Double Robust Marginal Estimation and Tabulation

This function provides a double robust approach for estimating causal effects. It first computes marginal effects using the `causal_contrast_marginal` function, then processes and tabulates these results using `tab_engine_marginal`. The function handles both continuous and categorical variables, and allows specification of the type of effect measure for both causal estimation and tabulation.

Usage

double_robust_marginal(
  df,
  Y,
  X,
  baseline_vars,
  treat_0,
  treat_1,
  nsims,
  cores,
  family,
  weights = TRUE,
  continuous_X = FALSE,
  splines = FALSE,
  vcov = "HC2",
  verbose = FALSE,
  delta = 1,
  sd = 1,
  new_name,
  estimand = c("ATE", "ATT"),
  type_causal = c("RR", "RD"),
  type_tab = c("RR", "RD")
)

Arguments

df

A data frame containing the dataset for analysis.

Y

The name of the outcome variable as a string.

X

The name of the treatment or exposure variable as a string.

baseline_vars

A vector of covariate names included in the model.

treat_0

The reference level of the treatment variable.

treat_1

The treatment level of the treatment variable.

nsims

The number of simulations to run, used in bootstrap or Monte Carlo methods.

cores

The number of processor cores to use for parallel computation.

family

A description of the error distribution and link function to be used in the model.

weights

The name of the weights variable in the data frame, or NULL if no weights are to be used.

continuous_X

Logical, indicating whether the treatment variable X is continuous.

splines

Logical, indicating whether to use spline functions for continuous variables.

vcov

The method to use for variance-covariance estimation.

verbose

Logical, indicating whether to print detailed output during computation.

delta

The assumed smallest worthwhile effect, used for E-value calculations in tabulation.

sd

The standard deviation of the effect estimate, used for E-value calculations.

new_name

A new name to assign to the tabulated output, typically describing the variable or model.

estimand

Specifies the target of the causal inference, such as "ATE" (Average Treatment Effect) or "ATT" (Average Treatment on the Treated).

type_causal

The type of effect size (e.g., "RR" for Risk Ratio or "RD" for Risk Difference) to be computed in the causal analysis.

type_tab

The type of effect size to be used in the tabulation of results.

Value

A list containing two elements: `causal_results` with the results from the causal analysis, and `tab_results` with the tabulated results including E-values and other statistics.

Examples

# Assuming you have a dataset `df_ate` and necessary variables defined
results <- double_robust_marginal(
  df = df_ate,
  Y = "t2_kessler_latent_anxiety_z",
  X = "treatment_var",
  baseline_vars = c("age", "gender"),
  treat_1 = "intervention",
  treat_0 = "control",
  nsims = 200,
  cores = 4,
  family = "gaussian",
  weights = TRUE,
  continuous_X = FALSE,
  splines = FALSE,
  estimand = "ATE",
  type_causal = "RD",
  type_tab = "RD",
  vcov = "HC2",
  new_name = "Test Model Effect",
  delta = 1,
  sd = 1
)
#> Error: object 'df_ate' not found