Parametric bootstrap of NOMINATE standard errors — morie_spatial_voting_nominate

Lewis & Poole (2004) parametric bootstrap: simulate roll-call matrices from fitted probabilities, re-estimate per bootstrap replicate, compute SE from the bootstrap distribution.

Usage

morie_spatial_voting_nominate_bootstrap(
  votes,
  ideal_points,
  normal_vectors_arr,
  cutpoints,
  n_boot = 100L,
  seed = 42L
)

Arguments

votes: Original vote matrix.
ideal_points: Fitted ideal points.
normal_vectors_arr: Fitted normal vectors.
cutpoints: Fitted cutpoints.
n_boot: Number of bootstrap replications.
seed: RNG seed.

Value

List with se_ideal_points, boot_means, n_boot.

References

Lewis, J. B. and Poole, K. T. (2004). "Measuring Bias and Uncertainty in Ideal Point Estimates via the Parametric Bootstrap." Political Analysis, 12(2).

Examples

set.seed(1)
v <- matrix(stats::rbinom(40, 1, 0.5), 5, 8)
fit <- morie_spatial_voting_dw_nominate(v, max_iter = 5L)
morie_spatial_voting_nominate_bootstrap(
  v, fit$ideal_points, fit$normal_vectors, fit$cutpoints, n_boot = 5L)
#> $se_ideal_points
#>            [,1]       [,2]
#> [1,] 0.03927144 0.08696910
#> [2,] 0.07038480 0.10135607
#> [3,] 0.08214763 0.11470586
#> [4,] 0.06887660 0.13162676
#> [5,] 0.13492223 0.07540698
#> 
#> $boot_means
#>            [,1]       [,2]
#> [1,]  0.3915152  0.5471552
#> [2,]  0.0532093  0.3298160
#> [3,]  0.2830456  0.2806046
#> [4,]  0.4296193  0.6784253
#> [5,] -0.1370382 -0.0269331
#> 
#> $n_boot
#> [1] 5
#>