Applied Longitudinal Analyses: Growth Curve Modeling and Group-Based Trajectory Modeling in Social Work Research

Traditionally, many social work researchers were taught to use repeated measures ANOVA to analyze longitudinal data. However, given the strict requirements for correctly using this method (i.e., equally spaced data collection points; complete data for all collection points), repeated measures ANOVA is often poorly suited for analysis of real-world longitudinal social work data.  Instead, because growth curve modeling does not require data to be collected at equally spaced intervals and does not require data for all time points, it is a preferred approach for analyzing longitudinal data. In addition to the less restrictive data requirements, growth curve modeling offers more flexibility (e.g., inclusion of polynomials to account for changes in acceleration of change, estimation methods to account for autocorrelation among the data, use of both time-invariant and time-varying predictors).  However, growth curve modeling also has limitations.  For example, even though growth curve modeling explicitly considers both intraindividual change and interindividual differences in change, this approach treats the data as if collected from a single population and that patterns of change are homogenous.  When the assumption of homogeneous change is not tenable, researchers can examine their longitudinal data using group-based trajectory modeling which assumes that the study population is made up of a finite number of sub-populations defined by distinctive patterns of growth. Through this approach, group-based modeling is person or subject-centered (not variable-centered), and group-based (not individual-based).

 This workshop emphasizes the types of research questions that can be answered using growth curve modeling and group-based trajectory modeling as well as how to estimate these models in SAS.  First, a research scenario, focused on substance use from early adolescence to early adulthood, is presented and research questions are examined using growth curve modeling.   Next, the same scenario and data are then examined using group-based trajectory modeling. After discussing both approaches, results from the two approaches are then compared.  Finally, after mastering the conceptual aspects of both models, we review the technical aspects of estimating both types of models. 

 Workshop Content. The workshop will focus on the following topics: review of longitudinal research design as a foundation; strength and limitations of growth curve and group-based trajectory models; research questions that can be examined using each of these modeling approaches; and  presentation of illustrative examples.

 Presentation plans and pedagogical techniques include: 1. An introduction to major concepts (presentation); 2. Interactive discussion of participants’ experiences with longitudinal analysis and how these techniques might apply to their research questions; 3. Presentation and discussion of two research examples that illustrate the applications of growth curve and multiple-group trajectory models; 4. Demonstration of analysis techniques in SAS; 5. Practice interpreting analysis results; and 6. Concluding questions and answers. Participants will receive a detailed handout containing Powerpoint slides, a suggested list of future readings, SAS syntax for conducting growth curve and group-based trajectory modeling analyses, and annotated SAS output.

Audience. This workshop will be suitable for researchers and PhD students with a general understanding of repeated measures ANOVA. Basic knowledge of SAS software is helpful, but not required.