-catplot- proportional bar graph problem

[Zhicheng Wang]

Dear Statalist,

I want to make a graph describing stack quartile constitution bar graph total to 100%.
My dataset consists of
area: area code 31 areas in total
quartile: social development quartile 1-Q1 2-Q2 3-Q3 4-Q4
q_num: number of sub-areas in different quartiles
percent: quartile composition

This is my dataset google drive link
https://drive.google.com/file/d/0Bz83KppKhKQ6WjlmRUVBUzlTNUU/view?usp=sharing


I tried the command below:

catplot quartile area [fw=q_num], asyvars stack percent(quartile) legend(rows(1) stack)

and only get this graph:

My problems:
1)the x-axis percent is not equal to 100,
2)the layout of areas label are too condensed, anyway to extend the y-axis space?
3)I want to change the order of areas according to the ranking of the proportion of first by quartile=4 and then by quartile=3, quartile=2, consequently, since not every area has quartile=4, quartile=3, quartile=2.


I am appreciated anyone can help me out~
Many thanks!

[Nick Cox]

Thanks for posting the dataset. See FAQ Advice #12 for advice to use dataex (SSC) and to explain where user-written programs you mention come from. Here catplot is from SSC.

Here is your data to save anyone downloading it.

clear
input byte area str1 quartile float(q_num percent)
34 `"1"' 23 21.90476226806641
34 `"2"' 15 14.2857141494751
34 `"3"' 31 29.5238094329834
34 `"4"' 36 34.28571319580078
11 `"1"' 18 100
50 `"1"' 10 25
50 `"2"' 7 17.5
50 `"3"' 12 30
50 `"4"' 11 27.5
35 `"1"' 28 33.33333206176758
35 `"2"' 42 50
35 `"3"' 14 16.66666603088379
62 `"1"' 11 12.64367771148682
62 `"2"' 9 10.34482765197754
62 `"3"' 15 17.24137878417969
62 `"4"' 52 59.77011489868164
44 `"1"' 56 45.5284538269043
44 `"2"' 31 25.2032527923584
44 `"3"' 33 26.82926750183106
44 `"4"' 3 2.439024448394775
45 `"1"' 19 17.27272796630859
45 `"2"' 17 15.45454502105713
45 `"3"' 27 24.54545402526856
45 `"4"' 47 42.72727203369141
52 `"1"' 6 6.818181991577148
52 `"3"' 5 5.681818008422852
52 `"4"' 71 80.68181610107422
46 `"1"' 4 19.0476188659668
46 `"2"' 6 28.5714282989502
46 `"3"' 11 52.38095092773438
13 `"1"' 36 20.93023300170898
13 `"2"' 50 29.06976699829102
13 `"3"' 59 34.30232620239258
13 `"4"' 27 15.69767475128174
23 `"1"' 52 39.39393997192383
23 `"2"' 57 43.18181991577148
23 `"3"' 19 14.39393901824951
23 `"4"' 4 3.030303001403809
41 `"1"' 36 22.64151000976563
41 `"2"' 27 16.98113250732422
41 `"3"' 35 22.0125789642334
41 `"4"' 61 38.36478042602539
42 `"1"' 29 28.15534019470215
42 `"2"' 30 29.12621307373047
42 `"3"' 23 22.33009719848633
42 `"4"' 21 20.38834953308106
43 `"1"' 28 22.95081901550293
43 `"2"' 27 22.13114738464356
43 `"3"' 33 27.04918098449707
43 `"4"' 34 27.86885261535645
15 `"1"' 28 27.7227725982666
15 `"2"' 27 26.73267364501953
15 `"4"' 18 17.8217830657959
32 `"1"' 56 52.8301887512207
32 `"2"' 32 30.18867874145508
32 `"3"' 17 16.03773498535156
32 `"4"' 1 .9433962106704712
36 `"1"' 16 16.1616153717041
36 `"2"' 36 36.3636360168457
36 `"3"' 37 37.37373733520508
36 `"4"' 10 10.1010103225708
22 `"1"' 20 33.33333206176758
22 `"2"' 24 40
22 `"3"' 13 21.66666603088379
22 `"4"' 3 5
21 `"1"' 50 50
21 `"2"' 19 19
21 `"3"' 23 23
21 `"4"' 8 8
64 `"1"' 5 22.72727203369141
64 `"2"' 8 36.3636360168457
64 `"3"' 4 18.18181800842285
63 `"1"' 8 17.39130401611328
63 `"2"' 5 10.8695650100708
63 `"3"' 9 19.56521797180176
63 `"4"' 24 52.17391204833984
61 `"1"' 17 15.88785076141357
61 `"2"' 27 25.23364448547363
61 `"3"' 44 41.12149429321289
61 `"4"' 19 17.75700950622559
37 `"1"' 31 22.14285659790039
37 `"2"' 42 30
37 `"3"' 48 34.28571319580078
37 `"4"' 19 13.5714282989502
31 `"1"' 16 88.88888549804688
31 `"2"' 2 11.11111068725586
14 `"1"' 20 16.80672264099121
14 `"2"' 30 25.21008491516113
14 `"3"' 48 40.33613586425781
14 `"4"' 21 17.64705848693848
51 `"1"' 25 13.81215476989746
51 `"2"' 31 17.12707138061523
51 `"3"' 46 25.41436386108398
51 `"4"' 79 43.64640808105469
12 `"1"' 11 68.75
12 `"2"' 3 18.75
12 `"3"' 2 12.5
54 `"1"' 1 1.369863033294678
54 `"2"' 4 5.479452133178711
54 `"3"' 6 8.219178199768066
54 `"4"' 62 84.93150329589844
65 `"1"' 25 25.51020431518555
65 `"2"' 49 50
65 `"3"' 24 24.48979568481445
53 `"1"' 6 4.651162624359131
53 `"2"' 13 10.07751941680908
53 `"3"' 31 24.03100776672363
53 `"4"' 79 61.24031066894531
33 `"1"' 27 30
33 `"2"' 41 45.55555725097656
33 `"3"' 20 22.22222137451172
33 `"4"' 2 2.222222328186035
end


