list_models: List model details

In JointAI: Joint Analysis and Imputation of Incomplete Data

List model details

Description

This function prints information on all models, those explicitly specified by the user and those specified automatically by JointAI for (incomplete) covariates in a JointAI object.

Usage

list_models(object, predvars = TRUE, regcoef = TRUE, otherpars = TRUE,
  priors = TRUE, refcat = TRUE)

Arguments

object	object inheriting from class 'JointAI'
predvars	logical; should information on the predictor variables be printed? (default is TRUE)
regcoef	logical; should information on the regression coefficients be printed? (default is TRUE)
otherpars	logical; should information on other parameters be printed? (default is TRUE)
priors	logical; should information on the priors (and hyper-parameters) be printed? (default is TRUE)
refcat	logical; should information on the reference category be printed? (default is TRUE)

Note

The models listed by this function are not the actual imputation models, but the conditional models that are part of the specification of the joint distribution. Briefly, the joint distribution is specified as a sequence of conditional models

\loadmathjax \mjdeqn

p(y | x_1, x_2, x_3, ..., \theta) p(x_1|x_2, x_3, ..., \theta) p(x_2|x_3, ..., \theta) ...ascii The actual imputation models are the full conditional distributions \mjeqnp(x_1 | \cdot)ascii derived from this joint distribution. Even though the conditional distributions do not contain the outcome and all other covariates in their linear predictor, outcome and other covariates are taken into account implicitly, since imputations are sampled from the full conditional distributions. For more details, see Erler et al. (2016) and Erler et al. (2019).

The function list_models prints information on the conditional distributions of the covariates (since they are what is specified; the full-conditionals are automatically derived within JAGS). The outcome is, thus, not part of the printed linear predictor, but is still included during imputation.

References

Erler, N.S., Rizopoulos, D., Rosmalen, J.V., Jaddoe, V.W., Franco, O.H., & Lesaffre, E.M.E.H. (2016). Dealing with missing covariates in epidemiologic studies: A comparison between multiple imputation and a full Bayesian approach. Statistics in Medicine, 35(17), 2955-2974.

Erler NS, Rizopoulos D, Lesaffre EMEH (2021). "JointAI: Joint Analysis and Imputation of Incomplete Data in R." Journal of Statistical Software, 100(20), 1-56. \Sexpr[results=rd]{tools:::Rd_expr_doi("10.18637/jss.v100.i20")}.

Examples

# (set n.adapt = 0 and n.iter = 0 to prevent MCMC sampling to save time)
mod1 <- lm_imp(y ~ C1 + C2 + M2 + O2 + B2, data = wideDF, n.adapt = 0,
               n.iter = 0, mess = FALSE)

list_models(mod1)

JointAI documentation built on April 27, 2023, 5:15 p.m.