Skip to main content

Table 1 INLA summaries for estimated posterior means of population-level parameters (together with 2.5% and 97.5% posterior quantiles) using full Laplace approximations

From: Bayesian model selection techniques as decision support for shaping a statistical analysis plan of a clinical trial: An example from a vertigo phase III study with longitudinal count data as primary endpoint

 

Parameter

Model

Intercept

time

dosage⋆time

Poisson, I

1.366 (1.123, 1.603)

-0.051 (-0.063, -0.038)

-0.130 (-0.150, -0.109)

Poisson, IS

1.638 (1.432, 1.837)

-0.189 (-0.275, -0.107)

-0.173 (-0.288, -0.061)

ZIP, I

1.375 (1.121, 1.620)

-0.049 (-0.062, -0.036)

-0.115 (-0.137, -0.093)

ZIP, IS

1.628 (1.421, 1.830)

-0.209 (-0.302, -0.119)

-0.198 (-0.323, -0.075)

NB, I

1.447 (1.193, 1.695)

-0.069 (-0.090, -0.050)

-0.127 (-0.156, -0.098)

NB, IS

1.642 (1.433, 1.840)

-0.190 (-0.289, -0.101)

-0.168 (-0.289, -0.049)

arcsinh∗, I

2.056 (1.853, 2.259)

-0.067 (-0.084, -0.051)

-0.074 (-0.096, -0.052)

arcsinh∗, IS

2.055 (1.854, 2.255)

-0.068 (-0.114, -0.022)

-0.073 (-0.134, -0.012)

  1. ∗ INLA posterior means and 95% credibility intervals for arcsinh-transformed outcome modeled as continuous response with Gaussian error terms.
  2. I: model with random intercept, IS: model with random intercept and slope associated with time.
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us