Sweep mediation Type I error across confounding strength levels

Usage

sweep_mediation_null_by_conf(
  conf_grid = c(0.2, 0.4, 0.6, 0.8, 1),
  n_iter = 100,
  n_samples = 500,
  n_features = 20,
  w_signal = 0.7,
  mo_confounding = 0.8,
  phi = 0,
  rho_G1 = 0,
  rho_G2 = 0,
  rho_pop = 0,
  lambda_XM = NULL,
  lambda_MY = NULL,
  omega_1 = NULL,
  omega_2 = NULL,
  feat_cor = 0,
  base_seed = 900,
  n_cores = 1,
  alpha = 0.05
)

Arguments

conf_grid: Numeric vector of confounding strength values. Default c(0.2, 0.4, 0.6, 0.8, 1.0).
n_iter: Replicates per conf_str value. Default 100.
n_samples: Observations per replicate. Default 500.
n_features: Features per replicate. Default 20.
w_signal: Proxy quality omega. Default 0.70.
mo_confounding: Strength of U1 -> M. Default 0.80.
phi: Strength of the mediator instrument Gm -> M. 0 = no mediator instrument. Default 0.
rho_G1: Correlation of G1 with conf_XM. Default 0.
rho_G2: Correlation of G2 with conf_MY. Default 0.
rho_pop: Shared population structure. Default 0.
lambda_XM: Optional per-path confounder loading vector (X->M path).
lambda_MY: Optional per-path confounder loading vector (M->Y path).
omega_1: Coverage of conf_XM by W1. NULL = use w_signal.
omega_2: Coverage of conf_MY by W2. NULL = use w_signal.
feat_cor: Within-module feature correlation. Default 0.
base_seed: Seed offset. Default 900.
n_cores: Parallel workers. Default 1.
alpha: Significance threshold. Default 0.05.

Value

A data frame with columns: conf_str, method, NIE_type1, NDE_type1.

Examples

t1e <- sweep_mediation_null_by_conf(c(0.2, 0.8), n_iter = 3,
                                    n_samples = 100)
head(t1e)
#>   conf_str method NIE_type1  NDE_type1
#> 1      0.2  UNADJ 0.7666667 0.01666667
#> 2      0.2 DIRECT 0.0000000 0.08333333
#> 3      0.2   COCA 0.0000000 0.00000000
#> 4      0.2 IV2SLS 0.1833333 0.05000000
#> 5      0.2    PGC 0.0000000 0.06666667
#> 6      0.8  UNADJ 1.0000000 0.05000000