`labelled` + `parameters` leads to `NA` variable labels

[vincentarelbundock]

Hi all,

We found some kind of incompatibility between labelled and paramters for some data frames.

Notice that the labels include several NA in the first example but not the second. In the first, I use the data supplied by the pscl package. In the second, I use a data frame from Rdatasets.

My hunch is that labelled does not apply the labels in the same way to both data frames, but I don’t know this package well enough to be sure.

This was first reported by @raffaem here: https://github.com/vincentarelbundock/modelsummary/issues/891

library(parameters)
library(labelled)
library(Rdatasets)

# Broken
mydf <- pscl::bioChemists
var_label(mydf) <- list(art="MyCount", fem="MyGender", kid5="Kids")
model <- lm(art ~ fem * factor(kid5), data = mydf)
p <- parameters(model, pretty_names = "labels")
attr(p, "pretty_labels")
#>             (Intercept)                femWomen           factor(kid5)1 
#>           "(Intercept)"      "MyGender [Women]"              "kid5 [1]" 
#>           factor(kid5)2           factor(kid5)3  femWomen:factor(kid5)1 
#>              "kid5 [2]"              "kid5 [3]" "MyGender [Women] * NA" 
#>  femWomen:factor(kid5)2  femWomen:factor(kid5)3 
#> "MyGender [Women] * NA" "MyGender [Women] * NA"

# Working
mydf <- rddata("bioChemists")
var_label(mydf) <- list(art="MyCount", fem="MyGender", kid5="Kids")
model <- lm(art ~ fem * factor(kid5), data = mydf)
p <- parameters(model, pretty_names = "labels")
attr(p, "pretty_labels")
#>              (Intercept)                 femWomen            factor(kid5)1 
#>            "(Intercept)"            "fem [Women]"               "kid5 [1]" 
#>            factor(kid5)2            factor(kid5)3   femWomen:factor(kid5)1 
#>               "kid5 [2]"               "kid5 [3]" "fem [Women] × kid5 [1]" 
#>   femWomen:factor(kid5)2   femWomen:factor(kid5)3 
#> "fem [Women] × kid5 [2]" "fem [Women] × kid5 [3]"
`