How to Place Brackets on Bar Graphs Created by Twoway Command for Representation of Statistical Significance between Groups in Stata

[smith Jason]

I have a small dataset below for the illustrative purpose,

clear
input str11 race str6 ses meanwrite sdwrite n
"hispanic" "low" 46.11111 9.58442 9
"hispanic" "middle" 46.09091 7.133787 11
"hispanic" "high" 48.25 10.24288 4
"asian" "low" 58 12.12436 3
"asian" "middle" 55.8 7.463243 5
"asian" "high" 61.66667 4.618802 3
"african-amer" "low" 48.09091 8.431542 11
"african-amer" "middle" 48.66667 8.594572 6
"african-amer" "high" 47.66667 16.86219 3
"white" "low" 52.54167 9.084048 24
"white" "middle" 52.80822 9.222041 73
"white" "high" 56.70833 8.7372 48
end

encode race, generate(race2)
encode ses, generate(ses2)
drop ses race
ren race2 race
ren ses2 ses
order race ses meanwrite sdwrite n

g sesrace = race if ses == 1
replace sesrace = race+5 if ses == 2
replace sesrace = race+10 if ses == 3


twoway (bar meanwrite sesrace if race==1) ///
(bar meanwrite sesrace if race==2) ///
(bar meanwrite sesrace if race==3) ///
(bar meanwrite sesrace if race==4) , ///
legend(row(4) position(right) order(1 "Hispanic" 2 "Asian" 3 "Black" 4 "White") ) ///
xlabel(0(5)15) ///
ylabel(0(20)70) ///
xlabel( 3 "Low" 8 "Middle" 13 "High", noticks) ///
xtitle("Socio Economic Status") ytitle("Mean Writing Score")

I know I can borrow some codes from this thread at https://www.statalist.org/forums/for...ificance/page2.
to add brackets to represent the significant difference between "Asian" and "Black" for the Low SES group students.
Can someone give me some help with the code?
Thank you!

[smith Jason]

The code I used here is incorrect,
tw (bar meanwrite sesrace if race==1) ///
(scatteri 60.55 2 60.55 3, recast(line) ///
lw(medthick) mc(none) lc(black) lp(solid)) (scatteri 60.55 2 60.55 3, ///
recast(dropline) base(60.52) lw(medthick) mc(none) lc(black) lp(solid)), ///
ysc(r(., 11)) scheme(s1mono) leg(off) text(60.57 2.5 "p<.001", place(n) size (0.3cm)) ///

(bar meanwrite sesrace if race==2) ///
(bar meanwrite sesrace if race==3) (bar meanwrite sesrace if race==4), ///
legend(label(1 "Hispanic") label(2 "Asian") label(3 "African-American") label(4 "White")) ///
xlabel(0(5)15) ylabel(0(20)70) xlabel(3 "Low" 8 "Middle" 13 "High", noticks) ///
xtitle("Socioeconomic Status") ytitle("Mean Writing Score")
How can I revise it?

[smith Jason]

Can someone help me? Thank you!

[Andrew Musau]

I am not really sure what the problem is. The y-axis positions are identified by your y-axis and each bar on the categorical axis is centered on an integer, starting at 1.

clear
input str11 race str6 ses meanwrite sdwrite n
"hispanic" "low" 46.11111 9.58442 9
"hispanic" "middle" 46.09091 7.133787 11
"hispanic" "high" 48.25 10.24288 4
"asian" "low" 58 12.12436 3
"asian" "middle" 55.8 7.463243 5
"asian" "high" 61.66667 4.618802 3
"african-amer" "low" 48.09091 8.431542 11
"african-amer" "middle" 48.66667 8.594572 6
"african-amer" "high" 47.66667 16.86219 3
"white" "low" 52.54167 9.084048 24
"white" "middle" 52.80822 9.222041 73
"white" "high" 56.70833 8.7372 48
end

