Date of Award
Doctor of Philosophy
David A. Patterson, Ph.D.
John G. Orme, Ph.D., Charles A. Glisson, Ph.D., Charles B. Hamilton, Dr. P.H.
Introduction: An important methodological consideration in the social sciences is the evaluation of the effectiveness of groups and specific group interventions. There is an increasing demand for service accountability in practice settings both in social services and public health services. Group services are rising as a practice modality. Emerging technology shows promise of providing the means for practitioners untrained in advanced research methods to gain useful information and improved decision- making capacities related to groups and group services. Computer based graphical representation of data patterns at multiple levels of analysis can provide the bases for data exploration and lead to further advances in the evaluation of complex group dimensions associated with group effectiveness.
Objectives: The purpose of this study was to evaluate group therapy experiential education outcomes using conventional data analytic methods for time series data. These include traditional methods of visual evaluation of single subject information, as well as, less common graphical representation methods that permit the simultaneous display of group process and outcomes and provide visual evaluation information across units of analysis.
Methods: Group level time series data for 16 experiential group therapy education groups were evaluated using a variety of graphical and statistical methods. This study demonstrates a range of graphical representations, which provide differing levels of evaluative information and time series statistical information. The limitations of inferences available when evaluating non-probability samples were addressed.
Results: Using widely ava ilable technology a number of graphical methods were demonstrated that present multilevel time series information to include group process and outcome simultaneously for both individuals and groups, as well as, for multiple variables of change. Data visualization evaluative methods were presented that illustrate levels of group participant concordance and variability over time. Graphical representations were generated that demonstrate the proportional contribution of multiple variables to group outcome over time. Graphical representations methods were also presented that represent multiple levels of analysis over time and for multiple groups with varying durations of group length for simultaneous comparison over time. The difficulties associated with identifying autocorrelation in time series data and with non-probability samples using graphical and statistical methods were addressed.
Basham, Randall E., "Data Visualization: Graphical Representation in the Evaluation of Experimental Group Therapy Education Outcomes. " PhD diss., University of Tennessee, 2002.