< Back to previous page
Project
The statistical power of meta-analyses of single-case experiments.
Despite the strong emphasis on large N designs in most statistical and methodological courses and handbooks, single-case designs are gaining popularity. They can provide a viable alternative or supplement to group studies to investigate causal research questions. This could be the case when it is unnecessary to collect data from a large group of subjects (e.g., generating pilot data) or when this is impossible (e.g., studying rare conditions). Although several data-analytic techniques have been suggested for the analysis of single-case experiments, visual analysisstill remains the predominant method. An important reason is the absence of these techniques in popular statistical packages.
The focus of this doctoral research project is on providing user-friendly tools for analyzing single-case data. As a statistical technique, randomization tests were selected, because they are promising for analyzing the specific data resulting from single-case experiments. For the software development, the statistical programming language R was chosen. To enhance the user-friendliness, an implementation in the window-based graphical user interface, the R commander, was also created.
The developed software is presented in six separate manuscripts and contains functions for three important steps in analyzing single-case data: visual exploration, statistical significance testing, and effect size calculation. Manuscript1 presents the SCVA package, which helps researchers in making graphical representations of single-case data and to transform graphical displays back to raw data. Manuscript 2 introduces the SCRT package for designing single-case phase and alternation experiments, as well as for conducting randomization tests on data resulting from these experiments. In manuscript 3, a systematic review on the use of multiple baseline designs is provided. Since it was found that most multiple baseline studies in the sample data relied solely on visual analysis of the data, in manuscript 4 an extension of the SCRT package is presented for use with multiple baseline data. Manuscript 5 explains the use of the SCMA package, with functions for probability combining and effect size calculation. In manuscript 6, all functionalities are combined in the RcmdrPlugin.SCDA package: a window-based graphical user interface, which overcomes the difficulties of the command line interface of R.By making these data-analytic tools available and assemble them in one overarching user-friendly system, we wish to contribute to their familiarity with applied researchers, and hence their increased use in empirical studies. The scripts could also be used in stochastic simulation studies, and in this way add to the knowledge on the statistical properties of effect sizemeasures and statistical techniques for single-case data.
The focus of this doctoral research project is on providing user-friendly tools for analyzing single-case data. As a statistical technique, randomization tests were selected, because they are promising for analyzing the specific data resulting from single-case experiments. For the software development, the statistical programming language R was chosen. To enhance the user-friendliness, an implementation in the window-based graphical user interface, the R commander, was also created.
The developed software is presented in six separate manuscripts and contains functions for three important steps in analyzing single-case data: visual exploration, statistical significance testing, and effect size calculation. Manuscript1 presents the SCVA package, which helps researchers in making graphical representations of single-case data and to transform graphical displays back to raw data. Manuscript 2 introduces the SCRT package for designing single-case phase and alternation experiments, as well as for conducting randomization tests on data resulting from these experiments. In manuscript 3, a systematic review on the use of multiple baseline designs is provided. Since it was found that most multiple baseline studies in the sample data relied solely on visual analysis of the data, in manuscript 4 an extension of the SCRT package is presented for use with multiple baseline data. Manuscript 5 explains the use of the SCMA package, with functions for probability combining and effect size calculation. In manuscript 6, all functionalities are combined in the RcmdrPlugin.SCDA package: a window-based graphical user interface, which overcomes the difficulties of the command line interface of R.By making these data-analytic tools available and assemble them in one overarching user-friendly system, we wish to contribute to their familiarity with applied researchers, and hence their increased use in empirical studies. The scripts could also be used in stochastic simulation studies, and in this way add to the knowledge on the statistical properties of effect sizemeasures and statistical techniques for single-case data.
Date:1 Oct 2008 → 8 May 2013
Keywords:Psychology, Meta-analyses
Disciplines:Psychological methods, Mathematical and quantitative methods, General pedagogical and educational sciences, Social theory and sociological methods, Political theory and methodology, Education curriculum, Education systems, Specialist studies in education, Other pedagogical and educational sciences, Biological and physiological psychology, General psychology, Other psychology and cognitive sciences
Project type:PhD project