The problem is that salary_var is a list containing a single-element. The call to sort() is then trying to sort a list, not an atomic element. You can see that salary_var is a list by running str(salary_var). If you omit the c(), you’ll instead end up with a data frame with a single column, which gives the same problem.
Two simple solutions:
To sort the values in the element of the list, use
head(sort(salary_var[[1]], decreasing=TRUE), 3)
where the [[1]] selects the first element of the list and sorts the values within it.
Alternatively, create salary_var explicitly as a numeric vector instead:
salary_var <- (irs_data[13]/irs_data[12])[[1]]
One note: in your post, you wrote new_var instead of salary_var in your call to sort() which may confuse other readers.