Sorting the order of bar graph values while using by

[Mohamad Cea]

Hello!

I want to order the results of a bar graph in descending order. I have two facets/panels as I want to disaggregate results by gender, male and female. But I want to order both panels/bars in descending order based on the order of how women responded. I am using this code:

graph hbar Q101A , over( Q102 ) by( Q1002 )

to get the attached graph but if I use the descending option the two panels are ordered differently. Any tips? Thanks!

[Andrew Musau]

gen order= -Q101A  
graph hbar Q101A , over(Q102, sort(order)) by(Q1002)

[Mohamad Cea]

Thanks Andrew! That helped.

Any one know how I can gen the order variable so that the plot appears in the order of the female plot for both?

[Andrew Musau]

I do not completely get what you are after. My advice is to present a data example using dataex and then explain what you mean in #3.

[Mohamad Cea]

Thanks for the follow up Andrew.

Here's the dataex export:

CODE]
* Example generated by -dataex-. To install: ssc install dataex
clear
input double(Q2061A Q1002)
2 1
8 1
2 1
2 1
1 1
14 1
1 1
1 1
1 2
1 2
2 2
14 2
3 2
2 2
8 2
1 2
2 2
1 2
1 1
12 2
1 2
3 2
11 2
11 2
1 2
14 2
1 2
12 2
2 2
2 2
1 2
1 2
2 1
6 1
12 1
1 1
1 1
12 1
1 1
6 2
1 2
2 2
1 2
12 2
1 2
7 1
7 1
12 1
7 1
2 1
12 1
1 1
12 2
12 2
2 2
1 2
1 2
1 2
2 1
12 1
1 1
1 1
1 1
11 1
1 1
2 2
2 2
3 2
1 2
11 2
12 2
1 2
3 1
1 1
1 1
2 1
1 1
1 1
1 1
2 2
1 2
12 2
2 2
2 2
1 2
1 2
7 1
6 1
14 1
1 1
1 1
1 1
2 1
1 1
2 1
3 1
1 1
7 1
7 1
8 1
end
label values Q2061A Q2061A
label def Q2061A 1 "economic situation", modify
label def Q2061A 2 "financial and administrative corruption", modify
label def Q2061A 3 "democracy and representation/governance", modify
label def Q2061A 6 "internal stability and security", modify
label def Q2061A 7 "foreign interference", modify
label def Q2061A 8 "religious extremism", modify
label def Q2061A 11 "fighting terrorism", modify
label def Q2061A 12 "public services", modify
label def Q2061A 14 "political/party issues", modify
label values Q1002 Q1002
label def Q1002 1 "male", modify
label def Q1002 2 "female", modify
[/CODE]
copy up to and including the previous line ------	-----------

I use this code: graph hbar, over(Q2061A) by(Q1002) graph attached for reference

What I want to add here is a way to sort both panels by the descending order of female respondents if that makes sense?

[Mohamad Cea]

I saw some mention of the egenmore axis and the seqvar commands but did not really comprehend how they could assist.

[Andrew Musau]

I think this will do what you want. As you are grouping categories, there is some redundancy in having labels for each by-graph. Look at tabplot from the Stata Journal that has a more appealing design and search the forum for code examples.

preserve
contract Q2061A Q1002
fillin Q2061A Q1002
replace _freq=0 if missing(_freq)
gen order=-_freq if Q1002==2
bys Q2061A: replace order= order[_N]
graph hbar _freq, over(Q2061A, sort(order)) by(Q1002)
restore

[Mohamad Cea]

This is exactly what I was looking for, thank you a lot. Any idea how to use aweights with this?

[Nick Cox]

This picks up on Andrew Musau's suggestion in #7. Ordering on the frequencies (equivalently percents) within a subgroup is programmable.

You have to find subgroup frequencies, inform the other group(s) what they are, and sort and make sure to copy back category names.

Some of this code is very similar to Andrew's, although he is missing the zero option of contract (perhaps misnamed).

labmask and tabplot are from the Stata Journal.

tabplot does often require some tweaking of where text goes and how much space you give to it. 0.23 and 0.85 here are the result of trial and error.

bysort Q2061A Q1002: gen freq = _N if Q1002 == 2
by Q2061A (Q1002) : replace freq = -freq[_N]

egen group = group(freq Q2061A)
labmask group, values(Q2061A) decode  

tabplot group Q1002 , percent(Q1002) showval(offset(0.23)) xsc(r(0.85 .)) yla(, labsize(small)) ytitle("") xtitle("") horizontal subtitle(% by gender) separate(Q1002) bar1(color(blue)) bar2(color(orange_red))

tabplot supports analytic weights. How that bears on your example I do not know.

[Mohamad Cea]

This picks up on Andrew Musau's suggestion in #7. Ordering on the frequencies (equivalently percents) within a subgroup is programmable.

You have to find subgroup frequencies, inform the other group(s) what they are, and sort and make sure to copy back category names.

