Draws one dataset from a named IV DGP via sim_iv_data(), then evaluates all
requested methods on that same draw.
Usage
sim_iv_one_rep(
n,
kx,
q,
k,
s,
design,
methods = "dnn_iv_oracle",
alpha = 0.05,
seed = NULL,
sigma = 0.5,
clip = 0.001,
n_folds = 5L
)Arguments
- n
Integer sample size.
- kx
Integer covariate dimension.
- q
Integer number of instruments requested from the DGP.
- k, s
Integer DNN tuning parameters.
- design
IV DGP name; passed to
sim_iv_data().- methods
Character vector of method names; see
sim_iv_methods().- alpha
Nominal test size for confidence intervals and J-tests.
- seed
Optional integer seed.
- sigma
Noise SD forwarded to
sim_iv_data().- clip
Clipping bound forwarded to DGPs and simulation nuisance fits.
- n_folds
Number of cross-fitting folds for cross-fit methods.
Value
A data.frame with one row per method and columns
method, theta_hat, theta0, bias, sq_err, se, cover,
ci_len, neff, J, df, pval_J, and reject_J.
Examples
sim_iv_one_rep(
n = 120, kx = 2, q = 2, k = 1, s = 10,
design = "dgp2_iv_homog_overid",
methods = c("dnn_iv_oracle", "iv_oracle"),
seed = 1
)
#> method theta_hat theta0 bias sq_err se cover
#> 1 dnn_iv_oracle 1.196561 1.211738 -0.01517726 0.0002303491 0.3300773 1
#> 2 iv_oracle 1.196561 1.211738 -0.01517726 0.0002303492 0.3300773 1
#> ci_len neff J df pval_J reject_J
#> 1 1.293879 21.94304 1.66955 1 0.1963189 0
#> 2 1.293879 21.94304 1.66955 1 0.1963188 0