Latent transition analysis as a sensitive outcome analysis for longitudinal data
Three secondary data analysis studies were conducted to provide methodological and theoretical examples of the application of Latent Transition Analysis (LTA) to three studies of smoking cessation among adults. Each study was designed to evaluate an interactive expert system intervention for smoking cessation in three different populations of smokers: (1) Chapter 2 evaluated the expert system intervention versus assessment only in smokers from the state of Rhode Island (N = 4,144); (2) Chapter 3 evaluated the expert system intervention versus a non-interactive manual-based intervention in a managed care setting (N = 2,882); (3) and Chapter 4 examined the expert system intervention versus assessment only for high school students' parents (N = 2,461) who were receiving interventions for up to three problem behaviors simultaneously. LTA in each study consisted of four parts: model specification, descriptive analyses of parameter estimates, tests of differential treatment effects using data augmentation procedures, and a descriptive process analysis. Major findings are: (1) The expert system intervention has better outcomes in terms of stage of change than the control conditions as evidenced by more progression to later stages and less regression to early stages; (2) The expert system intervention is particularly effective for the Precontemplation and Contemplation stages proximal to intervention delivery, while the effects weaken over time; (3) Data augmentation procedures lack sensitivity for detecting treatment effects. Recommendations are made for future research in this area. ^
Rosemarie Ann Martin,
"Latent transition analysis as a sensitive outcome analysis for longitudinal data"
Dissertations and Master's Theses (Campus Access).