Some of this code is very similar to Andrew's, although he is missing the zero option of contract (perhaps misnamed).

labmask and tabplot are from the Stata Journal.

tabplot does often require some tweaking of where text goes and how much space you give to it. 0.23 and 0.85 here are the result of trial and error.

bysort Q2061A Q1002: gen freq = _N if Q1002 == 2
by Q2061A (Q1002) : replace freq = -freq[_N]

egen group = group(freq Q2061A)
labmask group, values(Q2061A) decode

tabplot group Q1002 , percent(Q1002) showval(offset(0.23)) xsc(r(0.85 .)) yla(, labsize(small)) ytitle("") xtitle("") horizontal subtitle(% by gender) separate(Q1002) bar1(color(blue)) bar2(color(orange_red))

[ATTACH=CONFIG]n1595706[/ATTACH]


tabplot supports analytic weights. How that bears on your example I do not know.

Thanks so much Nick!

This is really helpful. For some reason this doesn't seem to work with more than two subgroups, so for example if I replace gender with age:

This function: "by Q2061A (Q1002) : replace freq = -freq[_N]"

replaces everything with missing, any tips?

Thanks again

[Nick Cox]

In your full dataset you may have values of Q2061A for which it is never true that Q1002 is 2.

tab Q2061A Q1002

will tell you that.

[Mohamad Cea]

In your full dataset you may have values of Q2061A for which it is never true that Q1002 is 2.

tab Q2061A Q1002

will tell you that.

Sorry for bothering you again with this but here's the dataex of the two variables I am now trying to plot:


----------------------- copy starting from the next line -----------------------

* Example generated by -dataex-. To install: ssc install dataex
clear
input double(Q609 Q2061A)
2  2
2  8
2  2
2  2
2  1
1 14
1  1
2  1
3  1
2  1
1  2
1 14
2  3
3  2
1  8
1  1
2  2
3  1
3  1
1 12
2  1
2  3
2 11
2 11
2  1
3 14
3  1
1 12
2  2
1  2
3  1
2  1
2  2
1  6
1 12
1  1
3  1
1 12
2  1
2  6
2  1
2  2
2  1
1 12
2  1
2  7
2  7
1 12
2  7
3  2
1 12
2  1
2 12
1 12
2  2
1  1
1  1
2  1
2  2
2 12
1  1
1  1
2  1
2 11
2  1
1  2
2  2
2  3
2  1
1 11
1 12
2  1
3  3
3  1
2  1
2  2
3  1
2  1
2  1
2  2
3  1
1 12
2  2
1  2
1  1
2  1
3  7
2  6
2 14
2  1
1  1
3  1
1  2
2  1
1  2
2  3
2  1
2  7
2  7
1  8
end
label values Q609 Q609
label def Q609 1 "religious", modify
label def Q609 2 "somewhat religious", modify
label def Q609 3 "not religious", modify
label values Q2061A Q2061A
label def Q2061A 1 "economic situation", modify
label def Q2061A 2 "financial and administrative corruption", modify
label def Q2061A 3 "democracy and representation/governance", modify
label def Q2061A 6 "internal stability and security", modify
label def Q2061A 7 "foreign interference", modify
label def Q2061A 8 "religious extremism", modify
label def Q2061A 11 "fighting terrorism", modify
label def Q2061A 12 "public services", modify
label def Q2061A 14 "political/party issues", modify

------------------ copy up to and including the previous line ------------------


And here's the code I'm using to group:

bysort Q2061A Q609: gen freq = _N if Q609 == 1
by Q2061A (Q609) : replace freq = -freq[_N]

egen group = group(freq Q2061A)
labmask group, values(Q2061A) decode

Any help would be appreciated!

Also is there a way for the tabplot subcommand showval to place the values to the right not left?

[Nick Cox]

To help further, I need to see the results of

 
 tab Q2061A Q1002 

tabplot, showval That is not supported. You'd need to write your own program.

[Mohamad Cea]

To help further, I need to see the results of

tab Q2061A Q1002 

tabplot, showval That is not supported. You'd need to write your own program.

of course! thanks in advance!

MOST IMPORTANT	LEVEL OF	RELIGIOSITY
CHALLENGE	religious	somewhat not	relig don't	kno	refused	Total
economic situation	3,327	4,949	1,144	124	41	9,585
financial and adminis	1,310	2,131	444	47	22	3,954
democracy and represe	172	275	78	7	2	534
internal stability an	417	649	154	12	1	1,233
foreign interference	721	1,128	206	18	8	2,081
religious extremism	280	357	99	11	5	752
fighting terrorism	666	705	254	18	4	1,647
public services	911	1,197	255	24	5	2,392
security	446	567	123	16	1	1,153
political/party issue	317	458	162	13	6	956
other	261	385	118	40	8	812
don't know	103	96	39	14	0	252
refused	22	9	5	0	0	36
Total	8,953	12,906	3,081	344	103	25,387

[Nick Cox]

These are different variables. But zeros in the table would be problematic. You have a choice. You can either pursue #10 in which case #11 and #13 still apply; or pursue #14 in which case we need to see (1) the graph code you tried and (2) the results of

contract , zero 
dataex

with the two variables in #14. So if the variables were a b (they aren't) you need to go

contract a b, zero 
dataex

See also https://www.statalist.org/forums/help#stata for how to present data examples here.