How to report row percentage and confidence interval using table1 template

[Ricardo Rodrigues Pereira]

#17595

is there a way to report rows percentage and confidence interval using table1 template? I want my table to look like this example.

         Has used sleeping pills in the last month
________________________________________________________________________
                   Yes                            No
________________________________________________________________________
 ________%_______95% CI_____________________________%_______95% CI______
Female 77.64 [75.22, 79.89]  ____________         22.36 [20.11, 24.78] 
Male   89.67 [88.15, 91.03]  ____________          10.33 [8.97, 11.85]
_______________________________________________________________________

[Attaullah Shah]

#17598

I have not personally used table1 in my research. Can you advise me if it is appropriate to report confidence intervals for categorical variables?

[Ricardo Rodrigues Pereira]

#17599

Yes, you can report the confidence interval for the proportion. I can get it when I use the command:

svy:tab variable1 variable2, ci percent

but it it doesn’t show up when use asdocx, I guess it’s not supported.

It would look like this: https://media.discordapp.net/attachments/1068130821283786793/1068130857040228433/image.png?width=806&height=473

[Attaullah Shah]

#17600

I have added the requested feature. First update asdocx with

asdocx_update

. Here is an example.

* Create a toy example
*generate random number of observations
clear 
set obs 200

* create variable for using sleeping pills
gen use_pills = round(runiform(0,1))
label define use_pills 0 "Yes" 1 "No"
label val use_pills use_pills
label var use_pills "Used sleeping pills in the last month"

 *create variable for male and female
gen gender = round(runiform(0,1))
label define gender 0 "Female" 1 "Male"
label val gender gender
replace use_pills = 1 if gender == 1 & use_pills == 0 & runiform()<0.7
replace use_pills = 0 if gender == 0 & use_pills == 1 & runiform()<0.3

Row percentage with confidence intervals

---

* if row percentages are required

asdocx tab use_pills gender, template(table1) row cat(% ci) replace 

Table 1: Demographics

Variables	Yes (n=79)	No (n=121)	Total (200)	P-value
gender				0.000
Female	64.444% (53.921 - 73.735)	35.556% (26.265 - 46.079)	100.000%
Male	19.091% (12.731 - 27.622)	80.909% (72.378 - 87.269)	100.000%
P-values by t-test for continuous variables and Chi2 test for binary/categorical variables.

* If the total column is not required, use option nototal
* Add option table_layout(auto) for auto adjust columns for contents
asdocx tab use_pills gender, template(table1) row cat(% ci) /// 
replace nototal table_layout(auto)

Table 1: Demographics

Variables	Yes (n=79)	No (n=121)	P-value
gender			0.000
Female	64.444% (53.921 - 73.735)	35.556% (26.265 - 46.079)
Male	19.091% (12.731 - 27.622)	80.909% (72.378 - 87.269)
P-values by t-test for continuous variables and Chi2 test for binary/categorical variables.

Column percentages

* If col percentages 
asdocx tab use_pills gender, template(table1) col categorical(% ci) replace 

Table 1: Demographics

Variables	Yes (n=79)	No (n=121)	Total (200)	P-value
gender				0.000
Female	73.418% (62.455 - 82.097)	26.446% (19.294 - 35.097)	45.0% (38.203 - 51.989)
Male	26.582% (17.903 - 37.545)	73.554% (64.903 - 80.706)	55.0% (48.011 - 61.797)
P-values by t-test for continuous variables and Chi2 test for binary/categorical variables.

[Ricardo Rodrigues Pereira]

#17601

Thank you, Professor. This new feature will be extremely beneficial for not only myself but also a significant number of users. It will greatly aid in my research.

[Ricardo Rodrigues Pereira]

#17602

One more thing, can you look if this can work with the svy command. It doesn’t seem to work for me.

[Attaullah Shah]

#17603

Support for confidence intervals with survey data has been added. Try this example after updating.

asdocx_update 

* Results from the Stata command  
svy: tabulate  diabetes race  , row  ci 

----------------------------------------------------------------------
Diabetes  |                            Race                           
status    |         White          Black          Other          Total
----------+-----------------------------------------------------------
 Not diab |         .8812          .0931          .0257              1
          | [.8437,.9107]   [.0705,.122]  [.0111,.0586]               
          | 
 Diabetic |         .8203          .1647          .0151              1
          | [.7476,.8755]  [.1203,.2213]  [.0033,.0668]               
          | 
    Total |         .8791          .0955          .0253              1
          | [.8407,.9093]  [.0725,.1249]  [.0108,.0585]               
----------------------------------------------------------------------

* In asdocx, we need to write the treatment variable first 
asdocx svy: tabulate race diabetes, row  ci  template(table1) replace cat(% ci)  

Table 1: Demographics

Variables	White (n=9065)	Black (n=1086)	Other (n=200)	P-value
Diabetes status				0.000
Not diabetic	88.121% (84.370 – 91.068)	9.308% (7.045 – 12.201)	2.571% (1.106 – 5.862)
Diabetic	82.028% (74.756 – 87.553)	16.467% (12.027 – 22.134)	1.505% (0.325 – 6.680)
P-values by t-test for continuous variables and Chi2 test for binary/categorical variables. * Design-based Pearson Chi2 test p-value

[Author]

Posts