论文信息 - A Comparison of Modelling Strategies for Value-Added Analyses of Educational Data

A Comparison of Modelling Strategies for Value-Added Analyses of Educational Data

SummaryModelling strategies for value-added multilevel models are examined. These types of models typically include an endogenous variable and this causes difficulties for the standard estimation techniques that are commonly used to analyse multilevel models. Two alternative estimation strategies are proposed: one using an instrumental variable approach and the other using a Bayesian analysis as available through the BUGS software. We conclude that the approach offered by the BUGS software has advantages over more classical estimation methods.

Neil Spencer | Antony Fielding

[1] Adrian F. M. Smith,et al. Bayesian computation via the gibbs sampler and related markov chain monte carlo methods (with discus , 1993 .

[2] Harvey Goldstein,et al. A Multi-Level Analysis of School Improvement: Changes in Schools’ Performance over Time , 1995 .

[3] Harvey Goldstein,et al. A user's guide to MLwiN, Version 1.0 , 1998 .

[4] H. Goldstein. Multilevel Statistical Models , 2006 .

[5] Philip Heidelberger,et al. Simulation Run Length Control in the Presence of an Initial Transient , 1983, Oper. Res..

[6] A. Raftery,et al. How Many Iterations in the Gibbs Sampler , 1991 .

[7] Neil Spencer,et al. Combining Modelling Strategies to Analyse Teaching Styles Data , 2002 .

[8] Walter R. Gilks,et al. BUGS - Bayesian inference Using Gibbs Sampling Version 0.50 , 1995 .

[9] Harvey Goldstein,et al. MULTILEVEL MODELS WHERE THE RANDOM EFFECTS ARE CORRELATED WITH THE FIXED PREDICTORS , 2002 .

[10] H. Goldstein,et al. Using Examination Results as Indicators of School and College Performance , 1996 .

[11] D. Rubin,et al. Inference from Iterative Simulation Using Multiple Sequences , 1992 .

[12] Neil H. Spencer,et al. Consistent Parameter Estimation for Lagged Multilevel Models , 2003, Journal of Data Science.