add_n & add_continous_stat for Likert tables

[larmarange]

fix #55

• [ ] add examples in documentation
• [ ] add tests

[larmarange]

Hi @ddsjoberg

I hope you are well. I have rewritten add_n.tbl_likert() to use all functionalities available in add_n.tbl_summary.

I have also added a add_continuous_stat() method (name could be changed) to allow the addition of continuous statistics from a Likert table. You may apply add_continuous_stat() several times.

library(tidyverse)
library(bstfun)

data(efc, package = "sjPlot")

cop <- sjmisc::find_var(efc, pattern = "cop", out = "df") %>%
  sjmisc::numeric_to_factor() %>%
  mutate(across(everything(), ~ fct_relabel(.x, str_to_lower))) %>%
  as_tibble() %>%
  mutate(c82cop1 = labelled::recode_if(c82cop1, c82cop1 == "never", NA))
names(cop) <- paste0("q", 1:ncol(cop))

cop %>% 
  tbl_likert(statistic = ~ "{p}%", sort = "desc") %>%
  add_n(statistic = "{n}/{N}", footnote = TRUE) %>%
  add_continuous_stat() %>%
  add_continuous_stat(statistic = "{sd}") %>%
  as_kable()

Characteristic	N	always	never	often	sometimes	Mean	SD
do you find caregiving too demanding?	902/908	4.3%	21%	14%	61%	3.31	0.94
do you feel trapped in your role as caregiver?	900/908	8.6%	37%	13%	42%	2.87	1.06
does caregiving have negative effect on your physical health?	898/908	6.5%	46%	9.5%	39%	2.80	1.03
do you feel supported by friends/neighbours?	901/908	12%	35%	27%	26%	2.67	1.00
do you feel you cope well as caregiver?	898/908	23%	0%	66%	11%	2.64	0.96
does caregiving cause difficulties in your relationship with your friends?	902/908	5.8%	57%	9.1%	28%	2.59	0.96
does caregiving cause difficulties in your relationship with your family?	902/908	1.7%	69%	5.5%	23%	2.51	0.87
do you feel caregiving worthwhile?	888/908	34%	8.6%	34%	24%	2.47	1.18
does caregiving cause financial difficulties?	900/908	1.9%	79%	4.3%	15%	2.32	0.74

^{Created on 2021-10-17 by the reprex package (v2.0.1)}

[larmarange]

For a Likert table, I have the feeling that "{p}%" would be a better and more common default value for statistic.

[larmarange]

I have updated the example (levels order was incorrect). In this example, the last variable has voluntarily a missing level.

Should we think about an example dataset to provide for tbl_likert?

library(tidyverse)
library(bstfun)

data(efc, package = "sjPlot")

cop <- sjmisc::find_var(efc, pattern = "cop", out = "df") %>%
  sjmisc::numeric_to_factor() %>%
  mutate(across(everything(), ~ fct_relabel(.x, str_to_lower) %>% fct_relevel("never", "sometimes", "often", "always"))) %>%
  as_tibble() %>%
  mutate(c82cop1 = labelled::recode_if(c82cop1, c82cop1 == "never", NA) %>% fct_drop()) %>%
  relocate(c82cop1, .after = last_col())
names(cop) <- paste0("q", 1:ncol(cop))

tbl <- cop %>% 
  tbl_likert(statistic = ~ "{p}%", sort = "desc") %>%
  add_n(statistic = "{n}/{N}", footnote = TRUE) %>%
  add_continuous_stat() %>%
  add_continuous_stat(statistic = "{sd}") 
tbl %>%
  as_kable()

Characteristic	N	never	sometimes	often	always	Mean	SD
do you feel you cope well as caregiver?	898/908	0%	11%	66%	23%	3.13	0.57
do you feel caregiving worthwhile?	888/908	8.6%	24%	34%	34%	2.93	0.96
do you feel supported by friends/neighbours?	901/908	35%	26%	27%	12%	2.16	1.04
do you find caregiving too demanding?	902/908	21%	61%	14%	4.3%	2.02	0.72
do you feel trapped in your role as caregiver?	900/908	37%	42%	13%	8.6%	1.92	0.91
does caregiving have negative effect on your physical health?	898/908	46%	39%	9.5%	6.5%	1.77	0.87
does caregiving cause difficulties in your relationship with your friends?	902/908	57%	28%	9.1%	5.8%	1.63	0.87
does caregiving cause difficulties in your relationship with your family?	902/908	69%	23%	5.5%	1.7%	1.39	0.67
does caregiving cause financial difficulties?	900/908	79%	15%	4.3%	1.9%	1.29	0.64

^{Created on 2021-10-17 by the reprex package (v2.0.1)}

[ddsjoberg]

hey hey @larmarange ! hope you're well! Am I meant to do something with this, or are you working on it?

[larmarange]

Dear @ddsjoberg I'm truly sorry. I'm currently overloaded with my regular work and I forgot to finalise the PR. If you have time, do not hesitate to work on it / finalise it.

The code is working. In the discussion, there are some remaining questions:

• For a Likert table, I have the feeling that "{p}%" would be a better and more common default value for statistic.
• Should we think about an example dataset to provide for tbl_likert?
• We could/should provide more examples in documentation.
• Unit tests are missing

Best regards

PS: another open question to be addressed in a separate PR: should we add a group option in tbl_likert() to produce sub-tables according to a factor and to stack them? If yes, that argument will have to be taken into account also in add_n() and add_continuous_stat()

[ddsjoberg]

@larmarange no worries, and on apology needed 😸

PS: another open question to be addressed in a separate PR: should we add a group option in tbl_likert() to produce sub-tables according to a factor and to stack them? If yes, that argument will have to be taken into account also in add_n() and add_continuous_stat()

Would this functionality be covered in tbl_strata()?

[larmarange]

Would this functionality be covered in tbl_strata()?

Yes it can be. Maybe a good idea to simply add an example with it.

[larmarange]

Let me know if you want to finalize the PR. Otherwise, I can try to find some time next week.

Regards

[ddsjoberg]

Let me know if you want to finalize the PR. Otherwise, I can try to find some time next week.

Regards

I won't have time in the next week either. Go ahead when you've got the time. Happy (American) Thanksgiving! 🦃 🦃 🦃

[ddsjoberg]

hey hey @larmarange ! I hope you've been well :)

I may have some time to work on this in the first or second week of April. FYI If you want to work on it, let me know and I won't touch it.

[larmarange]

Hey. :-)

I'm completely overloaded until mid-April. So do not hesitate to move forward. And thanks for all you are doing.

[ddsjoberg]

Hey hey @larmarange ! I think the PR is in a place that can be merged. But before that, I wanted to address some of your comments above:

• For a Likert table, I have the feeling that "{p}%" would be a better and more common default value for statistic.

That sounds reasonable to me. We can make this update in a future PR?

• Should we think about an example dataset to provide for tbl_likert?
• We could/should provide more examples in documentation.

I expanded the example in tbl_likert(), improved the dummy dataset, and added links from the all the tbl_likert() family of functions to all the other functions in the family. If you'd like to add a data set, I think we can do that in the future anytime.

• Unit tests are missing

Added!

[larmarange]

Thanks