Print glmmTMB conditional and zero-inflated model side-by-side with modelsummary

I'm trying to print the conditional and zero-inflated model of a zero-inflated Poisson model estimated with glmmTMB with modelsummary, but I'm unable to do so.

Here is my attempt so far, following the syntax described here

Owls <- transform(Owls,
                  Nest=reorder(Nest,NegPerChick),
                  NCalls=SiblingNegotiation,
                  FT=FoodTreatment)

fit_zipoisson <- glmmTMB(NCalls~(FT+ArrivalTime)*SexParent+
                           offset(log(BroodSize))+(1|Nest),
                         data=Owls,
                         ziformula=~1,
                         family=poisson)

ti <- list(
  broom.mixed::tidy(fit_zipoisson) |>
    filter(effect=="fixed" & component=="cond"),
  broom.mixed::tidy(fit_zipoisson) |>
    filter(effect=="fixed" & component=="zi")
)

gl <- list(
  broom.mixed::glance(fit_zipoisson),
  broom.mixed::glance(fit_zipoisson)
)

mod <- list(
  tidy = ti,
  glance = gl)
class(mod) <- "modelsummary_list"

modelsummary(mod)

which returns:

Error: `estimate` is not available. The `estimate` and `statistic` arguments must correspond to column names in the output of this command: `get_estimates(model)`

I guess this syntax doesn't support passing a list?

Solution

The catch is that a custom modelsummary list needs named items ("tidy" and "glance").


    library(glmmTMB)  
    ## your example data and model ...

create a helper function which returns a modelsummary list for given model, effects and component:


    get_modsum_list <- function(model, effects = "fixed", component = "cond"){
      mod <- list(
        ## list needs named items ('tidy' and 'glance')
        tidy = broom.mixed::tidy(model, effects = effects, component = component),
        glance = broom.mixed::glance(model)
      )
      class(mod) <- "modelsummary_list"
      mod
    }

get the desired modelsummary lists, combine to named list and feed into modelsummary:


    list(
      'conditional' = get_modsum_list(fit_zipoisson),
      'zero-inflated' = get_modsum_list(fit_zipoisson, component = "zi")
    ) |>
      modelsummary()