ROPE range defaults to [-0.1,0.1] instead of '0 +/- .1 * sd(y)' for stan_glm in describe_posterior

[strengejacke]

Fixes #587

[strengejacke]

library(rstanarm)
d <- data.frame(
  y = c(
    7, 38, 6, 11, 0, 8, 6, 5, 28, 2, 32, 14,
    46, 19, 20, 28, 51, 12, 10, 1, 2, 22
  ),
  x = c(
    46, 41, 30, 30, 61, 42, 63, 13, 52, 59, 42, 45,
    55, 47, 36, 48, 70, 67, 50, 34, 72, 46
  )
)
m <- stan_glm(y ~ x, data = d)
bayestestR::describe_posterior(m)
#> Summary of Posterior Distribution
#> 
#> Parameter   | Median |          95% CI |     pd |          ROPE | % in ROPE |  Rhat |     ESS
#> ---------------------------------------------------------------------------------------------
#> (Intercept) |   9.00 | [-15.60, 33.02] | 77.55% | [-1.48, 1.48] |     7.29% | 1.000 | 3213.00
#> x           |   0.16 | [ -0.32,  0.66] | 76.02% | [-1.48, 1.48] |      100% | 1.000 | 3240.00

^{Created on 2023-03-14 with reprex v2.0.2}

[mattansb]

What was this doing before?

[strengejacke]

It was working several months ago, but then Vincent and I came up with a performance improvement by converting model objects to data frames only once (https://github.com/easystats/bayestestR/pull/532). However, in .describe_posterior(), instead of model objects, a data frame of posteriors was passed to every function (like pd(), rope() etc.). This works well in most situations, except when you want to define the rope range. This PR fixes this issue

[strengejacke]

errors seem unrelated