Bias-Corrected and Accelerated Confidence Intervals

Bias-Corrected and Accelerated Confidence Intervals

bcaci(
  thetahatstar,
  thetahatstarjack = NULL,
  thetahat,
  theta = NULL,
  data,
  std = FALSE,
  complete = TRUE,
  alpha = c(0.001, 0.01, 0.05),
  par = TRUE,
  ncores = NULL,
  blas_threads = TRUE,
  mc = TRUE,
  lb = FALSE
)

Arguments

thetahatstar	Numeric vector. Sampling distribution of thetahat.
thetahatstarjack	Numeric vector. Jackknife vector of parameter estimates. If thetahatstarjack = NULL, thetahatstarjack is calculated using jack().
thetahat	Numeric. Parameter estimate.
theta	Numeric. Parameter. Optional argument.
data	n by 3 matrix or data frame. data[, 1] correspond to values for x. data[, 2] correspond to values for m. data[, 3] correspond to values for y.
std	Logical. Standardize the indirect effect \(\hat{\alpha} \hat{\beta} \frac{\sigma_x}{\sigma_y}\).
complete	Logical. If TRUE, create complete data set before bootstrapping. If 'FALSE`' bootstrap incomplete data set and fit the model with FIML.
alpha	Numeric vector. Alpha level. By default alpha = c(0.001, 0.01, 0.05).
par	Logical. If TRUE, use multiple cores. If FALSE, use lapply().
ncores	Integer. Number of cores to use if par = TRUE. If unspecified, defaults to detectCores() - 1.
blas_threads	Logical. If TRUE, set BLAS threads using blas_set_num_threads(threads = blas_get_num_procs()). If FALSE, set BLAS threads using blas_set_num_threads(threads = 1). If par = TRUE, blas_threads is automatically set to FALSE to prevent conflicts in parallel processing. This argument is useful when FUN can handle implicit parallelism when par = FALSE, for example linear algebra operations.
mc	Logical. If TRUE, use parallel::mclapply(). If FALSE, use parallel::parLapply() or parallel::parLapplyLB(). Ignored if par = FALSE.
lb	Logical. If TRUE use parallel::parLapplyLB(). If FALSE, use parallel::parLapply(). Ignored if par = FALSE and mc = TRUE.

See also

Other confidence intervals functions: bcci(), evalci(), len(), pcci(), shape(), theta_hit(), zero_hit()

Author

Ivan Jacob Agaloos Pesigan

Examples

B <- 5000
data <- jeksterslabRdatarepo::thirst
n <- nrow(data)
muthetahat <- colMeans(data)
Sigmathetahat <- cov(data)
thetahat <- fit.ols(data, minimal = TRUE)
thetahatstar <- pb.mvn(
  muthetahat = muthetahat, Sigmathetahat = Sigmathetahat,
  n = n, B = 5000, par = FALSE
)
bcaci(
  thetahatstar = thetahatstar,
  thetahat = thetahat,
  theta = 0.15, # assuming that the true indirect effect is 0.15
  data = data,
  par = FALSE
)
#>             est              se            reps         ci_0.05          ci_0.5 
#>    1.527185e-01    7.733199e-02    5.000000e+03   -2.433815e-02    7.725781e-03 
#>          ci_2.5         ci_97.5         ci_99.5        ci_99.95   zero_hit_99.9 
#>    3.826733e-02    3.513996e-01    4.127810e-01    5.210673e-01    1.000000e+00 
#>     zero_hit_99     zero_hit_95        len_99.9          len_99          len_95 
#>    0.000000e+00    0.000000e+00    5.454055e-01    4.050552e-01    3.131323e-01 
#>      shape_99.9        shape_99        shape_95  theta_hit_99.9    theta_hit_99 
#>    2.080401e+00    1.793624e+00    1.735947e+00    1.000000e+00    1.000000e+00 
#>    theta_hit_95 theta_miss_99.9   theta_miss_99   theta_miss_95           theta 
#>    1.000000e+00    0.000000e+00    0.000000e+00    0.000000e+00    1.500000e-01 
thetahat <- fit.ols(data, minimal = TRUE, std = TRUE)
thetahatstar <- pb.mvn(
  muthetahat = muthetahat, Sigmathetahat = Sigmathetahat,
  n = n, std = TRUE, B = 5000, par = FALSE
)
bcaci(
  thetahatstar = thetahatstar,
  thetahat = thetahat,
  theta = 0.15, # assuming that the true indirect effect is 0.15
  data = data,
  std = TRUE,
  par = FALSE
)
#>             est              se            reps         ci_0.05          ci_0.5 
#>      0.15303271      0.07212713   5000.00000000     -0.01705917      0.01281978 
#>          ci_2.5         ci_97.5         ci_99.5        ci_99.95   zero_hit_99.9 
#>      0.04354391      0.33414392      0.39591032      0.45179941      1.00000000 
#>     zero_hit_99     zero_hit_95        len_99.9          len_99          len_95 
#>      0.00000000      0.00000000      0.46885858      0.38309055      0.29060001 
#>      shape_99.9        shape_99        shape_95  theta_hit_99.9    theta_hit_99 
#>      1.75650188      1.73220550      1.65415281      1.00000000      1.00000000 
#>    theta_hit_95 theta_miss_99.9   theta_miss_99   theta_miss_95           theta 
#>      1.00000000      0.00000000      0.00000000      0.00000000      0.15000000