"Thinking Outside the Box: Using Data Visualization and Exploration to Identify Dynamic Psychological Processes in Intensive Longitudinal Data"

One of the challenges in working with innovative research methods is that new methods for working with data must be developed to understand findings. Intensive longitudinal data collection methods offer great promise for measuring the dynamics and patterns of responses to traumatic stress over time and in individuals as well as groups. But many readily available data analysis packages (e.g., HLM, SPSS) are still fairly limited in their capacity to automatically generate useful plots for exploring intensive longitudinal data. This means that it is not easy to see what is happening in a set of data. One can run planned multilevel analyses on group questions and get “answers” with no sense of what is going on with individuals. In addition, if one is exploring a new area of research, there is no easy way to “eyeball” the data. This presentation will demonstrate graphing methods using SPSS that allow one to visualize trends in symptoms over time for a group and/or individuals and to show relationships among variables over time. Examples will be presented from a study of early responses to traumatic stress in traumatically injured patients and family members to show how data visualization can present a very different picture from a statistical result. We will show examples in which HLM analyses appear to lead to one conclusion, while visualizations reveal some expected, some surprising, and some confusing results. We will also show how graphing methods can be used to depict interesting theoretical questions such as: Are early responses of those who develop PTSD characterized by high levels of arousal?; Do acute responses to traumatic stress correspond to the DSM diagnostic criteria for ASD?; Do reexperiencing and avoidance symptoms appear to cycle?; Are early PTSD symptoms the best predictors of longer-term PTSD outcomes?; Do early responses to traumatic stress predict longer term PTSD or depression outcomes?; Do men and women differ in their early responses to traumatic stress? Do survivors of events involving injury and threat of death to themselves differ from survivors of events involving injury and threat of death to loved ones in their early responses to traumatic stress? We will also provide some caveats that can help prevent misinterpretations of findings. These visualization methods allow exploration of data when patterns of findings do not appear to be what was expected, and they will be even more useful to identify hypotheses for statistical analyses in research that explores novel domains.