Skip to contents

Average treatment effects, marginal over the empirical covariate distribution

ate.causalmixgpd_causal_fit.Rd

ate() computes the posterior predictive average treatment effect.

Usage

# S3 method for class 'causalmixgpd_causal_fit'
ate(
  fit,
  newdata = NULL,
  y = NULL,
  type = c("mean", "rmean"),
  cutoff = NULL,
  interval = "credible",
  level = 0.95,
  nsim_mean = 200L,
  show_progress = TRUE
)

Arguments

fit

A "causalmixgpd_causal_fit" object from run_mcmc_causal().

newdata

Ignored for marginal estimands. If supplied, a warning is issued and training data are used.

y

Ignored for marginal estimands. If supplied, a warning is issued and training data are used.

type

Character; type of mean treatment effect:

  • "mean" (default): ordinary mean ATE

  • "rmean": restricted-mean ATE (requires cutoff)

cutoff

Finite numeric cutoff for restricted mean; required for type = "rmean", ignored otherwise.

interval

Character or NULL; type of credible interval:

  • NULL: no interval

  • "credible" (default): equal-tailed quantile intervals

  • "hpd": highest posterior density intervals

level

Numeric credible level for intervals (default 0.95 for 95 percent CI).

nsim_mean

Number of posterior predictive draws used by simulation-based mean targets. Ignored for analytical ordinary means.

show_progress

Logical; if TRUE, print step messages and render progress where supported.

Value

An object of class "causalmixgpd_ate" containing the marginal ATE summary, optional intervals, and the arm-specific predictive objects used in the aggregation. The returned object includes a top-level $fit_df data frame for direct extraction.

Details

The default mean-scale estimand is $$\mathrm{ATE} = E\{Y(1)\} - E\{Y(0)\},$$ where the expectation is taken with respect to the empirical training covariate distribution for conditional models.

When type = "rmean", the function instead computes a restricted-mean ATE using \(E\{\min(Y(a), c)\}\) for each arm. For outcome kernels with a finite analytical mean, the ordinary mean path is analytical within each posterior draw; rmean remains simulation-based.

For unconditional causal models (X = NULL), the computation reduces to a direct contrast of the unconditional treated and control predictive laws.

See also

att, cate, qte, ate_rmean, predict.causalmixgpd_causal_fit.

Examples

# \donttest{
N <- 25
X <- data.frame(x1 = stats::rnorm(N))
A <- stats::rbinom(N, 1, 0.5)
y <- abs(stats::rnorm(N)) + 0.1
mcmc_small <- list(niter = 100, nburnin = 50, thin = 1, nchains = 1, seed = 1)
cb <- build_causal_bundle(y = y, X = X, A = A, backend = "sb", kernel = "normal",
                         components = 3, mcmc_outcome = mcmc_small, mcmc_ps = mcmc_small)
fit <- run_mcmc_causal(cb, show_progress = FALSE)
#> Defining model
#> Building model
#> Setting data and initial values
#> Checking model sizes and dimensions
#>   [Note] This model is not fully initialized. This is not an error.
#>          To see which variables are not initialized, use model$initializeInfo().
#>          For more information on model initialization, see help(modelInitialization).
#> ===== Monitors =====
#> thin = 1: beta
#> ===== Samplers =====
#> RW sampler (2)
#>   - beta[]  (2 elements)
#> Compiling
#>   [Note] This may take a minute.
#>   [Note] Use 'showCompilerOutput = TRUE' to see C++ compilation details.
#> Compiling
#>   [Note] This may take a minute.
#>   [Note] Use 'showCompilerOutput = TRUE' to see C++ compilation details.
#> running chain 1...
#> Defining model
#> Building model
#> Setting data and initial values
#> Checking model sizes and dimensions
#>   [Note] This model is not fully initialized. This is not an error.
#>          To see which variables are not initialized, use model$initializeInfo().
#>          For more information on model initialization, see help(modelInitialization).
#> ===== Monitors =====
#> thin = 1: alpha, beta_mean, beta_ps_mean, sd, w
#> ===== Samplers =====
#> conjugate sampler (3)
#>   - sd[]  (3 elements)
#> categorical sampler (16)
#>   - z[]  (16 elements)
#> RW sampler (9)
#>   - alpha
#>   - beta_mean[]  (3 elements)
#>   - beta_ps_mean[]  (3 elements)
#>   - v[]  (2 elements)
#>   [Note] Assigning an RW_block sampler to nodes with very different scales can result in low MCMC efficiency.  If all nodes assigned to RW_block are not on a similar scale, we recommend providing an informed value for the "propCov" control list argument, or using the "barker" sampler instead.
#> Compiling
#>   [Note] This may take a minute.
#>   [Note] Use 'showCompilerOutput = TRUE' to see C++ compilation details.
#> Compiling
#>   [Note] This may take a minute.
#>   [Note] Use 'showCompilerOutput = TRUE' to see C++ compilation details.
#>   [Warning] To calculate WAIC, set 'WAIC = TRUE', in addition to having enabled WAIC in building the MCMC.
#> running chain 1...
#> Defining model
#> Building model
#> Setting data and initial values
#> Checking model sizes and dimensions
#>   [Note] This model is not fully initialized. This is not an error.
#>          To see which variables are not initialized, use model$initializeInfo().
#>          For more information on model initialization, see help(modelInitialization).
#> ===== Monitors =====
#> thin = 1: alpha, beta_mean, beta_ps_mean, sd, w
#> ===== Samplers =====
#> conjugate sampler (3)
#>   - sd[]  (3 elements)
#> categorical sampler (9)
#>   - z[]  (9 elements)
#> RW sampler (9)
#>   - alpha
#>   - beta_mean[]  (3 elements)
#>   - beta_ps_mean[]  (3 elements)
#>   - v[]  (2 elements)
#>   [Note] Assigning an RW_block sampler to nodes with very different scales can result in low MCMC efficiency.  If all nodes assigned to RW_block are not on a similar scale, we recommend providing an informed value for the "propCov" control list argument, or using the "barker" sampler instead.
#> Compiling
#>   [Note] This may take a minute.
#>   [Note] Use 'showCompilerOutput = TRUE' to see C++ compilation details.
#> Compiling
#>   [Note] This may take a minute.
#>   [Note] Use 'showCompilerOutput = TRUE' to see C++ compilation details.
#>   [Warning] To calculate WAIC, set 'WAIC = TRUE', in addition to having enabled WAIC in building the MCMC.
#> running chain 1...
ate(fit, interval = "credible", level = 0.90, nsim_mean = 100)
#> [ate] Preparing ATE inputs
#> [ate] Preparing propensity-score adjustment
#> [ate] Predicting treated-arm effects
#> [ate] Predicting control-arm effects
#> [ate] Aggregating ATE estimates
#> ATE (Average Treatment Effect)
#>   Prediction points: 1
#>   Conditional (covariates): NO
#>   Propensity score used: YES
#>   PS scale: logit
#>   Credible interval: credible (90%)
#> 
#> ATE estimates (treated - control):
#>    mean lower upper
#>  -0.017 -0.08 0.019
# }