 |
 |
 |
| Sunday, 15 January 2006: 10:30 AM-12:00 PM | |||
| Using Latent Class Analysis to Move beyond Aggregate Findings: A Conceptual and Practical Workshop Using MPlus Software | |||
| Speaker/presenters: | Rebecca J. Macy, PhD, University of North Carolina at Chapel Hill Mimi V. Chapman, PhD, University of North Carolina at Chapel Hill | ||
| Abstract Text: Purpose: Social work research has increasingly involved the use of large investigations and data sets to understand a range of complex biopsychosocial phenomena. Similarly, social work researchers have applied ever increasingly complex ordinary least squares regression (OLS) analyses to their research. Although advanced OLS regression techniques (hierarchical linear modeling, structural equation modeling) have tremendous utility for illuminating relationships among variables, these techniques can obscure potentially important diversity among participants. As the field moves toward increasing methodological and statistical sophistication, the risk increases that heterogeneity of participants, clients, and their experiences may be lost. This is worrisome because non-normative groups may require targeted social work interventions and policies. Person-centered techniques, such as cluster analysis, have the potential to capture heterogeneity and address this critical methodological gap. Results from person-centered analyses generalize to persons rather than variables. Thus, these analyses can show holistic relationships and complement OLS findings. Specifically mixture modeling, which is a particularly robust, semi-parametric type of cluster analysis, can be a powerful tool for investigating unique relationships in a given dataset. New software advances make the various types of mixture modeling- latent class, latent profile, latent growth modeling- accessible to applied researchers. Additionally software advances have further developed these techniques; now person-centered results can be related to covariates, outcomes, and longitudinal trajectories. These mixture modeling analytic techniques give social workers tools to investigate a range of person-centered research questions. Based on the presenters' own mixture modeling analysis experiences with diverse substantive areas (i.e., violent victimization, mental health, child maltreatment, clients' satisfaction with services) and using the MPlus analysis program, this workshop will provide examples of research questions addressed using mixture modeling techniques, demonstrate how to create and analyze latent classes and latent profiles, and demonstrate possible extensions of this type of cluster analysis. Contents: The workshop will focus on the following topics: 1.Introduction to key concepts related to mixture modeling analyses. 2.Examples of social work research questions in which the presenters used mixture modeling analyses. 3.Step-by-step information on using MPlus to run mixture modeling analyses. 4.Extensions of mixture modeling analyses using covariates. 5.Key specifications in MPlus to run such extensions. Pedagogical Techniques: Teaching methods include lecture, PowerPoint presentation, review of actual output, and computer analysis demonstration. | |||
See more of Workshop
See more of Meeting the Challenge: Research In and With Diverse Communities (January 12 - 15, 2006)