Structural Equation Modeling, A Bayesian Approach
暂无分享,去创建一个
would have been much more helpful to use graphical methods to display the data sets, perhaps along the lines of the elegant Trellis/Lattice graphics implemented by the nlme package in R and S, and described by Pinheiro and Bates (2004). Much material is devoted to explaining basic statistical inference topics like ML estimation, basic distributions, confidence intervals, and the like. Anyone contemplating working with mixed modeling should already have a good understanding of these concepts. There is no need to clutter an otherwise wonderful book with this material, especially when the zeal for simplicity stretches technical accuracy, as with the confidence intervals (Chap. 4), where the general approach given is β̂ ± SE β̂ , where β̂ is the point estimate and SE β̂ is its standard error. The author does not use a multiple (e.g., 2) of the standard error, so the confidence intervals have approximately 68% coverage, not the usual 95% coverage. There is no discussion of this fundamental issue, even though (unnecessary) detail is given on the background of interval estimation. Overall, this book would be useful for anyone who uses GenStat and/or R desiring an introduction to applied mixed modeling, and they should certainly have a look. But it should not be the sole resource.
[1] Helmut Küchenhoff,et al. Size and power of tests for a zero random effect variance or polynomial regression in additive and linear mixed models , 2008, Comput. Stat. Data Anal..