encode race, generate(race2)
encode ses, generate(ses2)
drop ses race
ren race2 race
ren ses2 ses
order race ses meanwrite sdwrite n

g sesrace = race if ses == 1
replace sesrace = race+5 if ses == 2
replace sesrace = race+10 if ses == 3

set scheme s1mono
twoway (bar meanwrite sesrace if race==1) ///
(bar meanwrite sesrace if race==2) ///
(bar meanwrite sesrace if race==3) ///
(bar meanwrite sesrace if race==4)  ///
(scatteri 65 2 65 3.15, recast(line) ///
lw(medthick) mc(none) lc(black) lp(solid)) (scatteri 65 2 65 3.15, ///
recast(dropline) base(62) lw(medthick) mc(none) lc(black) lp(solid)), ///
legend(row(4) position(right) order(1 "Hispanic" 2 "Asian" 3 "Black" 4 "White") ) ///
xlabel(0(5)15) text(66 2.6 "{it:p}<.001", place(n) size(vsmall)) ///
ylabel(0(20)70) ///
xlabel( 3 "Low" 8 "Middle" 13 "High", noticks) ///
xtitle("Socio Economic Status") ytitle("Mean Writing Score")

[smith Jason]

Thank you, Andrew. Also, I tried to add such bracket to Asian and Black.
I really need color graph instead of gray graph. Could you please help me?
This is my code below, and it doesn't work.

clear
input str11 race str6 ses meanwrite sdwrite n
"hispanic" "low" 46.11111 9.58442 9
"hispanic" "middle" 46.09091 7.133787 11
"hispanic" "high" 48.25 10.24288 4
"asian" "low" 58 12.12436 3
"asian" "middle" 55.8 7.463243 5
"asian" "high" 61.66667 4.618802 3
"african-amer" "low" 48.09091 8.431542 11
"african-amer" "middle" 48.66667 8.594572 6
"african-amer" "high" 47.66667 16.86219 3
"white" "low" 52.54167 9.084048 24
"white" "middle" 52.80822 9.222041 73
"white" "high" 56.70833 8.7372 48
end

* Initial Bar Chart.
graph bar meanwrite, over(race) over(ses) asyvars

encode race, generate(race2)
encode ses, generate(ses2)
drop ses race
ren race2 race
ren ses2 ses
order race ses meanwrite sdwrite n

g sesrace = race if ses == 1
replace sesrace = race+5 if ses == 2
replace sesrace = race+10 if ses == 3
sort sesrace
list sesrace ses race, sepby(ses)



tw (bar meanwrite sesrace if race==1) ///
(bar meanwrite sesrace if race==2 ) ///
(scatteri 62.55 2 62.55 3, recast(line) ///
lw(medthick) mc(on) lc(black) lp(solid)) (scatteri 62.55 2 62.55 3, ///
recast(dropline) base(61.50) lw(medthick) mc(none) lc(black) lp(solid)), ///
ysc(r(., 11)) text(62.02 2.5 "***", place(n) size (0.3cm)), ///
(bar meanwrite sesrace if race==3) ///
(scatteri 62.55 3 62.55 4, recast(line) ///
lw(medthick) mc(none) lc(black) lp(solid)) (scatteri 62.55 3 62.55 4, ///
recast(dropline) base(61.50) lw(medthick) mc(none) lc(black) lp(solid)), ///
ysc(r(., 11)) text(62.02 3.5 "***", place(n) size (0.3cm)), ///
(bar meanwrite sesrace if race==4), ///
legend(label(1 "Hispanic") label(2 "Asian") label(7 "African-American") label(8 "White")) ///
xlabel(0(5)15) ylabel(0(20)70) xlabel(3 "Low" 8 "Middle" 13 "High", noticks) ///
xtitle("Socioeconomic Status") ytitle("Mean Writing Score")

[Andrew Musau]

The color is determined by your scheme.

set scheme s1color

for example.

