How to avoid the adding of the column "ModelLik" when tab_df() is used to export the AIC table

Question

As example two LMM.

lme1 <- lme(mpg ~ cyl + disp,
           random = ~1|disp, 
           method = "ML", 
           data = mtcars)

lme2 <- lme(mpg ~ cyl *  disp,
           random = ~1|disp, 
           method = "ML",
           data = mtcars)

AIC table to select the best model.

library(AICcmodavg)

Cand.models <- list( )
Cand.models[[1]] <- lme1
Cand.models[[2]] <- lme2

aictab(Cand.models, sort = TRUE)

Model selection based on AICc:

     K   AICc Delta_AICc AICcWt Cum.Wt     LL
Mod2 6 164.40       0.00   0.94   0.94 -74.52
Mod1 5 169.87       5.46   0.06   1.00 -78.78

To export the AIC table I like to use the tab_df() function.

library(sjPlot)

tab_df(aictab(Cand.models, sort = TRUE))

Unnecessary the tab_df() functions add the column "ModelLik" to the table, how can I avoid this?

Answer 1

The reason is that the object returned by aictab has some more columns then printed with its print method. In the following, I assign the returned table to a variable tb and inspect it with str(). If you use RStudio, you can also see it in the Environment explorer.

Function tab_df just formats the data frame, so we can select, remove or even rename columns according to our needs. The following shows an example. As a small goody, I user-defined names for the models:

library("nlme")
library("AICcmodavg")
library("sjPlot")

lme1 <- lme(mpg ~ cyl + disp, random = ~1|disp, method = "ML", data = mtcars)
lme2 <- lme(mpg ~ cyl *  disp, random = ~1|disp, method = "ML", data = mtcars)

# alternative way to produce the list, can optionally provide speaking names 
Cand.models <- list( 
  'model 1' = lme1,
  'model 2' = lme2
)

# assign the table to a variable
tb <- aictab(Cand.models, sort = TRUE)
## look what is in

str(tb)
which_columns <- c("Modnames", "K", "AICc", "Delta_AICc", "AICcWt", "Cum.Wt", "LL")
tab_df(aictab(Cand.models, sort = TRUE)[which_columns])