One-step estimator for natural and interventional (in)direct effects

Usage

est_onestep(
  data,
  contrast,
  g_learners,
  h_learners,
  b_learners,
  q_learners,
  r_learners,
  u_learners,
  v_learners,
  d_learners,
  w_names,
  m_names,
  y_bounds,
  g_bounds = c(0.005, 0.995),
  effect_type = c("interventional", "natural"),
  svy_weights = NULL,
  cv_folds = 10L,
  cv_strat = FALSE,
  strat_pmin = 0.1
)

Arguments

data: A data.table containing the observed data, with columns in the order specified by the NPSEM (Y, M, R, Z, A, W), with column names set appropriately based on the input data. Such a structure is merely a convenience utility to passing data around to the various core estimation routines and is automatically generated by medoutcon.
contrast: A numeric double indicating the two values of the intervention A to be compared. The default value of NULL has no effect, as the value of the argument effect is instead used to define the contrasts. To override effect, provide a numeric double vector, giving the values of a' and a*, e.g., c(0, 1).
g_learners: A Stack object, or other learner class (inheriting from Lrnr_base), containing instantiated learners from sl3; used to fit a model for the propensity score.
h_learners: A Stack object, or other learner class (inheriting from Lrnr_base), containing instantiated learners from sl3; used to fit a model for a parameterization of the propensity score that conditions on the mediators.
b_learners: A Stack object, or other learner class (inheriting from Lrnr_base), containing instantiated learners from sl3; used to fit a model for the outcome regression.
q_learners: A Stack object, or other learner class (inheriting from Lrnr_base), containing instantiated learners from sl3; used to fit a model for a nuisance regression of the intermediate confounder, conditioning on the treatment and potential baseline covariates.
r_learners: A Stack object, or other learner class (inheriting from Lrnr_base), containing instantiated learners from sl3; used to fit a model for a nuisance regression of the intermediate confounder, conditioning on the mediators, the treatment, and potential baseline confounders.
u_learners: A Stack object, or other learner class (inheriting from Lrnr_base), containing instantiated learners from sl3; used to fit a pseudo-outcome regression required for in the efficient influence function.
v_learners: A Stack object, or other learner class (inheriting from Lrnr_base), containing instantiated learners from sl3; used to fit a pseudo-outcome regression required for in the efficient influence function.
d_learners: A Stack object, or other learner class (inheriting from Lrnr_base), containing instantiated learners from sl3; used to fit an initial efficient influence function regression when computing the efficient influence function in a two-phase sampling design.
w_names: A character vector of the names of the columns that correspond to baseline covariates (W). The input for this argument is automatically generated by medoutcon.
m_names: A character vector of the names of the columns that correspond to mediators (M). The input for this argument is automatically generated by medoutcon.
y_bounds: A numeric double indicating the minimum and maximum observed values of the outcome variable Y prior to its being re-scaled to the unit interval.
g_bounds: A numeric vector containing two values, the first being the minimum allowable estimated propensity score value and the second being the maximum allowable for estimated propensity score value.
effect_type: A character indicating whether components of the interventional or natural (in)direct effects are to be estimated. In the case of the natural (in)direct effects, estimation of several nuisance parameters is unnecessary.
svy_weights: A numeric vector of observation-level weights that have been computed externally. Such weights are used in the construction of a re-weighted estimator.
cv_folds: A numeric integer specifying the number of folds to be created for cross-validation. Use of cross-validation allows for entropy conditions on the one-step estimator to be relaxed. For compatibility with make_folds, this value specified must be greater than or equal to 2; the default is to create 5 folds.
cv_strat: A logical atomic vector indicating whether V-fold cross-validation should stratify the folds based on the outcome variable. If TRUE, the folds are stratified by passing the outcome variable to the strata_ids argument of make_folds. While the default is FALSE, an override is triggered when the incidence of the binary outcome variable falls below the tolerance in strat_pmin.
strat_pmin: A numeric atomic vector indicating a tolerance for the minimum proportion of cases (for a binary outcome variable) below which stratified V-fold cross-validation is invoked if cv_strat is set to TRUE (default is FALSE). The default tolerance is 0.1.