Ordered Probit Response Model

Ordered responses based on a latent Gaussian evidence variable. Binary response models are the 2-category special case.

Usage

ordered_probit()

Value

A model list with all the necessary functions to sample

Details

Model parameters are: location (unbounded), scale (log scale), and cut (first threshold free, remaining thresholds modeled as positive increments).

The cut ~ 1 specification yields the standard flexible K - 1 threshold parameterization from the ordinal regression literature. The final response category has an implicit upper cut at Inf. Internally, Ttransform leaves sampled cut values unchanged and returns the ordered thresholds in a derived cut_expanded column used by the likelihood and random-number generator.

Examples

dord <- design(
  Rlevels = c("left", "right"),
  factors = list(subjects = 1, S = c("left", "right")),
  formula = list(location ~ 0 + S, scale ~ 1, cut ~ 1),
  matchfun = function(d) d$S == d$lR,
  constants = c(scale = log(1)),
  model = ordered_probit
)
#> 
#>  Sampled Parameters: 
#> [1] "location_Sleft"  "location_Sright" "cut"            
#> 
#>  Design Matrices: 
#> $location
#>      S location_Sleft location_Sright
#>   left              1               0
#>  right              0               1
#> 
#> $scale
#>  scale
#>      1
#> 
#> $cut
#>  cut
#>    1
#>