tbl_summary and numeric variables

Question

tbl_summary [library (gtsummary)] does not treat all numeric variables in the same way and I can't figure out how to change it. For example.

mtcars only has numeric variables, so when I run this, I expect the means of every variable to be calcuated. Instead, it treats cyl, gear and carb as categorical.

tbl_summary(mtcars, statistic = list(all_numeric() ~ "{mean} ({sd})",
                                      all_categorical() ~ "{n} / {N} ({p}%)"))

I actually have a much bigger dataset and tbl_summary is treating some of the numeric variables as categorical. Would it be because there are such few N's (let's say I have a lot of missing rows) and tbl_summary does not want to calculate the mean for such a small N?

I can't wrap my mind around this!

Just a further example from my data. Q12_5_TEXT is a numeric variable, but this is the output from tbl_summary.

@Daniel D. Sjoberg please let me know if you have any suggestions! — NewBee
– NewBee, Commented Sep 24, 2020 at 0:24

Daniel D. Sjoberg · Accepted Answer · 2020-09-24 00:43:49Z

4

Variables with few unique levels are summarized categorically. For example, mtcars$cyl only has three unique levels: 4, 6, 8. With only three levels, a categorical summary is more appropriate than a mean or median.

Use the type= argument to change the default summary type.

answered Sep 24, 2020 at 0:43

Daniel D. Sjoberg

12.8k2 gold badges20 silver badges37 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

NewBee Over a year ago

How to do I change this default? type = all_continuous() ~ "continuous2", does not work

Daniel D. Sjoberg Over a year ago

continuous2 is in version 1.3.5 and higher. If downloading the new version doesn't solve your issue, then please post a reprex

James Cutler · Accepted Answer · 2021-01-21 01:22:56Z

3

I tried type = all_continuous() ~ "continuous2", and I have version 1.3.5, and it didn't change the summary type:

library(tidyverse)
library(gtsummary)

nrows <- 30

df <- tibble(
  a = sample(c(0,1,3.5,7.5),nrows,replace = T),
  b = sample(c("Group I","Group II"),nrows,replace = T)
)

df %>% 
  tbl_summary(
    by = b,
    type = all_continuous() ~ "continuous2",
    statistic = all_continuous() ~ "{mean} ({sd})"
  )

The output from this summarized variable 'a' as if it was a categorical variable in spite of the type argument. I also ran into this issue which is why I came here for the answer. If there is a different argument that I should be using that you could point me to, I would greatly appreciate it!

answered Jan 21, 2021 at 1:22

James Cutler

811 silver badge7 bronze badges

2 Comments

James Cutler Over a year ago

If 1.3.5 doesn't have this functionality yet, is there anyway you could add it in the upcoming 1.3.6 update?

Daniel D. Sjoberg Over a year ago

You need type = a ~ "continuous2". 😸

GClarke · Accepted Answer · 2022-01-31 23:02:26Z

2

I had this same issue and I fixed it by telling tbl_summary that the categorical variables are in fact continuous. Try:

df %>% 
  tbl_summary(
    by = b,
    type = list(all_continuous() ~ "continuous2",
                          all_categorical() ~ "continuous2"),
    statistic = all_continuous() ~ "{mean} ({sd})"
  )

answered Jan 31, 2022 at 23:02

GClarke

359 bronze badges

1 Comment

Kevin T Over a year ago

Thanks! Seems like a bug to have to declare that our continuous variables that are being treated as categorical are in fact continuous.

Collectives™ on Stack Overflow

tbl_summary and numeric variables

3 Answers 3

2 Comments

2 Comments

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

2 Comments

2 Comments

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related