I want to create multiple weighted crosstabs for a dataset and then bind them together. I have to do this for around 30 variables, with 13 or so crosstabs each, so I am trying to figure out a way to do it as automated as possible, using dplyr, but I'm getting a little lost between all the options for writing a function and loops, etc. I am using the pollster package to create my crosstabs, so I will use the illinois dataset in that package for my example.

Say I want to make a series of tables from this dataset, where each table displays a series of crosstabs (say, sex, educ6, and raceethnic) for a given variable (say I want the tables to be for martialstatus, rv, and voter).

I have created a function that can create each crosstab table with the format I want.

create_crosstabs <- function(var1, var2) {

  var1 <- enquo(var1)
  var2 <- enquo(var2)
  
  df2 <- illinois %>% 
  pollster::crosstab(!!var1, !!var2, weight = weight) %>% 
  mutate_if(is.numeric, round) %>% 
  select(-n)
  
return(df2)
}

This works, but I am struggling to get to the next part. So what this code makes if I run

create_crosstabs(educ6, maritalstatus) 

is a table that looks like:

I want to make it so I loop over for example, sex, educ6, and raceethnic, each time creating a crosstabs and then appending that crosstab table to the previous one, such that the table output looks like

I've previously done something like where looped over values of a variable, rather than over multiple variables, but trying to write it for this, nothing I do seems to work.

I will mention here that ideally I would like to further iterate over the second variable such that I get a series of these tables for the crosstabs for maritalstatus, rv, and voter, but if I could at least figure out how to create the individual tables themselves, that would be a big step.

CodePudding user response：

In the function, change the enquo to ensym

create_crosstabs <- function(var1, var2) {

  var1 <- ensym(var1)
  var2 <- ensym(var2)
  
  illinois %>% 
  pollster::crosstab(!!var1, !!var2, weight = weight) %>% 
  mutate(across(where(is.numeric), round)) %>% 
  select(-n)
  

}

so that it can take both quoted as well as unquoted input.

library(purrr)
library(pollster)
library(dplyr)
library(gt)
map_dfr(c(educ = "educ6", sex = "sex", race = "raceethnic"), 
   ~ create_crosstabs(!!.x, maritalstatus), .id = 'group') %>%
    group_by(group) %>%
    gt()

Or may also do

v1 <-c(educ = "educ6", sex = "sex", race = "raceethnic")
tmp <- imap_dfr(v1, 
                ~ create_crosstabs(!!.x, maritalstatus) %>% 
                 rename(group = 1), .id = "grp" ) 
tmp2 <- tmp %>%
      select(-grp) %>% 
      gt(rowname_col = "group")
for(nm in rev(names(v1)))
  tmp2 <- tmp2 %>%
  tab_row_group(label = nm, rows = which(tmp$grp == nm))
tmp2

-output