set scheme s1color

* #1
catplot quartile area [fw=q_num], asyvars stack percent(area) legend(rows(1) stack)

*  #2
gen group = area >= 40
catplot quartile area [fw=q_num], asyvars stack percent(area) legend(rows(1) stack) ///
by(group, note("")) nofill subtitle("", fcolor(none))

* #3
tabplot quartile area [fw=q_num], percent(area) subtitle(% in each area) ///
yreverse  xla(, labsize(small)) showval(mlabsize(vsmall) format(%2.0f))

#4
fillin area quartile
replace q_num = 0 if q_num == .

egen total = total(q_num), by(area)
gen prop = q_num / total

forval j = 1/4 {
    egen rank`j' = rank(-prop) if quartile == "`j'"
    bysort area (rank`j') : replace rank`j' = rank`j'[1]
}

egen area2 = group(rank4 rank3 rank2 rank1 area)
labmask area2, values(area)

tabplot quartile area2 [fw=q_num], percent(area) subtitle(% in each area) ///
yreverse  xla(, labsize(small)) showval(mlabsize(vsmall) format(%2.0f)) ///
xtitle(area)

Your most obvious problem is using the wrong variable for percent(). See #1 above.

The crowding problem can be relieved with a trick: see #2 above. Note from help graph bar that what you call the x-axis is in this case the y-axis!

In my view you get a clearer graph with tabplot (SJ), which you must install from the updated files.

SJ-17-3 gr0066_1  . . . . . . . . . . . . . . . .  Software update for tabplot
        (help tabplot if installed) . . . . . . . . . . . . . . . .  N. J. Cox
        Q3/17   SJ 17(3):779
        added options for reversing axis scales; improved handling of
        axis labels containing quotation marks

See #3 for a first result.

Your sorting is harder work but I think you see what you want. See #4. You need to install labmask (SJ) for that to work.

. search labmask, sj

Search of official help files, FAQs, Examples, SJs, and STBs

SJ-8-2  gr0034  . . . . . . . . . .  Speaking Stata: Between tables and graphs
        (help labmask, seqvar if installed) . . . . . . . . . . . .  N. J. Cox
        Q2/08   SJ 8(2):269--289
        outlines techniques for producing table-like graphs

Note that your percent variable seems inconsistent. You owe me about 20 minutes trying to find a bug in my code that was down to the data.

. egen totalpc = total(percent), by(area)

. tabdisp area if totalpc != 100, c(totalpc)

----------------------
     area |    totalpc
----------+-----------
       15 |   72.27723
       52 |   93.18182
       64 |   77.27273
----------------------

[Zhicheng Wang]

Dear Nick,
It is so nice of you. Thank you for pointing dataex (SSC) issues as well.
It is a stunning graph that can illustrate my research interests. But I think I prefer a graph like this shown below,
a stack 100% bar graph.
Where the age categories are the area code, while different colour indicates different quartiles.

Many thanks for your help!

Zhicheng

[Nick Cox]

I seriously recommend a field test with people who might be consumers of such a graph.

Otherwise good luck on making that legible and intelligible with 31 areas!

[Zhicheng Wang]

Hi Nick,
I learn to modify several options and make the graph meeting my requirement.
Thank you for your help again and I am sorry for the inappropriate requests made here.