Using Quantitative Visualization to Uncover Hidden Data Patterns in Social Work Research

PURPOSE: Information visualization, such as quantitative visualization advocated by Edward Tufte, has been used as a powerful tool for data exploration in a range of disciplines including epidemiology, economics, and genomics, but has been underutilized in social work research. This study demonstrates how quantitative visualization methods were used in conjunction with statistical analysis to explore data in an alcoholism treatment study. This approach revealed several findings that were not disclosed using traditional statistical analysis methods alone. This poster first shows the limitation of a traditional statistical approach that involved hypothesis testing using multivariate statistics. The main limitation is that the approach does not expose the details of data and can miss subtleties of information that can be of great value. This poster shows how quantitative visualization was used to explore an open question in the alcoholism treatment field. Previous studies show health and socioeconomic disparities exist between blacks and whites at intake in alcoholism treatment trials, suggesting blacks will fare more poorly than whites. However, both blacks and whites achieve equivalent drinking outcomes. The poster shows how quantitative visualization was used to uncover new information about the ways in which spirituality, religiosity, and drinking patterns vary by race.

METHOD: A subset of the Project MATCH Public Use Data Set was used to explore these questions (N=414). 324 participants were white and 90 were black. Purpose in life served as a measure of spirituality and was measured using the Purpose in Life Scale. Religiosity was measured using the Religious Background and Behavior Questionnaire. Drinking outcomes were measured by calculating drinks per drinking day and percent days abstinent on a monthly basis. Graphs were generated to show how those with high, medium high, medium low, and low purpose in life and religiosity fared in terms of drinking outcomes. Graphs were generated to show how participants who had positive change, negative change, and no change in purpose in life and religiosity fared in terms of drinking outcomes. T tests and ANOVAs were employed to test whether differences were statistically significant.

RESULTS: Visualizing the data brought three issues to light that would have remained hidden if multivariate regression were used exclusively. First, it was discovered that substantial improvement in drinking behavior occurred within the first month of alcoholism treatment. Second, a graph depicted that while blacks with high spirituality achieved best drinking outcomes, blacks with low spirituality had worst drinking outcomes suggesting both a cultural strength and a vulnerability. Third, for 87% of participants, religious behaviors decreased as drinking improved; however, for the minority of participants for whom religious behaviors increased, drinking outcomes were superior.

IMPLICATIONS: This study demonstrates the usefulness of quantitative visualization in social work research. Data graphing is particularly powerful in conjunction with statistical multivariate methods. While this technique is employed successfully in other fields, it is underutilized in social work research. Implications for research include fully employing this technique to help us in our efforts to promote sustainability, build strengths, and suggest new avenues of social work inquiry.