Estimation of a Nonlinear Intervention Phase Trajectory for Multiple-Baseline Design Data

• Published in: The Journal of experimental education Vol. 83; no. 4; pp. 514 - 546
• Main Authors: Hembry, Ian, Bunuan, Rommel, Beretvas, S. Natasha, Ferron, John M., Van den Noortgate, Wim
• Format: Journal Article
• Language: English
• Published: Washington Routledge 02.10.2015; Taylor & Francis Group, LLC; Taylor & Francis Inc
• Subjects: Computation; Data; Estimating techniques; Hierarchical Linear Modeling; Intervention; logistic models; Markov analysis; Markov Processes; MCMC estimation; MEASUREMENT, STATISTICS, AND RESEARCH DESIGN; Monte Carlo Methods; Monte Carlo simulation; multilevel models; multiple-baseline design; quadratic models; Research Design; Simulation; Statistical Analysis
• ISSN: 0022-0973; 1940-0683
• DOI: 10.1080/00220973.2014.907231

A multilevel logistic model for estimating a nonlinear trajectory in a multiple-baseline design is introduced. The model is applied to data from a real multiple-baseline design study to demonstrate interpretation of relevant parameters. A simple change-in-levels (ΔLevels) model and a model involving a quadratic function (Quadratic) for the nonlinear intervention phase data were also estimated. In addition, a simulation study was conducted to assess Markov chain Monte Carlo estimation of the logistic model and compare its trajectory recovery with use of the ΔLevels and Quadratic models. While most of the logistic model's parameter values were recovered well, trajectory recovery was very reasonable using the simpler Quadratic model. Results are discussed along with recommendations for practitioners and directions for future research.