`parameters.fixest`: Uninformative error

[vincentarelbundock]

This generates an uninformative error:

library(tidyverse)
mod <- fixest::fenegbin(height~mass+birth_year+factor(sex), data=starwars)
parameters::model_parameters(mod)

Error: Something went wrong... :-/

FWIW,

> parameters:::.model_parameters_generic(mod)
Error in data.frame(Parameter = params$Parameter, Statistic = as.vector(cs[,  :
  arguments imply differing number of rows: 6, 7

[strengejacke]

The initial issue is fixed, but of course, model_parameters() should not fail at all for this model. Will look into this.

data(starwars, package = "dplyr")
mod <- fixest::fenegbin(height~mass+birth_year+factor(sex), data=starwars)
#> NOTE: 51 observations removed because of NA values (LHS: 6, RHS: 51).
parameters::model_parameters(mod)
#> Error: Sorry, `model_parameters()` failed with the following error (possible
#>   class `fixest` not supported):
#> 
#>   arguments imply differing number of rows: 6, 7

^{Created on 2025-01-01 with reprex v2.1.1}

[strengejacke]

Ok, fixed in #1055 and easystats/insight#984

data(starwars, package = "dplyr")
mod <- fixest::fenegbin(height~mass+birth_year+factor(sex), data=starwars)
#> NOTE: 51 observations removed because of NA values (LHS: 6, RHS: 51).
parameters::model_parameters(mod)
#> # Fixed Effects
#> 
#> Parameter            |  Log-Mean |       SE |         95% CI | t(30) |      p
#> -----------------------------------------------------------------------------
#> (Intercept)          |      4.84 |     0.07 | [ 4.70,  4.99] | 67.66 | < .001
#> mass                 |  5.65e-03 | 9.14e-04 | [ 0.00,  0.01] |  6.18 | < .001
#> birth year           | -5.62e-04 | 1.83e-04 | [ 0.00,  0.00] | -3.08 | 0.004 
#> sex [hermaphroditic] |     -7.01 |     1.23 | [-9.53, -4.50] | -5.70 | < .001
#> sex [male]           |     -0.09 |     0.06 | [-0.22,  0.03] | -1.54 | 0.134 
#> sex [none]           |     -0.27 |     0.09 | [-0.46, -0.09] | -3.01 | 0.005
#> 
#> Uncertainty intervals (equal-tailed) and p-values (two-tailed) computed
#>   using a Wald t-distribution approximation.
#> 
#> The model has a log- or logit-link. Consider using `exponentiate =
#>   TRUE` to interpret coefficients as ratios.

^{Created on 2025-01-01 with reprex v2.1.1}

[vincentarelbundock]

Thanks, this looks great!!