CODEWITHSUNDEEP
rnamissing-data
Manolo
Manolo

Reputation: 73

error in finalfit::missing_compare: does not find an object

I am running tests on missing data using the finalfit package.

I have a dataset that has 11,046 obs and 27 variables. I have more than one dependent variable because I need later to develop a Confirmatory factor analysis with lavaan. The dataset can be found here.

explanatory_edu <- c("ch_edu", "a4g_4")

dependent <- "br_logical"

sl_cfa %>% 
   missing_compare(dependent, explanatory_edu)

I get the following error message:

Error in factor(g, levels = unique(g)) : object 'g' not found

What is this g object the error is referring to?

This is the output of ff_glimpse

> sl_cfa %>% 
+   ff_glimpse(dependent, explanatory_edu)
Continuous
# A tibble: 11,046 x 0

Categorical
                label var_type     n missing_n missing_percent levels_n levels levels_count
br_logical br_logical    <lgl>  6398      4648            42.1        2      -            -
ch_edu         ch_edu    <lgl> 11046         0             0.0        2      -            -
a4g_4           a4g_4    <lgl>  8723      2323            21.0        2      -            -
           levels_percent
br_logical              -
ch_edu                  -
a4g_4                   -

Not sure it is helpful, but I do not get any error with missing_pairs

sl_cfa %>% 
  missing_pairs(dependent, explanatory_edu, position = "fill")

which gives me this plot

enter image description here

where we can see that ch_edu seems to be MAR and a4g_4 seems to be MCAR.

PS

Would anyone with the adequate reputation create tags for finalfit and missing_compare function, please? Many thanks

Upvotes: 1

Views: 340

Answers (1)

Ewen
Ewen

Reputation: 1381

Thanks.

An underlying function used here is being re-written and the problems caused by tibbles will likely go away.

For now you can: 

library(finalfit)

explanatory_edu <- c("ch_edu", "a4g_4")
sl_cfa %>%
  data.frame() %>% 
  missing_compare(dependent, explanatory_edu)

#> Missing data analysis: br_logical       Not missing     Missing      p
#>                            ch_edu FALSE 4203 (65.7) 1730 (37.2) <0.001
#>                                    TRUE 2195 (34.3) 2918 (62.8)       
#>                             a4g_4 FALSE 2548 (50.5) 1774 (48.2)  0.034
#>                                    TRUE 2496 (49.5) 1905 (51.8)

As you mention, missingness in the variable ch_edu is strongly associated with br_logical.

Upvotes: 1

Related Questions