1 / 6

Skewed random effects in AD Model Builder

Skewed random effects in AD Model Builder. Goal. Analyse model in Section 14.2 of Skrondal and Rabe-Hesketh (2004) Generalized Latent Variable Modeling: Multilevel, Longitudinal and Structural Equation Models . Chapman & Hall

lave
Télécharger la présentation

Skewed random effects in AD Model Builder

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Skewed random effects in AD Model Builder

  2. Goal • Analyse model in Section 14.2 of Skrondal and Rabe-Hesketh (2004) Generalized Latent Variable Modeling: Multilevel, Longitudinal and Structural Equation Models. Chapman & Hall • Replace normal distribution with a skewed distribution for the random effects.

  3. 14.2 Diet and heart disease • Covariate measurement error model • 337 middle aged men (bank, transportation) • At requirement: weigh their food • Repeated measurement for 76 men • Heart disease or not? • Covariates: • age (numeric) • bus (0,1)

  4. Model description • Exposure model (true fiber intake) ηj = γ1·agej+γ2·busj + γ3·age*busj+ ζj, ζj ~ N(0, ψ) • Measured fiber intake (i = 1,2) yij = ηj + α0+(i-1)·α1 + ij , ij ~ N(0, θ) • Disease model(D = 0,1) logit(Dj|ηj) = β0 +β1·agej+β2·busj + β3·age*busj + λ·ηj

  5. Setting up and running the model • Prepare data in R: ”diet.s” • Compiling the model • admb -re diet • Running the model (Laplace approx.) • diet -est • Running the model (Gauss-Hermite appr.) • diet -est -gh 20

  6. Skewed distributions for ζ • Skrondal and Rable-Hesket: • Replace N(0,ψ) distribution for ζj by non-parametric distribution (Fig. 14.2) • Skewed distribution for ζj ζj = [a·uj + (1-a)[exp(uj)-exp(0.5)]/c1]/c2, 0 < a < 1 uj ~ N(0, ψ) c1 = sqrt[e(e-1)], c2 = sqrt(a2 + (1-a)2)

More Related