[smith Jason]

The color is determined by your scheme.

set scheme s1color

for example.

Thank you! I changed this setting; however, the color is not the same as the top. How can I keep the same color?
Moreover, I would like to remove the horizontal line at the top of the main graph and the vertical line on the right side.
How can I modify the code to achieve this?

[Andrew Musau]

Sorry, I cannot guide you on every little detail of your graph. Look at the documentation or get a colleague to help you.

[smith Jason]

Thanks for your advice.
I looked at this document at https://www.stata.com/manuals/g-2gra...ayscatteri.pdf and still cannot mimic and modify your code to figure out how to add bracket to the black and white for the low SES group.
As you have seen in my above code, once I inserted the corresponding code into any of the race group, although bracket was added to the graph and connected the two groups, there are unexpected legends took place.
I know that I need to solve the problem mainly independently, however, sometimes, it doesn't work at all even though I work hard and still don't know what to do and how to do.
Thanks for your help.

Below is my revised code based on your exemplary code,
clear
input str11 race str6 ses meanwrite sdwrite n
"hispanic" "low" 46.11111 9.58442 9
"hispanic" "middle" 46.09091 7.133787 11
"hispanic" "high" 48.25 10.24288 4
"asian" "low" 58 12.12436 3
"asian" "middle" 55.8 7.463243 5
"asian" "high" 61.66667 4.618802 3
"african-amer" "low" 48.09091 8.431542 11
"african-amer" "middle" 48.66667 8.594572 6
"african-amer" "high" 47.66667 16.86219 3
"white" "low" 52.54167 9.084048 24
"white" "middle" 52.80822 9.222041 73
"white" "high" 56.70833 8.7372 48
end

* Initial Bar Chart.
graph bar meanwrite, over(race) over(ses) asyvars

encode race, generate(race2)
encode ses, generate(ses2)
drop ses race
ren race2 race
ren ses2 ses
order race ses meanwrite sdwrite n

g sesrace = race if ses == 1
replace sesrace = race+5 if ses == 2
replace sesrace = race+10 if ses == 3
sort sesrace
list sesrace ses race, sepby(ses)



tw (bar meanwrite sesrace if race==1) ///
(bar meanwrite sesrace if race==2 ) ///
(scatteri 62.55 2 62.55 3, recast(line) ///
lw(medthick) mc(on) lc(black) lp(solid)) (scatteri 62.55 2 62.55 3, ///
recast(dropline) base(61.50) lw(medthick) mc(none) lc(black) lp(solid)), ///
ysc(r(., 11)) text(62.02 2.5 "***", place(n) size (0.3cm)), ///
(bar meanwrite sesrace if race==3) ///
(scatteri 62.55 3 62.55 4, recast(line) ///
lw(medthick) mc(none) lc(black) lp(solid)) (scatteri 62.55 3 62.55 4, ///
recast(dropline) base(61.50) lw(medthick) mc(none) lc(black) lp(solid)), ///
ysc(r(., 11)) text(62.02 3.5 "***", place(n) size (0.3cm)), ///
(bar meanwrite sesrace if race==4), ///
legend(label(1 "Hispanic") label(2 "Asian") label(7 "African-American") label(8 "White")) ///
xlabel(0(5)15) ylabel(0(20)70) xlabel(3 "Low" 8 "Middle" 13 "High", noticks) ///
xtitle("Socioeconomic Status") ytitle("Mean Writing Score")

[smith Jason]

