Alpha-NOMINATE (stub)

Carroll et al. (2013) mixture model between Gaussian and quadratic utility, sampled via slice sampling (Neal 2003). Porting the slice sampler is beyond this session's budget.

Usage

morie_spatial_voting_alpha_nominate(
  votes,
  n_dims = 2L,
  n_samples = 500L,
  burn_in = 100L,
  seed = 42L
)

Arguments

votes

Vote matrix. @param n_dims Latent dimensions.

n_dims

Integer; latent ideal-point dimensionality (default 2).

n_samples

MCMC samples. @param burn_in Burn-in length.

burn_in

Integer; MCMC burn-in iterations to discard before summarising the posterior.

seed

RNG seed.

Value

List with ideal points, discrimination, difficulty and an engine tag (EM-IRT closed-form approximation).

References

Carroll, R., Lewis, J. B., Lo, J., Poole, K. T., and Rosenthal, H. (2013); Neal, R. M. (2003) Annals of Statistics.

Examples

morie_spatial_voting_alpha_nominate(matrix(0, 5, 5))
#> $ideal_points
#>             [,1]     [,2]
#> [1,]  0.03451343 1.016539
#> [2,] -0.63786317 1.331414
#> [3,] -0.26056360 1.041006
#> [4,] -0.28837748 1.383241
#> [5,] -0.25001500 1.043871
#> 
#> $discrimination
#>              [,1]      [,2]
#> [1,]  0.608461237 -3.971283
#> [2,]  0.922349912 -4.011777
#> [3,] -0.001279056 -4.704735
#> [4,]  0.306571288 -4.643598
#> [5,]  0.562128133 -4.241045
#> 
#> $difficulty
#> [1] -7.024192 -6.922670 -6.373856 -6.376845 -6.749693
#> 
#> $n_dims
#> [1] 2
#> 
#> $n_samples_target
#> [1] 500
#> 
#> $burn_in_target
#> [1] 100
#> 
#> $engine
#> [1] "morie_spatial_voting_em_irt (deterministic EM approximation; full alpha-NOMINATE Gibbs not yet ported -- see Carroll et al 2013)"
#>