AuxSurvey (version 1.0)

svyBayesmod: Bayesian Survey Model Estimation

Description

This function fits a Bayesian model using Stan for survey data. It allows you to specify the outcome formula, the function for Stan, and apply different types of survey analysis, including weighted or unweighted models, for both sample and population data. The function supports posterior estimation, confidence intervals (CIs), and MCMC diagnostics.

Usage

svyBayesmod(
  svysmpl,
  svypopu,
  outcome_formula,
  BayesFun,
  subset = NULL,
  family = gaussian(),
  invlvls,
  weights = NULL,
  nskip = 1000,
  npost = 1000,
  nchain = 4,
  printmod = TRUE,
  doFigure = FALSE,
  useTrueSample = FALSE,
  stan_verbose = FALSE,
  HPD_CI = FALSE,
  seed = NULL
)

Value

A list containing the Bayesian estimates and confidence intervals (CIs) for each subset or the entire dataset. Each element in the list includes: - estimate: The Bayesian point estimate for the outcome. - CI: The credible intervals for the outcome estimate. - Other elements based on the specified confidence levels in invlvls.

Arguments

svysmpl

A dataframe or tibble representing the sample data (samples). This should contain the outcome variable and any additional covariates.

svypopu

A dataframe or tibble representing the population data (population). This should contain all variables in the model.

outcome_formula

A formula for Stan, specifying the outcome and predictors in the model.

BayesFun

The name of the Stan function to be used for fitting the Bayesian model.

subset

A character vector representing filtering conditions to select subsets of the sample and population. Default is NULL, in which case the analysis is performed on the entire dataset. If specified, estimates for both the whole data and the subsets will be calculated.

family

The distribution family for the outcome variable. Currently, the following options are supported: gaussian for continuous outcomes and binomial for binary outcomes.

invlvls

A numeric vector specifying the confidence levels for the credible intervals (CIs). If more than one value is specified, multiple CIs will be calculated.

weights

A numeric vector of case weights. The length of this vector should match the number of cases in svysmpl. These weights will be used in the Bayesian model for weighted estimation.

nskip

An integer specifying the number of burn-in iterations for each chain in the MCMC for Stan models. Default is 1000.

npost

An integer specifying the number of posterior sampling iterations for each chain in the MCMC for Stan models. Default is 1000.

nchain

An integer specifying the number of MCMC chains for Stan models. Default is 4.

printmod

A logical scalar; if TRUE, posterior estimates will be printed.

doFigure

A logical scalar; if TRUE, MCMC diagnostic plots will be generated.

useTrueSample

A logical scalar; if TRUE, the estimator will use true sample information.

stan_verbose

A logical scalar; if TRUE, MCMC information will be printed during Stan model fitting.

HPD_CI

A logical scalar; if TRUE, the calculated credible intervals will be highest posterior density intervals (HPD). Otherwise, symmetric intervals will be used. Default is FALSE.

seed

An integer specifying the random seed for reproducibility. Default is NULL.

Examples

# \donttest{
## Example usage with survey data:
## Simulate sample and population data
data = simulate(N = 3000, discretize = 3, setting = 3, seed = 123)
population = data$population  # Get population data
samples = data$samples        # Get sample data
ipw = 1 / samples$true_pi    # Compute inverse probability weights

## Define outcome formula and Stan function
outcome_formula = "Y1 ~ Z1 + Z2 + Z3 + (1|auX_3)"
BayesFun = "stan_glmer"

## Fit Bayesian model using weighted survey data
bayes_model = svyBayesmod(svysmpl = samples, svypopu = population,
                          outcome_formula = outcome_formula,
                          BayesFun = BayesFun, weights = ipw,
                          family = gaussian(), nskip = 2000, npost = 2000,
                          nchain = 2, printmod = TRUE, invlvls = 0.95, stan_verbose = TRUE)
# }