The problem is solved by myself with the Stata code below,
clear
input str11 race str6 ses meanwrite sdwrite n
"hispanic" "low" 46.11111 9.58442 9
"hispanic" "middle" 46.09091 7.133787 11
"hispanic" "high" 48.25 10.24288 4
"asian" "low" 58 12.12436 3
"asian" "middle" 55.8 7.463243 5
"asian" "high" 61.66667 4.618802 3
"african-amer" "low" 48.09091 8.431542 11
"african-amer" "middle" 48.66667 8.594572 6
"african-amer" "high" 47.66667 16.86219 3
"white" "low" 52.54167 9.084048 24
"white" "middle" 52.80822 9.222041 73
"white" "high" 56.70833 8.7372 48
end

encode race, generate(race2)
encode ses, generate(ses2)
drop ses race
ren race2 race
ren ses2 ses
order race ses meanwrite sdwrite n

g sesrace = race if ses == 1
replace sesrace = race+5 if ses == 2
replace sesrace = race+10 if ses == 3

set scheme s1color
twoway (bar meanwrite sesrace if race==1) ///
(bar meanwrite sesrace if race==2) ///
(bar meanwrite sesrace if race==3) ///
(bar meanwrite sesrace if race==4) ///
(scatteri 65 2 65 3.05, ///
recast(line) lw(medthick) mc(none) lc(black) lp(solid)) ///
(scatteri 65 2 65 3.05, ///
recast(dropline) base(62) lw(medthick) mc(none) lc(black) lp(solid)) ///
(scatteri 65 3.05 65 4.10, ///
recast(line) lw(medthick) mc(none) lc(black) lp(solid)) ///
(scatteri 65 3.05 65 4.10, ///
recast(dropline) base(62) lw(medthick) mc(none) lc(black) lp(solid)), ///
text(66 2.6 "{it:p}<.001", place(n) size(vsmall)) ///
text(66 3.8 "{it:p}<.001", place(n) size(vsmall)) ///
legend(row(4) position(right) ///
order(1 "Hispanic" 2 "Black" 3 "White" 4 "Other") ) ///
legend(region(lstyle(none))) ///
xlabel(0(5)15) ylabel(0(20)70) ///
xlabel( 3 "Low" 8 "Middle" 13 "High", noticks) ///
ytitle("Predicted Probability of Disiciplinary Action") ///
title("Medium Infractions") ///
plotregion(style(none))

[smith Jason]

I have a small dataset below for the illustrative purpose,

clear
input str11 race str6 ses meanwrite sdwrite n
"hispanic" "low" 46.11111 9.58442 9
"hispanic" "middle" 46.09091 7.133787 11
"hispanic" "high" 48.25 10.24288 4
"asian" "low" 58 12.12436 3
"asian" "middle" 55.8 7.463243 5
"asian" "high" 61.66667 4.618802 3
"african-amer" "low" 48.09091 8.431542 11
"african-amer" "middle" 48.66667 8.594572 6
"african-amer" "high" 47.66667 16.86219 3
"white" "low" 52.54167 9.084048 24
"white" "middle" 52.80822 9.222041 73
"white" "high" 56.70833 8.7372 48
end

encode race, generate(race2)
encode ses, generate(ses2)
drop ses race
ren race2 race
ren ses2 ses
order race ses meanwrite sdwrite n

g sesrace = race if ses == 1
replace sesrace = race+5 if ses == 2
replace sesrace = race+10 if ses == 3


twoway (bar meanwrite sesrace if race==1) ///
(bar meanwrite sesrace if race==2) ///
(bar meanwrite sesrace if race==3) ///
(bar meanwrite sesrace if race==4) , ///
legend(row(4) position(right) order(1 "Hispanic" 2 "Asian" 3 "Black" 4 "White") ) ///
xlabel(0(5)15) ///
ylabel(0(20)70) ///
xlabel( 3 "Low" 8 "Middle" 13 "High", noticks) ///
xtitle("Socio Economic Status") ytitle("Mean Writing Score")

I know I can borrow some codes from this thread at https://www.statalist.org/forums/for...ificance/page2.
to add brackets to represent the significant difference between "Asian" and "Black" for the Low SES group students.
Can someone give me some help with the code?
Thank you! [ATTACH=CONFIG]n1714663[/ATTACH]

The information of the setting of this color is available at https://imagecolorpicker.com/