Meta-Analyses of Multiple Baseline Time Series Design Intervention Models for Dependent and Independent Series

• Activity Number: 196
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Education
• Author(s): Oluwagbohunmi Adetunji Awosoga*+ and Joseph W. McKean and Bradley E. Huitema
• Companies: University of Lethbridge and Western Michigan University and Western Michigan University
• Keywords: Level Change ; AB-Design ; Meta Analysis ; Double Bootstrap ; Autocorrelation ; Contaminated Normal distribution

Abstract

This paper develops a traditional meta-type analysis for multiple baseline series. Robust methodology for multiple baseline series is also developed. The procedures are almost efficient as the traditional method on "good" data and are generally much more efficient on data containing outliers. The diagnostic procedures for the analysis of these data are also developed. The design matrices provided for the two-phase (AB) design allow for change in level and change in slope between each phase and the subsequent phase. Our methodology can be extended to more than two Phased time series design intervention models. Our parametric procedures are based on least squares (LS) estimation. The robust procedures are similar to the parametric procedures except another norm rather than the Euclidean norm is used. Illustrative examples are discussed. A Monte Carlo study of the methods are provided.