is analogous to repeated measures in group designs).
Inter-participant replication can be accomplished by using different subjects in different single-case designs and comparing finds across subjects on the variable of interest. There are several methods to accomplishing this, such as using different subjects in a simple (AB) design and comparing the effectiveness of an intervention. Using a multiple baselines across participants design allows for even more diversity in findings that is difficult to accomplish in group research. By taking multiple…